Initial accelerometer calibration

Question

Initial accelerometer calibration

Civilduino opened this issue 5 years ago · comments

Hello Kris!

First of all thank you for your great work and for sharing knowledge so other can learn!

I am new with the MPU-6050 and I am trying to understand the auto calibration process that you have implemented. If I understand correctly, all is based on ideal situations, e.g. the sensor is located with one axis (z axis) along the gravity vector and other axis having 0g, the sensor have ideal performance characteristics and is temperature is stable, among others.

Then if we believe that we positioned the sensor in the ideal position, we then adjust the measurements by adding offsets to match the ideal case. Is this what you are doing? is this the same as this https://github.com/jrowberg/i2cdevlib/blob/master/Arduino/MPU6050/examples/IMU_Zero/IMU_Zero.ino

Can this be called calibration? I am really confuse if it is necessary to calibrate MEMS accelerometers or if they are adjusted from factory. Could you please elaborate more on this topic of calibration and what the sensor is given from factory?

Thank you once again!

Kris Winer · Answer 1 · Mon Jan 07 2019 04:12:59 GMT+0800 (China Standard Time)

All sensors need calibration. The accel and gyro in particular need offset calibration. So lay flat and motionless and collect data. Average the data and subtract such that the readings in this condition show 0 0 1 for the accel and 0 0 0 dps for the gyro. The residual is the offset bias that needs to be subtracted from all subsequent data.

…

On Sun, Jan 6, 2019 at 12:06 PM Civilduino ***@***.***> wrote: Hello Kris! First of all thank you for your great work and for sharing knowledge so other can learn! I am new with the MPU-6050 and I am trying to understand the auto calibration process that you have implemented. If I understand correctly, all is based on ideal situations, e.g. the sensor is located with one axis (z axis) along the gravity vector and other axis having 0g, the sensor have ideal performance characteristics and is temperature is stable, among others. Then if we believe that we positioned the sensor in the ideal position, we then adjust the measurements by adding offsets to match the ideal case. Is this what you are doing? is this the same as this https://github.com/jrowberg/i2cdevlib/blob/master/Arduino/MPU6050/examples/IMU_Zero/IMU_Zero.ino Can this be called calibration? I am really confuse if it is necessary to calibrate MEMS accelerometers or if they are adjusted from factory. Could you please elaborate more on this topic of calibration and what the sensor is given from factory? Thank you once again! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#38>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGY1qiUR96b13YKULFQC--AfxK7ug06aks5vAldJgaJpZM4ZylVk> .

Civilduino · Answer 2 · Tue Jan 08 2019 04:42:37 GMT+0800 (China Standard Time)

I agree that all sensors need calibration. However, this is not a "real" calibration but an offset correction. This method assumes that the surface where the sensor lies is perfectly aligned with the gravity vector and therefore the offset from the readings are subtracted to match the "reality".

However, it is extremely difficult to have an ideal surface aligned to the gravity vector. So the true angle with respect to gravity is hard to obtain unless you have expensive equipment or do a more proper calibration.

A method of calibration is explained in this paper https://ieeexplore.ieee.org/document/6015525

Kris Winer · Answer 3 · Tue Jan 08 2019 04:57:03 GMT+0800 (China Standard Time)

Again, if all you have is an accelerometer you will be very limited in the accuracy and kinds of orientation estimation you can do. But why not use a 9 DoF sensor and get accurate absolute orientation estimates in all orientations?

…

On Mon, Jan 7, 2019 at 12:42 PM Civilduino ***@***.***> wrote: I agree that all sensors need calibration. However, this is not a "real" calibration but an offset correction. This method assumes that the surface where the sensor lies is perfectly aligned with the gravity vector and therefore the offset from the readings are subtracted to match the "reality". However, it is extremely difficult to have an ideal surface aligned to the gravity vector. So the true angle with respect to gravity is hard to obtain unless you have expensive equipment or do a more proper calibration. A method of calibration is explained in this paper https://ieeexplore.ieee.org/document/6015525 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGY1qkoUULYF6asvTdpsq0M1NGD6SNX6ks5vA7E_gaJpZM4ZylVk> .

Sofia Assis · Answer 4 · Wed Jan 09 2019 04:56:38 GMT+0800 (China Standard Time)

To build on what @Civilduino mentioned, I would also suggest the following approach to gyroscope and acceleration calibration

https://github.com/Kyle-ak/imu_tk

which I found pretty nice as it did not require much resources in terms of equipment. :)

Mike Collins · Answer 5 · Sat Jul 03 2021 04:09:13 GMT+0800 (China Standard Time)

Hi Kris,
Did the vendor ever release some pseudo-code to do an initial setup ?
O am i so late to the game, that it was a pay-to-learn thing initially ?

Any communication would be greatly appreciated.

Kris Winer · Answer 6 · Sat Jul 03 2021 04:16:36 GMT+0800 (China Standard Time)

Huh? Which vendor? Setup for what exactly?

…

On Fri, Jul 2, 2021 at 1:09 PM Mike Collins ***@***.***> wrote: Hi Kris, Did the vendor ever release some pseudo-code to do an initial setup ? O am i so late to the game, that it was a pay-to-learn thing initially ? Any communication would be greatly appreciated. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKSNZDGCZUDQQX4CTVLTVYMHHANCNFSM4GOKKVSA> .

Mike Collins · Answer 7 · Sun Jul 04 2021 01:44:34 GMT+0800 (China Standard Time)

The vendor of mpu6050, TDK I think.

Setup for initial calibration / zero-ization (lol)

I used your code here and converted/modified to my language of preference (TCL)

I'm curious, should i be reading the values from the "normal" registers after this "calibration" ?

Kris Winer · Answer 8 · Sun Jul 04 2021 01:54:29 GMT+0800 (China Standard Time)

If you want the accel and gyro data then, yes, you need to read it from the accel and gyro data registers and then properly scale it to get acceleration in gs or m/s^2 and rotation rate in degrees per sec, etc.

…

On Sat, Jul 3, 2021 at 10:44 AM Mike Collins ***@***.***> wrote: The vendor of mpu6050, TDK I think. Setup for initial calibration / zero-ization (lol) I used your code here and converted/modified to my language of preference (TCL) I'm curious, should i be reading the values from the "normal" registers after this "calibration" ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKTY2UE5BQLP7RIWFTLTV5EAZANCNFSM4GOKKVSA> .

Mike Collins · Answer 9 · Sun Jul 04 2021 03:51:15 GMT+0800 (China Standard Time)

Hmmmm, yep am doing so.
I calibrate/zeroize, from a standing position, no movement, using the procedure you created, recoded into TCL,.
And then when i reused my original code reading from the regular registers I'm still not getting a value (even adjusted) to what i expect.
I would assume that i should have (x:0 y:0 z:1), after calib, when not moving ?

Kris Winer · Answer 10 · Sun Jul 04 2021 03:54:57 GMT+0800 (China Standard Time)

Yes, 0, 0, 1000 mg to within +/- 5 millig or so. And 0, 0, 0 millidegree per sec to within 10 mdps or so. How are you using the calibration offsets? These have to be subtracted from the data, of course....

…

On Sat, Jul 3, 2021 at 12:51 PM Mike Collins ***@***.***> wrote: Hmmmm, yep am doing so. I calibrate/zeroize, from a standing position, no movement, using the procedure you created, recoded into TCL,. And then when i reused my original code reading from the regular registers I'm still not getting a value (even adjusted) to what i expect. I would assume that i should have (x:0 y:0 z:1), after calib, when not moving ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKXU766LRHG7GWGPDFTTV5S33ANCNFSM4GOKKVSA> .

Mike Collins · Answer 11 · Sun Jul 04 2021 04:02:51 GMT+0800 (China Standard Time)

lol, well i believe i am subtracting correctly, lol

I had my procedure to perform the calibration, but didn't know it had to be loaded back into the sensor, and rather than try to figure all that out, i used your calib procedure.

I only reused mine to get the calibrated values into my own variables.
I would rather store the value that you create, and just haven't studied it out enough, yet, to figure out where they are, and capture them, into my variables.

Mike Collins · Answer 12 · Sun Jul 04 2021 04:05:39 GMT+0800 (China Standard Time)

Loading the value into sensors offsets, don't cause it to provide more calib/zeroized values ?

Hmmm I would've thought that was the whole point, of doing that.

Kris Winer · Answer 13 · Sun Jul 04 2021 04:08:14 GMT+0800 (China Standard Time)

This is not rocket science. You calculate the offset biases, then these are either stored in the sketch (or equivalent) or written to the offset bias registers (but these are volatile, so need to be reloaded at every power on). So it is easy to check what the average uncalibrated values are when the sensor is motionless and flat. It is easy to cmpare the offet biases you calculate with the average uncalibrated values. So easy to check if everything is happening as you expect/program.

…

On Sat, Jul 3, 2021 at 1:03 PM Mike Collins ***@***.***> wrote: lol, well i believe i am subtracting correctly, lol I had my procedure to perform the calibration, but didn't know it had to be loaded back into the sensor, and rather than try to figure all that out, i used your calib procedure. I only reused mine to get the calibrated values into my own variables. I would rather store the value that you create, and just haven't studied it out enough, yet, to figure out where they are, and capture them, into my variables. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKUESXNECVMPI5BQJNLTV5UHLANCNFSM4GOKKVSA> .

Kris Winer · Answer 14 · Sun Jul 04 2021 04:08:56 GMT+0800 (China Standard Time)

Not understanding your question here...

…

On Sat, Jul 3, 2021 at 1:05 PM Mike Collins ***@***.***> wrote: Loading the value into sensors offsets, don't cause it to provide more calib/zeroized values ? Hmmm I would've thought that was the whole point, of doing that. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKX6W3CSPX5RCRLUIRTTV5UR3ANCNFSM4GOKKVSA> .

Mike Collins · Answer 15 · Sun Jul 04 2021 04:19:48 GMT+0800 (China Standard Time)

Not understanding your question here...
…
On Sat, Jul 3, 2021 at 1:05 PM Mike Collins @.***> wrote: Loading the value into sensors offsets, don't cause it to provide more calib/zeroized values ? Hmmm I would've thought that was the whole point, of doing that. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKX6W3CSPX5RCRLUIRTTV5UR3ANCNFSM4GOKKVSA .

Maybe i'm assuming to much.
I wass thinking the point of storing that in the sensor, somehow adjusted the values the sensor provides when it is read (from the register).

Is it just a torage location tat is just commonly acceible, or does it actually affect the values the sensor provides ?

Mike Collins · Answer 16 · Sun Jul 04 2021 04:22:18 GMT+0800 (China Standard Time)

This is not rocket science. You calculate the offset biases, then these are either stored in the sketch (or equivalent) or written to the offset bias registers (but these are volatile, so need to be reloaded at every power on). So it is easy to check what the average uncalibrated values are when the sensor is motionless and flat. It is easy to cmpare the offet biases you calculate with the average uncalibrated values. So easy to check if everything is happening as you expect/program.
…
On Sat, Jul 3, 2021 at 1:03 PM Mike Collins @.***> wrote: lol, well i believe i am subtracting correctly, lol I had my procedure to perform the calibration, but didn't know it had to be loaded back into the sensor, and rather than try to figure all that out, i used your calib procedure. I only reused mine to get the calibrated values into my own variables. I would rather store the value that you create, and just haven't studied it out enough, yet, to figure out where they are, and capture them, into my variables. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKUESXNECVMPI5BQJNLTV5UHLANCNFSM4GOKKVSA .

Yea this turned into a whole mouthful for me, smh, lol, sorry.

Kris Winer · Answer 17 · Sun Jul 04 2021 04:24:43 GMT+0800 (China Standard Time)

You can choose to load the offset bias registers with your calibration data. In this case, the raw data should include the offset bias. But in my experience, this doesn;t work well for the accel, so I usually just do this manually. On Sat, Jul 3, 2021 at 1:19 PM Mike Collins ***@***.***> wrote:

…

Not understanding your question here... … <#m_-5220157750711000405_> On Sat, Jul 3, 2021 at 1:05 PM Mike Collins *@*.***> wrote: Loading the value into sensors offsets, don't cause it to provide more calib/zeroized values ? Hmmm I would've thought that was the whole point, of doing that. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKX6W3CSPX5RCRLUIRTTV5UR3ANCNFSM4GOKKVSA . Maybe i'm assuming to much. I wass thinking the point of storing that in the sensor, somehow adjusted the values the sensor provides when it is read (from the register). Is it just a torage location tat is just commonly acceible, or does it actually affect the values the sensor provides ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKTIPS3EBOX77SKAIP3TV5WG5ANCNFSM4GOKKVSA> .

Mike Collins · Answer 18 · Sun Jul 04 2021 04:25:42 GMT+0800 (China Standard Time)

So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement).
Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ?

Kris Winer · Answer 19 · Sun Jul 04 2021 04:28:36 GMT+0800 (China Standard Time)

Don't need to read the gyro then

…

On Sat, Jul 3, 2021 at 1:25 PM Mike Collins ***@***.***> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA> .

Mike Collins · Answer 20 · Sun Jul 04 2021 04:31:19 GMT+0800 (China Standard Time)

This is my first experience using a motion sensor.
I had no idea the values jumped so wildly.
That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol.
Probably to be expected.

Don't need to read the gyro then
…
On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.***> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA .

Ok

Mike Collins · Answer 21 · Sun Jul 04 2021 04:31:55 GMT+0800 (China Standard Time)

Don't need to read the gyro then
…
On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.***> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA .

Ok

Kris Winer · Answer 22 · Sun Jul 04 2021 04:40:48 GMT+0800 (China Standard Time)

They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins ***@***.***> wrote:

…

This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins *@*.***> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA> .

Mike Collins · Answer 23 · Sun Jul 04 2021 04:46:18 GMT+0800 (China Standard Time)

They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @.> wrote:
…
This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA .

Ahhh crud.

Yea sitting still, the values may jump from positive to negative, its just ridiculous.

Kris Winer · Answer 24 · Sun Jul 04 2021 04:52:38 GMT+0800 (China Standard Time)

As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins ***@***.***> wrote:

…

They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins *@*. *> wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.*> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA> .

Mike Collins · Answer 25 · Sun Jul 04 2021 05:00:14 GMT+0800 (China Standard Time)

They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @.> wrote:
…
This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA .

lol, i don't have great vision, and that craps written so small, i'm not even sure there is something wworthwhile written on it.

The other name of the whole sensor is GY-521, are you familiar with that ?

Mike Collins · Answer 26 · Sun Jul 04 2021 05:10:27 GMT+0800 (China Standard Time)

As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins @.***> wrote:
…
They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @. > wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA .

lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet.

I've reverted to reading the raw values to try and figure this out / understand.

last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768.
Is that ridiculous, or just fairly normal, i have no frame of refernce.

Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy.
Problem is i got them all off Ebay, so they all probably crappy.

I already swapped 1 out first because i thought it was defective reading so wild.

Kris Winer · Answer 27 · Sun Jul 04 2021 05:14:49 GMT+0800 (China Standard Time)

sounds like you are not doing the 2-s complement conversion properly? On Sat, Jul 3, 2021 at 2:10 PM Mike Collins ***@***.***> wrote:

…

As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins *@*.***> wrote: … <#m_4503423486452604556_> They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins *@*. *> wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.*> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> < #38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA . lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet. I've reverted to reading the raw values to try and figure this out / understand. last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768. Is that ridiculous, or just fairly normal, i have no frame of refernce. Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy. Problem is i got them all off Ebay, so they all probably crappy. I already swapped 1 out first because i thought it was defective reading so wild. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKR7PI75FWMA6HKJYI3TV54E3ANCNFSM4GOKKVSA> .

Mike Collins · Answer 28 · Sun Jul 04 2021 05:15:17 GMT+0800 (China Standard Time)

So thats an intersting comment about the reading time.

I couldn't find even great refernce to what is good sampling time.

I did find 1 routine "suggestion" to read 1100 samples every 100ms (or something like that, then thwout the first 100, and calcutlate the rest.

Hasn't helped :(

Kris Winer · Answer 29 · Sun Jul 04 2021 05:22:05 GMT+0800 (China Standard Time)

Set the sample rate to 200 Hz, the LP filter to 42 Hz and the full scale range to 2 g. Forget about calibration for now... What is the average and standard deviation of the three axes when the sensor lies motionless and flat?

…

On Sat, Jul 3, 2021 at 2:15 PM Mike Collins ***@***.***> wrote: So thats an intersting comment about the reading time. I couldn't find even great refernce to what is good sampling time. I did find 1 routine "suggestion" to read 1100 samples every 100ms (or something like that, then thwout the first 100, and calcutlate the rest. Hasn't helped :( — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKS3PGTQDMBTCXBHPULTV54W7ANCNFSM4GOKKVSA> .

Mike Collins · Answer 30 · Sun Jul 04 2021 05:26:50 GMT+0800 (China Standard Time)

sounds like you are not doing the 2-s complement conversion properly? On Sat, Jul 3, 2021 at 2:10 PM Mike Collins @.> wrote:
…
As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins @.> wrote: … <#m_4503423486452604556_> They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @. > wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> < #38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA . lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet. I've reverted to reading the raw values to try and figure this out / understand. last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768. Is that ridiculous, or just fairly normal, i have no frame of refernce. Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy. Problem is i got them all off Ebay, so they all probably crappy. I already swapped 1 out first because i thought it was defective reading so wild. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR7PI75FWMA6HKJYI3TV54E3ANCNFSM4GOKKVSA .

2s compliment, is that reading the *OUT_H and then left shifting ?

Or is there another calculation i missed in there ?

Mike Collins · Answer 31 · Sun Jul 04 2021 05:32:23 GMT+0800 (China Standard Time)

Set the sample rate to 200 Hz, the LP filter to 42 Hz and the full scale range to 2 g. Forget about calibration for now... What is the average and standard deviation of the three axes when the sensor lies motionless and flat?
…
On Sat, Jul 3, 2021 at 2:15 PM Mike Collins @.***> wrote: So thats an intersting comment about the reading time. I couldn't find even great refernce to what is good sampling time. I did find 1 routine "suggestion" to read 1100 samples every 100ms (or something like that, then thwout the first 100, and calcutlate the rest. Hasn't helped :( — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKS3PGTQDMBTCXBHPULTV54W7ANCNFSM4GOKKVSA .

so what does 200 hz turn into with TCL delays ? Are you familiar with TCL ?

I had not done anything with LP filter, before getting closer to your calib.
I don't remember if i stripped out and made a routine out of just the LP filter.

And why does full-scale range seem to be backwards for me, lol ?
Seems like full-scale would be the 16384 value not, 2048 , but again, i probably don't understand all the refernce terms.

Kris Winer · Answer 32 · Sun Jul 04 2021 05:32:25 GMT+0800 (China Standard Time)

The raw sensor data is two bytes per axis which needs to be combined to make a signed 16-bit integer, i.e., stored as 2's complement. So a signed value that fluctuates a bit about zero will look like wild swings in the data if not handled properly. On Sat, Jul 3, 2021 at 2:27 PM Mike Collins ***@***.***> wrote:

…

sounds like you are not doing the 2-s complement conversion properly? On Sat, Jul 3, 2021 at 2:10 PM Mike Collins *@*. *> wrote: … <#m_6879422066615639933_> As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins @.*> wrote: … <#m_4503423486452604556_> They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins *@*. *> wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.*> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> < #38 <#38>> < #38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA . lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet. I've reverted to reading the raw values to try and figure this out / understand. last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768. Is that ridiculous, or just fairly normal, i have no frame of refernce. Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy. Problem is i got them all off Ebay, so they all probably crappy. I already swapped 1 out first because i thought it was defective reading so wild. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR7PI75FWMA6HKJYI3TV54E3ANCNFSM4GOKKVSA . 2s compliment, is that reading the *OUT_H and then left shifting ? Or is there another calculation i missed in there ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKWMY3GYSL7JDZ2C7RTTV56CLANCNFSM4GOKKVSA> .

Mike Collins · Answer 33 · Sun Jul 04 2021 05:36:08 GMT+0800 (China Standard Time)

The raw sensor data is two bytes per axis which needs to be combined to make a signed 16-bit integer, i.e., stored as 2's complement. So a signed value that fluctuates a bit about zero will look like wild swings in the data if not handled properly. On Sat, Jul 3, 2021 at 2:27 PM Mike Collins @.***> wrote:
…
sounds like you are not doing the 2-s complement conversion properly? On Sat, Jul 3, 2021 at 2:10 PM Mike Collins @. > wrote: … <#m_6879422066615639933_> As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins @.> wrote: … <#m_4503423486452604556_> They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @. > wrote: … <#m_4641202338694893310_> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @.> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> < #38 <#38>> < #38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA . lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet. I've reverted to reading the raw values to try and figure this out / understand. last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768. Is that ridiculous, or just fairly normal, i have no frame of refernce. Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy. Problem is i got them all off Ebay, so they all probably crappy. I already swapped 1 out first because i thought it was defective reading so wild. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR7PI75FWMA6HKJYI3TV54E3ANCNFSM4GOKKVSA . 2s compliment, is that reading the *OUT_H and then left shifting ? Or is there another calculation i missed in there ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKWMY3GYSL7JDZ2C7RTTV56CLANCNFSM4GOKKVSA .

Yea i think i finally understood that too, and have it in 1 of my read functions.

I can see when i move my sensor through the range of motion i want to deal with, it goes from a postive in 1 direction and negative the other.
But the starting point is typically always a negative number.

    # Accelero and Gyro value are 16-bit
    set high [read_byte_data $register]
#     set high 0
    set low [read_byte_data [expr {$register+0x01}]]
#     set low 0

    #concatenate higher and lower value
    set value [expr {($high << 8) | $low}]

    #to get signed value from mpu6050
    if {$value > 32768} {
      set value [expr $value - 65536]
    }

Mike Collins · Answer 34 · Sun Jul 04 2021 05:43:46 GMT+0800 (China Standard Time)

The raw sensor data is two bytes per axis which needs to be combined to make a signed 16-bit integer, i.e., stored as 2's complement. So a signed value that fluctuates a bit about zero will look like wild swings in the data if not handled properly. On Sat, Jul 3, 2021 at 2:27 PM Mike Collins @.***> wrote:
…
sounds like you are not doing the 2-s complement conversion properly? On Sat, Jul 3, 2021 at 2:10 PM Mike Collins @. > wrote: … <#m_6879422066615639933> As I mentioned, if you calibrate the accel by subtracting off the offset bias (average of values while sitting still) you should see 0, 0, 1000 milligs with a jitter of +/- 10 millig, +/- 20 at the most. Said differently, the uncalibrated jitter should be +/- 10 - 20 millig. If you average the readings for ~1 second or so and calculate the standard deviation (use Excel), the deviations shouldn't be more than ~10 - 20 millig. If you do this (and you should) what values do you get for the average and standard deviation for each axis? On Sat, Jul 3, 2021 at 1:46 PM Mike Collins @.> wrote: … <#m_4503423486452604556> They shouldn't jump wildly. Maybe your sensor is crap. I would guess a super cheapo breakout from China or EBay...Probably a fake MPU6050 too... On Sat, Jul 3, 2021 at 1:31 PM Mike Collins @. > wrote: … <#m_4641202338694893310> This is my first experience using a motion sensor. I had no idea the values jumped so wildly. That's when i started searchng about how to calm that down, and got into this whole calibration discussion, lol. Probably to be expected. Don't need to read the gyro then … <#m_6404295785107076203_> On Sat, Jul 3, 2021 at 1:25 PM Mike Collins @._> wrote: So in my usage, i'm pretty sure i'm only interested in the accel values ( just movement). Are those the only registers i need to be reading or do i need gyro as well and do all the same calculations every time ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> < #38 <#38>> < #38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQCHNVAZMIEICVQEDTTV5W5BANCNFSM4GOKKVSA . Ok — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTHMOVLUXGKVM3GLEDTV5XSDANCNFSM4GOKKVSA . Ahhh crud. Yea sitting still, the values may jump from positive to negative, its just ridiculous. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR37XRKPFU2M2CFZH3TV5ZKJANCNFSM4GOKKVSA . lol, "great minds, or fools never differ" ,lol, i was just looking at doing a simple deviation but wasn't getting a very good handle on it., yet. I've reverted to reading the raw values to try and figure this out / understand. last thing i was trying was smoothing out the value some, looks like it jumps more than half the maximum +/- value of 32768. Is that ridiculous, or just fairly normal, i have no frame of refernce. Good or bad thing is have more than 1 of the sensors, so i can swap out if they seem to be crappy. Problem is i got them all off Ebay, so they all probably crappy. I already swapped 1 out first because i thought it was defective reading so wild. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR7PI75FWMA6HKJYI3TV54E3ANCNFSM4GOKKVSA . 2s compliment, is that reading the *OUT_H and then left shifting ? Or is there another calculation i missed in there ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKWMY3GYSL7JDZ2C7RTTV56CLANCNFSM4GOKKVSA .

Yea i think i finally understood that too, and have it in 1 of my read functions.

I can see when i move my sensor through the range of motion i want to deal with, it goes from a postive in 1 direction and negative the other.
But the starting point is typically always a negative number.

Mike Collins · Answer 35 · Tue Jul 06 2021 01:56:15 GMT+0800 (China Standard Time)

Hi Kris.
Well i had to backup a little bit and add some additional debuging info capture, and here's what i have.

You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are.

After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me.

`
cmd="twowire writeregbyte twowire5 0x6B 0x80"
cmd="twowire writeregbyte twowire5 0x6B 0x01"
cmd="twowire writeregbyte twowire5 0x6C 0x00"
cmd="twowire writeregbyte twowire5 0x38 0x00"
cmd="twowire writeregbyte twowire5 0x23 0x00"
cmd="twowire writeregbyte twowire5 0x6B 0x00"
cmd="twowire writeregbyte twowire5 0x24 0x00"
cmd="twowire writeregbyte twowire5 0x6A 0x00"
cmd="twowire writeregbyte twowire5 0x6A 0x0C"
cmd="twowire writeregbyte twowire5 0x1A 0x01"
cmd="twowire writeregbyte twowire5 0x19 0x00"
cmd="twowire writeregbyte twowire5 0x1B 0x00"
cmd="twowire writeregbyte twowire5 0x1C 0x00"
cmd="twowire writeregbyte twowire5 0x6A 0x40"
cmd="twowire writeregbyte twowire5 0x23 0x78"
cmd="twowire writeregbyte twowire5 0x23 0x00"
fifo_count="1024"
packet_count="85"
accel_bias is array:
accel_bias(0) =5542251
accel_bias(1) =5566359
accel_bias(2) =5465702

gyro_bias is array:
gyro_bias(0) =5513694
gyro_bias(1) =1404804
gyro_bias(2) =5546861

inList is dict:
mean min max count stdev sampvar popstdev popvar

accel_bias is array:
accel_bias(0) =65202
accel_bias(1) =65486
accel_bias(2) =64302

gyro_bias is array:
gyro_bias(0) =64866
gyro_bias(1) =16527
gyro_bias(2) =65257

accel_bias is array:
accel_bias(0) =65202
accel_bias(1) =65486
accel_bias(2) =47918

gyro_bias is array:
gyro_bias(0) =64866
gyro_bias(1) =16527
gyro_bias(2) =65257

wdata is array:
wdata(0) =192
wdata(1) =167
wdata(2) =239
wdata(3) =220
wdata(4) =192
wdata(5) =69

cmd="twowire writeregbyte twowire5 0x13 0xc0"
cmd="twowire writeregbyte twowire5 0x14 0xa7"
cmd="twowire writeregbyte twowire5 0x15 0xef"
cmd="twowire writeregbyte twowire5 0x16 0xdc"
cmd="twowire writeregbyte twowire5 0x17 0xc0"
cmd="twowire writeregbyte twowire5 0x18 0x45"
accel_bias_reg is array:
accel_bias_reg(0) =-1100
accel_bias_reg(1) =-355
accel_bias_reg(2) =1990

mask_bit is array:
mask_bit(0) =0x0
mask_bit(1) =0x01

accel_bias_reg is array:
accel_bias_reg(0) =-9250
accel_bias_reg(1) =-8540
accel_bias_reg(2) =-3999

Ax is array:
Ax(bias) =5542251
Ax(breg) =-9250
Ax(calb) =3
Ax(count) =85
Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779}
Ax(dbias) =65202
Ax(max) =65229.0
Ax(mean) =65202.95294117647
Ax(min) =65182.0
Ax(popstdev) =10.802947182139594
Ax(popvar) =116.70366782009779
Ax(sampvar) =118.09299719890848
Ax(stdev) =10.867060191188253
`

Kris Winer · Answer 36 · Tue Jul 06 2021 02:16:25 GMT+0800 (China Standard Time)

Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number.

…

On Mon, Jul 5, 2021 at 10:56 AM Mike Collins ***@***.***> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. ` cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 ` — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA> .

Mike Collins · Answer 37 · Tue Jul 06 2021 02:20:51 GMT+0800 (China Standard Time)

Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number.
…
On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.***> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA .

That's very possible.
The raw value that i initially see is something like -1012 ( or something like that)
I guess "small" is relative, and i would hope its something close to zero.

Kris Winer · Answer 38 · Tue Jul 06 2021 03:06:40 GMT+0800 (China Standard Time)

~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins ***@***.***> wrote:

…

Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins *@*.***> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA> .

Mike Collins · Answer 39 · Tue Jul 06 2021 06:20:22 GMT+0800 (China Standard Time)

~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins @.> wrote:
…
Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA .

Hmmmm, so when do the values get changed to +/- 32768 ?

I mean it's so close right now ?

Kris Winer · Answer 40 · Tue Jul 06 2021 06:50:54 GMT+0800 (China Standard Time)

In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins ***@***.***> wrote:

…

~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins *@* . *> wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.*> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA> .

Mike Collins · Answer 41 · Tue Jul 06 2021 09:12:34 GMT+0800 (China Standard Time)

In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins @.***> wrote:
…
~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins @ . > wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA .

Yep, i showed you that earlier in the thread.

#concatenate higher and lower value
set value [expr {($high << 8) | $low}]

Kris Winer · Answer 42 · Tue Jul 06 2021 09:56:03 GMT+0800 (China Standard Time)

Does this manage twos complement conversion correctly? On Mon, Jul 5, 2021 at 6:12 PM Mike Collins ***@***.***> wrote:

…

In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins *@*.***> wrote: … <#m_-8533214321366424106_> ~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins *@* . *> wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.*> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA . Yep, i showed you that earlier in the thread. #concatenate higher and lower value set value [expr {($high << 8) | $low}] — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKQ3NUGO6G2ZU4UJQTDTWJKA3ANCNFSM4GOKKVSA> .

Mike Collins · Answer 43 · Tue Jul 06 2021 10:30:01 GMT+0800 (China Standard Time)

Does this manage twos complement conversion correctly? On Mon, Jul 5, 2021 at 6:12 PM Mike Collins @.> wrote:
…
In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins @.> wrote: … <#m_-8533214321366424106_> ~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins @ . > wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA . Yep, i showed you that earlier in the thread. #concatenate higher and lower value set value [expr {($high << 8) | $low}] — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQ3NUGO6G2ZU4UJQTDTWJKA3ANCNFSM4GOKKVSA .

Far as i know.
I don't use it ALL the time, but never had any problem, when i did.

i mean its the same functionality, left shifting 8 bits, and then OR'ing the other bits in their normal positions.

Kris Winer · Answer 44 · Tue Jul 06 2021 10:45:18 GMT+0800 (China Standard Time)

Nope...likely does not handle two's complement properly. Suggest you check, and if not, do it manually. On Mon, Jul 5, 2021 at 7:30 PM Mike Collins ***@***.***> wrote:

…

Does this manage twos complement conversion correctly? On Mon, Jul 5, 2021 at 6:12 PM Mike Collins *@*. *> wrote: … <#m_3261014820833263279_> In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ <https://www.exploringbinary.com/twos-complement-converter/> google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins @.*> wrote: … <#m_-8533214321366424106_> ~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins *@* . *> wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.*> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub < #38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA . Yep, i showed you that earlier in the thread. #concatenate higher and lower value set value [expr {($high << 8) | $low}] — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQ3NUGO6G2ZU4UJQTDTWJKA3ANCNFSM4GOKKVSA . Far as i know. I don't use it ALL the time, but never had any problem, when i did. i mean its the same functionality, left shifting 8 bits, and then OR'ing the other bits in their normal positions. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKSJLFNUUJ7YMJAQLP3TWJTDJANCNFSM4GOKKVSA> .

Mike Collins · Answer 45 · Tue Jul 06 2021 11:56:25 GMT+0800 (China Standard Time)

Nope...likely does not handle two's complement properly. Suggest you check, and if not, do it manually. On Mon, Jul 5, 2021 at 7:30 PM Mike Collins @.***> wrote:
…
Does this manage twos complement conversion correctly? On Mon, Jul 5, 2021 at 6:12 PM Mike Collins @. > wrote: … <#m_3261014820833263279_> In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ https://www.exploringbinary.com/twos-complement-converter/ google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins @.> wrote: … <#m_-8533214321366424106_> ~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins @ . > wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub < #38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA . Yep, i showed you that earlier in the thread. #concatenate higher and lower value set value [expr {($high << 8) | $low}] — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQ3NUGO6G2ZU4UJQTDTWJKA3ANCNFSM4GOKKVSA . Far as i know. I don't use it ALL the time, but never had any problem, when i did. i mean its the same functionality, left shifting 8 bits, and then OR'ing the other bits in their normal positions. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSJLFNUUJ7YMJAQLP3TWJTDJANCNFSM4GOKKVSA .

Really Kris ?

TCL is a major scripting language, you think they wouldn't support that capability ?
Its fine if you don't want to assist me.
I'll just continue to look around.

Thanks.

Kris Winer · Answer 46 · Tue Jul 06 2021 12:28:30 GMT+0800 (China Standard Time)

You can assume, or just check that it works. But you are right, if you can't be bothered to simply verify then you should ask someone else for help. On Mon, Jul 5, 2021 at 8:56 PM Mike Collins ***@***.***> wrote:

…

Nope...likely does not handle two's complement properly. Suggest you check, and if not, do it manually. On Mon, Jul 5, 2021 at 7:30 PM Mike Collins *@*.***> wrote: … <#m_2126317938110217686_> Does this manage twos complement conversion correctly? On Mon, Jul 5, 2021 at 6:12 PM Mike Collins *@*. *> wrote: … <#m_3261014820833263279_> In C++ (no idea what TCL is) one converts two 8-bit bytes into one signed 16-bit integer like this: uint8_t rawData[2] = {0, 0}; readData(rawData); // function to read data int16_t output = (int16_t) ( (int16_t) rawData[1] << 8) | rawData[0]; This takes advantage of a trick or built in functionality of C++ in that the compiler knows that a 16-bit integer (int16_t) is signed so does the proper conversion. The old school way to do this is to form the uint16_t combination of the rawData and then test for a 1 in the 15th bit position. So: if(output & 0x8000) { output = ~output + 1; output = - output; } so that 0xFFFF would become -1, the smallest negative value for 16-bit signed data. . 0x7FFF would be +32767, the maximum positive value for 16-bit signed data. https://www.exploringbinary.com/twos-complement-converter/ <https://www.exploringbinary.com/twos-complement-converter/> https://www.exploringbinary.com/twos-complement-converter/ <https://www.exploringbinary.com/twos-complement-converter/> google twos complement conversion On Mon, Jul 5, 2021 at 3:20 PM Mike Collins @.*> wrote: … <#m_-8533214321366424106_> ~ -60 milligs or something like this at 2 g FS, typical and easily corrected offset bias.... On Mon, Jul 5, 2021 at 11:21 AM Mike Collins *@* . *> wrote: … <#m_-8435496152261155125_> Ax(min) =65182.0 This is not properly cast in 2's complement. 65182 is 0xFE9E, which is a very small negative number. … <#m_-4677973895798972836_> On Mon, Jul 5, 2021 at 10:56 AM Mike Collins @.*> wrote: Hi Kris. Well i had to backup a little bit and add some additional debuging info capture, and here's what i have. You will notice some of the variable names are basically the same as in your library, so won't be hard to see what some values are. After you asked about std deviation, i went looking for a TCL package thst provided some of those functions. And they are captured in the TCL "array" named after the axis. Since i hadn't got this to work, sort of after main recoding, i'm only testing with 1 axis at this poiunt, dealing with all 3 when just 1 doesn't work, just seems silly, to me. cmd="twowire writeregbyte twowire5 0x6B 0x80" cmd="twowire writeregbyte twowire5 0x6B 0x01" cmd="twowire writeregbyte twowire5 0x6C 0x00" cmd="twowire writeregbyte twowire5 0x38 0x00" cmd="twowire writeregbyte twowire5 0x23 0x00" cmd="twowire writeregbyte twowire5 0x6B 0x00" cmd="twowire writeregbyte twowire5 0x24 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x0C" cmd="twowire writeregbyte twowire5 0x1A 0x01" cmd="twowire writeregbyte twowire5 0x19 0x00" cmd="twowire writeregbyte twowire5 0x1B 0x00" cmd="twowire writeregbyte twowire5 0x1C 0x00" cmd="twowire writeregbyte twowire5 0x6A 0x40" cmd="twowire writeregbyte twowire5 0x23 0x78" cmd="twowire writeregbyte twowire5 0x23 0x00" fifo_count="1024" packet_count="85" accel_bias is array: accel_bias(0) =5542251 accel_bias(1) =5566359 accel_bias(2) =5465702 gyro_bias is array: gyro_bias(0) =5513694 gyro_bias(1) =1404804 gyro_bias(2) =5546861 inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar inList is dict: mean min max count stdev sampvar popstdev popvar accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =64302 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 accel_bias is array: accel_bias(0) =65202 accel_bias(1) =65486 accel_bias(2) =47918 gyro_bias is array: gyro_bias(0) =64866 gyro_bias(1) =16527 gyro_bias(2) =65257 wdata is array: wdata(0) =192 wdata(1) =167 wdata(2) =239 wdata(3) =220 wdata(4) =192 wdata(5) =69 cmd="twowire writeregbyte twowire5 0x13 0xc0" cmd="twowire writeregbyte twowire5 0x14 0xa7" cmd="twowire writeregbyte twowire5 0x15 0xef" cmd="twowire writeregbyte twowire5 0x16 0xdc" cmd="twowire writeregbyte twowire5 0x17 0xc0" cmd="twowire writeregbyte twowire5 0x18 0x45" accel_bias_reg is array: accel_bias_reg(0) =-1100 accel_bias_reg(1) =-355 accel_bias_reg(2) =1990 mask_bit is array: mask_bit(0) =0x0 mask_bit(1) =0x01 accel_bias_reg is array: accel_bias_reg(0) =-9250 accel_bias_reg(1) =-8540 accel_bias_reg(2) =-3999 Ax is array: Ax(bias) =5542251 Ax(breg) =-9250 Ax(calb) =3 Ax(count) =85 Ax(data) ={65202.95294117647 65182.0 65229.0 85 10.867060191188253 118.09299719890848 10.802947182139594 116.70366782009779} Ax(dbias) =65202 Ax(max) =65229.0 Ax(mean) =65202.95294117647 Ax(min) =65182.0 Ax(popstdev) =10.802947182139594 Ax(popvar) =116.70366782009779 Ax(sampvar) =118.09299719890848 Ax(stdev) =10.867060191188253 — You are receiving this because you commented. Reply to this email directly, view it on GitHub < #38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>> <#38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>>> (comment) <#38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSYM6P77TN7XXSFKNDTWHW4VANCNFSM4GOKKVSA . That's very possible. The raw value that i initially see is something like -1012 ( or something like that) I guess "small" is relative, and i would hope its something close to zero. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> <#38 <#38> <#38 <#38>>> (comment) <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKTLRH6RVMFN7CG636TTWHZY3ANCNFSM4GOKKVSA . Hmmmm, so when do the values get changed to +/- 32768 ? I mean it's so close right now ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 <#38> <#38 <#38>> (comment) <#38 <#38> (comment) <#38 (comment) <#38 (comment)>>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKR73WNUHMX4AUMC27LTWIV3FANCNFSM4GOKKVSA . Yep, i showed you that earlier in the thread. #concatenate higher and lower value set value [expr {($high << 8) | $low}] — You are receiving this because you commented. Reply to this email directly, view it on GitHub < #38 <#38> (comment) <#38 (comment) <#38 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKQ3NUGO6G2ZU4UJQTDTWJKA3ANCNFSM4GOKKVSA . Far as i know. I don't use it ALL the time, but never had any problem, when i did. i mean its the same functionality, left shifting 8 bits, and then OR'ing the other bits in their normal positions. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment) <#38 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSJLFNUUJ7YMJAQLP3TWJTDJANCNFSM4GOKKVSA . Really Kris ? TCL is a major scripting language, you think they wouldn't support that capability ? Its fine if you don't want to assist me. I'll just continue to look around. Thanks. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#38 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKWJ45PJKG6AZFO43BDTWJ5HJANCNFSM4GOKKVSA> .