Copyright 2022 Douglas P. Fields, Jr. All Rights Reserved.
Decode the output from a Symbolics 19" Console to the CPU using a Terasic Cyclone V GX Starter Kit board and a Digilent RS-422 PMOD.
The signal is RS-422 based, but uses a Biphase encoding. (This seems also to be called Differential Manchester encoding.)
- Marking (1) is a long pulse either high or low
- Spacing (0) is a pair of shorrt pulses, either high then low or vice versa
This allows a clock to be recovered from the signal.
Per the 19" Console Mantenance Guide (August 1991):
- "75 Kbaud bidirectional link over a serial line"
- Two pairs of "Serial biphase" for txd and rxd to/from console
- Also: 75 ohm coax line for "phase encoded video"
Per the 19" Premium Console Hardware Technical Support Maintenance Guide (November 1993):
- "receives keyboard codes, mouse clicks, and mouse movements, translates them to Lisp Machine keycodes, and sends them via RS422 serial out the console port."
75 "KBaud" means 13⅓ μs per bit.
Measurements with the oscilloscope:
- Short biphase pulse: ~6.4 μs
- 2x Long biphase pulse: ~26 μs
- Oscilloscope measured frequency: 38.5KHz; period: 26.00 μs
- Long pulses when no input - so this is likely "marking" (logic 1)
Short biphase pulse
Two long biphase pulses
Conclusions:
- Use 6 μs for the minimum length of a short pulse and 2x that for long pulse
- Ignore pulses that go, say, 3x the length of the short pulse
- Ignore pulses that go less than, say, 0.5 μs
- At 50 MHz system clock, 6 μs is 300 clocks (20 ns or 0.02 μs per clock)
Additional equipment used:
- Digilent Analog Discovery 2
- Rigol DS1074Z+ Oscilloscope
- Tek TDS754A Oscilloscope
- Custom Hirose to DB9/SMA cables, using 26AWG twisted pairs and 75 ohm RG-179 coax
TODO
With the NRZ data, it's possible to decode it using a logic analyzer set to UART mode at 75 kbaud. (The logic analyzer actually thinks it's running around 76.92 kbaud. This seems reasonable as it is almost a multiple of 19.2kbps which would be 76.8kbps.)
Set to 8 data bits, one stop bit and no parity, console seems to send two bytes of data for each keypress and release. The keyboard has 88 keys on it. Two keys have a light (CAPS, MODE locks).
TODO: See if it is 7 data bits + parity? TODO: What's the bit order? (WaveForms decodes UART LSB first.)
Record various keypresses using Waveforms.
Waveforms settings:
- Samples: 100M
- Rate: 500 kHz
- Mode: Record
- Trigger: Normal, Simple (on UART data falling edge)
Key Down Up
-------- ----- -----
a C0 29 A0 00
b C0 12 A0 00
c C0 11 A0 00
d C0 2A A0 00
e C0 3A A0 00
f C0 22 A0 00
g C0 2B A0 00
h C0 23 A0 00
i C0 34 A0 00
j C0 2C A0 00
k C0 24 A0 00
l C0 2D A0 00
m C0 13 A0 00
Key event Code
-------- -----
a down C0 29
b down C0 12
b up D0 12
a up A0 00
Key event Code
-------- -----
a down C0 29
b down C0 12
a up D0 29
b up A0 00
Key event Code
-------- -----
a down C0 29
b down C0 12
c down C0 11
b up D0 12
a up D0 29
c up A0 00
Original Reversed NOT Reversed NOT
-------- -------- ------- ------------
A0 05 5F FA
B0 0D 4F F2
C0 03 3F FC
Code Meaning
----- ----------
A0 00 All keys up
C0 xx Key down
D0 xx Key up
xx = see *symbolics-keyboard-mapping*
Things to look at:
si:*kbd-auto-repeat-enabled-p*- sys.sct/l-sys/console.lisp (search around for keyboard and kbd)
- console-send-command - tell us current brightness & volume
- reset console command
- initialize-keyboard
- console-hardware-char-available and console-get-hardware-char
- sys.sct/sys/console.lisp - generic console support
- console-convert-to-software-char
- tv:default-screen
- console-serial-read-byte
- There is a bunch of Keyboard definition stuff here, constants, shift bits
- %%kbd-hardware-char-key-number
- 4 mouse button names (:mouse-l, -m, -r, and -4)
- Some keys cannot autorepeat (#\Function, #\Select, Network, Suspend, etc.)
- ;; When the repeat key is pressed, we want to start repeating ;; the last character only if its key is still held down.
- %type-key-up %type-key-down %type-all-keys-up
- %%kbd-hardware-char-opcode
*symbolics-keyboard-mapping*- this is the jackpot of 88 keys in order, counted in octal
a - 51, 29, 41 (octal, hex, dec)
b - 22, 12, 18
c - 21, 11
d - 52, 42
e - 72, 3A
f - 42, 22
m - 23, 13
- sys.sct/l-sys/wired-console.lisp
wired-slb-console-process-byte- seems to handle console decoding- Look up functions: ldb, ofb, ofb
sys.sct/l-sys/sysdef.lisp
Defines these 8
(DEFENUMERATED *SLB-CONSOLE-BYTE-TYPES*
(%TYPE-MOUSE-SWITCH
%TYPE-MOUSE-MOVE
%TYPE-ALL-KEYS-UP
%TYPE-STATUS
%TYPE-KEY-DOWN
%TYPE-KEY-UP
%TYPE-SERIAL-WINDOW
%TYPE-SERIAL-IN))If these had binary assignments:
raw with 1bbbxxxx
%TYPE-MOUSE-SWITCH 000 8x
%TYPE-MOUSE-MOVE 001 9x
%TYPE-ALL-KEYS-UP 010 Ax
%TYPE-STATUS 011 Bx
%TYPE-KEY-DOWN 100 Cx
%TYPE-KEY-UP 101 Dx
%TYPE-SERIAL-WINDOW 110 Ex
%TYPE-SERIAL-IN 111 Fx
Seems like the leading 1 means "command" of some sort per wired-slb-console-process-byte.
This same function shows that the status can return:
- 0 - Brightness
- 1 - Volume
- 2 - Switchpack (?) whatever that is
- 4 - Modem status
Mouse stuff:
- Mouse is processed 60 times a second per sys.sct/l-sys/wired-console.lisp
- mouse-kbd-button-transition
Mouse movemends seem to send 1 byte at a time, hex codes like 91 94 95 90
Mouse movements seem to be:
Up: 92
Down: 91
Right: 94
Left: 98
Down/right: 95 (and 91, 94 mixed in)
So, movements are 9x with the four directions as individual bits.
Mouse buttons: down then up (with mouse ball in)
Left 84 80
Middle 82 80
Right 81 80
Took mouse ball out: LM down then up: 04 02 00 (why 0 and not 8?)
Mouse seems to be unable to properly detect multiple mouse buttons pushed down simultaneously, or at least doesn't report multiple.
- LR down then up: 84 81 80
- MR down then up: 82 84 81 80
Or, maybe left is internally wired the same as middle & right?
ldb-test bytespec integeris in sys.sct/sys2/lmmac.lisp- not zero (ldb arg1 arg2)
ldb./sys.sct/i-sys/opdef.lisp ./sys.sct/clcp/functions.lisp- Manual: Symbolics Common Lisp Dictionary (Book 9) page 312
- Summary of Byte Manipulation Functions in Language Concepts manual
- p108, section 4.2.4.9
- Similar to Common Lisp
- Also see: Bitsavers manual PDF pages 107-108 (listed as pages 111-112)
- It can beep
- It has brightness and volume
- It can have audio
- We will need to handle the "status request" from the computer to the console, and then it responding with the info per the %type-status above:
There is a conflict between the Altera USB-Blaster and the Digilent Analog Discovery 2. They both seem to use FTDI USB interface chips, and plugging both into the same USB port (by way of a hub) can make the Altera (Quartus) jtagserver/jtagconfig not see the Altera USB Blaster. So, plug them in in different ports or orders until you find one that allows both to be used concurrently. Or, just unplug the analog discovery when you want to program your Altera (Intel) FPGA over the USB-Blaster.
Got an odd error once from Waveforms: The device is being used by another application SERC: 1001 JTAG-IDs h44002093 h00000000 Device programming failed.
https://forum.digilent.com/topic/8797-analog-discovery-2-and-altera-usb-blaster-conflict/
- Atmel Manchester Coding Basics includes a discussion of Biphase encoding as well; see figure 2-1
- Digilent PMOD RS485
- Terasic Cyclone V GX Starter Kit
- Intel Quartus 21.1