Project Logs

Collapse

• Firmware Source Committed

  Michael Wessel • 12/12/2023 at 02:47 • 0 comments

  https://github.com/lambdamikel/picoram2090/tree/main/src/picoram_2090 
  I am omitting the TTS code for now though (i.e., the SPI Interface to the Epson S1V30120).
  

• PicoRAM 2090 got a link on the official Busch homepage!

  Michael Wessel • 12/05/2023 at 02:43 • 0 comments

  https://www.busch-modell.de/information/Microtronic-Computer.aspx 

• Final Demo Video: https://youtu.be/dFnKhUI47KU

  Michael Wessel • 11/26/2023 at 18:30 • 0 comments

  It's the end of the project as we know it - thanks for hanging on, folks!
  

  
  

• GitHub updated! https://github.com/lambdamikel/picoram2090 ​

  Michael Wessel • 11/25/2023 at 17:44 • 0 comments

  The GitHub repo and page have been updated. The repo also includes the Gerbers and firmware by now; firmware sources will come a bit later as I am still refactoring things a bit: 
  

  https://github.com/lambdamikel/picoram2090

  Enjoy!
  
  

• Final PicoRAM 2090 PCB

  Michael Wessel • 11/24/2023 at 05:15 • 0 comments

  The third revision of the PCB is the final one for now - I have fixed all shortcomings and am satisfied with it for now. Now, I'll make one more final demo video, and then upload of the sources and Gerbers to GitHub.
  

• First PCB Assembled

  Michael Wessel • 11/19/2023 at 16:50 • 0 comments

  The good news is - it works!
  

  The bad news is: I'll need one more round of PCB revisions; problems are: 

  • the power switch is blocking the Pico's USB port - I hence had to put in a temporary micro slide switch which is very flimsy and a bit of a hack.
  • I also learned that the double throw 6pin DPDT power blue & white switches come in 7x7 and 8x8 mm; the 7x7 that I currently have don't fit the KiCAD footprints without pin bending. So I ordered the 8x8 ones.
  • the silkscreen text for buttons and button legend is too small. 
  • the SRAM 2114 data lines are in reverse bit order - I had to change the firmware for now by including a reverse_bits on the data lines, but have fixed this already in the PCB. 
  • the voltage divider for the Microtronic's DISP line is 1k-1k instead of 1k-2k... I had to cut a trace and use one of the spare 2k resistors in one of the isolated resistor network DIPs with a bodge wire. 
  • I made a stupid mistake when I crimped the 2114 connector cable - it took me 2 hours before I finally identified the badly crimped cable shorting every adjacent pair of wires, questioning my design, as the main culprit why it wouldn't work at all at first. I had to de-solder the crimp connector from the board and messed up a trace, requiring one more bodge wire.

  Well, the 2nd revision of the PCB is already in production, and this should be the final one! 

  Here are a few pics of the mess ups:
  

• PCB Designed and in Production!

  Michael Wessel • 11/12/2023 at 16:19 • 0 comments

  Waiting for the PCB to arrive... 

• RetroChallenge 2023/10 Log Entry #6 - Final Wrap-Up Demo Video

  Michael Wessel • 10/29/2023 at 07:48 • 0 comments

  I achieved everything I wanted for this year's RetroChallenge 2023/10 - here is a final wrap up demo video:

  So the last week I mostly spent time writing the demo programs you see in the video. Some of these graphics demos (involving trig) were created with the help of a bunch of little Lisp functions (see below) - the Microtronic cannot do trig functions (sin, cos) reasonably fast, so these were pre-computed with Lisp and materialized in the code. The Lisp-program generated Microtronic instructions were just put in a .MIC file and then put on the PicoRAM's SD card. In the end, it turned out to be a good design decision to utilize an SD card interface for memory dumps / .MIC file storage, as opposed to, say, an EEPROM - given that I can easily create and exchange .MIC programs / content with and from the PC using the SD card.
  

  It might be a good idea to actually add extended op-codes just for trig, i.e., if I wanted to add simple Logo-turtle like graphics capabilities as op-codes, I will need fast trig from register memory as well. For now, I've run out of available vacuous op-codes, but it is of course possible to implement different sets of vacuous op-codes. There could be one set which includes op-codes for sound and speech, like the current set, and another set that utilizes the same vacuous op-codes for trig functions instead. I could use one of the PicoRAM push buttons to cycle through different op-code sets, just like I am doing now to swap memory banks.
  

  Interesting side note: running some of these graphics demos, I noticed that the 2nd core, which is running the op-code extensions, display, user interface, SD card interface, speech, sound, .etc., wasn't fast enough during certain display line drawing operations to keep up with the Microtronic. As the PicoRAM has no way of halting the Microtronic (it only snoops the RAM bus!), if the extended op-codes such as drawing a line to the display take too long, the 2nd core might miss the next extended op-code if it's still busy executing the current operation (e.g., drawing the line). Hence resulting in (graphics) glitches.
  

  Solution: I overclocked the Pico to 250 Mhz - all glitches are gone! This problem never occurred for the SRAM emulation running on the 1st core, only the 2nd core was problematic as it was implementing the extended op-codes, driving the display, etc. In general, the Pico is a great platform for this kind of work - fast and powerful. I don't think this would have been possible with the old AVR MCUs - the 2nd core is a blessing. And overclocking and using the 2nd core is insanely easy, even from C! It's definitely a platform / MCU I am going to use for future projects. Despite the slight annoying required level shifters, it's close to perfect.
  

  Speaking of the op-codes - writing all these demo programs I tweaked the firmware and its op-codes here and there, just to make the programming more convenient. You don't really know how well a programming language or a set of op-codes works until you eat your own dog food! For example, I removed the original
  

     50D SOUND PLAY FREQ <X><Y> 

  op-codes because I found it more useful to have an op-code that enables auto-echo to the speech synthesizer of text. With speech echo enabled, I can use the
  

     506 DISP SHOW CHAR <LOW><HIGH>

  op-code to drive the text display as well as the speech synthesizer.
  

  The final (hah - or let's say latest, current) set of op-codes is hence the following: 
  
  

     0xx ENTER LITERAL DATA x
  
     3Fx ENTER DATA FROM REG x 
  
     500 HEX DATA ENTRY MODE
     501 DEC DATA ENTRY MODE
     502 DISP CLEAR SCREEN 
     503 DISP TOGGLE UPDATE  
     504 DISP REFRESH
     505 DISP CLEAR LINE  
     506 DISP SHOW CHAR 
     507 DISP CURSOR SET CURSOR LINE 
     508 DISP SET CURSOR  
     509 DISP PLOT   
     50A DISP LINE    
     50B DISP LINE FROM   
     50C DISP LINE TO     
     50D SOUND PLAY NOTE  (SOUND OFF FIRST) 
     50E SPEAK DISP ECHO
     50F SPEAK BYTE 
  
     70x SWITCH MEMORY BANK x
  

  And here is the Lisp program that I used to generate...

  Read more »

• RetroChallenge 2023/10 Log Entry #5 - DS3231 for Microtronic

  Michael Wessel • 10/22/2023 at 17:12 • 0 comments

  In my original RetroChallenge 2023/10 participation announcement, I promised a battery backed-up Real Time Clock (RTC) for the Microtronic. Well, here it is! I have hooked up the DS3231 RTC.

  
   I did not use any of the DS3231 Pico SDK C libraries found online literally, but I had used this module before in some of my previous projects (e.g., https://hackaday.io/project/177784-lambdaspeak-futuresoft-edition) and hence didn't have a lot of issues getting it to work. Most helpful I found https://github.com/bnielsen1965/pi-pico-c-rtc/blob/master/ds3231.c and I took bits and pieces of it, but heavily refactored - thanks, good code, Bryan Nielsen (https://github.com/bnielsen1965)! Nice and simple - I really dislike when people create a huge CPP framework for something that is not rocket science.
  

  So how does that work? See, the Microtronic already has an internal real time clock in its firmware, and there is even an op-code F06 ("load time") that can be used to load the current time to register memory (registers for A-B for hours, C-D for minutes, and E-F for seconds). Two built-in ROM programs (PGM 3 and PGM 4) also allow you to set and see the current time (without using the F06 op-code).
  

  However, the Microtronic RTC is not battery backed-up, and hence if you want to use the real time in your programs, you will always have to set the clock manually when you turn it on (using the ROM program PGM 3). This of course means limited utility.
  

  How can the DS3231 RTC help here? Well, using similar ideas as explained in my previous log entry https://hackaday.io/project/192655-picoram-2090/log/224467-retrochallenge-202310-log-entry-3, instead of adding extra side-effects / semantics to vacuous op-codes, I can also intercept a non-vacuous op-code such as F06 ("load time"), and do anything I want by switching first into a dedicated memory bank to not interfere with the current program, then execute this program with the Microtronic, and finally return to the original program. When F06 is intercepted, the Pico first retrieves the current time from the DS3231, and then materializes a set of MOVI nx = Move Immediate Value n to Register x (op-code 1nx) into the current memory bank that will, when executed by the Microtronic, write the current DS3231 time into the A to F registers, overwriting the values that F06 had already left there. After returning from the banked-in sub-program, the registers A-F contain the correct time.
  

  Here is a demo video:
  

  Again, the underlying mechanism that loads the DS3231 time is entirely transparent to the calling program - from a user program's perspective, there is no difference in accessing the DS3231 vs. accessing the internal Microtronic RTC - the same single op-code is used, nothing else is required. This is the program from the demo: 

  00 F6A # display "time" registers A..F on the LED display 
  01 F06 # load time from RTC 
  02 FF0 # wait for keyboard input 
  03 F08 # clear registers
  04 C00 # goto 00 
  

• RetroChallenge 2023/10 Log Entry #4 - DEMO

  Michael Wessel • 10/21/2023 at 14:06 • 0 comments

  Here is a first "all-in-one" demo - SD card interface, graphics, text, speech!
  

  Here is the program:
  

  00 F08 # clear registers 
  01 F20 # display registers 0 and 1 on the LED display 
  02 50A # enable draw line (x1, y1) - (x2, y2) mode; display (0,0) - (x7F,x1F) 
  03 000 # send x1 low nibble: 0 
  04 3F0 # send x1 high nibble: value in register 0 
  05 000 # send y1 low nibble: 0 
  06 000 # send y1 high nibble: 0 
  07 000 # send x2 low nibble: 0
  08 3F1 # send x2 high nibble: value in register 1 
  09 0FF # send y2 low nibble: F  
  0A 011 # send y2 high nibble: 1 -> x1F (32 4*8 = 32) 
  0B 520 # add to to register 0 
  0C 980 # reg 0 = 8? 
  0D E0F # yes, then goto x0F 
  0E C02 # no, goto x02 
  0F 100 # reg 0 = 8: clear register 0 (to 0) 
  10 521 # add 2 to register 1 
  11 981 # reg 1 = 8? 
  12 E14 # yes, then goto x14
  13 C02 # no, goto x02 
  14 50E # register 2 = 8, end of nested reg 0, 1 loop. enable speak echo mode.  
  15 506 # enable print ASCII char to display with speech echo 
  16 0DD # send ASCII codes for MICROTRONIC (x4D, x49, x43, ...); M 
  17 044 
  18 099 # I 
  19 044
  1A 033 # C 
  1B 044
  1C 022 # R 
  1D 055
  1E 0FF # O 
  1F 044
  20 044 # T
  21 055
  22 022 # R 
  23 055
  24 0FF # O 
  25 044
  26 0EE # N
  27 044
  28 099 # I
  29 044
  2A 033 # C 
  2B 044
  2C 0AA # LF (0x0A) -> speech synth will speak MICROTRONIC 
  2D 000
  2E F00 # end of program, halt 
  

View all 17 project logs