Recently (Apr 7, 2024)...

Have been getting maindec stuff going again, by updating the github repo (no hardware changes).

Also did a little with the TC08 dectape emulator, though it doesn't boot OS/8.  Reading on the web about TC08 and dectape, sounds like the drivers do a lot of cycle dependent stuff.  For example, the OS/8 boot block switches the field number in the middle of the transfer.  So I'm not going to put much more effort in it, sounds like too much of a tar-baby for me.  The RK8JE emulator works with OS/8 so that's good enough.

Source code breakdown...

  • asm - pdp8 assembler (allows long labels) and linker
  • cc,ctrl - left-overs from hode, no pdp-8 stuff
  • de0 - out-of-date DE0 FPGA implementation
  • driver - autoboardtest, raspictl programs that run on PC,RasPI to control the tubes
  • filesys - an aborted attempt to write a PDP-8 file system
  • focal - focal-69
  • freqcounter - arduino-based front panel frequency counter (pcb in kicads/freqctr)
  • iowarrior - makefile that downloads and builds IO warrior library - not used
  • kicads - various PCBs
  • maindec - scripts to run maindec programs on simulator and on tubes
  • modules - processor source code
  • netgen - takes modules files and generates PCBs, real Verilog, C++ simulator, TCL simulator
  • scheme - a scheme interpreter
  • silly - silly programs
  • zynq - up-to-date Zynq/ZTurn FPGA implementation

About the tubes...

So back in June or July 2022 I didn't know what tubes to use, I was thinking 12AX7 but they are $10..$20 apiece.  I went to MIT flea market and saw someone with a couple tables of tubes, and looked at what he had.  I noticed a carton of 5 6J6s and wondered what they were.  So I looked on my phone and saw that they were dual triode, common cathode.  I thought I could use them.  So I gave him the $10 he wanted.  I put one on perfboard and did measurements and it seemed to work pretty much as the datasheet said.  Then I had some small circuit boards made that would have one tube and places for the resistors, diodes, capacitors and connectors to make a pair of and-or-invert gates.  When they arrived, I hooked it all up and tried it out, seemed to work ok as gates.  Then I took three of the tubes (6 gates) and made a D-FF and that seemed to work.  IIRC, clocked around 100kHz (ahem kcps).

Meanwhile I wondered how common are the 6J6s?  Looking on the Internet and eventually ebay of course, I found someone selling cartons of 200 6J6s for $150.  This is too much of a co-incidence and probably one of those ebay too-good-to-be-true deals.  But it was worth the risk of $150.  And amazingly, a box arrived and had some label for some DoD installation in Utah, dated October 1985.  Opened it up and little plain-white cartons of 5 tubes, each box stamped 6J6A.  So I tried them in my D-FF setup and they worked!  Then I thought hope there are still more of those cartons available.  Fortunately, yes, so I bought two more cartons, thinking one would be enough (giving me 400 tubes), but good to have an extra carton so I have plenty of spares for an extra $150.  Unfortunately seems those cartons of 200 are long gone, at least from that source.

Fortunately though the design isn't too dependent on the tube type.  I was originally thinking that I might even end up mixing tube types.  All I would have to do is use different eight-tube board layouts, like one for 6J6s, one for 12AX7s etc.  I would need to have different resistors and cathode bias zeners for the different tube types.  The main logic boards would still be the same cuz they just work on 5V logic.  One could even make the tube boards from JFETs or MOSFETs.