Byte Hamr Labs

Pulling a 32 MHz clock mod out of a CoCo 3

/ Tandy CoCo 3 / updated 2026-09-25 / from dino-homebrew coco3 6309 gime repair basic09 nitros-9

The CoCo 3 with the 6309 in it boots normally now and holds up to tapping and light jostling. At some point someone had added a 32 MHz speed mod to it, and the machine was always flaky. I pulled the mod, put the oscillator back to what the schematic calls for, and cleaned a corroded GIME socket. It hasn’t had a long burn-in yet.

I sat down to do software on it. The flakiness kind of interrupted that, so most of the session went to the repair.

Someone had added a 32 MHz crystal and cut around R65

The mod was a switch between the stock 28.63636 MHz crystal and a 32 MHz one, with a jumper wire underneath bypassing R65. The trace from one leg of X1 to the resistor network feeding the GIME had been scratched off, and the jumper added in its place. Maybe that was intentional, maybe not. High mode never worked. The machine was always a bit unstable: bump it or use it for a few minutes and it glitched or froze, and a light tap sometimes brought it back.

I couldn’t find this mod written up anywhere, or any reliable article on running a CoCo 3 at 32 MHz with R65 bypassed. So I didn’t try to work out what it was meant to do. I just wanted it back to what the schematic calls for.

When I pulled it, one leg fell off the 32 MHz oscillator. Maybe that was the problem all along, maybe it wasn’t.

Back to the schematic, and a corroded GIME socket

I went through the CoCo 3 schematic, found the oscillator, and tested the parts around it, including the GIME, before pulling anything.

  1. Removed the 32 MHz oscillator, the switch wiring and the R65 jumper.
  2. Cleaned up the oscillator area, resoldered the stock 28.63636 MHz crystal, and bridged the scratched-off X1 trace with a clean jumper.
  3. The GIME had blue corrosion on it, which is copper, probably moisture or old flux. I pulled it from its PLCC socket by working alternate corner notches.
  4. Cleaned the socket and the chip leads with DeoxIT, rinsed with isopropyl, let it dry, and reseated it with the chamfered corner matching the socket.

It boots, and tapping doesn’t do anything to it any more. The mod, the trace and the socket could each explain the old symptoms, so I can’t say which one it was.

BASIC09 ran its first program

BASIC09 shares a name with Color BASIC but it’s much closer to Pascal: named procedures, typed variables, parameters, structured loops and no line numbers. Motorola had Microware build it around 1979 to show off the 6809, and OS-9 grew up partly as its host.

This was the first procedure, entered with e hello and a leading space on each line to insert it:

DIM i:INTEGER
FOR i=1 TO 3
PRINT "hello from basic09 "; i
NEXT i
CommandPromptDoes
e nameB:create or edit a procedure
l*E:list the whole procedure
qE:leave the editor, errors show up here
run nameB:run it
save name / load nameB:source to and from the data directory
pack nameB:compile to I-code as a shell command
byeB:back to the OS-9 shell

make stays silent

The native assembler is fine by hand:

asm /dd/dev/assem/mon.asm o=/dd/dev/assem/mon l

That produced the listing and the mon module. make on the same file does nothing at all. Line endings aren’t it, since the makefile was written on the CoCo. The command line in it starts with a space (20) where make wants a tab (09), and the editor turns CTRL+I into four spaces, so I can’t type a tab. Getting make working normally is next, starting with make -? to see what filename and options it expects.

The Speech/Sound Cartridge never got a fair try

The SSC has never worked on this CoCo 3. Every session I tried to use it, the machine flaked out, so I don’t know yet whether the cartridge has a problem of its own. With the machine stable it gets another go.

Still open