labnotes/_source/_posts/2024-05-01-cosmac.markdown

layout	title	date	comments	categories
post	Building a COSMAC/RCA CDP1802 learning computer	2024-05-01 13:00:00 +0200	true

PCB making

Printing the transfer

I tried several methods to prepare the toner transfer intermediate:

• baking parchment paper with double sided tape
• glossy magazine with scotch tape
• glossy magazine with glue

What seemed to work best is the glossy magazine glued to a sheet of paper strong enough to not have irregularities. ⚠️ Don't forget to mirror the image before printing 🤦

Preparing the board

Grain #500 sandpaper -> steel wool -> acetone

Transferring the toner

• 1st attempt (glossy paper taped to carrying paper, max heat, 1.5 min): failure the paper sticks to the PCB, no transfer
• 2nd attempt (parchment paper glued to carrying paper outside of printing zone, max heat, 3 min moving across surface, let cool): much better, good enough for silkscreen but probably not for etching
• 3rd attempt (glossy paper glued, pre-scrubbed with isopropanol):
  • very good print, almost no defects
  • transfer with acteone/alcool (3:8): total failure, did not stick to copper at all.
• parchment paper, taped to carrying paper on all sides (outside of printing zone), pre-cleaned with isoprop
  • iron 2 min on each part of the surface, temp setting 2, lots and lots of pressure
  • let cool completely
  • transfer failed :-(

Looking for a laminator...

Laminator (Pavo Vancouver A4) / parchment paper : utter failure, almost no transfer at all

:check: Great result with stickers backing sheet + cloth iron (max temp) + lots of pressure

• press for 2 minutes without moving iron much
• shift weight to distribute pressure
• high pressure finish over details (iron side/tip)

Reference: Youtube

Fix broken traces with Lumicolor permanent marker

⚠️ Missed a broken trace

Etching

Etchant solution:

• 50 mL 23% hydrochloric acid
• 50 mL 12% hydrogen peroxide

Etches in a few minutes. Reaction is exothermic. Don't take it out too early. The protection by toner and marker both work well.

Soldering

Checking regularly for continuity, expected resistance values, and absence of shorts was key to success.

Debugging

System worked on first try except for LOAD mode flip-flop. The 4027 from AliExpress behaves in a weird way, as though J and K were reversed.

The expectation is:

• ~DMA-IN is cabled to ~Q (i.e. the active low DMA IN signal is inactive when Q=0 ~Q=1)
• at startup Q=0 ~Q=1
• upon a raising edge of CLOCK, the flip flop transitions to Q=1 / ~Q=0 / DMA IN is active
• upon a positive signal on RESET, the flip flop transitions to Q=0 / ~Q=1 / DMA IN is inactive

According to both the truth table and the circuit design, this is the behaviour that should be observed with J=1 K=0. In practice this is the behaviour observed with J=0 K=1.

If I set J=1 K=0 and set Q=1 by presenting a positive signal on SET, then on a raising edge of CLOCK I see the flip flop going to Q=0 ~Q=1.