This is a follow-up to my other thread, where the CFFA1 corruption was solved by replacing the four 74S257 with 74LS257:https://www.applefritter.com/content/cffa1-—-data-corrupted-only-cards-sram-range-1000–8fff
The machine ran fine after that. Meanwhile, I got another cool card, the Vossi's IECmem card, which can supply both the RAM and the ROM from its own SRAM/EPROM: https://github.com/vossi1/Apple1-IECmem . I wrote a whole new DOS and menu for it and it worked well. Today, I removed the E–W wire on the patch area and made a W–1 bridge instead, to get contiguous RAM (bank W at $1000), as the IEC board provide a basic rom.
And the machine refused to boot after that....
Symptom since then: power-up shows the normal _@ pattern, CLEAR SCREEN gives a single blinking @, and RESET does nothing at all ,no keyboard response. Same with the keyboard unplugged and RESET done by shorting pin 1 to pin 9.
Restoring the original E–W jumper did not fix it. This is when I began to panic....
What I have checked:
- All four supplies correct, no ripple, +5V at the CPU stays good under load.
- CPU: /RES high and drops on reset, clock present, R/W, A13–A15 and D0–D7 all toggling. A second 6502 behaves the same.
- PIA (6520): supplies, /RES, E and chip selects all fine, but PB0–PB6 all sit around 1.85V, i.e. still configured as inputs — so Wozmon never gets as far as initialising them I think.
- Patch area: continuity correct, no shorts between 0/1, W/X or 1/2.
The interesting part. I plugged Vossi's IECmem card in the machine, which can supply both the RAM and the ROM from its own SRAM/EPROM: https://github.com/vossi1/Apple1-IECmem
with all 16 DRAMs and both PROMs removed, the X–0, W–1 and Y–F pads open, and the card supplying RAM (DIP 1) and the Wozmon (DIP 7), the machine does reset, prints "" and accepts keystrokes — but the characters come out wrong and erratic, and nothing can be run. With the keyboard unplugged, a reset produces random characters on screen.
Putting back either the DRAMs alone or the PROMs alone — even with those pads still open, so they are never selected — kills it again: no reset.
So: CPU, PIA, terminal and keyboard all work, every testable chip passes, supplies are clean, and yet the machine only comes to life (sort of) when the motherboard supplies neither RAM nor ROM — and even then the data is unreliable.
ha, and today, it's my birthday, what a nice present, I was so happy to have a working Apple 1 at last, well, it's life :-)
Solved. It turned out to be three separate dead components, which is why it was a pain to pin down.
Background: the machine stopped working right after I soldered a jumper in the RAM bank patch area. No prompt after reset, reset itself doing nothing.
Step 1 — nothing ran at all, not even with vossi's IECmem card:https://github.com/vossi1/Apple1-IECmemThat card emulates ROM and RAM, switchable by region with DIP switches. The machine should have booted from it with no PROMs and no motherboard RAM at all, but it didn't.
So I pulled the TTLs and tested them one by one on my programmer. One 7410 in B2 failed. That gate sits in the DRAM RAS network and enables the 8T97 buffers. Measuring confirmed it: pin 1 of the 8T97s sat at 1.8 V with the bad part, 4.06 V with a good one. Replaced it, and the machine finally booted from the IEC card. From there I could test methodically.
Step 2 — vossi's ramtest:https://github.com/vossi1/Apple1-ramtestIt replaces the Wozmon at FF00 and needs no working RAM at all, no zero page, no stack — same principle as Uncle Bernie's diagnostic PROMs. It run rom the IEC card. I had to burn a rom with this test instead the Wozmon. It found a dead DRAM in bank X immediately. Swapped it and the machine boots normally again. Out of 16 chips only that one was gone.
One thing that nearly sent me down the wrong path: with only a few chips left in the second bank, testing the fully populated bank throws errors on exactly the bits that are populated in the other bank. At first the test showed half the bank as dead. One bank must be fully populated and the other completely empty.
Step 3 — the PROMs. The machine still hung whenever they were fitted, even with the y patch open. I moved the y patch to B so they wouldn't collide with the IEC card's ROM, then just dumped B000.B0FF in the Wozmon and compared it to a known good Wozmon image.
149 of 256 bytes wrong — and every single error is in the upper nibble. The lower nibble is correct across all 256 bytes without one exception. Repeated several times with power cycles in between, identical every time.
One nibble completely gone, the other perfect, is the signature of a single dead PROM. And it explains the hang precisely: at FF06 the byte reads 80 instead of D0, so STY $D012 becomes STY $8012 and the display is never initialised.
So: 7410 in B2, one DRAM, one PROM. My guess is still ESD while soldering, though I can't prove it.
New PROMs are on the way — a friend built an Arduino-based burner and has already verified that it work.
Can you share any info on the Arduino-based burner?
Thanks
Wow, great work, glad you were able to figure it out and get it running again so quickly.
ESD issues with BPROMs? They’re usually "relatively" forgiving.
You should check your soldering iron tip for dangerous potentials.
I recently had a nasty surprise with a supposedly high-end soldering station.