A4000 rebuild - DiagRom works, Kickstart boots to black screen

[PacificState]

I've just rebuilt a battery-rotted A4000 by building up a clone motherboard and transferring all the critical ICs over - and I've also recapped the A3640 board that came with it.

By burning and installing Chucky's DiagROM, I've verified that the machine 'works' and passes all the tests - however, any attempt to install Kickstart ROMs (either the original ones that came with the machine, or new ones from the images on the 3.2 AmigaOS CD) results in a black screen (and yes - I've tried with a floppy drive connected).

Few things:

- I'm waiting on a new Ricoh chip as I'm not sure if the original one is functional (the original battery damage was extensive); however, the machine doesn't boot with it installed or out

- I've tried swapping the CIAs around, but to no avail

- I can tap the Caps Lock key and flash the LEDs about 8 times before they hang

I also have Amiga Forever, so I'm thinking of burning a Logica ROM

Can anyone suggest next steps? I can only think that the Ricoh chip matters.

[jbenam]

Originally Posted by PacificState

Can anyone suggest next steps? I can only think that the Ricoh chip matters.

Hello there, the A4000 can boot without the Ricoh chip with no issue (it just won’t have RTC) at least that’s what I remember happening on mine.

You might be seeing an issue with one of the chips non tested by DiagROM (Buster, Zorro GALs, etc). You can try removing Buster as IIRC the A4000 will boot fine with it, you will just lose all Zorro-related functionality.

Also I would check that A3640 on another Amiga if possible - just to be sure.

[giantclam]

Originally Posted by jbenam

also i would check that a3640 on another amiga if possible - just to be sure.

+1

[PacificState]

Thanks:

- I do have access to another A4000, but it'll be some months away!

- I'm a bit uncertain that the EPROMs I have are working correctly, so I'm going to burn the DiagROM again and check that it works in a different set.

I actually tried burning the Logica ROMS just now and, again - I get a black screen. Can anyone who has the images to hand post the md5's of the files?

[PacificState]

I was able to try the A3640 on another A4000 board I'm restoring - albeit one temporarily without SIMM slots:

- With the DiagROMs installed, I get a green/red flash at boot

- With Kickstart ROMS installed, I again get a black screen

So I think this points to the card. Having recapped it already, I don't see any obvious rot, so I guess the next options are (per http://wordpress.hertell.nu/?page_id=514) to:

- buzz out the traces

- desolder the GALs and verify their content (I've never done this before, but I assume my TL866ii and adapter board will handle it)

It's odd to me that the DiagROM is fine, though - there's clearly some difference. Any other ideas?

[hooverphonique]

Originally Posted by PacificState

- desolder the GALs and verify their content (I've never done this before, but I assume my TL866ii and adapter board will handle it)

I've programmed those GALs using a TL866CS before, so should work..


But indeed, the cpu clearly works since diagrom "works", so it is a really strange one..

[AmigaHope]

Dumb question, but do you have a hard drive properly jumpered and installed on the IDE port? The A4000 will blackscreen for an entire minute if the IDE port is left empty. Some people made IDE resistor terminators for the IDE port to stop this behavior (for people who booted off of a SCSI controller, etc.)

[PacificState]

Thanks! Connecting the floppy seems to make no difference. Actually, I *didn't* connect the floppy in the test on the second board as I had no chip ram installed - I think you should get the green screen before the hang.

[AmigaHope]

Originally Posted by PacificState

Thanks! Connecting the floppy seems to make no difference. Actually, I *didn't* connect the floppy in the test on the second board as I had no chip ram installed - I think you should get the green screen before the hang.

It's good to have a floppy drive connected to the floppy port, but I'm not talking about the floppy, I mean the *HARD DRIVE* connected to the hard drive port. Kickstart really really wants there to be a hard drive on the IDE port, jumpered to be the only drive, or two drives master and slave.

[PacificState]

I've tried a drive, too ;( Even with it connected, and no Chip RAM installed, it's 50/50 with a Kickstart ROM as to whether I get a green screen at boot, whereas the DiagROM boots consistently.

[Jope]

Hold up, are you testing without the Chip RAM SIMM? Perhaps keep that in there. If you suspect it is broken, you should be able to get away with a larger double sided SIMM.

[PacificState]

I'm testing without chip RAM first, to make sure I can get consistently to a green screen - which only happens about 50 percent of the time.

[PacificState]

And... it doesn't work with a CPU card that works in an A3000. Something is clearly up here that the DiagROM doesn't detect!

[giantclam]

...and what does diagrom serial port output look like?...

[PacificState]

So there are now two problems here when DiagROM boots:

- The keyboard sometimes doesn't work at all, then works fine after I reset

- I had thought that serial didn't work as I had the wrong 9->25 adapter, but I now think the cabling is right - and it seems that the Paula lines aren't swinging between 5V Peak-Peak, which seems to mean the line driver TxD pins aren't hitting +/- 12V.

I'm starting to look at the serial issue next as I bet it's related (bad Paula?).

[PacificState]

Hmm:

- When the keyboard fails, the data line toggles continually and the clock stays high when hitting a key; then, after a reset sometimes, if the keyboard starts to work the clock line starts toggling during a keypress. I’m not familiar with the keyboard logic, otherwise I’d presume the lines would be the other way around. Maybe this is a keyboard fault?

- The serial *does* seem to be toggling / albeit the peak-peak voltage halves when I connect it to serial. Continuing to look at this one on the scope…

[PacificState]

I also swapped Paula and the CIA without any improvement.

[hooverphonique]

Originally Posted by PacificState

if the keyboard starts to work the clock line starts toggling during a keypress. I’m not familiar with the keyboard logic, otherwise I’d presume the lines would be the other way around. Maybe this is a keyboard fault?

Why would you not expect the clock to toggle when pressing a key?


The keylock feature disables the keyboard clock, maybe it's acting up (depending on where you measure the static clock state) ?


I can't deduce from you posts whether you think Paula is involved in keyboard input - it isn't.

[PacificState]

I'm definitely learning here, and now understand how the keyboard wiring works - the clock isn't continuous (as I'd thought) and toggles to clock the data out from the keyboard. It presumably looks like the clock line isn't being pulled low. And no, I don't think Paula is involved in the keyboard issue, but I was trying to figure out if it was related to the serial problem. To recap, I have three problems with the machine:

- It boots with the DiagROMs and checks out fine, but fresh 3.2 Kickstart ROMs result in a black screen even with a hard drive connected
- The serial lines appear to be toggling as expected, but the host computer doesn't detect anything
- The keyboard works 50% of the time

And there's now a fourth problem - which may be the same as the first: I extracted the Logica ROMs from my Amiga Forever installed and burned those instead. Now I see that the machine starts to boot, but I end up hanging on a white screen - which the manual says is a vertical blanking failure. Any ideas as to what could cause that?

[DigitalKeeper]

-> the Board boots without the RTC (no clock!)
-> the Board boots without Buster (no Zorro Bus!)
-> the Board boots without a keyboard

Diagrom -> Red/Green Flashing -> indicates a Problem with ChipRAM

So......

.....without Pictures - difficult to say more....

But - have a look at U891 and the traces there...

Here is a Picture with Areas to check:


-----

unnecessary to say, that the RAM Socket *MUST" be removed!