GVP A2000-030 Red Screen

[raskulous]

Hello everyone!


I have an a2000 that I installed a GVP a2000-030 combo card in, and I'm only getting a red screen on boot with the card enabled.

I've tried my system with a romulator (with 1.3, 1.4, and 2.1 kickstarts), as well as two different known working 1.3 ROMs, all produce the same red screen result.

As soon as I remove the GVP card, it's back to working normally.

I normally run a RAM expansion card, and a SCSI controller, but both of those are currently removed from the system.

Any ideas what I should try next? Should I get a USB EPROM programmer and test this ROM on the GVP?

The label on my GVP auto-boot ROM is here: https://i.imgur.com/EDDfbW5.jpeg

[mark_k]

Does the card have any RAM installed? Maybe check whether the jumper settings correspond to the amount of RAM?

[thebajaguy]

You can test without any RAM enabled on the 030 card. Set the board to all-high RAM setting and remove the EPROM. The EPROM software-maps the high memory, and that won't happen with it out (The SCSI interface will be inactive also). There is a jumper you can also pull to disable the ROM being seen, but I am not 100% sure of the board model you have (it varies between the Combo22/33 and the G-Force models).

I'd also temporarily pull all expansion cards if you have anything else installed (particularly RAM, RTG, or Bridgecard). The 030 card, with memory enabled, will claim some space in the 8M Zorro II expansion area with these Zorro II cards now needing to share it. If you already have 16-bit memory or other use for that that space, you could be going past the limit, and some expansion cards don't behave well when there is no space for them. There is also the motherboard revision (mentioned in the next paragraph) that might come into play.

I've seen red-screen errors when the CPU edge connector or slot is dirty (Isopropyl Alcohol can help here), damaged 8520's (try swapping CIA-A and CIA-B to start). and sometimes on motherboards that are very early revs lacking the more stable field revs. The 4.4/4.5 and 6.2+ are generally stable and most common, but 4.1-4.3 can be marginal, and the 6.0-6.1 were known to have expansion issues (the terminator packs on the side of the buffer chips between CPU slot and Z2 slots are the easy thing to spot for the 6.2 update - note that C= pulled in most of the problem Rev 6/6.1 motherboards and fixed them, but there are a few stragglers). If you still have red-screen with no 32-bit RAM enabled, give all motherboard DIP chips a mild lift and re-seat - just in case oxidation has affected connection/continuity. (Give Agnus a general thumb-press down only - I don't recommend trying to get it to move more than that without proper tools and knowledge).

If there was battery damage in the past, there may be signals which were damaged that can matter to the CPU card, but not the usual 68000 operation.

[QuikSanz]

@thebajaguy,


How good are the caps on these cards, getting old?


Chris

[thebajaguy]

On any GVP card that is CPU, memory, SCSI, Serial: Any of the end-of-chip small ceramics (TH or SMD) are merely for digital switch noise - they tend to be okay, but you can replace them 1:1 if you feel up to it. If they are TH type Electrolytic and scattered around, they also do the same digital-noise cleanup job. They are of little concern if they are TH type. The cards can run without them.

Any of the cards with the larger SMD Tantalum/Ceramics, particularly near the 5V/12V connector area, but sometimes scattered around some cards, are used for FCC noise and some filtering on the power lines entering the card, and have been known to burn up/short in more recent times when they fail. You can replace these with good quality SMD parts, and there is no reason not to try to use the Panasonic Electrolytic/Hybrid in these places. They have no logic-circuit purpose, but do clean up noise from the system/prevent it from entering the system board. I've had my Spectrum 28/24 fry at these parts, and @Acill repaired them for me.

I don't have any direct experience with the audio products or the IV/24, but I would suggest carefully replacing 1:1 with what is there, as many on those will probably have analog properties, and the values are important.

[QuikSanz]

I'll be having him take care of my A2000 board, may as well send him the GVP also.

[raskulous]

Originally Posted by mark_k

Does the card have any RAM installed? Maybe check whether the jumper settings correspond to the amount of RAM?

I have combed through all the jumper settings, and they are correct. There is no additional RAM installed in the SIMM slots.

[raskulous]

Originally Posted by thebajaguy

There is a jumper you can also pull to disable the ROM being seen, but I am not 100% sure of the board model you have (it varies between the Combo22/33 and the G-Force models).

I've tested this, as I am not using the on-board SCSI. The boot from SCSI is disabled (J9), so it's doing it's auto-boot to the storage in the A2000 (which is a Bhudda IDE)

Originally Posted by thebajaguy

I'd also temporarily pull all expansion cards if you have anything else installed (particularly RAM, RTG, or Bridgecard).

I've tried this as well, all cards are removed from the system currently.

Originally Posted by thebajaguy

I've seen red-screen errors when the CPU edge connector or slot is dirty (Isopropyl Alcohol can help here), damaged 8520's (try swapping CIA-A and CIA-B to start). and sometimes on motherboards that are very early revs lacking the more stable field revs. The 4.4/4.5 and 6.2+ are generally stable and most common, but 4.1-4.3 can be marginal, and the 6.0-6.1 were known to have expansion issues (the terminator packs on the side of the buffer chips between CPU slot and Z2 slots are the easy thing to spot for the 6.2 update - note that C= pulled in most of the problem Rev 6/6.1 motherboards and fixed them, but there are a few stragglers). If you still have red-screen with no 32-bit RAM enabled, give all motherboard DIP chips a mild lift and re-seat - just in case oxidation has affected connection/continuity. (Give Agnus a general thumb-press down only - I don't recommend trying to get it to move more than that without proper tools and knowledge).

If there was battery damage in the past, there may be signals which were damaged that can matter to the CPU card, but not the usual 68000 operation.

I will try this. My board is a 4.3, so it's definitely in the "older" category. I'll clean the slots up with IPA, but I already went over the pins on the card itself to make sure it's nice and clean. The battery damage on my board was very minimal, and it's been cleaned up and working well for years now. I have a lithium 2032 battery installed on it now.

I can also try re-seating all the chips on my board, thanks.

[thebajaguy]

Default for Combo22 and G-Force 030/25 is 1MB soldered on.
Default for Combo33 and G-Force 030/40 and 50 is 4MB soldered on.
There is a jumper that will flip the memory config from AutoConfig of that bae amount to high/software mapped. The latter position, with no ROM active (remove ROM or disable ROM with the proper jumper), means no RAM will be seen. Takes the possibility of bad memory out of the equation.
- Note there is a jumper that tells the driver (in the ROM) not to poll the SCSI. It's not that jumper. It's another that makes the ROM disappear off the card. We want that jumper.

[raskulous]

Further to my other replies above, I've also tested the GVP with disabling the FPU by removing it, removing the crystal oscillator, and changing the required jumper. Same red screen result.


One other thing about this card: It's labelled on the board as "40MHZ", but there is a 50MHZ 030 installed in the CPU slot. Is this abnormal? Was it upgraded at some point, or is the little sticker referring to the FPU (Which is 40MHZ)?



This is a photo of my card:

[thebajaguy]

Originally Posted by raskulous

I will try this. My board is a 4.3, so it's definitely in the "older" category. I'll clean the slots up with IPA. The battery damage on my board was very minimal, and it's been cleaned up and working well for years now.

I can also try re-seating all the chips on my board, thanks.

I have found that some battery damage neutralizations are not a thorough as they may seem. I have had an A2000 6.2 on my bench for a while that I had to remove the keyboard connector, mouse port, 68000 socket, and re-run a couple of lines. Had corrosion still present in those hiden points. The system was 'unstable' but had been running. Took another tech (Thomas of Amiga of Rochester) to re-flow several chip solder points to get the thing stable again.

If this doesn't work, then you need the A2000 schematics (Rev 6 is fine) and to trace out each signal from the 86-pin slot to each of it's termination points. Battery damage can cut some of these lines, but there are a few that are unique to the CPU slot, and some are extensions to the slot that are not in the normal 68000's path.

[thebajaguy]

Regarding the pic - ignore the '40MHz' - they were all capable of being built into either speed (25/40/50), but the programable parts on the board made the clock speed. Some cards were shipped with a 50MHz 68030 when 40MHz were not in hand to build the cards originally. Others have replaced them in the field and tried for more.

FPU is made to be clocked Async of the CPU with the other oscillator socket. In all the years at GVP, we never had an FPU presence cause issues with the CPU. This is likely due to 5v parts being used everywhere on the card. Modern accelerator cards today use 5v tolerant 3.3v parts in many areas, and may be more sensitive to the 'load' the FPU places on some lines.

It's possible the card was being forced up to 50MHz (some of them came with 50MHz PAL sets, others were with 40MHz parts but the tolerance was enough to make the speed). I can't read the codes and speeds off the PALs, so not able to give a hint.

To see if that may be the issue, put the 40MHz clock in the main position and clock both FPU and CPU at 40MHz.

[raskulous]

Originally Posted by thebajaguy

Default for Combo22 and G-Force 030/25 is 1MB soldered on.
Default for Combo33 and G-Force 030/40 and 50 is 4MB soldered on.
There is a jumper that will flip the memory config from AutoConfig of that bae amount to high/software mapped. The latter position, with no ROM active (remove ROM or disable ROM with the proper jumper), means no RAM will be seen. Takes the possibility of bad memory out of the equation.
- Note there is a jumper that tells the driver (in the ROM) not to poll the SCSI. It's not that jumper. It's another that makes the ROM disappear off the card. We want that jumper.

Hmm okay.


J9 is to disable the ROM, which I currently have set to OPEN.
J14 is to tell the card if there is a SCSI drive connected, the manual says it should be closed if there are no drives connected. It's currently closed.


In the manual it states that if you already have 16-bit wide RAM in auto-config space, it's probably wisest to config all the combo board RAM as Extended by removing the jumper on J12. So, I should remove J12 and test again?

[raskulous]

Originally Posted by thebajaguy

To see if that may be the issue, put the 40MHz clock in the main position and clock both FPU and CPU at 40MHz.

Oh that's interesting! With the 50MHz clock removed and the 40MHz clock in the main position it boots to my Romulator screen, but doesn't seem to want to get to a kickstart screen after a selection is made. I've tried 1.3, 1.4, and 2.1.


I put the clocks back in their respectful positions just as a test, and it goes back to red screen as expected.



It's progress, though...

Oh and I'm in Canada. So it should be NTSC, although I don't think that really matters for our troubleshooting.

[Jope]

Have you tried a physical 1.3 ROM when you clock to 40MHz?

[thebajaguy]

NTSC/PAL Amiga won't matter. Kickstart version won't matter. Works the same against all versions.

The Romulator shouldn't matter, either.

I agree the 40MHz setting may be where you want to be for the time being. It may be safe to set the card for AutoConfig memory (4MB). These should be accurate settings for the cards if your manual is not handy: https://bigbookofamigahardware.com/b...ct.aspx?id=177 J8 is the ROM visible/disable jumper. J14 controls SCSI bus poll (you said you have no drive on it).

Semi-Unrelated to your troubleshooting - what is the Rev of the G-Force card? (top edge, above the RAM), and can you give me a closeup pic of the PALs in sockets in the lower-right region? I hope to see part numbers (hex checksums) and part speeds where possible. U34/U36 might tell me what board speed this originally was.

[raskulous]

Originally Posted by Jope

Have you tried a physical 1.3 ROM when you clock to 40MHz?

Just trying that now. 1.3 stock kickstart installed, Buddha IDE removed, still does not get to the workbench insert screen.


With both the stock 1.3 Kickstart ROM and the Romulator it behaves "better" in the sense it bypasses the ROM check and cycles through the other colours on the screen, but just fails to reach the workbench disk insert screen.


Last colour on the screen is Grey-ish..



"LIGHT GRAY CIA (U8/U301) problem Stops at gray, CIA defective"


I'm going to swap CIA chips and try again.

[Jope]

Originally Posted by thebajaguy

The Romulator shouldn't matter, either.

Quite a few reports in the announcement thread about problems with the Romulator and CPU slot cards in the A2000

http://eab.abime.net/showthread.php?...ight=romulator

[raskulous]

Originally Posted by thebajaguy

NTSC/PAL Amiga won't matter. Kickstart version won't matter. Works the same against all versions.

The Romulator shouldn't matter, either.

I agree the 40MHz setting may be where you want to be for the time being. It may be safe to set the card for AutoConfig memory (4MB).

Semi-Unrelated to your troubleshooting - what is the Rev of the G-Force card? (top edge, above the RAM), and can you give me a closeup pic of the PALs in sockets in the lower-right region? I hope to see part numbers (hex checksums) and part speeds where possible. U34/U36 might tell me what board speed this originally was.

Rev 3.


Getting photos of those PALs is tough, but I think I got it all. Let me know if there's anything you can't read.


https://imgur.com/a/oOjkfkR

[raskulous]

Swapped CIAs, same thing. Without the GVP it boots to the workbench insert screen right away, with the GVP installed it stalls on a grey screen (I believe light grey)