Closed Super DMAC for Amiga 3000

[Toaster4000]

I've got a Rev2 Super DMAC in my 16 MHz A3000, and it doesn't seem to like my A3640. SCSI mostly works, but it randomly writes corrupt data, so I can't actually use it.

I'd love a Rev4, but since I'm unlikely to find that, I'd also be interested in a (later run?) Rev2 that has enough tolerance to work with the A3640. I'd be happy to buy outright, trade, or trade+cash, as long as I end up with a Super DMAC that doesn't randomly corrupt my data while using an A3640.

For those considering a swap because they don't care about A3640 compat, I've yet to repro any issue with my Rev2 Super DMAC while using the stock 68030, and the machine was used for years for music production on two SCSI hard disks using that same original Super DMAC. So it works, but not how I'd like to be using it.

Thanks!

 

[screemo]

Hello,

I'm not really an expert, but did you consider trying the newer scsi chip instead of the WD Proto that is already in the machine?

It was known to cause lock ups and wierd things if the scsi was being tasked.

Anyway, it might be worth a try - chip is AM33C93A-16PC (this is a US seller: https://www.ebay.com/itm/203353258866?epid=1608780855&hash=item2f58cc7b72:g:ZIUAAOSwn7JYDO3s)

Best
Stefan

 

[Toaster4000]

Hello,

I'm not really an expert, but did you consider trying the newer scsi chip instead of the WD Proto that is already in the machine?

It was known to cause lock ups and wierd things if the scsi was being tasked.

Anyway, it might be worth a try - chip is AM33C93A-16PC (this is a US seller: https://www.ebay.com/itm/20335325886...UAAOSwn7JYDO3s)

Best
Stefan

Already using the updated SCSI chip, unfortunately. Everything is stable with the 030. But thanks for the idea!

 

[screemo]

Arh bugger, had hoped it was an easy fix ?

 

[Toaster4000]

Arh bugger, had hoped it was an easy fix ?

You and me both! ?

 

[anightin]

I'm trying to get my Super DMAC working with my A3660 board as it happens.

My issue presents itself as mostly checksum errors when doing reads.

I'm just trying to establish if this is because of the Super DMAC and A3660, or whether it's to do with SCSI termination of which there is none on my A3000 motherboard (but enabled on my SCSI2SD) and the external connector.

 

[Toaster4000]

I'm trying to get my Super DMAC working with my A3660 board as it happens.

My issue presents itself as mostly checksum errors when doing reads.

I'm just trying to establish if this is because of the Super DMAC and A3660, or whether it's to do with SCSI termination of which there is none on my A3000 motherboard (but enabled on my SCSI2SD) and the external connector.

The easy test there would be to pull the accelerator and try again, assuming the data on the drive isn't already corrupted. For me, with the A3640 in, if I kept writing the same file to disk over and over, after a few minutes the MD5 of the written data would stop matching the source, indicating random data corruption.

I'm not sure what you mean about termination. The acceptable options are termination on the SCSI2SD + resistor packs on the A3000 system board/nothing connected externally, or no resistor packs on the system board/external termination.

It's worth nothing that on some A3000s, including mine, the termpower diode was installed backwards by Commodore. This should be fixed if you have the same problem on your machine, but it wasn't the cause of any of my issues. If the diode is installed in the right direction, you shouldn't need to connect power to your SCSI2SD.

 

[anightin]

Thanks Toaster4000 ! I tested the termination power diode and then played with external and internal termination settings and got much the same results with the A3660 i.e. when the board gets warm after about 5-10 mins I start to get the occasional checksum error on reads.

Swapping the 030 back in and the problems went away again.

I noticed that my A3660 was running with a 5ns delay line setting (thanks to hese for the prompt) so changed this to 10ns and now everything is working fine.

 

[hese]

I noticed that my A3660 was running with a 5ns delay line setting so changed this to 10ns and now everything is working fine.

When you asked for help regarding the A3000 SCSI issues you had a few months ago, I mentioned to check the A3660 delay line setting and set it to 10ns if it was not that already.

If you use the card in an A3000 with 5ns delay line setting, you are asking for trouble. Better just to keep it in the stock 10ns setting.

 

[anightin]

When you asked for help regarding the A3000 SCSI issues you had a few months ago, I mentioned to check the A3660 delay line setting and set it to 10ns if it was not that already.

If you use the card in an A3000 with 5ns delay line setting, you are asking for trouble. Better just to keep it in the stock 10ns setting.

You did indeed, and it was this prompt that caused me to check and change it, thanks hese , should have given you credit for that -- it has been running fine since then.

 

[Toaster4000]

If you use the card in an A3000 with 5ns delay line setting, you are asking for trouble. Better just to keep it in the stock 10ns setting.

Interesting. Is there an equivalent change for a stock A3640 that might fix SCSI compatibility? I'd guess from what you're saying a stock A3640 would already be at 10ns, and therefore there'd be nothing for me to try changing.