Anyone got a STE with DMA issues ?

[JezC]

Finally dug out the 3x working STE's - only one has the -38 DMA chip (it also has the external blitter)...

I can't easily check which brand of CPU it is as there are several clips holding it down & the dual TOS board (non-Exxos) in there has wires taped across the top of the clips

Is there any date/serial number range that might help narrow down what CPU brand it might be?

I'll try it with a hard drive sometime tomorrow...but I'd be surprised if it doesn't work as it was bought as a bundle with an ICD FaST hard drive (and was used by the previous owner for MIDI stuff, primarily Cubase IIRC)

Hope to have more news soon...

[RetroPassionUK]

I have tried several STEs now of various " disposition & revisions". It's just typical, I cannot get any of them to fail DMA transfers now I have copied endless thousands of files about. Previously I would only have to create a folder or save desktop. It was always repeatable.

I think the problem is tolerances everything is just so borderline that even just using a different power supply or a different ROM board could be enough to make or break things. Problem is a lot of my old test boards of been dismantled and rebuilt that many times over the years.. I did not specifically save a broken STE specifically for testing DMA issues anymore.

I mean last time I was doing tests the MOT brand CPU would always fail but the SGS ones wouldn't. I have tried 5 different MOT CPU's and they all work. I likely use the same set of CPUs when testing previously. Maybe the things only fail on extremely cold days or extremely hot days or something

I will have a hunt for some more of the STE boards. But as it stands I will have to put out a plea to UK folks. That if any of you have a failing ultrasatan with your STE motherboard, to basically loan it to me to investigate it. I cannot develop a hard drive test program when I do not have a faulty system

I have a mini ultrasatan you can borrow if that helps? Not had a chance to use it at all!!

[exxos]

Is there any date/serial number range that might help narrow down what CPU brand it might be?

Not what I am aware of unfortunately. There must be some cutoff when it Atari started using the 001 DMA. But I have no idea how that correlates to serial numbers.

Really it is just a matter of trying hard drives if it fails then investigate further..They generally I think the problem is likely only be apparent with more modern drives like ultrasatan.

[exxos]

I have a mini ultrasatan you can borrow if that helps? Not had a chance to use it at all!!

Thanks, though it is the STE itself which suffers from the problems more than the hard drives.

[exxos]

Still looking

[JezC]

Still looking

Well, I just tried the potential "dodgy DMA" STE with the Megafile 30 I bought off you a few years back (wit a Mega ST system unit) & it allowed me to copy the sysinfo.prg from floppy onto the HDD & run it from the HDD without error.

I also dug out a DataPulse hard drive which I'll try tomorrow (in case there is anything 'special' in the Megafile drives that would counteract any potential issue).

After that, I have a System Solutions MINIS drive which is reasonably accessible but the others are all packed away in the garage, so they won't be accessible for a week or three.

So, barring problems with those two mechanical drives, I guess we're no further forward - unless anyone thinks it's worth me testing this STE with a more modern non-mechanical HDD on the off chance...?

[exxos]

So, barring problems with those two mechanical drives, I guess we're no further forward - unless anyone thinks it's worth me testing this STE with a more modern non-mechanical HDD on the off chance...?

Why do things never break when you want them to

I'm not sure how much of a problem it is with older drives. In theory they should suffer the same. Mostly it wasn't until everyone started using ultrasatan that things really got noticed.

[Steve]

So, barring problems with those two mechanical drives, I guess we're no further forward - unless anyone thinks it's worth me testing this STE with a more modern non-mechanical HDD on the off chance...?

Why do things never break when you want them to

I'm not sure how much of a problem it is with older drives. In theory they should suffer the same. Mostly it wasn't until everyone started using ultrasatan that things really got noticed.

Yeah cause 'back in the day' the amount of home users (not business users) who actually had a hard disk would have been miniscule, like less than 5%. Businesses on the other hand may have had a hard disk, but if they lost all their data they probably would have just binned the Atari and bought an Amstrad or something. So when you think about it like this, it's no surprise Atari shipped so many faulty units without really knowing about it.

[Steve]

@JezC in my experience every single STe with -38 chip I have ever tried has had the problem. Of course with modern drives like Ultrasatan though. I have a feeling that the manufacturers of older hard disks may have built in certain tolerances that help with the issue (I think only recently someone was disecting an ICD adaptor and finding very strange design decisions inside) So personally I'd be fairly certain that STe you have would have a problem if tested with Ultrasatan.

[JezC]

@JezC in my experience every single STe with -38 chip I have ever tried has had the problem. Of course with modern drives like Ultrasatan though. I have a feeling that the manufacturers of older hard disks may have built in certain tolerances that help with the issue (I think only recently someone was disecting an ICD adaptor and finding very strange design decisions inside) So personally I'd be fairly certain that STe you have would have a problem if tested with Ultrasatan.

Well, quite happy to try one & see if it gives problems or not! I can't afford to buy one just to test it out at the moment though, so it'd have to be a loan for the test (or try it when the next retro event is organised).

The other option might be to build one of those STM32-based kits & try one of those instead...but are they likely to show problems as much as the U/S do...?