VOGONS


First post, by polpo

User metadata
Rank Member
Rank
Member

I have a 386sx system that I'm building with this motherboard and I seem to be having problems with anything that uses low DMA channels 1-3. This first manifested with the system not being able to boot or access floppy drives. It would hang on booting from floppy, or if booted from a CF card would give errors in DOS when accessing the floppy. I thought I had a bad super IO card, so I tried another and had the same problem. I continued on, not minding too much that I wouldn't be able to use a floppy drive.

Fast forward a bit to when I built my own Snark Barker SB clone card and I got highly distorted and slow output when playing digitized audio with DMA (thanks to TubeTime for writing SBDIAG to help narrow down that it's DMA-related). At first I thought it was my fault as it's a DIY board, but to make sure, I put in my Gravis Ultrasound board and configured it to use DMA 1, and got distorted samples. However, the GUS works perfectly fine when using any high DMA. At this point, when reading about DMA channels, I noticed that the floppy controller uses DMA 2. So yeah, now I'm pretty sure that all low DMAs don't work reliably on this board.

Some more info: the motherboard has a Chips & Technologies NEAT chipset (datasheet here) which includes the common 82C206 peripheral controller which has integrated DMA controllers (datasheet here).

So what can I do now? I'd much rather fix this problem than find another 386 board as they're getting expensive these days. I'm hoping it's a bad 74LS series chip that I can swap out, or perhaps I could swap out the C&T 82C206 as there seem to be plenty available on eBay but that's a more complicated rework. I have an oscilloscope and the willingness to poke around on the board – what should I look out for?

Reply 1 of 10, by weedeewee

User metadata
Rank l33t
Rank
l33t

From previous threads on this forum regarding similar situations, I kinda recall that the dma might be multiplexed, going through a logic chip, before reaching the 82c206.
Going from that recollection, you might have a bad logic chip on he board.

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 2 of 10, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

DMA, interrupt, clock timers etc are in 82C206 chipset. Replace this will fix the low DMA issues.

Cheers,

Great Northern aka Canada.

Reply 3 of 10, by polpo

User metadata
Rank Member
Rank
Member

So I've purchased a 82C206 and hot air rework station, as the chip is not socketed on this motherboard. Wish me luck, I've never done SMT soldering of this scale before! I'm going to watch plenty of tutorials on YouTube first.

Reply 4 of 10, by polpo

User metadata
Rank Member
Rank
Member

I desoldered the 82C206, put in a PLCC socket, and inserted the 82C206 I got on eBay. Good news: the system boots and high DMA still works. Bad news: low DMA still doesn’t work. Next I’m going to start tracing signals that are DMA related around the board and test every logic chip they go through.

I am a little bit worried about that PAL chip that’s nearby. If it’s bad I’m out of luck since it’s likely I won’t be able to recover the programming from it and flash it to a GAL.

Attachments

Reply 5 of 10, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

A '206 chip should not have the DMA signals multiplexed, so first check if you get connections between it and ISA slots - both for request and ack. If not, see if there is some 74 buffer in between, could be it died. That would be an easy fix.
If you do get direct connections then possibly the issue is with address lines between main chipset and '206 (or data lines, but that's less likely as it would affect more stuff). Pinout for '206 is on the net, all of these chips are identical no matter the manufacturer.

Reply 8 of 10, by polpo

User metadata
Rank Member
Rank
Member

So, egg on my face: I traced the DREQ1 and DACK1 lines based on my photo above with the chip off. I checked for continuity and found that there was no continuity on DACK1 to where I was expecting it to go on the board. After I popped the 82C206 chip back out and gave things a good look, I finally spotted it: the only pin I missed soldering was the DACK1 pin. Soldered it up, popped the chip in, and DMA1 is now working perfectly! I did later verify that the DACK and DREQ pins on the ISA bus are connected straight through to the '206 chip, with no buffering via any of the 74xxx chips on the board. I should have noticed something was odd when DMA 1 was failing in a different way after I first tried the new 82C206: SBDIAG was producing no sound at all, and samples uploaded to the GUS were even worse sounding than before.

I also tried out a floppy drive and that also works.

Thanks for the ideas, all, I now finally have a perfectly functioning 386 board!

Attachments

  • dack-dreq.jpg
    Filename
    dack-dreq.jpg
    File size
    265.4 KiB
    Views
    544 views
    File license
    Public domain