floppy needs working DMA
try a sound card in DMA mode

I will see what I have, can I check for DMA in checkit ?

rasz_pl wrote on 2022-08-02, 15:59:

floppy needs working DMA
try a sound card in DMA mode

I have made a video for the first part of the repair, I am no expert in these motherboard so just used basic fault finding and worked the damaged areas.
https://www.youtube.com/watch?v=S1IrG-e3O4I

rasz_pl wrote on 2022-08-02, 15:59:

floppy needs working DMA
try a sound card in DMA mode

I tried my aztec sound galaxy and FM works, but no digital sounds. Checkit will see the floppy drive DMA when installed and checkit just gave errors for every sector of a disc that worked on the pentium with the same controller.
Checkit has shown that the RTC and dos times are not running at the same speeds and Omnidisk a tool that can read floppy without DMA said the system was running too slow/ couldn't keep up.

This is all I have found so far and plan to dig a bit deeper over the weekend.

I can see activity on dack2 when I try to load a program and then the system to locks up, I am thinking the data it's reading is corrupted

am I correct in thinking the dack2 is requesting data from the isa slot and sending it to the chipset and this is passing to ram while the cpu does other things ?
the data lines are the same physical lines for all data so the fact none dma routines are working says they are ok. Ram is tested and is ok so it could be a problem in the data buffer controller ? again I am just guessing at all of this.

My guess is there´s still some damaged traces / vias left that haven´t been found and fixed.
The battery leakage has been creeping under the first 2 ISA sockets (greenish colorations), the 3 DIL circuits between AT-connector and oscillator seem to be affected also and there´s probably more to be found.

I would do a second revision here, remove the ISA slots etc. and check everything carefully for corrosion / discontinuities. It´s a very time consuming procedure, but I often found it was the only way to go.

leejsmith wrote on 2022-08-05, 16:40:

I tried my aztec sound galaxy and FM works, but no digital sounds.

dma is not running/broken in some subtle way

leejsmith wrote on 2022-08-05, 16:40:

Checkit will see the floppy drive DMA when installed

it probably checks for DMA controller presence, but doesnt perform a test memory-to-memory transfer
https://www.os2museum.com/wp/the-danger-of-datasheets/ http://www.os2museum.com/wp/more-fun-with-isa-dma/

leejsmith wrote on 2022-08-05, 16:40:

Checkit has shown that the RTC and dos times are not running at the same speeds

is dos time significantly slower?

leejsmith wrote on 2022-08-05, 16:40:

Omnidisk a tool that can read floppy without DMA said the system was running too slow/ couldn't keep up.

a faulty DMA could be constantly triggering and pausing CPU (like slomo), would explain dos time running slower

leejsmith wrote on 2022-08-05, 17:13:

am I correct in thinking the dack2 is requesting data from the isa slot and sending it to the chipset

no, dack acknowledges the request is being executed and data is already on the bus
I highly recommend The_Undocumented_PC_Second_Edition_Frankvan_Gilluwe.pdf (https://archive.org/details/The_Undocumented_ … rankvan_Gilluwe 600MB pdf is OCRed and text searchable)
page 1036 "A Typical I/O-to-Memory Transfer" describes dma mechanism quite well.

leejsmith wrote on 2022-08-05, 17:08:

and this is passing to ram while the cpu does other things ?

cpu is paused (Halted)

leejsmith wrote on 2022-08-05, 17:08:

I can see activity on dack2 when I try to load a program and then the system to locks up, I am thinking the data it's reading is corrupted

hard to guess. I would search for headland chipset documentation, find which chip implements dma and trace all of the responsible pins ... or the other way around, look at ISA pinout and trace all DMA tracks until you find corrosion/break/short

leejsmith wrote on 2022-08-05, 16:40:

Checkit has shown that the RTC and dos times are not running at the same speeds

is dos time significantly slower?

it is slower

I started to check connections from the ISA to the gc101sx chip and drq2,dack2 and tc all had good signal path. I plan to check more and setup a logic analyzer and compare to my other pc with the same isa controller.
Thanks for the link I will take a look at that today

I setup my pentium and 286 with the XT-IDE and IDE-IDE controller. I just have a text file with 1 in it and can read the file ok on both systems.
So this says to me all signals are present and it must be a timing problem ?
The 286 has no activity on memw and lots on ALE compared to the pentium. memr also has a lot more activity on the 286.

I can read small text files no trouble so a few bytes, but then one that was 1210 bytes locks up the system.

reading the smaller file with 1234 is ok, then reading the 1.2kb file locks up the system

This is the point I think it fails, it's a lot shorter than the others and has activity in other places I do not see when it reads the smaller file, so this would be sending bad data ?

Now with added 16bit memr memw and sbhe
I can see the memw at the end of the segment when it works on this one.

something is wrong with those 286 captures, how does it read anything if on 286.jpg memw is never active?

http://wearcam.org/ece385/lecture6/isa.htm

both memw and ior should be active at the same time, otherwise no transfer is taking place

>lots on ALE compared to the pentium. memr also has a lot more activity on the 286.

because on pentium ram is on a separate 64bit bus, in 286 for all intends and purposes ISA is the system bus and ram accesses go thru it
at least we dont see any other spurious DMA cycles so my theory of crazy dma running and slowing cpu down is busted

-if you have a scope look at memw signal, maybe its marginal/weak and glitchy to the point your LA doesnt capture it
-try writing to a floppy, that would engage IOW and MEMR

rasz_pl wrote on 2022-08-06, 14:50:

both memw and ior should be active at the same time, otherwise no transfer is taking place

that's not how it works, only one strobe can be active at a time

rasz_pl wrote on 2022-08-06, 14:50:

something is wrong with those 286 captures, how does it read anything if on 286.jpg memw is never active? […]
Show full quote

something is wrong with those 286 captures, how does it read anything if on 286.jpg memw is never active?

http://wearcam.org/ece385/lecture6/isa.htm

both memw and ior should be active at the same time, otherwise no transfer is taking place

>lots on ALE compared to the pentium. memr also has a lot more activity on the 286.

because on pentium ram is on a separate 64bit bus, in 286 for all intends and purposes ISA is the system bus and ram accesses go thru it
at least we dont see any other spurious DMA cycles so my theory of crazy dma running and slowing cpu down is busted

-if you have a scope look at memw signal, maybe its marginal/weak and glitchy to the point your LA doesnt capture it
-try writing to a floppy, that would engage IOW and MEMR

I had to drop the LA down to 1ms to see the memr data on the 286, but it was ok at 16ms on the pentium.
i will test them both again at 1ms

It just doe not have much activity on the memw until the end and it was the same when I looked with my scope.
should I be seeing a signal on memr when it has one on iow like the pentium.

someone has said the io ch ck at the top could be a problem, when it's active low it indicates a none-correctable system error.