PDA

View Full Version : Trying to repair an IBM 5160 System board, could do with some input



A Fox
October 30th, 2015, 04:33 PM
Hi there. I have recently come across an IBM 5160 (and 5170 but thatís another story, it almost works) and Iíve decided to try and get it operational, now the first thing I must say is Iím new here and I donít really know what Iím doing here so bear with me if Iím doing something stupid, Iím familiar with modern computers but everything I know about this particular machine and everything Iíve tried so far Iíve really only come across in the past few days while trying to repair it, or through guess work.

Now the main problem Iím having is with the system board, itís not working and I have reached a point in trying to fix it where I think I need some input.

Now Iím not completely clueless, the machine hasnít been powered on since the late 80s so I didnít just try turning it on, the first thing I did before I connected any power to it was completely disassemble the machine, clean off the parts and inspect for any obvious damage and apart from the RAM expander card that had a RTC and a battery (which had leaked and corroded one end of the card but thankfully not the mainboard) everything was in reasonable condition apart from a lot of dust, I then plugged in just the PSU on its own (with a couple of old failed IDE HDDís so there would be a load on it)and tested the voltages with a scope, all of which were correct and clean then left it running for an hour to make sure it was still stable, which it was. So I know the PSU to be fully working.

I then tried powering up the system board. the first time I tried to turn the board on nothing happened, this was with just the board itself, the power supply and the speaker connected.

I have read about the possible issue of failed tantalum capacitors on these boards shorting them out, so I have measured the PSU with a scope while it was connected to the board and all voltages were still present and correct so the issue wasnít with the caps shorting (though for all I know they could have failed open circuit).

The next thing I tried was probing the CPU to check it was receiving power and its clock, which it was, 5V and exactly 4.77MHz. Now after some searching I found online (on this forum in fact) a diagnostic ROM for the machine so the next day I Ďborrowedí some old 27c256 from the back of the storage cupboard at the office and fished out a device programmer and burned the diagnostic ROM (and a full set of all three BIOS versions for the machine just in case the issue was the BIOS chip) this turned out to be a good move and with the diagnostic ROM the machine now powered up, the GPU it came with seems to have a broken character ROM but as luck would have it I actually have had a monochrome 8bit ISA adaptor lying around that did work in it (I must admit Iíd always thought it was a serial/parallel card till I saw the one in the XT). With the diagnostic ROM I found most parts were passing but I was getting memory errors so after looking around I found, again here, schematics and code for a RAM tester, which I built up and found 3 of the RAM chips to be defective. I replaced these chips with some spares off the RAM expansion card and I now have a full set of known good RAM.

Now hereís the problem, now Iím getting no memory errors but a parity error at every address (or at least the start address of each bank) I know all the RAM is good, including the parity RAM so the question is, where should I be looking next? Does anyone have any suggestions as to where a parity issue might be? I have both a scope and a 34 channel logic analyser and Iíve ordered a chip clip which has yet to arrive so I can put the analyser on each chip in turn and look at its operation but any opinions on likely candidates? If I have to check them all it would take me a very long time. I did think the problem may have something to do with U20 which is a 74S280 parity chip but, with the scope at least where I can only monitor one pin at a time it seems to be operating, at least there is a signal coming out of both outputs though without knowing whatís on all the inputs I canít say if it outputting correctly. Is there any common part that fails that may cause this issue? The board is a 64-256KB system board which had the 1982 BIOS in it when I got it, I have attached both a picture of the board and of the diagnostic screen showing the error if either of these help.

Also, when I do find the chip at issue does anyone have any suggestions on the best ways to remove the old chip without causing damage to the board? The solder on the bottom of the board is still shiny and flows well however the solder on top has turned dull and doesnít want to reflow.

Any help in fixing this would be much appreciated!

27581

27582

TJ_Mossman
October 30th, 2015, 05:34 PM
I vagely remember reading about this issue a few years ago, I can't remember the problem IC in question but the one you mentioned sounds about right.

Personally I'd desolder the chip and test it with one of those cheap chinese IC testers (www.ebay.co.uk/itm/IC-tester-LED-Test-optocoupler-LM339-KT152-DIP-CHIP-Detector-TESTER-/171923459410). It doesn't specifically mention 74S series ICs but I've had it work with them.

modem7
October 30th, 2015, 11:07 PM
Now after some searching I found online (on this forum in fact) a diagnostic ROM for the machine ...
Be aware that the Supersoft/Landmark diagnostics can sometimes be misleading. We have had people here unknowingly 'go down the wrong track'.


Now hereís the problem, now Iím getting no memory errors but a parity error at every address (or at least the start address of each bank) ...
The "PARITY ERROR AT ADDRESS 05040" in the photo equates to address 20 KB. Your motherboard has banks of 64 KB, so perhaps the Supersoft/Landmark diagnostics is checking at the start of each 4 KB block.


I did think the problem may have something to do with U20 which is a 74S280 parity chip but, with the scope at least where I can only monitor one pin at a time it seems to be operating, at least there is a signal coming out of both outputs though without knowing whatís on all the inputs I canít say if it outputting correctly.
The parity circuitry is shown [here (http://www.minuszerodegrees.net/temp/2/j86fg936fhw897.png)]. If the PCK signal is going high, then you know that the Supersoft/Landmark diagnostics is truthful.


Is there any common part that fails that may cause this issue?
Not that I am aware of.

A Fox
October 31st, 2015, 09:47 AM
Thanks for the help so far, I've ordered that tester, though its going to take a while to arrive. in the meantime I have looked at PCK and it is going high at about the same time as each parity error report so the diagnostic seems to be correct in respect to the issue. I have also looked at the other chips in the parity circuit and seem to have found something wrong. U87, specificly the gate ANDing MDPOUT and +XMEMR seems to be incorrect, the output on pin 6 is following the input on pin 5 regardless to the level of pin 4, I have captured this behaviour on a scope, at least the state of pins 4 and 6, the state of pin 5 is the same as pin 6 but I cannot show all three on the scope at the same time, I have attached an image showing this the red trace being MDPOUT on pin 4 and the yellow trace being the output on pin 6

27600

obviously when the input pin 4 is low I would expect the output to be low but as you can see it is not. I have tried replacing U87 however the issue persists even on the new chip (which is now socketed) the trace image was in fact captured from the new IC. the issue seems to be the low level, which is at 1.6V (the high is at 4.8V) the low level for a 74S series is 0.8V so it seems to be always seeing the input as a 1 but I can't see why the signal would be so high, as far as I can tell from the schematics MDPOUT is only connected to the outputs of the parity memory, and these chips are known to be working, they even work on this board if swapped with other bits. is the MDPOUT line connected somewhere else I haven't noticed? it doesn't seem to be at lest according to the schematics, any ideas?

retrogear
October 31st, 2015, 02:36 PM
Yes, 1.6V is definitely not an active low. I've been pouring over the schematic and it sure looks like you've covered all the circuit points for MDPOUT. Might have to look on the traces for any "crud" ?
You have a resistive leakage to the 5v line somewhere.

Larry G

modem7
November 1st, 2015, 01:41 AM
Also, when I do find the chip at issue does anyone have any suggestions on the best ways to remove the old chip without causing damage to the board?
I have drafted something and put it at [here (http://www.minuszerodegrees.net/soldering/ic_removal/ic_removal.htm)].

Chromedome45
November 1st, 2015, 08:30 AM
Definitely the best way of removing a suspect chip Modem7. If you have a spare available. And of course socket the new chip. I have always done it this way.

retrogear
November 1st, 2015, 08:47 AM
One other quick thought. Put your scope on the offending waveform in DC mode and increase the vertical amplitude (on the waveform) to watch that 1.6v low section and see if it "creeps" with temperature. You might find the fault with heating / cooling
components or circuit areas (or even flexing the board) ...


Larry G

A Fox
November 3rd, 2015, 03:21 PM
Ok, ive now tested the track on its own with all chips on it removed and its down at zero and definitely floating, the outputs of the RAM chips is definitely going down to zero when their not connected to the line, and I've also tried bending the pins in question out of their sockets and connecting them directly together, and the problem persists, so its not the track and the same is happening with any set of 4164s, the ones that came on the motherboard, the ones from the ram expander and even the ones from my graphics adaptor so either ALL the 4164s I own have the same issue with their outputs or its the 7408. now since I have already replaced it and have the same issue abet with 1.2V instead of 1.6V all I can think of is that the 74S08 was in fact broken and the circuit just cant work with the 74LS08 I had on hand to replace it, I cant quite see how, but the LS series DOES take more current to pull its inputs down than the S series, only 0.6mA more but still more and since the S before had a higher voltage than the LS it must have failed in a way to take more current? I'm really not sure at this point, I have ordered a 74S08 to see if it will make any difference but other than that? well, ive replaced every chip connected to the input and it obviously worked originally. Ill update on what happens when I receive the new IC.

A Fox
November 23rd, 2015, 11:22 AM
Ok! so I've (finally) received the ICs I ordered and have had some success! the 74S08 turned out not to be the problem, the strange level is actually the pin tristating, or rather floating when there is no input to it, the lows could be seen but they were very short and very rare so I hadn't managed to catch any.

the chip that was the issue was the 74S280 (U20), (replacements for which came from Romania, hence the delay, there was one person here in England selling them, but for £9 each whereas I got 50 shipped from Romania for £7, even if I do now have a box of them that I don't really need more of) with the replacement in place the parity issue is gone and without the floppy controller connected all, non-floppy related tests pass. so firstly I must thank everyone here who's helped me get this far!

my new issue is with the floppy drive. Now I know the drive, disk and controller work as putting them in any other machine and it boots into IBM DOS and all three were the original parts supplied with the computer (even the disk that magic still reads) but when used in the 5160 mainboard, the floppy controller test passes but the floppy read test just goes on forever, the drive starts turning and the access light turns on and it just stays doing that forever, without the diagnostic ROM the computer will post, run its memory test then gets to a flashing underscore and starts reading the floppy but this too then runs on forever never booting into DOS, without the FDD controller the machine can be booted into basic, which seems to work, any ideas?

retrogear
November 23rd, 2015, 04:41 PM
How's your +12VDC supply? It supplies the mechanics of the drive including the motor.

A Fox
November 23rd, 2015, 04:52 PM
How's your +12VDC supply? It supplies the mechanics of the drive including the motor.

12.16V, I've also tried other supplys as well as the original XT one with the same effect, plus I have tried other FDD cards (both the original IBM one and a 16bit FDD/HDD/Serial/Parallel card I know to work in 8bit mode) with the same effect (FDD controller passes but floppy read test hangs forever), the drive motor spins and the light is on but no attempt to read seems to be made, at least the head servo dosen't move.

retrogear
November 23rd, 2015, 06:20 PM
Any other ISA cards in use besides video? Not necessarily on XT, but I've had a video card conflict with a floppy card on ISA bus before. No clue why. Worked fine with a different video card.
Do you have another video card to try? Don't have to have a compatible display for it, just see if floppy seems to read better.

A Fox
November 24th, 2015, 01:24 AM
Any other ISA cards in use besides video? Not necessarily on XT, but I've had a video card conflict with a floppy card on ISA bus before. No clue why. Worked fine with a different video card.
Do you have another video card to try? Don't have to have a compatible display for it, just see if floppy seems to read better.

Ok, I tried with a different video card, the one that actually came in the system, and while I can't actually see anything with it (dead character ROM) the same thing happened with the floppy, I also tried with no video card at all, and while obviously that wont work with the actual BIOS with the diagnostic ROM it will still run to the floppy test and the same thing happens, motor running but no drive access. I don't have any other cards in at the moment, the board isn't even in the case yet so just PSU system board, Video and floppy controller. I'm pretty sure at this point that the problem must be with the main board, some resource used on the ISA bus to access the floppy disk that isn't used to access the video card must be dead, unless the wrong floppy type is selected? but I don't see any way of setting that on this board, I know you need to on all later boards but I can't see how you do that on the XT the only switches are for the number of drives, which is currently set to one the drive is a Tandom TM100-2A if that makes any difference? its just a 5 1/4" 360KB one, I'm guessing that's the only type that works anyway?

retrogear
November 24th, 2015, 04:09 AM
Reading thru the IBM tech ref so I'll just keep speculating.
As far as ISA bus, it looks like signals unique to a floppy controller vs like a video adaptor are:
-DACK2, +DRQ2 these operate on DMA channel 2
+T/C (or TIC) but this applies to any DMA channel
+IRQ6 definitely this one unique to diskette adaptor.

- means active low, + means active high
I don't think a video adaptor would request DMA since it's only an output device?
Look up the ISA bus pinout for corresponding pins. Since you have a scope, look for activity on these lines.
You could even compare signals to another functioning ISA floppy system.

Larry G

A Fox
November 24th, 2015, 10:51 AM
I was looking at that too, I also looked at the document for the video adapter, and while mine is a clone not a real IBM one it appears to be the same design and it doesn't use DMA at all. now as luck would have it, I also have another 5160 mainboard I acquired two weeks ago for a tenner on ebay and while it also doesn't work, its only error is the keyboard controller (everything else passes the diagnostics and it boots but always showing a keyboard error, I had even assumed from the description that the seller just didn't know there was a difference between XT and AT keyboards but obviously not the case, I believe the issue with that one is U27 but have yet to receive a replacement IC) and on the other mainboard the floppy read test passes so I have looked at the differences between the signals on those pins on each board.

IRQ6 is identical on both, as is the INT pin on U25. -DACK2 and +DRQ2 both produce a series of pulses 16.4ms long during the floppy read test on both boards, as far as can be seen they are also identical. +T/C is the different one, on the (sort of) working board it stays low constantly during the diagnostic tests, never triggers the scope. however on the board with the floppy issue there is a constant short pulse roughly twice a second for the entire time the machine is running however this is where my knowledge runs out, any ideas what this could mean?

retrogear
November 24th, 2015, 02:11 PM
Yea my knowledge is thin here too. TC signals end of DMA transfer so maybe more requests are getting repeated on your bad floppy so you see it on the scope?
Your DMA controller is an 8237. You could check for presence of RESET on pin 13, CLK on pin 12, RDY on pin 6.
Actually, since TC shows a transfer was successful maybe the DMA is working? It keeps getting repeated because some criteria is not met?
I guess you would have to compare signals on the DMA chips on good/bad systems to see where to go next.

PS - why repeating twice a second? That's an eternity in computer time. That might be the boot cycle to re-read track 0 ?
PSS - try opening the drive door while it's reading and see if pulse stops.

A Fox
November 24th, 2015, 03:36 PM
Ok, all three signals are present on both, however the clock is weaker on the non working board, at 3.6V peek to peek instead of 4.2V, but that should still be in tolerance for logic inputs.

I can't measure the timing between pulses exactly because they are too short and too far apart and my scope doesn't have the memory depth to display two at once but it is roughly twice per second, opening the door changes nothing either but the pulses start way before the FDD is accessed, I'm running only the diagnostic ROM at the moment and it doesn't check the FDD controller at all until its about to start the read test and the T/C pulses start right after the DMA chip is accessed for the first time during test on the DMA chip which is completed right at the start of the diagnostic, this cannot be coincidence as its that chip generating the pulses, however the test on the chip that starts the pulses reports it as passed. plus the pulses happen even if the FDD controller is removed. I really wish the chip was socketed, while I'm having to specially order each 74 series chip I try to replace, for some reason I really don't understand I actually have a draw of brand new 8237s (along with most of the 82xx support ICs and 8088s, no idea where they came from) I just really don't want to try and remove a 40 pin chip unless I'm sure its the one at fault as I'll more than likely ruin the board. does it at all seem likely to be that chip? or is the clock actually too week at 3.6V?

retrogear
November 24th, 2015, 06:13 PM
Hmmm - if the diagnostic passes the dma chip then it should be ok. Not sure about the difference in CLK amplitude. 3.6V sounds suspicious. The CLK is provided from U1, an 8284A clock generator
pin 8 so could compare the waveforms there between the two. There is some TTL gating before it arrives at the DMA chip. Like you said, the CLK amplitude could be functionally ok but tracing back
the difference in amplitude could reveal a problem.

A Fox
November 25th, 2015, 12:02 PM
the clock gens strong on both boards, and its literally the last gate before the DMA causing the drop but i really don't think its the issue since the DMA still passes its test, also those tc pulsed are on both boards, I must just have managed to look at the wrong pin or not press hard enough to get through the dirt on the other before, I'm guessing its the RAM refresh finishing, that uses the DMA, kinda proving its working actually as the ram is working so obviously the refresh is. so I'm back to being totally lost, I'm really not sure what to do now, I mean after all I've managed to fix I don't want to just give up but I have no clue where to start now, since both DMA and ints seem to be working and those are the only things the MDA adaptor doesn't use as well.

retrogear
November 25th, 2015, 12:09 PM
One longshot thought. You say you are bread-boarding this w/o a chassis? What about proper grounding? Ground the metal chassis of the floppy, metal cage of the power supply to the metal ground of the motherboard like the keyboard jack area. I forget if the XT has any motherboard ground screws or just the nylon pins. Your display is not near the floppy ribbon is it?

Stone
November 25th, 2015, 12:34 PM
What about the four ground wires on P8 and P9? Don't they count for anything?

A Fox
November 25th, 2015, 12:41 PM
gave it a go, no luck unfortunately same issue, and the floppy cable isn't anywhere near the display no, but surely if that was the only issue the other board wouldn't work either? I guess things could be on the edge. the only possible thing I can think right now is something to do with U10, its on the floppys DMA line and its right next to the other chip that was dead and has the same dark tarnishing to its pins, though Im not even sure what its doing apart from that it must have something to do with dma, maybe I should try and boot to something else other than a floppy? something that uses a different DMA line? I do have an XTIDE card but I need to bootstrap it with a floppy first to install the firmware, which I cant do on either board, I will try to set it up in the 5170 I also got with the 5160 that has at least got a working mainboard and see If I can get it to boot (I got it broken for 99p on ebay, so even with the new flash chip it still may not work)

A Fox
November 25th, 2015, 12:43 PM
What about the four ground wires on P8 and P9? Don't they count for anything?


for grounding the power yes, but I get where he was coming from, a good ground plane across all devices really helps remove noise from any electronic system.

retrogear
November 25th, 2015, 12:55 PM
Humor me one more time. You say it was the last gate where the drop in CLK amplitude was? What is the Vcc on pin 14? That's U74? It should be >4.5 and hopefully 5V. You could scope the signal lines on U10 for good 5V signals. Hmmm, U10 is fed by U74 ...

SpidersWeb
November 25th, 2015, 12:57 PM
Sorry if this is a stupid question - but have the floppy drive and cable in question been tested in another system? Or another type of drive attempted? (even a 3.5" 1.44 drive would do for this test - and you can boot these). I'm just not sure I noticed this being covered off earlier in the thread.

A Fox
November 25th, 2015, 01:09 PM
Sorry if this is a stupid question - but have the floppy drive and cable in question been tested in another system? Or another type of drive attempted? (even a 3.5" 1.44 drive would do for this test - and you can boot these). I'm just not sure I noticed this being covered off earlier in the thread.

Yes, I tried the drive and controller on another 5160 board (keyboard doesn't work on it but it still reads the FDD) and it works fine on that board I can even fully boot and use DOS from that drive and controller on a 486 board I have(when I disable the onboard controller on it of course) , I have also tried a different controller card with the same result.


Humor me one more time. You say it was the last gate where the drop in CLK amplitude was? What is the Vcc on pin 14? That's U74? It should be >4.5 and hopefully 5V. You could scope the signal lines on U10 for good 5V signals. Hmmm, U10 is fed by U74 ...


5.4V on VCC and I have looked at the signals on U10 but the majority of the pins are on the address and data bus so there is a lot of activity anyway but no real way to see from where, there is a v4.48V signal on the pin from U74 though

retrogear
November 25th, 2015, 01:29 PM
Nothing real definitive. Could replace U10 and/or U74 but ??? However, I've learned from my real job in IT hardware support that passing diagnostics is not a sure thing.
The proximity of a known bad chip is a good possibility.

Larry G

A Fox
November 25th, 2015, 04:26 PM
I think I'm gonna try U10, will have to order one though so will take a bit of time. the xtide was a bust, it will boot with it, it comes up with the xtide BIOS, booting to the floppy from that produces the same result of the system hanging on read, and booting from the xtide... well, I tried 15 different IDE HDDs ranging from 40MB to 160GB and it didn't recognise even ONE of them! the ONLY IDE device I could get it to recognise was a 128MB DOM (which happened to have XP embedded on it so it wasn't going to boot as is) only problem is, I can't format it, while it was the ONLY IDE device recognised by the xtide, the xtide was the only thing to recognise the disk! all other IDE interfaces on any machine ranging from a 486 to a core 2 duo just see nothing when its connected! I can see the device on my 5170 in the xtide and fdisk sees one dos partition of 3MB and one non-dos partition of 120MB but since fdisk can't remove a non-dos partition I can't do anything with it. if I DO try booting from it on the 5160 I get the message "missing operating system" which is obviously true but I'm not sure if this says anything, as its either from the xtide BIOS, the 5160 BIOS or the boot sector of the disk, if the message IS from the boot sector, maybe that says something about what's working? if it can at least read a boot sector from a fixed disk?

Malc
November 26th, 2015, 12:19 AM
....it comes up with the xtide BIOS, booting to the floppy from that produces the same result of the system hanging on read, and booting from the xtide... well, I tried 15 different IDE HDDs ranging from 40MB to 160GB and it didn't recognise even ONE of them! the ONLY IDE device I could get it to recognise was a 128MB DOM.....

Just curious What "XTIDE" card do you have ?, If it's the Lo-Tech ISA CF Card (https://www.lo-tech.co.uk/product/lo-tech-isa-compactflash-pcb/) it'll work with CF / Microdrive or DOM and probably ATA-2 compliant hard drives.

A Fox
November 26th, 2015, 05:42 AM
it is one of those yes, its just I'd assumed out of my random pile of IDE disks at least ONE would have supported ATA-2 is all. and I believe I have worked out now why the DOM only works on the XTIDE, it takes power from 5V on the IDE key pin, which no normal IDE controller provides, guessing from what I have found online here but seems likely, when I get home ill try providing that and formatting/installing dos on another machine and see if I can get the 5160 to boot from it.

A Fox
November 26th, 2015, 01:28 PM
Ok, getting somewhere, I finally managed to install an OS on the XTIDE (will install but can't be booted from with the original PC DOS 3.00 I have, had to mess around a lot to get a 5 1/4 " 360K msdos 6.22 boot disk to do it) and when I put this into the XT I can boot into DOS! and can even run programs without a problem from the fixed disk, now when I try to switch to the FDD (i.e. typing "A:") I get the same issue, the machine just locks up, except for the very first time when I tried this when I just got "divide overflow" so the machine works and can run programs from the fixed disk but just can't read the floppy even though the controller and drive are definitely working, this has to mean something? I have already ordered a replacement for U10 and as I said I already happened to have replacements for the DMA controller on hand but I really feel like i'm stabbing in the dark here, the issue may not have anything to do with DMA, does any of this give someone here any idea? is there something I can do from within a working dos install to try and shed some more light on this?

A Fox
November 26th, 2015, 02:14 PM
Ok, one more thing, I decided to try some other cards in the machine and set up a sound blaster 2.0 on the AT and got that working on the xtide, when I put it in the XT and try to run a program that uses it I get exactly the same issue as I get with the floppy controller, the machine just locks up as soon as it is accessed, so the issue is something used both by a standard IBM floppy controller and a sound blaster 2.0 (specificly I used the midi player PLAY.EXE that comes with the card), so it definitely sounds like its DMA? but that's used by the DRAM refresh too... could it still be an issue? could it be the interrupt controller?

A Fox
November 26th, 2015, 03:57 PM
ok right, definitely DMA there is a diagnose.exe for the soundblaster that tests all the parts in turn, address 220 , successfully found, irq 7 successfully found, dma channel 1, error dma channel not found, so IRQs are working (and the address but we already knew that for the other cards to work) the issue must be DMA related yes?

retrogear
November 26th, 2015, 04:16 PM
Wow that's some progress !! I was thinking the hard drive would also use dma but from what I read it depends on the controller. Does XTIDE manual indicate dma is used? I assume not.
Odd that your diagnostic rom indicated dma ok, but dma has several modes so maybe rom can't test all modes. Can you get diagnostic software on your hard disk via putting in the other pc?
Maybe others could suggest the best diagnostic to use for dma testing from DOS. Maybe hammering dma with diagnostics while scoping the pins will reveal a problem.

PS - actually I could compare waveforms on my 5150 which also uses the 8237. I have more fun doing this to a pc than actually using it !!!

Larry G

per
November 26th, 2015, 05:14 PM
XT-IDE does NOT use DMA due to how the 16-to-8 bit bus conversion is done. It's solely based on IO polling, and the port you poll alternates between two ports for every byte read. This way you can read both the hi and low byte of 16 bits over the 8 bit bus.

A Fox
November 27th, 2015, 12:03 PM
Wow that's some progress !! I was thinking the hard drive would also use dma but from what I read it depends on the controller. Does XTIDE manual indicate dma is used? I assume not.
Odd that your diagnostic rom indicated dma ok, but dma has several modes so maybe rom can't test all modes. Can you get diagnostic software on your hard disk via putting in the other pc?
Maybe others could suggest the best diagnostic to use for dma testing from DOS. Maybe hammering dma with diagnostics while scoping the pins will reveal a problem.

PS - actually I could compare waveforms on my 5150 which also uses the 8237. I have more fun doing this to a pc than actually using it !!!

Larry G

yes, I can transfer files to it, by copying files to a flash drive, putting that in my XP core 2 duo machine with a 5.25" FDD in it, copying the files to a 360K disk (both that drive and the AT's drive are 1.2MB but I only have 360K disks), putting that disk in the AT, booting the XTIDE in my AT, copying the files over to the XTIDE, then disconnecting everything from the AT, connecting it all back to the XT and booting the XT, so easy! but yeah I can copy stuff over with no issues if someone can point me to some diag software. and as for that last part, I understand! when I actually get it working I probably wont spend nearly as much time on it! (I will spend some of course!)

on another note I got the other XT motherboard I got on ebay working now! I was right about it, it just had U27 dead (interestingly BIOS version 1 still give a RAM error code, but runs anyway, V2 gives a floppy error code, but boots from it anyway but V3 works fine so I'm using that on it) I connected that up to the XTIDE and the floppy works on that board I ran the soundblaster diagnostic.exe and on that machine it reports the DMA as found successfully! so it really IS DMA on the other board that's wrong as this is an identical board (almost, this one has a chip in the "spare" slot and the other doesn't) interestingly enough though, it still doesn't actually work in that machine, the same thing as happens on the other board happens on this one if I try to play anything with the card, it just hangs, is the SB 2.0 just not compatible with the 5160? it works in a 5170? could it just be that it needs more RAM? I only have 256KB in the XT

A Fox
November 30th, 2015, 04:43 PM
Well, huge success! I have finally got the mainboard working! the final issue was U11, though I only found it after having replaced U10 and U28, luckily I had managed to remove U28 without damaging it (lots of solder wick applied to the pins of the IC on top rather than the board) so I just put the original back in the socket I added to the board.



That means the final damage on that board was:-

-U20 (parity calculator in RAM circuit) failed so board always gave parity error even with known good memory.
-U11 (octal latch in DMA circuit) failed so board passed DMA controller test but hung on any (non-RAM refresh) use of the DMA, for example reading from a FDD.
-(multiple) 6 4164 DRAM ICs failed and gave memory errors.

for anyone's reference the board was a 64-256KB SYSTEM BOARD and came with the V1 BIOS installed.



well, I must thank everyone here who gave me a hand on this, I probably wouldn't have got this far without you! and while I haven't finished restoring this machine by a long shot, the mainboard is fixed and the other problems are for another thread!

oh, and the SB 2.0 still doesn't work in this board either, even though it finds its DMA now, can't work that one out...

SpidersWeb
November 30th, 2015, 05:01 PM
Awesome work, I love threads like these.

RE: SB2.0 - make sure it's set to IRQ 7 (not IRQ 5), and make sure to set your blaster variable (SET BLASTER=A220 I7 D1 etc).

retrogear
November 30th, 2015, 05:28 PM
Great !! Quite a compound set of failures. It would be interesting if your 6 DRAM chips were actually good since there were other failures involving shared address and data lines but those chips are cheap so best to leave well enough alone.

Larry G

A Fox
December 1st, 2015, 04:55 AM
Great !! Quite a compound set of failures. It would be interesting if your 6 DRAM chips were actually good since there were other failures involving shared address and data lines but those chips are cheap so best to leave well enough alone.

Larry G

I did test all the RAM chips in a DRAM tester and those were the ones that failed that test so I'm guessing they are all bad


Awesome work, I love threads like these.

RE: SB2.0 - make sure it's set to IRQ 7 (not IRQ 5), and make sure to set your blaster variable (SET BLASTER=A220 I7 D1 etc).

I have done both of those things, I think it just needs more that the 256K of RAM I have, it works with the same setup an AT with 640K