IBM XT 5155, XT-IDEv4, Parity Check 2 on pressing enter key - advice needed!

[mrmanse]

Hello and thank you all for contributing to this wonderful forum.

Situation:
I get Parity Check 2 error message almost every time i press the enter key, but only when system is equipped with a XT-IDEv4 and HDD.

Background:
I recently purchased an IBM 5155, equipped with a NIC and a 384kB RAM expansion board, otherwise near original condition. DOA, but simply replacing three bad tantalum capacitors fixed that (C56, C58 on mobo and a +5V line on the RAM card). I also ran the Supersoft/Landmark Diagnostic ROM and found a faulty 41256 DRAM on the expansion board, and replaced it. The test now completes without errors. The bios is the 11/08/82 version.
I want to use a 2,5" HDD in the case, controlled by a XT-IDEv4 (Glitch works type), but I cannot get it to work.

Error details:
* When equipped with a HDD, that is detected by the XT-IDEv4, the computer boots to DOS from HDD, but then halts with a Parity Check 2 message upon pressing the enter key. Pressing other keys, like typing a DOS command, works normally. A few times the Parity Check 2 also occurred spontaneously, but only rarely.
* If booting from floppy without a HDD, either no HDD attached or XT-IDE-card wrong address set, the computer works completely normally.
* Powering the HDD externally has no effect, the Parity Error remains. Voltage on a spare drive molex connector reads a perfect 5,05 volts with Fluke multimeter with HDD connected to the 5155 power supply.
* Removing all non essential ISA cards have no effect, now only remains CGA card, 384KB RAM and the XT-IDEv4 card.
* Running the Supersoft/Landmark Diagnostic ROM with the same configuration of cards and a HDD still reveals no errors.
* The HDDs, the cables and XT-IDE cards are known good. I have several, and this set I took from fully working 5150 system.
* I have tried different addresses on the XT-IDE (300, 310, 320, 340) and ROM (C800, D000), but always the same result. Also tried different ISA slots (not port 8, of course).
* Using another known good Model F keyboard gives same results.
* A few times I did get a littly bit further by disabling "chuck mod" and flashing the XT-IDE v1 ROM image. I ran fdisk and partitioned the drive, but after a reboot the partition table was corrupt, containing Xenix type partitions with partition parameters completely off (FDISK from Dos 3.3 and 6.2 versions same result). It doesn't seem to be a general error of reading or writing though, because if I manually edit the partition table with Norton Diskedit, my changes stick as expected.
Update:
* If I remove the RAM expansion card from ISA slot AND run the XT-ide in compatible mode i don't seem to get any Parity Check 2. However, writing operations on the drive are slightly corrupt. I can't produce a working partition with FDISK, I get NON-DOS and XENIX type partitions etc.
* If i remove the RAM expansion card and run the XT-ide in chuck mod I immediately get Parity Check 1 (yes, that's a one!).

My thoughts at this point
It seems XT-IDE compatible mode does something the mobo can't handle, and chuck mod does something that my 384KB RAM card can't handle. Is this a problem with the ISA slots?? Addressing of some kind?
Upgrading BIOS? I have a few spare EPROMs, but I seems to me unlikely to be the problem.
Faulty keyboard controller? but why only problems when the HDD is detected by XT-IDE bios?
Faully DMA controller? I could replace it, but I don't know how to be sure.
Local voltage drops on the motherboard due to pre-failing tantalum capacitors? I can measure som more on the IC pins.

Thank you for reading this far - any helpful advice greatly appreciated!

/Måns

 

[mrmanse]

Update:
Regarding the adresses indicated by Parity Check 1 and 2, they are 1C000 (S) or rarely 00000 (S). That seem to indicate bank 0 and 1 of motherboard RAM, so i swapped these. I swapped Bank 2 to Bank 0, Bank 3 to Bank 1 and tried to test things with only 128KB RAM. I then got Parity Check 10000 (S) (bank 0), and an unstable system, still not writing reliably. So i put new DRAMS in bank 2 and 3, and also the RAM expansion card and tried again. Same error! Parity Check 2 1C000 (S).
There must be something wrong with RAM circuitry on the motherboard, because error remains on the same address when all DRAMs have been moved or replaced. Another indication is the unreliable writing to HDD.

Any suggestions? Anyone? I'm soon out of ideas here.

 

[modem7]

My suggestions:

1. The check of RAM (motherboard and expansion card) by the motherboard's POST and by the Supersoft/Landmark diagnostic ROM is expected to be crude. Try something like CheckIt.

2. Try the third revision of motherboard BIOS (i.e. 05/09/86).

 

[mrmanse]

My suggestions:

1. The check of RAM (motherboard and expansion card) by the motherboard's POST and by the Supersoft/Landmark diagnostic ROM is expected to be crude. Try something like CheckIt.

2. Try the third revision of motherboard BIOS (i.e. 05/09/86).

Thank you, will do later on. Late last night the computer ran without errors in this configuration:
* External power to hdd
* A different CGA-card (same type IBM original)
* NO floppys or even a floppy controller
* 384KB ram card

Replacing the CGA-card back to the original one immediately produced the Parity Error, and while doing so I registered a voltage dip to 4,8v (using multimeter with recording function, not sure how accurate it is).

So... I'm back to thinking it's a power issue after all. I find it very strage though that everything works well without the XT-ide card, because I don't see how it could cause the voltage drop. It's well built using mostly HCT family logic, one ACT and two F family.

Do you think it would be meningful to replace all tantalum capacitors on the affected CGA-card? the motherboard? Decoupling capacitors? Do they age?

/Måns

 

[Malc]

What hard drive are you using, 2.5" laptop drive ?, What make / model, Just curious, I have a R4 XTIDE in my 5155 and have no problems with CF or SD cards.

 

[mrmanse]

What hard drive are you using, 2.5" laptop drive ?, What make / model, Just curious, I have a R4 XTIDE in my 5155 and have no problems with CF or SD cards.

I have tried several, just to be sure. Two drives that absolutely work well in another computer are a Conner CFS540A (3,5" 540 MB) and the drive that I intend to use in the computer when i figure this out, a Hitachi IC25N040ATMR05-0 (2,5" 40GB).

/Måns

 

[glitch]

If your 5V supply is dropping to 4.8V I'd also suspect a power issue. If possible, load the supply down with a dummy load and measure. Automotive tail lamp bulbs work well. If it's falling out of regulation, you may just have old caps that are drying out and losing capacity.

 

[Malc]

Yes 4.8v is a bit low, Some of these 2.5" drives can be a bit power hungry on spin up, I had a look at my stash of 2.5" drives and found a Hitachi 80Gb with a rating of 5v 1.0A, I hooked it up to my 5155 with R4 XTIDE, Partitioned / Formatted and installed DOS 6.22 just fine, I measured the 5v line from the P10 - P11 connectors and got a steady 5.06v with the 5155 powered up. Looking on Modem7's website for the 5155 PSU I see that the 5v line in the P10 - P11 connectors are rated at 0.7A max. I'd pull the PSU and load it up on the bench, When i first got my 5155 i pulled the PSU and loaded it up on the bench leaving it running for hours on and off over 2 days checking the output voltages over a period of time. Hugo Holden did a write up on the 5155 PSU and it's a very good read, There is a link to it on Modem7's Website, Be aware of the P12 connector to CRT loading requirements if you do pull the PSU.

 

[mrmanse]

Thank you all for assistance.

I have rechecked the power levels and I was in error. The reading 4,8v could only be reproduced by setting the fluke multimeter to one digit precision, using auto scale it measures a very steady 5,08-5,10 volts. Checked with oscilloscope, no significant fluctuations. The other power levels are also fine.

BIOS is now upgraded to the latest, but no change of symptoms. The whole system have been thoroughly tested with CheckIt 4.1, no errors, especially not in RAM.

I have also checked the electrolyte capacitors of the power supply, and they read perfectly good capacitance.

I also tried swapping the 7404 and 7432 logic circuits of the XT-ide card to LS-type, but it too made no difference (7404 was HCT, and 7432 was ACT).

I reiterate, the system runs just fine except when there is a HDD that has been detected by the XT-ide universal bios. By simply disconnecting the IDE-cable and rebooting, it runs fine, with the XT-ide still installed, and a HDD drawing power. The serial drive works also.

I have noticed a few additional things, however:

* I have two identical CGA cards. With one of them the computer always gives parity check at first enter key pressed at dos prompt (after setting date and time, they never fail). With the other CGA card I usually can run a few minutes if I pull out serial card and floppy controller.
* After reassembly of the computer after working with the power supply the address indicated by parity check changed a litte, and while it previously gave one of two addresses, it now gives two other ones. I did touch the voltage calibration of the supply, not sure if relevant.
* I can not boot from known good floppy if HDD is detected. Again, if pulling the IDE cable, the computer boots fine from the same floppy and runs fine.

Tomorrow I'll get some test leads and try to hook it up to a small logic analyzer I bought off ebay a few years back, but I'm a beginner at that. Could it be a timing problem with i/o and mem read/write lines?

/Måns

 

[rittwage]

I had bizarre problems where my XTIDE v4 did not work in my 5150 (or 5160 I think) but it did work in newer PC's (386/486). I sent it to Glitch and he fixed it... I think he said one of the chips was out of spec or something for older PC's. Not sure if it's the same issue or not - my problem was just corrupted read/write, I never got a parity error.

 

[glitch]

I had bizarre problems where my XTIDE v4 did not work in my 5150 (or 5160 I think) but it did work in newer PC's (386/486). I sent it to Glitch and he fixed it... I think he said one of the chips was out of spec or something for older PC's. Not sure if it's the same issue or not - my problem was just corrupted read/write, I never got a parity error.

If he's bought his board any time in the last year, that shouldn't be the issue -- that was a bad batch of chips from a supplier I stopped using, over that very issue.

 

[Malc]

If mrmanse took the XTIDE from a 5150 and it was working perfectly then a bad chip shouldn't be an issue, Unless the board got zapped by static in the process of moving it, I can see a bad 573 giving partition / corruption issues I've had that before myself but not the memory issue, I do remember someone having a problem with partition / corruption issues and it turned out to be a bad 5v from the PSU, Noise in the 5v line i believe, But no memory error.

 

[mrmanse]

Thank you for suggestions! I'll test the xt-ide in a working 5160, but since I have three if the all giving this problem I do believe it's the computer.

Could anyone comment on this capture of logic levels during read operation?

Strobes on IOW_ISA should be DRAM refreshing (66kHz), the duplets on IOR_ISA is because two bytes are read, but what are the spikes interferring with CS_IDE? The appear random, duration <4ns, intervall 1us to 200ms. Is that normal?

/Måns