On again, off again

[clh333]

Set up the 5170 to test a new monitor. First had the CPU, M keyboard and 5151 Display connected to set a baseline. The AT hadn't been on in a while and the CMOS battery was low, as I was informed on power-up. As I was going through the BIOS configuration (this machine has an AMI BIOS substituted for the IBM ROMs; done so I could run an XT-IDE) the machine suddenly shut down; it had been on for less than two minutes, I would guess.

I let it sit for a few minutes, turned the switch off and then back on again. I got a flash of green at the keyboard lights and then nothing more.

Before I go inside the machine I'd like to ask for diagnostic advice: I'm hoping it's a fuse in the PSU. There was no "crack" or smoke to indicate failure of a capacitor, but then again I don't know as much about IBMs as, say, Modem7.

Thanks for your advice.

-CH-

 

[Stone]

The PSU is easy to check on its own. Just load it and see.

 

[modem7]

I let it sit for a few minutes, turned the switch off and then back on again. I got a flash of green at the keyboard lights and then nothing more.

It is normal for the keyboard LED's to flash when the keyboard gets +5V. But maybe the +5V is only present briefly (e.g. PSU starting then shutting down). Is the PSU's fan starting, and then stopping ?

Do you have a multimeter?

The IBM technical reference indicates a requirement for the (IBM-supplied) PSU to have a very significant load on the +12V line. IBM did not indicate why. Some people say that it is required for the PSU to properly regulate its output. Nonetheless, try connecting the hard drive to the PSU, the hard drive providing the required +12V loading.

 

[clh333]

I replaced the CMOS batteries with live ones and retested, with no change in the result; brief flash at the keyboard and nothing more. I disconnected and removed the PSU from the chassis and took off the cover; I cleaned off the dust inside and checked both fuses for continuity, using the Ohmeter, not the continuity setting on the multimeter. Both were good.

I attached a Seagate ST-225 and ST-251 to P11 and P12 power leads and powered up the PSU, which started without hesitation. I checked voltages of all pins of the remaining connectors, which read as follows (all values VDC relative to the black, ground lead):

P8: Orange 5.05
Red 5.20
Yellow 11.81
Violet -11.77

P9: White -5.05
Red 5.20
Red 5.20
Red 5.20

P10: Yellow 11.81
Red 5.20

The power supply appears to be functional.

-CH-

 

[Stone]

The power supply appears to be functional.

That's what I thought.

Now you should check PSU connector on the 5170's board for shorts. Check each one against ground (no need to check the four grounds, obviously). If any turns up short you'll know which rail has the short on it.

 

[pinkdonut666]

My guess is a shorted tantalum capacitor on the planar board for one of the main voltage lines. Check out
http://https://www.youtube.com/channel/UCAfYntJkyJdnYbzDsOGu0qQ

He's got lots of in depth videos on repairing IBM PC and compatibles, can be kind of dry but EXTREMELY informative.

 

[Chuck(G)]

What are the PSU voltages when connected to the motherboard?

 

[Stone]

What are the PSU voltages when connected to the motherboard?

How would zero sound to you?

 

[Chuck(G)]

That's what I'd expect, but it's always good to verify. Could be that a single power rail has gone gilgongo. It'd be nice to rule that one out.

 

[clh333]

In the absence of this abundance of information (for which I thank one and all) I reinstalled the PSU in the chassis and removed all but the video card, the memory card and the HD/FD card, which is a SIIG multifunction. I powered only the board, the HD (a Quantum Bigfoot) and a 1.2 FD. Attached a keyboard and monitor of the appropriate era and powered up once again.

I was again reminded that CMOS settings were lost and I went into the AMI BIOS config utility. Set the time and date and fudged some numbers to represent a 32 MB HD. This time I was able to save to CMOS and reboot. When I did I got the message that there was an HD controller failure. Since the POST routine had signaled the 1.2 A: drive I tried inserting a floppy with the MS-DOS 6.22 OS on it into the drive. I wanted to see if I could bypass the HD and get to a prompt. I could not; I got the same HD controller failure message.

So before I chase my tail any more I'll go back and test the board's connectors to which P8 and P9 attach (I do know enough to put the black wires together in the middle but fortunately IBM has keyed them already) and also test the voltages there with the PSU on.

Thanks again for your help.

-CH-

P.S. this all started because I wanted to test a Leading Edge amber monitor I had just received. Fortunately the XT still worked.

 

[clh333]

With the power supply disconnected from the board I measured the following resistances to ground at each pin.

Starting at the board connector pin closest to the rear of the chassis ("pin 1", representing the range wire of connector P-eight) and going through to "pin 12", representing a red wire on connector P9:

pin 1: infinite (at 1 Ohm range on the multimeter)
pin 2: 60 Ohms
pin 3: infinite
pin 4: infinite
pin 5: 1.4 Ohms
pin 6: 1.6 Ohms
pin 7: grounded to Ohmeter
pin 8: 1.4 Ohms
pin 9: infinite
pin 10: 60 Ohms
pin 11: 60 Ohms
pin 12: 60 Ohms

-CH-

 

[modem7]

With the power supply disconnected from the board I measured the following resistances to ground at each pin.

Nothing stands out as unusual.

... and also test the voltages there with the PSU on.

And the result was ?

If you have an IBM supplied PSU, then we expect that all voltages will be in tolerance, because your motherboard started ("I was again reminded that CMOS settings were lost ..."). The PSU's 'POWER GOOD' signal is required to start the motherboard, and the IBM supplied PSU will only generate POWER GOOD if the PSU's output voltages are in tolerance.

Set the time and date and fudged some numbers to represent a 32 MB HD.

Careful. With the IBM BIOS, 'fudging' the hard drive type incorrectly can lead to a '1790-Disk 0 Failure' error. It is not about hard drive size; it is about cylinder and head count. As a fictitious example, if I have a hard drive fitted that has 500 cylinders, but I set the drive type in CMOS SETUP to a same-MB-capacity type that has 1000 cylinders, the POST in the IBM BIOS will do a test read of cylinder 1000, which will fail, and the POST will display '1790-Disk 0 Failure'. Same applies for head count.

Your AMI BIOS may do something similar.

When I did I got the message that there was an HD controller failure.

What exactly is the message shown ?

 

[clh333]

In my initial diagnostics of the machine, when first acquired about two years ago, I went through the power supply and motherboard; everything passed then. I remember having to replace the CMOS backup battery, and struggling with creating the disk used for BIOS configuration. I think I recall it made a difference to the IBM BIOS which was Drive A: a 3.5 or 5.25, and what density. Once I had the machine booting from a floppy I wanted to introduce an XT-IDE, as I had done on my two 5160s. I discovered the IBM 5170 BIOS is incompatible with the XT-IDE.

At the time the machine had no hard drive, if I recall, and since I had the Quantum drive sitting around I decided to use it. That decision required an ISA controller and a different BIOS; I opted for the AMI BIOS from MinusZeroDegrees. I either had or acquired the SIIG controller. Malc was kind enough to point out that I could occupy the two openings at U17 and U37 with EEPROMs loaded with the XT-IDE BIOS, so I burned four EPROMS with the appropriate code and installed them.

The AMI Bios does not require configuration from a floppy, so I was able to turn on the machine and set it up from the console. The Bigfoot is far larger than the 32MB limit of the day but I found some documentation on STASON and entered some figures that seemed to work. Like an idiot, I did not write them down. But the machine was able to boot and I loaded the drive with MS-DOS 6.22 and set the machine aside. That was a mistake but there were bigger issues that intervened: A year or more passes.

Fast forward to a few days ago when I drag out the machine and try to use it to drive a monitor for a test. The machine complains the CMOS backup is down; as I am trying to address this from within the AMI BIOS config the machine abruptly halts and will not restart. I removed and checked the power supply output voltages and checked the board for shorts.

Which brings us to today:

I didn't want to go poking around the board with meter probes so I took time to find some leads with hooks. To make more room I moved the controller farther away from the power supply. This morning I was prepared to continue the investigation. I decided to reboot the machine to document the controller error I encountered yesterday. At the same time I could photograph the BIOS config as I had set it up yesterday. I powered up the machine and opted to enter config; this option is offered right after a memory test. Configuration settings held from yesterday and I did not change any setting. I exited config, saving, and expected to see the controller error as I had yesterday.

Instead I got an XT-IDE BIOS message offering to boot from the "foreign" hard disk or from the BIGFOOT! I selected the Bigfoot and the machine booted into a DOS C: prompt. So? Apparently the XT-IDE BIOS could determine the proper geometry from the drive itself, which the AMI BIOS could not.

The machine has been on for half an hour or more without a hiccup. I will make the power supply test at the motherboard and report the findings, but I suspect there will not be anything unusual in the results. I can only guess that moving the controller to a different slot may have affected the machine's ability to boot. It is possible that I had the controller, which is a 16-bit card, in an 8-bit slot and caused the PSU to shut down in defense of its integrity. If that's the case I'm not going to repeat the experiment to prove the hypothesis. I've already demonstrated how little I know: As one of my teachers once advised me, "It is better to keep your mouth shut and be thought a fool than to open your mouth and remove all doubt."

Thanks to everyone for their suggestions. I'll let you know how this goes.

-CH-

Pictures:

 

[Malc]

.....Instead I got an XT-IDE BIOS message offering to boot from the "foreign" hard disk or from the BIGFOOT! I selected the Bigfoot and the machine booted into a DOS C: prompt. So? Apparently the XT-IDE BIOS could determine the proper geometry from the drive itself, which the AMI BIOS could not.

As you are using the XUB go into the AMI bios setup and set Hard drives to NONE, As in as far as the AMI bios is concerned there are no Hard drives fitted, Save the settings, This will give the XUB full control of the hard drive.

A while back my 5162 developed an issue, It would power up and run for a few seconds or a minute or two and shut down or just not power up, It turned out to be the IBM supplied parallel / serial card was faulty, I replaced it and problem solved.

 

[Stone]

A while back my 5162 developed an issue, It would power up and run for a few seconds or a minute or two and shut down or just not power up, It turned out to be the IBM supplied parallel / serial card was faulty, I replaced it and problem solved.

I have several of those cards, some even NOS and they all have developed a short tantalum (or two) in that cluster of three on the 12V rail.

 

[Malc]

That's probably the issue with mine though i never checked, As i had spares it got put in the growing pile of jobs for another day which never arrives.

 

[clh333]

Does it look like this? This card was in the machine when it died; was subsequently removed.

-CH-

 

[Stone]

Yes, that's it.

The offending tantalums are C11 and C12.

When I removed them the card worked.

 

[Malc]

Yep that's the bugger.

 

[clh333]

Zut alors! C11 and C12 are three-legged and may be hard to find. Are you sure they can be removed?

I reconfigured the SIIG card to provide COM1: and COM2:, LPT1: and a game port, as well as HD and FD control. Haven't really tested anything yet, though.

Two other questions:

The FDs are on rails. It looks as though there ought to be a tab on the front of the rail to secure the rail (and thus the drive) to the chassis. Have I got the wrong rails?

I configured the RAMPAGE 286 card that was in the machine to provide 128 conventional and 386 extended memory. That shows up in the AMI BIOS. I'm going to need the AST driver to access the expanded memory. I think there's 2 MB on board. What software is that, exactly? I got the manual from MZD, thanks!

Malc: made your changes to the AMI BIOS and the "foreign" drive disappeared; defaults to Bigfoot.

Thanks again for everyone's help.

-CH-