• Please review our updated Terms and Rules here

Problem IBM AT 5170 mainboard

So POWER GOOD and all voltages okay.

So if there is a partial short on the +5V line of your 5162 motherboard, then both power supplies are dealing with it.

I haven't time to look through the 5162 technical reference now (on [this] link), but I have no doubt that the 1700 series checks (hard drive) will be the same as those for a 5170. The first check is the 1782, where the POST asks the controller to do a self check. A 1782 is posted if the controller responds in the negative, or the controller does not respond. See [here] for details.

The new information about the controller working in a 486 DX2 system in very interesting.

The symptoms/results though can be explained by the slot on the 5162 not having all signals. Damaged by corrosion? Does use of all 16-bit slots result in the 1782 error?
 
Last edited:
modem7 said:
Does use of all 16-bit slots result in the 1782 error?
Click to expand...

I swapped the MFM controller and VGA card trough the 5 available 16-bit slots. The results are interesting.

At first it gave me no errors at all on post using the MFM card in slot 2 or 3! Although there were no errors, it would not boot from disk.
I continued using other slots and the 1782 and/or the 601 errors reappear.
I changed back to using slot 2 and 3. It gave me again a 1782 and/or 601 error. After further testing it didn’t work anymore and keeps displaying 1782 In all slots sometimes accompanied by error 601.
The VGA card does work in all slots in all conditions.

I don’t know if it is a problem but the 5162 boots very fast when powering up. The ST-225 spins up slowly and needs a few seconds to complete its power on self-test. (seek test etc.).
When the ST-225 has to power up from a complete halt it appeared to me it didn’t complete its self-test yet at the point that the controller gets the command to execute its own self-test.
 
My experiments on a 5170 with first generation controller reveal:

1. If the controller is placed in an 8-bit slot, the 1782 passes, but the next test, the 1780, fails.

2. The 1782 test passes even when there are no hard drive cables attached to the controller (but the next test, the 1780, fails).
That suggests that the controller's self test does not rely on the presence of a hard drive. This is what I would expect.

huubwen said:
I don’t know if it is a problem but the 5162 boots very fast when powering up. The ST-225 spins up slowly and needs a few seconds to complete its power on self-test. (seek test etc.).
When the ST-225 has to power up from a complete halt it appeared to me it didn’t complete its self-test yet at the point that the controller gets the command to execute its own self-test.
Click to expand...
From item 2 above, if the drive were not ready, we would expect a 1780 error, not a 1782.

huubwen said:
At first it gave me no errors at all on post using the MFM card in slot 2 or 3! Although there were no errors, it would not boot from disk.
Click to expand...
The failure to boot when there were no 1700 series errors may be due to an unrelated cause.
Best to first hunt down the cause of the 1782.

huubwen said:
After further testing it didn’t work anymore and keeps displaying 1782 In all slots sometimes accompanied by error 601.
Click to expand...
So you don't have any choice now but hunt down the cause of the 1782.

It's like you have a bus related issue, one that affects the hard/floppy controller, but not the VGA card.
 
Since there's been nothing but jabber about the 1782 error, I decided to take a couple of minutes and look at the 5162 BIOS and see where the 1782 originates.

Well, if you look in the DISK-SETUP routine, anything that jumps to the tag CTL_ERRX will cause the error message. That appears to be only the section that issues an INT 13H, function 14H, "execute controller diagnostic". That routine, CTRLR_DIAGNOSTIC, is very simple. It sets up the PICs for an interrupt, waits for a busy status on the controller to clear and then issues a 90H command to the card and waits for it. If the card clears busy and returns no error, the code declares the controller good.

So, what might cause the controller to fail? The NOT_BUSY routine is a simple CPU loop waiting for the busy bit to clear, so no problem there.

Going to the schematics of the hard/floppy controller, about the only thing that jumps out is the requirement for +12 and +5 on the bus. If your +12 is wonky, then things are not going to go well.

About the only thing that could cause both a 601 and a 1782 error is that the PIC isn't getting the interrupts. Since the diskette section uses IRQ 6 and the hard disks section uses IRQ 14, they're not even on the same connector; the likelihood of both failing is small.

Have you looked at the CPU crystal? If someone wanted to "hot rod" this board (very common back then), that would cause all sorts of CPU-dependent timing loops to go screwy.
 
Chuck(G) said:
Since there's been nothing but jabber about the 1782 error, I decided to take a couple of minutes and look at the 5162 BIOS and see where the 1782 originates.
Click to expand...
Thank you for sorting out.
Chuck(G) said:
Have you looked at the CPU crystal? If someone wanted to "hot rod" this board (very common back then), that would cause all sorts of CPU-dependent timing loops to go screwy.
Click to expand...
I have visually inspected the component which I believe is the CPU crystal (see picture). It seems untouched to me.
Somehow I believe the partly shorted +5V on the board must be related to the problems with the MFM controller card. As you described a stable +5V and +12V on the bus is required by the card.
Any suggestions how to find the partly +5V short?
20120724_213320.jpg
 
Have you tried cleaning the edge contacts on the MFM card ? I've had several problems
recently with ISA and MCA cards that a simple cleaning with a new pencil eraser , solved
the problem.

Just rub the eraser over the contacts a few times, and wipe off with a clean rag.
 
huubwen said:
I have visually inspected the component which I believe is the CPU crystal (see picture). It seems untouched to me.
Somehow I believe the partly shorted +5V on the board must be related to the problems with the MFM controller card. As you described a stable +5V and +12V on the bus is required by the card.
Any suggestions how to find the partly +5V short?
Click to expand...

The crystal is the correct frequency. What bothers me is your statement that the POST is very fast. That's not my experience with the 5170; I don't think it should be any different with the 5162.

Have you tried running a RAM diagnostic for a few hours on the system? There's something that's not quite right.

My next step would be to inspect the PCB for a possible bad solder joint. After that, it'd be removing each of the socketed ICs one by one to see if a marked difference in the resistance reading is noticed. After that, it'd be to replace the electrolytic and tantalum capacitors one by one. Not something for the faint-of-heart.

Can you burn your own EPROMs? If so, I have a diagnostic ROM set for the 5170 that might turn something up.
 
mikey99 said:
Just rub the eraser over the contacts a few times, and wipe off with a clean rag.
Click to expand...
Miket99 you made my day :D

I grabbed a pencil eraser and I thoroughly cleaned the contact of the MFM card. And guess what; it didn’t give any error at all. It will not fail anymore even after multiple power offs.

The self-test of the ST-225 changed to. It spins up and does just two seeks. That’s it. No complete seek test anymore.

I successfully installed FreeDOS. Man; what is that 286-6 way faster than my 5160.

I'm still not completely convinced yet but it looks very promising.

Altought the MFM card works now I am still determent to find the partly short at the +5V line.
I will follow the steps as suggested by Chuck(G).
 
As I said earlier, I didn't think there was a partial short. Normally, if this were the case, you could identify the component just by feeling around for heating.

Don't use the pencil eraser trick more than once--the plating on those contacts is very, very thin (a few microns) and an abrasive eraser will take it off in only a few iterations. Get some good contact cleaner, such as DeOxIt. You'll also be able to clean the edge connectors themselves that way.
 
Chuck(G) said:
As I said earlier, I didn't think there was a partial short.
Click to expand...
Because I do not experience a problem (except for the board not working with all my power supplies) I decided not to take any further action with the board and just enjoy it.

Next step for is finding an original IBM XT286 case to put the 5162 board in. Finding it would be a real challenge.

Thank you all for your help
 
Glad to hear that cleaning the edge contacts on the MFM card fixed the problem !

I just happen to have a 5162 motherboard here, and I measured the +5 pin resistance
with no cards plugged into the motherboard. Using my Simpson 260 (old analog meter)
I read about ten (10) ohms between the +5 connector pins and the ground pin.
Using Ohms law I=E/R, this translates to about .5 amps or 500 ma. This does sound
a bit high for current draw on the motherboard, but still well within spec.... I would
think the +5 V power supply would be capable of handling several amps draw on the +5
line. Using the resistance value measured with a meter may not be that accurate of
a way to calculate the true current.

I last powered this board up about two years ago, and it worked fine then.
 
Last edited:
The resistance is irrelevant if it's not a short or open. You're measuring a non-linear circuit, so the resistance is going to depend upon the supply voltage in your meter and the input impedance (the 260 will have a much lower impedance than almost any DMM). So, you'll get two very different readings depending on the polarity of the measurement (which probe goes to which side) and the measurement voltage (does your meter have a "diode" setting that uses higher voltage?). Add to that the time elapsed to allow the decoupling capacitors to charge, any leakage resistance, the forward drop of any protective diodes, etc. and you're not going to wind up with a "If it measures X ohms, it's good" type of test.

A dead short's a dead short, however.
 
remeber if the second wire on one of the connecors is not there try another supply
the drive will need to be llf for an AT not an xt -different format
this runs agains the grain but ANY similar or eben an advanced wd11006 ( 1:1 card will work. One morning I was called as someone had been robbed? I just plugged in the nearest wd1003 type care in and the system worked. The typical wd card is very close.
just dont try a 1:1 llformatted drive with an older card.
sometimes it is simpler to use a third party program like wdformat wdfmt.exe ? if the cmose is set right.,
this next may be heresy. but later bios chips work in AT and AT clones. did it not theory.
 
You must log in or register to reply here.
Back
Top