Early 5170 Not POSTing

[KLund1]

Hi again,
I have a Type 1 5170 motherboard with U27=6181028, U47=6181029 BIOS Roms.
Sadly if does not boot.
I have a Post code card I just remember I had. I get 0201 on power up. I get no beeps
There was a 16mhz crystal installed, and a 12mhz taped to one of the chips. I put the 12mhz in for this testing. The 16mhz legs were bent in half to fit in and one broke off when I was gently removing it. But since I get some post code the 12mhz should be working ?
I looked at the always helpful munuszerodegrees site and there was no troubleshooting for a 5170.
what to test for next?
Thanks

 

[modem7]

I have a Post code card I just remember I had. I get 0201 on power up. I get no beeps

See the 'Example of a modern POST card' section at [here]. If you have the same (or very similar) card, the [02][01] means that 02 was the last POST code received by the card, and 01 is simply the code that was received before that.

There is a list of POST codes for the IBM 5170 at [here]. You have the 01/10/84 BIOS, so 02 means that a problem happened when the POST went to verify/test the shutdown byte in the 146818 RTC chip.

If your MC146818 is in a socket, try re-seating it. Could be the chip; could be something that supports its operation.

Deja vu. This happened very recently to member xavicampa. See post #24 and on of [here].

 

[KLund1]

Thanks
Very little on this board is socketed. The MC14618 is not socketed. Per your linked thread, I checked the voltages on this chip, pin 13=0v, 18=5v, 24=5v.
I'm amusing that is good, but the chip may still be bad, or the support chips. Which are the support chips? How best to check them.
Looking very closely to chips in that area, U115 had a old sharpy pen check mark on it, that rubbed of when I touched it.
FYI I am using a modern ATX psu with an AT adapter. I have exactly the same post code card you referenced.
Here is a better pic of the board for reference. the broken leg 16mhz crystal in taped on a chip.

 

[Makefile]

Hello

Assuming your BIOS is consistent with these listings (https://github.com/kaneton/appendix-bios/blob/master/src/test1.asm#L298) the only thing that can go wrong between 02 and 03 is if the BIOS ROM has an invalid checksum, in which case it halts without beeping.

 

[modem7]

Assuming your BIOS is consistent with these listings (https://github.com/kaneton/appendix-bios/blob/master/src/test1.asm#L298)

The OP's BIOS is not consistent with that listing.

That listing is titled "IBM PC AT v2 bios source code". There were three versions of the BIOS, and the OP has the first version. As shown at [here], IBM swapped POST tests 02 and 03 between versions. It can be seen in the source code for the first version.

 

[modem7]

I'm amusing that is good, but the chip may still be bad, or the support chips. Which are the support chips? How best to check them.

There are quite a few chips that support the MC146818. Diagnosis is going in with a logic analyser, with the knowledge of how to use it.

You have indicated that you are using an ATX power supply, presumably to indicate to us that you are aware of note 6 at the bottom of [here]. But if you have one, try another power supply, just in case. Or, if you have the ability, upgrade the BIOS as a test.

There are now four faulty 146818 recorded at [here]. Perhaps try your luck replacing the MC146818. Add an IC socket while you are at it. Get two MC146818, in case one is bad.

 

[KLund1]

There are quite a few chips that support the MC146818. Diagnosis is going in with a logic analyser, with the knowledge of how to use it.

The chips around the MC146818 are TTL. I am not good O-scopes or L.A.'s. But I can easily de-solder and check them in a chip tester. Then put back in with a socket. I know this adds possible risk to the board. But I have become pretty go at this.

I mention the PSU to cover that I have a high quality source for power. I am not using an old AT 5170 ear PSU that may have bad ripple, or have a voltage fall too low due to loads.

"if you have the ability, upgrade the BIOS as a test."
I have 2 other 5170 MB's (one known to work), and I and try those Bios chips and see what happens. Then decide if I need to remove the MC146818 or the other TTL's. But it might be helpful to know which are the MC146818 support chip location numbers.

One more time, I very much appropriate all the input and help!

 

[KLund1]

Tested with other 2 sets of roms. They were the same version, same make chips.
No change in the the post codes for both sets.
Where might I find MC146818 chips to purchase?
Thanks

 

[modem7]

Tested with other 2 sets of roms. They were the same version, same make chips.

I thought that you may have an 'upgrade' set (i.e. later version).

But it might be helpful to know which are the MC146818 support chip location numbers.

I gave up. See my partial diagram at [here], which shows some of the chips that are involved in generating signals that go to the MC146818.

Where might I find MC146818 chips to purchase?

I see some on eBay. Be sure to get the 24-pin DIL package.

 

[xavicampa]

Where might I find MC146818 chips to purchase?

I got a few spares from the time I had mine fail, let me know if you can't source them locally, I can ship a couple (from Norway, mind you).

 

[KLund1]

Ok, back to this boards.
I got a MC146818. I de-soldered the original and put in a socket. Put in the 'new' chip, powered on. Nothing. no screen, no beep.
While I was waiting for this chip to arrive. I found I had a working Exerex 286 10mhz working board that has the same chip. In fact, a lot of the exerex board is very similar to the ibm board. I tested the MC146818 old, and 'new' in the already socketed exerex board, and all chips are working. so this chip is not the problem with the ibm board.
While putting in the socket on the IBM board, I got to wondering it the crystal might be the problem.
I think the crystal should be checked. I recently purchased a handheld oscilloscope, in order to learn how to use it. This might be a good place to start. I connected the neg to the crystal holder, and the probe to a random leg, powered on. I got a ready of 12 mhz. Ok that works.
I looked at the diagram in post #9. I can find the chips and probed the legs. But I have no idea what a good readout should be, let alone if i have the scope set right. (the scope is a FNIRSI-1C15 110mhz single probe unit). I did check voltage to all those chips and they all get 5v. The AUTO button seems helpful. I probed pin 20 on MC146818 and saw some waves, pressed the auto button and saw a nice square wave. All other pins were set high or low. No waveforms from the auto button. There was a repeating ripple on the low or high line, but not sure how to 'zoom in'

 

[modem7]

While I was waiting for this chip to arrive. I found I had a working Exerex 286 10mhz working board that has the same chip. In fact, a lot of the exerex board is very similar to the ibm board. I tested the MC146818 old, and 'new' in the already socketed exerex board, and all chips are working. so this chip is not the problem with the ibm board.

Okay, MC146818 ruled out.

I think the crystal should be checked. I recently purchased a handheld oscilloscope, in order to learn how to use it. This might be a good place to start. I connected the neg to the crystal holder, and the probe to a random leg, powered on. I got a ready of 12 mhz. Ok that works.

If the 12 MHz crystal wasn't working, the POST would not get executed, and therefore, the POST codes of 01 and 02 not get generated.

I looked at the diagram in post #9. I can find the chips and probed the legs. But I have no idea what a good readout should be, let alone if i have the scope set right. (the scope is a FNIRSI-1C15 110mhz single probe unit).

An oscilloscope is inadequate for a lot of what is going on in that diagram within post #9. What that diagram shows is that this isn't a case of replace or test the three or so support chips - there are lots of chips 'supporting' the MC146818.

It would be really good if you could swap out the IBM BIOS ROM's for the AT version of the Supersoft/Landmark diagnostic ROM's, to see what that diagnostic makes of the situation. Are you in a position to do that? Someone on the list at [here] could supply you with a set of ROM's. Note the video card requirements.

 

[modem7]

You have indicated that you are using an ATX power supply, presumably to indicate to us that you are aware of note 6 at the bottom of [here]. But if you have one, try another power supply, just in case. Or, if you have the ability, upgrade the BIOS as a test.

I mention the PSU to cover that I have a high quality source for power. I am not using an old AT 5170 ear PSU that may have bad ripple, or have a voltage fall too low due to loads.

We really don't know what is going on re note 6.

- Is it all XT-class PSU's or just some?
- Are some AT-class PSU's also a problem?
- Why did IBM swap POST tests 02 and 03? Was it because it seemed more logical for the POST to test the integrity of the BIOS (which includes the POST itself) first? Or did IBM resellers sometimes encounter a problem when replacing faulty IBM PSU's with third party ones, and investigation revealed that delaying the early 146818 related POST test was found to be a fix?

It could be some kind of timing issue.

I think it would be good for you to try out an upgraded BIOS.
Note too that if you should later plan to fit an XT-IDE/XT-CF card, note 1 at the bottom of [here] indicates that your 01/10/84 dated BIOS is not going to 'play ball'.

 

[KLund1]

I have a set of original Landmark/SUpersoft chips or PC/XT and AT Still in the the original box. I totally forgot about them until you suggested it. Thx. But they are untested. I put the AT set in the 5170 board, in the correct orientation, and correct chip in the correct socket. No joy. no display. No post code just "- - - -". So as you indicated above, this board does not really even start to post. (I put them in the Everex board just to see, and that board would not post. I put it's original chips back and works normally)
Several of the larger chips get pretty warm after about 4 minutes, U102, U124, U80, U111, the 286 cpu gets almost too hot to touch. The OS, and ram chips stay cold.
I can program a new OS rom set(s). Which version would you suggest for this board version? Perhaps a one for testing, another later for performance and features?

note:
The PSU works for the Everex 286 board, my working test XT board, and my test 486 board, along with numerous other clone PC to 486 boards over the last couple years. This is not an issue. Same for the video card. Understood about the BIOS and XT-ide/cf card.

 

[modem7]

I have a set of original Landmark/SUpersoft chips or PC/XT and AT Still in the the original box. I totally forgot about them until you suggested it. Thx. But they are untested. I put the AT set in the 5170 board, in the correct orientation, and correct chip in the correct socket. No joy. no display. No post code just "- - - -". So as you indicated above, this board does not really even start to post. (I put them in the Everex board just to see, and that board would not post. I put it's original chips back and works normally)

The Supersoft/Landmark diagnostics does not output codes to a POST card. (If it did, the codes would be listed in the manual.)

The POST in the IBM 01/10/84 dated BIOS is starting and outputting 01 and 02, and so, looking at the Supersoft/Landmark diagnostic screen output at [here], I did expect at least a display with the first few tests passing.
I don't know what version you have, but the version at [here] has caught me out quite a few times, because of its 20 second delay before anything is displayed.
And you will have read that VGA cards (well, the vast majority) are not supported.

Maybe your AT set of Supersoft/Landmark ROM's are faulty.

Several of the larger chips get pretty warm after about 4 minutes, U102, U124, U80, U111, the 286 cpu gets almost too hot to touch.

See the 'CPU appears to get very hot' section of [here].

I can program a new OS rom set(s). Which version would you suggest for this board version? Perhaps a one for testing, another later for performance and features?

Either of the two '85 dated ones will be good for testing the 'note 6' thing. I have used both on a type 1 motherboard. Per [here], we know for fact that the 06/10/85 BIOS is supported on the type 1 motherboard. I have not heard of people reporting a problem using the 11/15/85 BIOS on a type 1 motherboard.

If your problem is not the 'note 6' issue, then per [here], we expect that the display on your POST code will change to [03][02].

Note that the two '85 dated BIOS' have a 'speed' test in the POST. See note 1 at [here]. Good thing that you have the original crystal fitted.

 

[KLund1]

I had forgotten about using only a 5151 on these tests, and to wait a little before it shows on screen. I put notes in the manual for future reference.
I can see the test results now. My SUpersoft version is 1.1 (1991) (I'll make a set of 1.2 later, as these seem to work unless you think 1.2 may give a more detailed result?)
Understood that supersoft does not show post codes, I had the post code card also installed for post #14. sorry for not mentioning that.
Understood about hot CPU, but other chips getting a little hot to the touch?
See picture for test results after 6 or 7 passes. Note I do not hear any beeps during the tests.
Looks like several chips might be bad. But is it possible that one chip could be causing the other to fail a test, if so which.?
Many thanks once again for all the assistance !

 

[modem7]

Understood about hot CPU, but other chips getting a little hot to the touch?

Maybe quite warm. For example, on my IBM 5150 motherboard, the 8253 timer chip is very warm. Very subjective.

See picture for test results after 6 or 7 passes. Note I do not hear any beeps during the tests.

Note the 8254 related (repeat: related) failure. Channel 2 of the 8254 is involved in speaker operation.

Looks like several chips might be bad. But is it possible that one chip could be causing the other to fail a test, if so which.?

Give me some time to do some analysis. I'm about to head out the door to do some shopping.

 

[modem7]

Looks like several chips might be bad. But is it possible that one chip could be causing the other to fail a test, if so which.?

Give me some time to do some analysis. I'm about to head out the door to do some shopping.

The 8254 (fail) and 74LS163 (fail) both use 8 bits of the external data bus, but so do the two 8237's (pass) - so, for now, a problem on the external data bus is discounted.

The 8254 (fail) and 74LS163 (fail) both use some bits of the external address bus, but the the two 8237's (pass) use those same bits and more - so, for now, a problem on the external address bus is discounted.

The 8254 (fail) and 74LS163 (fail) both get their 'chip select' signal from the same chip, a 3-to-8-decoder in socket U123. But so do the two 8237's (pass). Thinking about a partial failure of U123, I could not come up with a failure mode that explained the Y0 and Y6 outputs working, but not Y2 and Y4.

What does the +5V rail on the motherboard measure?

Maybe for now, we just focus on failure of the 8254 timer. Do you see a 1.19 MHz clock on the three CLK pins? Do you see activity on the /CS pin during the time that the three 8254 tests are being run ?

 

[KLund1]

Hope shopping went well.
I get 1.19mhz steady on all three CLK 8254 pins. I see a positive square wave on the /CS pin a half second after power up then goes down, then up when I hear the monitor starts, and then down, then a lot after the supersoft runs. The VCC is steady at 5.1v

 

[modem7]

The 8254 tests will be reading/writing from/to the 8254. Is there activity on the /RD and /WR pins?

All three timer channels are being tested, i.e. read and writes of I/O ports 40h, 41h, 42h, and 43h. Is there activity on both the A0 and A1 pins?

I see a positive square wave on the /CS pin a half second after power up then goes down, then up when I hear the monitor starts, and then down, then a lot after the supersoft runs.

That sounds a bit weird. I expect (repeat: expect) that it would be mostly high, with you only seeing high-low-high transitions during the three 8254 tests.
When I find time, I might see what the 8254's /CS looks like with the Supersoft/Landmark ROM's fitted to one of my 5170 motherboards.
Sometimes, reality turns out to be different to expectation, and then investigation reveals that one's exception was based on not knowing all the facts.

I don't know whether or not you are following it, but there is a timer related thread active at [here] - post #14 onward. Different in that it's a 5150 motherboard, and Ruud's diagnostic ROM is being used.