Help with 5150 rev A Supersoft ROM results

[Old_hitech]

I have a IBM PC 5150 Rev A 16K -64K motherboard that appears "dead" with original BIOS rom. I previously found a +12V cap that was shorted and have removed it. I installed Supersoft diagnostic ROM and get the following results. I have ram chips installed in Banks 0, 1, 2 and 3. Item 1- is when the NO MEMORY ERROR 14000 yields bit position Xs (in 7, 6, 2 and 0 position). Item 2 -Shortly after that the screen changes to MEMORY ERROR AT ADDRESS 1C000 and bit position Xs (in 7, 6, 2 and 0 position). Item 3 - After about 3 1/2 minutes the first pass ended with memory error at 98000 and bit position Xs (in 7, 6, 2 and 0 position). There were a few other memory error noted, but always with bit position Xs (in 7, 6, 2 and 0 position). I'm thinking that I should replace Bank 0 (soldered in) RAM for Bit 7 first, then re-run the diagnostics, but the first MEMORY ERROR AT ADDRESS 1C000 is not indicating a Bank 0 location. Looking for recommendations on what I should do next?

P.S. I tried to upload some pictures of the screen for items 1-3, but the upload tool kept stating that file sizes were too large. When I adjust picture size to acceptable size, I could not read any of the screen information. I would like to know what I am doing wrong when uploading pictures. Thanks.

 

[chris_nh]

14000 hex = 81920 decimal /1024 = 80k Hmmm ...

Check this out:
http://www.minuszerodegrees.net/supersoft_landmark/5150/16KB-64KB.htm

"WARNING: If your 16KB-64KB motherboard has a RAM related problem, the Supersoft/Landmark Diagnostic ROM is known to display incorrect and misleading information.
It is though the Supersoft/Landmark Diagnostic ROM is not intended for the 16KB-64KB version of IBM 5150 motherboard.
Click here for a list of known bugs."

Not booting with the regular BIOS installed? Have you ruled out a bad BIOS chip perhaps? If there is a memory error, it might be in bank 1??. A few possibilities here I think. Have you tried booting the board w/ BIOS and only bank 0 installed? This might boot to BASIC in that configuration.

 

[modem7]

What does Ruud's diagnostic ROM report ?

You can get it from [here].

 

[Stone]

P.S. I tried to upload some pictures of the screen for items 1-3, but the upload tool kept stating that file sizes were too large. When I adjust picture size to acceptable size, I could not read any of the screen information. I would like to know what I am doing wrong when uploading pictures. Thanks.

Adjusting the picture size can have more than one possibility. The number of bytes is what the forum actually measures. You don't need to decrease the dimensions of the picture to decrease the number of bytes. You can also do that by decreasing the picture quality when resaving it thereby leaving it large enough to read the details but using fewer bytes so the forum will accept it. IIRC, the max size allowed is 250k.

 

[Old_hitech]

I made another attempt at reducing the size of the screen shots for Items 1-3 that I mentioned in my initial post. Thanks to Stone for the tips on reducing the size.

 

[modem7]

U36 8255 PARITY DETECTED
---------------------------

The '8255 parity detected' test is described on "Page 20" of the manual for the SuperSoft/Landmark diagnostics. Refer to the diagram [here]. So, before RAM refresh is started, and RAM used, the diagnostics are interrogating the 8255 as to the state of its PC6 and PC7 pins, both pins expected to be reflecting no RAM parity errors.

As expected, on my 16KB-64KB motherboard, if I simulate a failed RAM chip in bank 0, I do not see failure of the U36 8255 PARITY DETECTED test.

As an experiment, I took the PC7 pin high on my motherboard. As expected, the U36 8255 PARITY DETECTED test failed. But the following 16K CRITICAL MEMORY REGION also failed, reporting parity errors, but no bits in the 'FAILING BITS' row.
That is not what you see; you see failing bits in the 'FAILING BITS' row. You do see parity errors, and so perhaps your failure of the U36 8255 PARITY DETECTED test is causing those parity errors.


16K CRITICAL MEMORY REGION
--------------------------------

As explained earlier, for the 16KB-64KB motherboard, the Supersoft ROM can be misleading in regard to RAM chip failure. In my experience, bit identification is usually accurate (except for bit 0); the problem is in address/bank identification.

Did Ruud's diagnostic ROM (see post #3) report a critical error when checking the first 2KB of RAM ?

Where multiple bits are shown in error, another consideration is a failure of RAM refreshing. I disabled RAM refresh, and observed (on the first pass) that the 16K CRITICAL MEMORY REGION test passed, then the MEMORY REFRESH test failed. As expected.

If Ruud's diagnostic ROM shows a problem in the first 2KB of RAM, then I think that you can safely assume bank 0.

 

[Old_hitech]

modem7, Thanks for the analysis of my screen shots. I am not sure if I am following everything in your explanation, but I think you are saying that my problem could be in the bank 0 RAM. Is that correct? Or do you think I need to replace U36 8255 in addition to the Bank 0 RAM?
I do not have a Ruud diagnostic ROM. Where can I purchase one? I purchased the Supersoft/Landmark ROM from eBay.

 

[modem7]

I do not have a Ruud diagnostic ROM. Where can I purchase one? I purchased the Supersoft/Landmark ROM from eBay.

From what I see on eBay, you received both a ROM and an adapter.

I was expecting the adapter to be the one advertised at [here], but the one you were supplied with looks slightly different. Therefore, I am uncertain as to the range of ROM/EPROM types that your adapter can work with. But I think that it highly likely that your adapter will work with a 27C256 EPROM (or equivalent).

If you are willing to accept the risk (risk that your adapter may or may not work with a 27C256 EPROM), then what you could do is use the list at [here] to find someone local who will supply you with a programmed 27C256 EPROM (or equivalent).
Specify:
* 27C256 EPROM (or equivalent). Target is an IBM 5150, and so 27C256 speed (access time) is unimportant.
* The code for the EPROM would come from [here].
* The code is 8KB sized, and needs to be replicated many times in order to fill the 27C256.

If you want to 'play safe', you could buy the adapter at [here] as well. It will definitely work with a 27C256 EPROM (or equivalent).

( A response to your other questions currently being drafted. )

 

[modem7]

modem7, Thanks for the analysis of my screen shots. I am not sure if I am following everything in your explanation, but I think you are saying that my problem could be in the bank 0 RAM. Is that correct? Or do you think I need to replace U36 8255 in addition to the Bank 0 RAM?

First, there are no guarantees. Software like the Supersoft ROM diagnostics are tools. You end up with possibilities and associated probabilities.

Was your motherboard working, then one day, discovered not to be? Or did you acquire the motherboard as is? Because, if the former, one would not expect multiple faults.

Based on my earlier experiments, I believe that (for now), your motherboard has two distinct issues:
1. The hardware that reports RAM parity errors is faulty. The 8255 chip is only part of the subject hardware.
2. Faulty RAM chips.


1. The hardware that reports RAM parity errors is faulty.
--------------------------------------------------

Reference diagram at [here]. A logic probe or multimeter is required. First step would be to see if the PC6 and PC7 pins on the 8255 chip are low (up until the time of the U36 8255 PARITY DETECTED test). If they are, a (partially) faulty 8255 is highly likely. If one is not, 'backtracking' is done (under direction) to work out what is the source cause of the pin being high.

Also, if you tie both the PC6 and PC7 pins on the 8255 to ground, and the U36 8255 PARITY DETECTED test then passes, that would suggest that the 8255 is good (and the cause elsewhere).



2. Faulty RAM chips.
--------------------------------------------------

At this time, we are not sure which bank they are in. Some people may say, "Because the motherboard appeared dead when the IBM BIOS ROM was fitted, the bank must be 0", but for all we know, you also have a faulty IBM BIOS ROM.

The faulty RAM chips may be in multiple banks. If I simulate that, bit 5 in bank 0 and bit 4 in bank 2, the very first RAM error that the Supersoft ROM shows, is at address 05040 (yes, misleading, and known about), and shows both bits 5 and 4. There is no bit 5 appearing first and then 4 being added to the bit list later.

That suggests that your 'failing bits' of 7, 6, 2 and 0, could be scattered among all banks.

Experiment

As an experiment, I removed all RAM chips in banks 1/2/3, leaving only bank 0. In bank 0, I still had simulated a 'bad chip in bit 5'. No switch changes required for this particular activity. The Supersoft ROM then showed bit 5 in error at address 05040. It did not appear to go on to test the other banks.

When I then removed the simulated 'bad chip in bit 5' (ie. only bank 0 fitted, and that bank good), the Supersoft ROM then showed all bits in error, except for bit 0, at address 05040. The Supersoft ROM is known to sometimes show bit 0 as good when it is not. Presumably, the Supersoft ROM was this time, going on to test bank 1.

Which bits in error are shown when you only have bank 0 populated ?

 

[Old_hitech]

I recently acquired this rev A 5150 with the 16-64K motherboard from eBay. The seller stated that he heard a single tone from the speaker when he powered on the computer back a few months ago. When I received it I powered it on and heard no tones. I followed the trouble-shooting procedures on the minuszerodegrees website. I found a +12V short in one of the tantalum caps. Removed the bad cap and found that the motherboard still did not power up with tones. I purchased the Supersoft ROM and that is where am I at now. I did try a working 1501476 BIOS ROM from my working 5150 RevB computer. But when installed into the 16K-64K motherboard and powered it on I still did not hear tones. I will try your trouble-shooting tips for the 8255. What is the best method to ground the PC6 and PC7 pins?

 

[Stone]

Just for the hell of it try starting up with the P8 plug disconnected from the motherboard.

 

[fs5500]

Great.
I wonder your original BIOS ROM part no. is 5700051 or 5700671

 

[modem7]

I will try your trouble-shooting tips for the 8255.

And perform the same experiment that I did in post #9 ?

What is the best method to ground the PC6 and PC7 pins?

What I use is shown at [here] - two IC hooks (supplied with wire) that I have joined.
To take a pin low, the second hook is connected to the chip's ground pin.
To take a pin high, the second hook is connected to the chip's +5V pin.

 

[pinkdonut666]

I found a +12V short in one of the tantalum caps. Removed the bad cap and found that the motherboard still did not power up with tones.

did you replace the cap or just leave it out? leaving it out could cause issues.

at least with my 5150, when it had bad ram in bank 0 it wouldn't post. it would just sit there are behave like it was dead. i just piggybacked a 4116 ram chip on each chip of bank 0 until it fired up, then it started giving me 201 errors for the other banks. I actually ended up replacing every single MOSTEK ram IC on my 16-64k board because I kept having more and more die one after the other and got fed up with the 201 errors.

Don't get me wrong, it could be something more complicated and the diagnostic roms are a good tool. But I am lazy and burning roms & wiring jumpers to test circuits is alot of work. so I usually try the very simple things first, and i've been lucky so far.

Hopefully you can get this board up and running. just my $0.02

 

[Old_hitech]

Here are some updates:

The ROM adapter that came with my Supersoft ROM is shown in the attached picture.

I did a power on with the P8 plug disconnected and got a similar result (hear speaker noise at power on, but no beeps after that, monitor screen goes from white background to dark, but no cursor or text on screen)

My original BIOS ROM is a 5700671, but I have tried a 1501476 and get the same no-start, no beeps result.

I have not tried the 8255 tests as I am trying to remember where I put my micro-clip wires.

The cut one end of the cap to remove the short. I did this for all +12V caps. I have not yet installed new caps.

I did try a different setting for SW1 (1-off, 2-on, 3-off, 4-off, 5-on, 6-off, 7-off, 8-on) and SW2 (1 thru 5 - on, 6,7,8-off) for a 64K system and ran the Supersoft ROM diagnostics (see picture). I pretty much got the same Parity and bad memory bit result. The test still ran for about 3minutes and checked memory above 64K (A0000). Not sure why this happens.

I thought it would be a good idea to get another 16K-64K motherboard so I bought a 5150 rev A computer that was on sale locally. It also powered on with no beeps nothing on monitor screen. I removed all the cards and checked the power supply (good) and motherboard for shorts (good - no shorts). I inserted the Supersoft ROM and installed the video card and got the results shown in the picture (Refresh error). I referenced the Supersoft ROM manual and it simply states “replace the timer” — the timer is soldered on, so it is not that easy. Anyone have experience/recommendations with/for this result?

 

[modem7]

The ROM adapter that came with my Supersoft ROM is shown in the attached picture.

To determine the range of ROM's that would work with it, a circuit digram would be required, or a pointer to a web page that has the information.

I thought it would be a good idea to get another 16K-64K motherboard so I bought a 5150 rev A computer that was on sale locally. It also powered on with no beeps nothing on monitor screen. I removed all the cards and checked the power supply (good) and motherboard for shorts (good - no shorts). I inserted the Supersoft ROM and installed the video card and got the results shown in the picture (Refresh error). I referenced the Supersoft ROM manual and it simply states “replace the timer” — the timer is soldered on, so it is not that easy. Anyone have experience/recommendations with/for this result?

Please create a separate thread for that, because two-threads-within-one-thread makes for confusing reading.

(Note: Do not rush off to replace the timer chip. The timer chip is only part of the refresh circuitry.)

I have not tried the 8255 tests as I am trying to remember where I put my micro-clip wires.

U36 8255 PARITY DETECTED
---------------------------

For now, what do you measure (high or low) on the PC6 and PC7 pins ?

I did try a different setting for SW1 ...

16K CRITICAL MEMORY REGION
--------------------------------

* On a 5150 motherboard, the memory switches on switch bank SW1 do not disable/enable RAM banks.
* From my experiments, the Supersoft ROM only reads the settings of the switches on switch bank SW1 to:
- Display the settings on-screen.
- From the switch 2 setting, determine if an 8087 is fitted.

The cut one end of the cap to remove the short. I did this for all +12V caps. I have not yet installed new caps.

All !
On a 16KB-64KB motherboard, the +12V is used for the 4116 class RAM chips.
You need to replace those caps before further RAM related diagnosis.

 

[Old_hitech]

modem7, Thanks for the information.
I went to order some replacement capacitors on Digikey website and found a candidate replacement for the 10uf 16V cap. --- KEMET Tantalum Through-Hole Capacitors – Radial Dipped
T350, T351, T352, T353, T354, T355 & T356 - These part numbers are close in size - not sure which one to order without measuring the original cap. Anyone have experience ordering this replacement? If some can you provide weblink or vendor\manufacturer\part number?

 

[modem7]

I went to order some replacement capacitors on Digikey website and found a candidate replacement for the 10uf 16V cap. --- KEMET Tantalum Through-Hole Capacitors – Radial Dipped
T350, T351, T352, T353, T354, T355 & T356 - These part numbers are close in size - not sure which one to order without measuring the original cap. Anyone have experience ordering this replacement? If some can you provide weblink or vendor\manufacturer\part number?

(Info for others: On the 16KB-64KB type of 5150 motherboard, the subject caps are 2-legged, not 3-legged.)

It sounds like you are after a close physical match to what is fitted. I have not yet had a need to replace a tantalum cap on either of my two 16KB-64KB motherboards. And if that time comes, I will probably just use something out of my spares box.

 

[mcs_5]

Any reason not to use 10uF 25V X7R MLCC caps?

 

[Old_hitech]

Ok. It has been a while, but I have replaced the +12V caps (all 5 of them) and I am still getting no beeps at power on. I installed the Supersoft ROM and got the following results. Same as before. Not sure what the next step should be. I'm thinking that I should unsolder the Bank 0 RAM and install sockets and new RAM chips. I am hoping that this is the main reason for the fails and that the U36 8255 parity failure will stop appearing. What do you think? Also, where can I get sockets for the Bank 0 RAM?