IBM 5170 Issue

[flyinghi5]

Hey all, I know I'm new to this forum and found this when I was searching for help on my 5170. I believe it is one of the first type 1 boards since it has several jumpers on the back of the board for last minute changes before release.

My issue with this board is it is failing on a DMA failure. I get post code 06 with the newer IBM BIOS. It had the first version of the IBM BIOS but I wanted to be able to use a 1.44 disk drive but it failed on DMA also. I have replaced the DMA chips twice, thinking the first set I replaced them with were also defective. I have replaced U5 and U113 thinking this was the route the BIOS took during the POST. I had no luck there either.

I have tried looking at the source code but I'm not real proficient with assembly and have tried to follow it. It appears it is basically writing FF, 55 and AAh to DMA address and comparing the write to ensure it is correct. I assume this is a direct write to the DMA controller. Can someone confirm this for me? Is there a way I can confirm what is being written and read to verify this? I have also followed things on minuszero but burning a set of Supersoft diagnostic ROM but nothing ever shows on the screen. The video card, which is a straight mono card from the 80's, is good because I tried it in a newer system and it worked flawlessly. So in addition to the IBM BIOS stopping at 06 and Supersoft giving me the code for DMA controller and no video, I have to assume the DMA must pass before I can get video??

Any help would be much appreciated.

 

[Trixter]

It seems odd that you would get a DMA error for bad RAM if the RAM itself passes the POST. Try checking www.minuszerodegrees.net for troubleshooting tips. If you have access to an EEPROM burner, you can download the Supersoft diag ROMs for the AT which might give you better POST information.

Worse comes to worse, there are some reasonably-priced motherboards out there: http://www.vcfed.org/forum/showthre...ot-of-reasonably-priced-IBM-5170-motherboards

 

[flyinghi5]

Well I'm not sure it passes a RAM test yet because it never gets that far into the POST. I have burned a set of the Supersoft and that is, in addition to POST card, what tells me about the DMA failure. I have another board that works and that is how I know the RAM is good and I'm actually working on this as a project to learn a little about the workings of the 5170 board. I just don't understand how the BIOS talks to the DMA during the POST. I have a little understanding of the communication between I/O and DMA but not with POST.

 

[modem7]

Hey all, I know I'm new to this forum ...

Welcome to these forums.

I have tried looking at the source code but I'm not real proficient with assembly and have tried to follow it. It appears it is basically writing FF, 55 and AAh to DMA address and comparing the write to ensure it is correct. I assume this is a direct write to the DMA controller. Can someone confirm this for me? ...

I do not know what you mean by "direct write". I/O type writes and writes are being done. Looking at the source code, I agree with your "basically writing FF, 55 and AAh to DMA address and comparing the write to ensure it is correct." There are two DMA controllers, with check/test #6 targeting the first controller, and check/test #7 targeting the second controller.

... Is there a way I can confirm what is being written and read to verify this?

A logic state analyser.

What test equipment do you have? For example, do you have the ability to verify that the RESET pin on the first DMA controller is not stuck high? On the first DMA controller, are you seeing the /IOW and /IOR pins temporarily dropping low (multiple times) during check/test #6 ?

I have also followed things on minuszero but burning a set of Supersoft diagnostic ROM but nothing ever shows on the screen. The video card, which is a straight mono card from the 80's, is good because I tried it in a newer system and it worked flawlessly.

It does not matter that the video card is fully functional. The Supersoft/Landmark diagnostic has certain video card requirements; certain compatibility needs to be met. Per [here], even the IBM MDA is known not to work for the AT version of the Supersoft/Landmark Diagnostic. My clone MDA works, but that does not mean that all clone MDA cards will work.

On the aforementioned web page, note the, "There is a 20 second delay before anything is seen on the display."

So in addition to the IBM BIOS stopping at 06 and Supersoft giving me the code for DMA controller and no video, I have to assume the DMA must pass before I can get video??

For the IBM BIOS', yes.

For the Supersoft/Landmark diagnostic, no. An incompatible video card? Not waiting more than 20 seconds? Motherboard fault causing this issue?

 

[flyinghi5]

Welcome to these forums.

Thanks and thank you for your response. Lots of good info.

I do not know what you mean by "direct write".

I guess what I was trying to ask is does the BIOS write the FF, 55, AA directly to the DMA address or does it write it to RAM and then the controller read it from RAM.

A logic state analyser.

I do not have a logic analyzer per se. Maybe something I need to invest in.

What test equipment do you have?

I do have a Rigol DS1054Z that I know how to use fairly well but not perfect. I don't know everything it can do but it does have, I believe, a two channel logic analyzer that can decode SPI, Parallel, RS232 and I2C. I also have basics like Fluke 87 multimeter, logic probe, solder station etc.
.

For example, do you have the ability to verify that the RESET pin on the first DMA controller is not stuck high?

Actually, I did this and reset goes high for a second or two on startup and goes low.

On the first DMA controller, are you seeing the /IOW and /IOR pins temporarily dropping low (multiple times) during check/test #6 ?

I'm not sure exactly when Supersoft tests the DMA but approx. 10 seconds after powerup I show both of these pins dropping low many times when on a 10us setting. It also shows a period of between 5.8 and 9.8us on /IOR and 15.8 on /IOW. They will go high for several seconds and then repeat.

It does not matter that the video card is fully functional. The Supersoft/Landmark diagnostic has certain video card requirements; certain compatibility needs to be met. Per [here], even the IBM MDA is known not to work for the AT version of the Supersoft/Landmark Diagnostic. My clone MDA works, but that does not mean that all clone MDA cards will work.

OK, that makes sense now. When I use the IBM "black and white" card I get nothing on the screen but when I use an unknown brand monochrome card I get junk showing on the screen.

Appears I'll have to learn how to use the quote system better too.

 

[modem7]

I guess what I was trying to ask is does the BIOS write the FF, 55, AA directly to the DMA address or does it write it to RAM and then the controller read it from RAM.

No, not via RAM. They are I/O reads and writes between the CPU and DMA controller (with various chips between the two).

IBM BIOS - POST (power-on self test)

First, the POST is not a full-functional test. It is a crude test only.

Test/check #6 is not testing to see if the first DMA controller can do DMA operations. Its intention is to test that the controller's registers can be written to and then read back without problem. So, hardly testing any of the chip's functionality at all. Essentially, a 'Can I communicate with the DMA controller, and are its I/O registers okay' test.

Supersoft/Landmark diagnostic

Quoting the user manual, "This test writes, reads, and verifies that the DMA controller registers can hold all possible bit combinations, providing a basic functional test of the DMA controller." I disagree with the use of "basic". This is the same test that the POST in the IBM BIOS is doing.

I do not have a logic analyzer per se. Maybe something I need to invest in.

I remember that mine was quite expensive (relatively speaking).

I do have a Rigol DS1054Z that I know how to use fairly well but not perfect...

A DSO. Most of my diagnosis work is done using my DSO. I only use my logic state analyser when I have to, because it is a pain to set up.

Actually, I did this and reset goes high for a second or two on startup and goes low.

Good.

I'm not sure exactly when Supersoft tests the DMA but approx. 10 seconds after powerup I show both of these pins dropping low many times when on a 10us setting. It also shows a period of between 5.8 and 9.8us on /IOR and 15.8 on /IOW. They will go high for several seconds and then repeat.

The signal send to /IOW drops for every write operation of type I/O. The signal send to /IOR drops for every read operation of type I/O. You should be able to find I/O timing diagrams for the 80286 on the Internet.

But the two signals are common to all chips that the CPU does I/O operations with, and so only a few of your sightings are expected to be associated with the 'test' of the DMA controller. But you now have confidence that the signals are getting through.

Which of the multiple drops of those lines the DMA controller responds to is controlled by the signal coming into the /CS pin ('chip select'). Are you seeing /CS temporarily go low?

And while you're there, do you see a clock signal coming into the CLK pin ?

OK, that makes sense now. When I use the IBM "black and white" card I get nothing on the screen but when I use an unknown brand monochrome card I get junk showing on the screen.

One of my clone CGA cards also displays 'junk' when it is used with the Supersoft/Landmark diagnostic.

 

[flyinghi5]

Modem7, thanks for all the great info. This definitely gives me a path to follow in troubleshooting this board. I will check the /CS pin and coordinate that with the data etc. I do have a good CLK signal to pin 12.

 

[flyinghi5]

Well, based on modem7's comments I checked the DMA chips /CS, /IOR, and /IOW. Seems it may be a timing issue although all the timings appear to be right on. I can never get a read or write signal at the same time as a /CS signal. When using Supersoft, the /CS is always 158ns apart with the /CS coming and going before a /IOR or /IOW signal.

 

[flyinghi5]

I finally had time to sit down and do more troubleshooting on the 5170. As per my last post, modem7 put me into the /CS, /IOR and /IOW lines on the DMA chip. I was indeed getting the signals at the chip but after looking at the timings on the datasheet, I realized they were not at the right time. The /CS would never match with any read or right and then figured it had to be a timing issue.

I proceeded to track things down and everything lead me to the D82284-6 (U82) chip. The chip was replaced and the 5170 is back to life.

Modem7, thanks for your help, it lead me down the right road.

 

[flyinghi5]

I've got one more question concerning capacitors on the 5170. I have a few that need replaced but am having a hardtime with the coding. I understand that K=10% etc. but I also understand that not every manufacture codes are in the same order.

The board has K5K 453, K5J 300, and K2J 270. The thing I'm having issues with is decoding the 1st three. I find thing in searches that state the 1st K=Kemet, 5= 50V, and 2nd K=10%. I find that believable but on the K2J it they say 2=200V and J=5%. I couldn't imagine needing a 200V capacitor on the motherboard. Can anyone confirm this coding. Thanks

 

[modem7]

... but on the K2J it they say 2=200V and J=5%. I couldn't imagine needing a 200V capacitor on the motherboard. Can anyone confirm this coding. Thanks

That is C83, a 27pF ceramic cap used as part of the oscillator circuit for the RTC chip. On an IBM 5170 board that I am looking at right now, its C83 is also marked "K2J" on the rear.

Assuming a KEMET made cap, the datasheet at [here] indicates 200V. Maybe KEMET or their dealer was doing a fire sale of the 200V version due to lack of demand, and the motherboard maker 'bought up big'.

 

[bear]

Or maybe ceramic disc capacitors are, by their very physical characteristics, naturally low-capacitance, high-voltage devices. I can't even find a ceramic disc currently in production and for sale on Mouser that is rated less than 500 V.

 

[Chuck(G)]

I hope that was a typo: 50V disc ceramics are quite common

I don't care for them in precision timing circuits because they can be microphonic.

 

[Malc]

.... I can't even find a ceramic disc currently in production and for sale on Mouser that is rated less than 500 V.

Check these 3 part No's on Mouser: 80-C322C270J2G / 80-C315C270J2G / 80-C317C270J2G I did not check measurements so you will need to check.