PDA

View Full Version : IBM 5170 6mhz board won't boot.. "dead"



offensive_Jerk
April 12th, 2016, 12:37 PM
I've got an earlier version IBM 5170 board with the piggy-back ram and 6mhz 286 that appears dead. No video, no beep. I got it as parts and put them all together.

I've tried swapping the PSU with a different PSU that works and same results. I've also tried reseating the socketed chips including the CPU.

It had two banks of RAM filled. I removed one bank and set the jumper to 256k setting and tried swapping the ram modules around and still didn't boot.

I bought some extra RAM on ebay types MT1128-15 4128P and TMS41128A-15NL. Not enough to completely fill one whole bank, but I tried different combinations of new and old and still didn't boot. Unless most of the RAM chips failed, I suspect something else is hosed.

I then installed an ISA post card. Not sure how to read it though. The directions for the card are poorly translated from China. There are two sets of LED displays and it shows 02 01 when I turn the computer on. not sure what that means in IBM POST code lingo. Am I only supposed to read one of the segmented LED displays?
30616

SpidersWeb
April 12th, 2016, 12:49 PM
201 - Memory Test Failed

If you can write an EPROM, I strongly recommend the Supersoft/Landmark diagnostic ROMs as a time saver.

IndirectX9
April 12th, 2016, 01:20 PM
Sounds like some chip(s) in Ram Bank 0 died; I suggest filling the bank with the new RAM first, and after that, there should be video and then a 201 error
If that doesn't work, then there may be a contact problem within the socket(s), and I would check for corrosion or loose socket pins on that bank. Also, Swapping chips within the bank will not identify a faulty one.

offensive_Jerk
April 12th, 2016, 02:31 PM
Yeah I've tried swapping chips until my fingers bled. The only way I can burn a eeprom is with my xt-ode card, would that work?

modem7
April 12th, 2016, 11:39 PM
I then installed an ISA post card. Not sure how to read it though. The directions for the card are poorly translated from China. There are two sets of LED displays and it shows 02 01 when I turn the computer on. not sure what that means in IBM POST code lingo. Am I only supposed to read one of the segmented LED displays?

The output of each 'checkpoint' in the POST of the IBM 5170 is a byte, not a 16-bit word. My understanding of the operation of POST cards containing dual 7-segment displays is that one display shows the 'current' POST code, and the other display shows the previous POST code.

If that is your scenario, then the "02" is the relevant code.

Per below (source: IBM 5170 Technical Reference), what checkpoint 02 does varies according to the the revision of IBM BIOS fitted:
01/10/84: 02 = About to "VERIFY AND CLEAR SHUTDOWN FLAG" (this is a location in the CMOS/RTC chip)
06/10/85: 02 = About to verify that the checksum of the BIOS ROM chips are good.
11/15/85: 02 = About to verify that the checksum of the BIOS ROM chips are good.

offensive_Jerk
April 13th, 2016, 05:31 AM
The output of each 'checkpoint' in the POST of the IBM 5170 is a byte, not a 16-bit word. My understanding of the operation of POST cards containing dual 7-segment displays is that one display shows the 'current' POST code, and the other display shows the previous POST code.

If that is your scenario, then the "02" is the relevant code.

Per below (source: IBM 5170 Technical Reference), what checkpoint 02 does varies according to the the revision of IBM BIOS fitted:
01/10/84: 02 = About to "VERIFY AND CLEAR SHUTDOWN FLAG" (this is a location in the CMOS/RTC chip)
06/10/85: 02 = About to verify that the checksum of the BIOS ROM chips are good.
11/15/85: 02 = About to verify that the checksum of the BIOS ROM chips are good.

According you your site I have the 1st revision BIOS chips. Now I have to look at the CMOS/RTC chip?
30620

offensive_Jerk
April 13th, 2016, 03:47 PM
Also, I have some bios chips I yanked from a dead corroded 286 board somewhere. Could I put those in this AT? I think I remembered something about a low and high bios chip. Which one would go where?
Also, I could probably pull the BIOS chips I have from the last model 5710 type 3. Would it even be worth messing with that since I most likely have some sort of CMOS / RTC error?

modem7
April 14th, 2016, 01:41 AM
POST card

A reminder of my earlier, "if that is your scenario". I cannot guarantee that your POST card is behaving as I suspect that it does. Before proceeding on a presumption of code 02, how about trying the following experiment.

Plug the POST card into your (working) type 3 motherboard. That will have the third revision of IBM BIOS. Disconnect everything from the motherboard except for the POST card and power supply. Turn on the power supply. On my type 3 motherboard, in the same situation, the codes flashing past on the POST card stop at [3F][40] after about 25 seconds.

If I power off then remove one of the two parity RAM chips ("BIT PH" or "BIT PL") and then power on, then I see the displays alternating between [DD][00] and [00][DD]. This is the POST sending [DD] and [00] in a continuous alternating pattern, e.g. [DD] then [00] then [DD] then [00] then [DD] then [00] ...

If you are seeing the above too, then your POST card behaves like mine (current POST code and previous POST code). Therefore, with your type 1 motherboard, the 02 that you see in [02][01] is the final POST code output by the BIOS.

modem7
April 14th, 2016, 01:54 AM
Also, I have some bios chips I yanked from a dead corroded 286 board somewhere. Could I put those in this AT?
Might the board's BIOS chips be part of the "corroded" damage, i.e. in the 5170, you would be replacing ROMs of unknown status (yet to be checksummed) with ROMs of unknown status (could be 'dead').


Also, I could probably pull the BIOS chips I have from the last model 5710 type 3. Would it even be worth messing with that since I most likely have some sort of CMOS / RTC error?
Yes. Worth a try, particularly because it is easy to do.

modem7
April 14th, 2016, 02:10 AM
According you your site I have the 1st revision BIOS chips. Now I have to look at the CMOS/RTC chip?
Assuming a 02 POST code from a 01/10/84 BIOS, my opinion is "not just yet".

Analogies are never perfect, but here goes: Imagine you finding a vintage car (IBM 5170) in a completely unknown working order. Unsuccessfully, you try to start the car - the starter motor does not turn at all. You suspect the battery. You bring out your trusty voltmeter, and measure battery voltage (CMOS/RTC chip) from the cigarette lighter socket (CPU). You measure 0 volts. A dead battery may be the cause, but so are the things between the battery (CMOS/RTC chip) and the lighter socket (CPU), e.g. cigarette lighter fuse, damaged cigarette lighter wiring.

I think that you should proceed in obtaining the Supersoft/Landmark Diagnostic ROMs, and seeing what they indicate.

offensive_Jerk
April 15th, 2016, 12:22 AM
I swapped the BIOS chips with my other AT with the type 3 motherboard. Nada.
I plugged the post card back in and this time the display shows 0302.
3065130652

offensive_Jerk
April 15th, 2016, 01:05 AM
POST card

A reminder of my earlier, "if that is your scenario". I cannot guarantee that your POST card is behaving as I suspect that it does. Before proceeding on a presumption of code 02, how about trying the following experiment.

Plug the POST card into your (working) type 3 motherboard. That will have the third revision of IBM BIOS. Disconnect everything from the motherboard except for the POST card and power supply. Turn on the power supply. On my type 3 motherboard, in the same situation, the codes flashing past on the POST card stop at [3F][40] after about 25 seconds.

If I power off then remove one of the two parity RAM chips ("BIT PH" or "BIT PL") and then power on, then I see the displays alternating between [DD][00] and [00][DD]. This is the POST sending [DD] and [00] in a continuous alternating pattern, e.g. [DD] then [00] then [DD] then [00] then [DD] then [00] ...

If you are seeing the above too, then your POST card behaves like mine (current POST code and previous POST code). Therefore, with your type 1 motherboard, the 02 that you see in [02][01] is the final POST code output by the BIOS.


After swapping the bios and getting code 03, I tried the above. My post card gave the exact results you described.

modem7
April 15th, 2016, 01:40 AM
After swapping the bios and getting code 03, ...
Good. That is what we expect for the third revision of BIOS - IBM have the "VERIFY AND CLEAR SHUTDOWN FLAG" check at 03.


... I tried the above. My post card gave the exact results you described.
We are now 100% confident in the interpretation of your card's display.

offensive_Jerk
January 5th, 2017, 08:52 PM
I think that you should proceed in obtaining the Supersoft/Landmark Diagnostic ROMs, and seeing what they indicate.

OK.......So...... (necropost..I know it's been a long time, but I finally got around to playing with this.)

Current setup: Only a monochrome card is inserted, and the switch flipped to the rear position on the motherboard. I have only one bank of RAM inserted from when I messed with it last time.

I had someone burn and send me the Supersoft ROMS. I put them into their respective sockets.
Turn the board on, and......Nothing. Keyboard lights blink, go out, nothing else happens. The power LED does illuminate, however. I then swap the bank of ram into the other bank and insert other RAM into the first bank. Now, all RAM banks are populated.

Flip the power. Same thing.... Nothing.

I then scratch my head and consider what to do next, so on a whim, I removed the monochrome card and power up with no video card at all installed. No cards are plugged in.
Flip the switch, and I get an alarm type sound from the speaker. Two-tone beep from the speaker that repeats very quickly.

I then swap out for a VGA card, and played with different combinations on the motherboard switch. Motherboard continues to beep with the VGA card.

Put a different monochrome card in (Persyst Mono II) and connect the 5151 monitor just in case the other MDA card was bad.
Flip the switch and, nothing. No sounds, no beeps, no video.

I put the original BIOS chips back in and powered up without a video card to see what happened. No beeps.

So, in short, when there is no video card installed, or a VGA card installed, the computer beeps when the Supersoft BIOS is installed. With the IBM BIOS, I cannot get it to do anything.

Another Question: Would it hurt anything to put a faster 286 in the board (10 or 8mhz) just to see if the processor is bad?

modem7
January 6th, 2017, 03:03 AM
With the IBM BIOS, I cannot get it to do anything.
Via a POST card, you have already deduced that the POST in the IBM BIOS fails on a very early test (test 02 in first BIOS revision, test 03 in later BIOS revisions).
We expect no beeps nor video, because that is the nature of failures of early tests in IBM BIOS' - the POST simply halts the CPU.


I had someone burn and send me the Supersoft ROMS. I put them into their respective sockets.
Turn the board on, and......Nothing. Keyboard lights blink, go out, nothing else happens. The power LED does illuminate, however. I then swap the bank of ram into the other bank and insert other RAM into the first bank. Now, all RAM banks are populated.
Note that the Supersoft ROM diagnostics display something on-screen even if the motherboard has no RAM fitted.


I then scratch my head and consider what to do next, so on a whim, I removed the monochrome card and power up with no video card at all installed. No cards are plugged in.
Flip the switch, and I get an alarm type sound from the speaker. Two-tone beep from the speaker that repeats very quickly.
With my good early 5170 motherboard, if I insert the Supersoft ROM diagnostics, remove the video card, then power up the motherboard, I hear the beep pattern of [six hi-lo tones followed by a short tone]. That is expected. It is the Supersoft ROM diagnostics telling me that it could not initialise a compatible video card. I know that because the beep patterns for the AT version of the Supersoft ROM diagnostics are listed on pages 40 and 41 of the Supersoft manual at [here (http://minuszerodegrees.net/supersoft_landmark/Supersoft%20Landmark%20ROM.htm)].

If you listen long enough, is that what you hear - six of those "Two-tone" beeps followed by a single short beep? Then, the same pattern of beeps repeats over and over.


I then swap out for a VGA card, and played with different combinations on the motherboard switch. Motherboard continues to beep with the VGA card.
No point in trying a VGA card. The Supersoft ROM diagnostics does not support it. I would expect the same symptoms as if no video card was installed.

Note too that the Supersoft ROM diagnostics are not reliant on motherboard switches/jumpers being set in any particular way.


Put a different monochrome card in (Persyst Mono II) and connect the 5151 monitor just in case the other MDA card was bad.
Flip the switch and, nothing. No sounds, no beeps, no video.
The lack of beeps suggests that the Supersoft ROM initialised the card, then started running its tests (but for some unknown reason, you see nothing on the display).


I put the original BIOS chips back in and powered up without a video card to see what happened. No beeps.
As expected. (Per my first comment above.)


Another Question: Would it hurt anything to put a faster 286 in the board (10 or 8mhz) just to see if the processor is bad?
No.

Stone
January 6th, 2017, 04:39 AM
It had two banks of RAM filled. I removed one bank and set the jumper to 256k setting and tried swapping the ram modules around and still didn't boot.

I bought some extra RAM on ebay types MT1128-15 4128P and TMS41128A-15NL. Not enough to completely fill one whole bank, but I tried different combinations of new and old and still didn't boot. Unless most of the RAM chips failed, I suspect something else is hosed.
Since it only takes one bad chip to kill the bank you really need at least one good bank of known good chips to really be able to determine anything concrete. Otherwise you might just be spinning your wheels.

offensive_Jerk
January 6th, 2017, 07:03 AM
Since it only takes one bad chip to kill the bank you really need at least one good bank of known good chips to really be able to determine anything concrete. Otherwise you might just be spinning your wheels.

Wasn't that the point of getting the Supersoft ROMS?


Note that the Supersoft ROM diagnostics display something on-screen even if the motherboard has no RAM fitted.

offensive_Jerk
January 6th, 2017, 12:41 PM
If you listen long enough, is that what you hear - six of those "Two-tone" beeps followed by a single short beep? Then, the same pattern of beeps repeats over and over.


Yes, this is what I hear. Anything else I can try? The 10mhz cpu didn't work either. I'll try a CGA card/Monitor and see what happens, but after that I'm stumped.


https://youtu.be/DGtDVucYrGQ

modem7
January 6th, 2017, 12:56 PM
I just tried the combination of the Supersoft ROMs and a clone MDA card in one of my early (and functional) 5170s, and I am hearing the 'no video card, or cannot initialise video card' beep pattern.
I then tried an IBM CGA card and an IBM EGA card - same beep pattern.
!!!!!
So right now, I appear to be seeing (hearing) what you are. Of course, I am going to do some experimentation to work out what is going on.

offensive_Jerk
January 6th, 2017, 02:13 PM
I just tried the combination of the Supersoft ROMs and an IBM MDA card in one of my early (and functional) 5170s, and I am hearing the 'no video card, or cannot initialise video card' beep pattern.
I then tried an IBM CGA card and an IBM EGA card - same beep pattern.
!!!!!
So right now, I appear to be seeing (hearing) what you are. Of course, I am going to do some experimentation to work out what is going on.
I d been fooling with this thing. I can't get it to boot with the ibm chs card or a Persydtm color II. Just makes the alarm noise

modem7
January 6th, 2017, 02:42 PM
I d been fooling with this thing. I can't get it to boot with the ibm chs card or a Persydtm color II. Just makes the alarm noise
I am still experimenting, but a few times, I fell for the trap of not waiting enough time. The AT version of the Landmark ROMs takes about 20 seconds before it displays anything.

modem7
January 6th, 2017, 03:09 PM
Background: If I am using the Landmark ROMs, it is normally the XT ones, and I do no bother with a speaker connection.

After experimentation with the AT version of the Landmark ROMs, some preliminary results follow:


VIDEO BEEP PATTERN

Even on a fully functional IBM 5170 (of type 1 motherboard), using a compatible video card, I get the 'no video card, or cannot initialise video card' beep pattern. So, I am seeing a display (after the 20 second delay), and the Landmark tests run successfully to completion. It is just that for whatever reason, I am getting the misleading video beep pattern at the start. It happened for:
* IBM CGA card
* IBM EGA card (note: EGA works but the display is poor)
* Clone MDA card
* Clone CGA card

For now, I consider this to be a bug in the AT version of the Landmark ROMs. Need to verify the above on other 5170 motherboards that I have.


IBM MDA

The IBM MDA card results were interesting. I have two spares, both functional (tested today). I do not hear the 'no video card, or cannot initialise video card' beep pattern. I do not see a display. Both IBM MDA cards present the same symptoms. If I then simply substitute my clone MDA card, beeps and display happen.

Above verified on a second 5170 motherboard.

Surprising to me was that if there was going to be a problem, I would have expected it to have been with the clone MDA card and not the IBM one.

Presumably, some clone MDA cards (being better clones that mine) will behave the same as for the IBM MDA.

offensive_Jerk
January 6th, 2017, 04:33 PM
So odd...
That beeping totally threw me off. I thought I waited long enough, but I guess I didn't.
So, I got something on the screen!!!

It passes all the tests, then reports an error, then another error, but it doesn't tell me what.


************EDIT*******
I just realized there is a keyboard test while reading the manual. I don't think I had the keyboard plugged in. Going to rerun the test.


https://youtu.be/WgQu9axaaQs

offensive_Jerk
January 6th, 2017, 05:07 PM
Ignore that first video. I ran it again with the battery plugged in. I got a LOT of errors!

https://farm1.staticflickr.com/459/32034187011_6771ebf389_z_d.jpg

modem7
January 6th, 2017, 05:43 PM
So odd...
That beeping totally threw me off.
It certainly distracted me.


Ignore that first video. I ran it again with the battery plugged in.
Hmmm. Shouldn't need one.

Is that repeatable? I.e. If battery missing, PROTECTED MODE CPU stalls. If battery present, PROTECTED MODE CPU fails but tests continue.


I got a LOT of errors!
But some of those are probably understandable (e.g. no floppy controller + drive plugged in).

The ones always expected to pass are the PROTECTED MODE CPU test, the REAL TIME CLOCK INTERRUPT, and CMOS RAM TEST.

All three of those correlate nicely with the failure of the VERIFY SHUTDOWN STATUS BYTE test within the POST of the IBM BIOS ROM.

Supersoft test: PROTECTED MODE CPU

Within the MC146818 CMOS RTC chip is some RAM, and one of the RAM bytes is used to store the SHUTDOWN STATUS BYTE (shown [here (http://stanislavs.org/helppc/cmos_ram.html)]). One use of that byte is when the BIOS switches the CPU out of protected mode (done by resetting the CPU via the keyboard controller, per [here (http://minuszerodegrees.net/5170/motherboard/5170_reset_flow.jpg)]).

Supersoft test: REAL TIME CLOCK INTERRUPT

An interrupt generated by the MC146818 CMOS RTC chip.

Supersoft test: CMOS RAM TEST

RAM within the MC146818 CMOS RTC chip.

IBM ROM POST test: VERIFY SHUTDOWN STATUS BYTE

RAM within the MC146818 CMOS RTC chip.


So everything is pointing to the CMOS MC146818 RTC chip, or its supporting circuitry. (partial diagram (http://minuszerodegrees.net/images2/5170_battery_circuitry.jpg))

Does a very close visual inspection of the MC146818 RTC chip and the surrounding area reveal anything odd ?

offensive_Jerk
January 8th, 2017, 11:40 AM
Is that repeatable? I.e. If battery missing, PROTECTED MODE CPU stalls. If battery present, PROTECTED MODE CPU fails but tests continue.
Yes, it is repeatable. I just tested this again.


Does a very close visual inspection of the MC146818 RTC chip and the surrounding area reveal anything odd ?
Not that I can "see" visually :(

https://farm1.staticflickr.com/738/32193022405_a59f7b159f_k_d.jpg

modem7
January 8th, 2017, 11:47 PM
After experimentation with the AT version of the Landmark ROMs, some preliminary results follow:
After doing more experimentation, I have listed the results in the notes at [here (http://minuszerodegrees.net/supersoft_landmark/5170/Supersoft%20Landmark%20ROM%20-%20IBM%205170.htm)].

modem7
January 8th, 2017, 11:53 PM
Does a very close visual inspection of the MC146818 RTC chip and the surrounding area reveal anything odd ?

Not that I can "see" visually
That general area sits beneath the battery, and therefore is often the first area to be affected by a leaking battery.

modem7
January 9th, 2017, 12:29 AM
Is that repeatable? I.e. If battery missing, PROTECTED MODE CPU stalls. If battery present, PROTECTED MODE CPU fails but tests continue.

Yes, it is repeatable. I just tested this again.

So,
Battery absent: During the PROTECTED MODE TEST, the Supersoft/Landmark ROMs restart from scratch (per your video in post #23).
Battery present: The PROTECTED MODE TEST completes (the test takes about 25 seconds), then the remaining tests are done.

I have two type 1 motherboards, and for both, the presence/absence of a battery has no effect on the output of the Supersoft/Landmark ROMs.

At first I thought, I know what the cause is. This will be failure of the area that I have marked as '+5V SWITCH' in the diagram at [here (http://minuszerodegrees.net/images2/5170_battery_circuitry.jpg)] - with no battery present, the CMOS/RTC circuitry is not getting power when the motherboard is powered on.

But I am not convinced, e.g. if the switch had failed, then with a battery present, surely the motherboard would start up normally with the IBM BIOS ROMs fitted.

Anyhow, the switch can be quickly ruled out. Do you have the ability to measure DC voltage? If so, remove the battery, then measure pin 24 of the MC146818P chip (U117) when the motherboard is powered on. You should measure about +5 volts.

offensive_Jerk
January 9th, 2017, 04:56 PM
So,
Battery absent: During the PROTECTED MODE TEST, the Supersoft/Landmark ROMs restart from scratch (per your video in post #23).
Battery present: The PROTECTED MODE TEST completes (the test takes about 25 seconds), then the remaining tests are done.

I have two type 1 motherboards, and for both, the presence/absence of a battery has no effect on the output of the Supersoft/Landmark ROMs.

At first I thought, I know what the cause is. This will be failure of the area that I have marked as '+5V SWITCH' in the diagram at [here (http://minuszerodegrees.net/images2/5170_battery_circuitry.jpg)] - with no battery present, the CMOS/RTC circuitry is not getting power when the motherboard is powered on.

But I am not convinced, e.g. if the switch had failed, then with a battery present, surely the motherboard would start up normally with the IBM BIOS ROMs fitted.

Anyhow, the switch can be quickly ruled out. Do you have the ability to measure DC voltage? If so, remove the battery, then measure pin 24 of the MC146818P chip (U117) when the motherboard is powered on. You should measure about +5 volts.

I get no voltage reading off of pin 24 of U117 when the system is powered up.

offensive_Jerk
January 9th, 2017, 04:57 PM
Just for giggles I plugged the battery back in when the system was powered off and did get a 5v reading off of pin 24 of U117

mikey99
January 9th, 2017, 05:32 PM
I also have a 'dead' AT motherboard ......this one shows very minor battery damage around the area of the MC146818 chip.
Mainly just some greenish color on several of the chips pins. No obvious damaged traces. It may be that the MC146818 is easily
susceptible to electrolyte damage from the battery.
This thread has rekindled my interest in repairing this board ...so I will be burning the SuperSoft ROMS and checking this
out.....

modem7
January 9th, 2017, 10:51 PM
This thread has rekindled my interest in repairing this board ...so I will be burning the SuperSoft ROMS and checking this
out.....
Be sure to open up a new thread.

modem7
January 9th, 2017, 10:53 PM
I get no voltage reading off of pin 24 of U117 when the system is powered up.
Definitely abnormal. But I cannot match that to the symptoms.

But perhaps check again. There have been times when I have been caught out when my multimeter probe did not pierce the oxide layer on the pin.


Just for giggles I plugged the battery back in when the system was powered off and did get a 5v reading off of pin 24 of U117
That will be the battery voltage minus about 0.8V
At least it verified that you were measuring the right pin.

So, if you check again, and again see no +5V on pin 24 when no battery is present, and the motherboard is powered, then do a close visual inspection of the components that make up the +5V switch. Include the solder side of the board:

http://minuszerodegrees.net/temp/3/temp_76gfj4w8734.jpg

offensive_Jerk
January 10th, 2017, 06:40 AM
I checked that area again, and nothing looks odd. When I got it, I believe it was the origonal battery and I don't think it leaked at all.
Looked on the solder side too and didn't see anything on those points except for possible cold solder joints. I reflowed the solder on the areas marked in the image above just in case. No change.

Here's a video of what happens when I put a mulitmeter on pin 24 of U117 with the battery plugged in and turn the system on.

The voltage changes from 4.76v when powered off, to 0v when I flip the power on.



https://youtu.be/o531UWmmD8c

mikey99
January 10th, 2017, 07:19 AM
Check if you have +5v making it to the 5v switch circuit with the board powered on....... you can check for +5v on R25.
If +5v is present then you may have a bad component in that switch circuit.

offensive_Jerk
January 10th, 2017, 09:04 AM
Check if you have +5v making it to the 5v switch circuit with the board powered on....... you can check for +5v on R25.
If +5v is present then you may have a bad component in that switch circuit.

Yup, R25 has 4.99v.

offensive_Jerk
January 10th, 2017, 09:22 AM
Also R25 does not test as 51K ohms when I test it in-circuit... it measures 8.5k....
R24 is right on the money at 10k but R26 is also at 8.3k.....

mikey99
January 10th, 2017, 09:53 AM
The resistance measurements sound about right considering the other components in the circuit.

With the battery disconnected, and the board powered on, also check for 5V on the collector of Q2......
That would be the pin on the right towards Q1 (shown in the picture above). That's the output of the
switch circuit.

AND are you sure you don't have 5v on Pin 14 of U108 OR Pin 24 of U117 ?

offensive_Jerk
January 10th, 2017, 10:01 AM
Look at my last YouTube video. I didn't move the leads on my multimeter at all when I turned the system on and off. The meter was on pin 24 of u117

offensive_Jerk
January 10th, 2017, 10:31 AM
OK, here's what I've tested and the voltages. I put the pin numbers on the IC's in this pic to make sure I have it right, haha. Correct me if I have the pins wrong.

Anyway, pin 28 on U117 reads 0v when board powered on.
Pin 14 on U108 reads 2.5v when board powered on and battery disconnected.
Collector of Q2 reads 5v (That would be the pin on the right towards Q1 shown in the picture above)

https://farm1.staticflickr.com/644/31423324403_b4a2decba7_o_d.png

mikey99
January 10th, 2017, 10:51 AM
5v on Q2 collector is good news, but the pins you checked on the two chips are incorrect.

Check the datasheets on those two chips.... pin 24 on U117 is directly across from Pin 1
And Pin 14 on U108 is directly across from Pin 1 . Pin numbering runs in a 'U' shape.

offensive_Jerk
January 10th, 2017, 10:58 AM
5v on Q2 collector is good news, but the pins you checked on the two chips are incorrect.

Check the datasheets on those two chips.... pin 24 on U117 is directly across from Pin 1
And Pin 14 on U108 is directly across from Pin 1 . Pin numbering runs in a 'U' shape.

Ugh, must have read some bunk information. Boy do I feel dumb, providing wrong information. At least we got that sorted out.

Anyway, the pins check out OK, both have 5v on both of the chips.

mikey99
January 10th, 2017, 04:57 PM
Hey, don't worry about it, at least now we know the +5 supply voltages are okay on those two chips !
This is all a learning game..... :-)

It would be a good idea to check the OSC (blue) line in the diagram Pin 10 on U108.....if you have a scope or logic probe.

modem7
January 10th, 2017, 10:42 PM
It would be a good idea to check the OSC (blue) line in the diagram Pin 10 on U108.....if you have a scope or logic probe.
Using a DC voltmeter, that pin measures about 2.3V on my motherboard (with the 2.3V being the average of the clock signal swinging between approx. 0V and approx. 5V).

modem7
January 10th, 2017, 10:54 PM
On the MC146818, the pins to check (motherboard powered on) are:

-----------------------------------------
PIN 24 - VDD (power)

So, you have just concluded that pin 24 is in fact getting +5V via the switch.

-----------------------------------------
PIN 13 - CHIP ENABLE (active low)

This pin should be at about 0V. (I.e. chip is enabled.)
Looking at the photo in post #41, you have already measured that pin (incorrectly labelled "24") as 0V.

-----------------------------------------
PIN 18 - RESET (active low)

This pin should be at about +5V. (I.e. the chip not being held in a reset state.)

Is it about +5V ?

-----------------------------------------

offensive_Jerk
January 12th, 2017, 05:23 PM
On the MC146818, the pins to check (motherboard powered on) are:

-----------------------------------------
PIN 24 - VDD (power)

So, you have just concluded that pin 24 is in fact getting +5V via the switch.

-----------------------------------------
PIN 13 - CHIP ENABLE (active low)

This pin should be at about 0V. (I.e. chip is enabled.)
Looking at the photo in post #41, you have already measured that pin (incorrectly labelled "24") as 0V.

-----------------------------------------
PIN 18 - RESET (active low)

This pin should be at about +5V. (I.e. the chip not being held in a reset state.)

Is it about +5V ?

-----------------------------------------
Yup, pin 18 is 4.9v

modem7
January 12th, 2017, 08:48 PM
There are other inputs to the MC146818, but they are not static - a multimeter is not good enough.
At this point, tools such as logic state analysers and logic probes are required.

Of course, you could always take a punt and replace the MC146818 with new (IC socket recommended). That's easy for me to say; I have spares, sockets, and soldering skill.

offensive_Jerk
January 13th, 2017, 06:56 AM
There are other inputs to the MC146818, but they are not static - a multimeter is not good enough.
At this point, tools such as logic state analysers and logic probes are required.

Of course, you could always take a punt and replace the MC146818 with new (IC socket recommended). That's easy for me to say; I have spares, sockets, and soldering skill.

I have a oscilloscope I bought at a garage sale for $1, but I don't know how to even use one, so that's out.

I actually am going to just replace the MC146818 chip. Fingers crossed! Now I just need a goat to sacrifice.

Chuck(G)
January 13th, 2017, 07:24 AM
If you can find a working Dallas RTC chip (e.g. DS1287), one can usually substitute that for the MC146818 and eliminate a bunch of potential problems in the support circuitry.

offensive_Jerk
January 13th, 2017, 08:08 AM
If you can find a working Dallas RTC chip (e.g. DS1287), one can usually substitute that for the MC146818 and eliminate a bunch of potential problems in the support circuitry.

Interesting. I suppose you mean by "working" the battery in the chip needs to be good?

Chuck(G)
January 13th, 2017, 09:18 AM
Yes, that's the idea, though I suspect even a dead-battery DS1287 will enable your machine to get past the CMOS issues when powered up. Consider that the chip eliminates the issues of the oscillator crystal and battery backup circuitry, as well as the CMOS clock chip itself.

Your basic "shotgun" approach.

mikey99
January 13th, 2017, 09:20 AM
Is a DS1287 a direct replacement for the existing MC146818 chip ? If so then adding a socket when you remove the
MC146818 chip will allow you to plug in either one.

Chuck(G)
January 13th, 2017, 09:27 AM
Works for me on my Faraday "A Tease" 6MHz 80286 board, which normally takes a MC146818, so yes, it should.

yuhong
January 13th, 2017, 02:49 PM
Maxim article on this:
http://pdfserv.maximintegrated.com/en/an/app521.pdf

modem7
January 13th, 2017, 04:12 PM
Maxim article on this:
http://pdfserv.maximintegrated.com/en/an/app521.pdf
I see that that covers the DS12887, not the DS1287. The DS1287 would have been out-of-production.

By my reading of the referenced document and the MC146818 datasheet and the DS12887 datasheet:

MC146818

According to the Motorola MC146818 datasheet that I have, the MC146818 has a "MOTEL" circuit that automatically caters for the MC146818 being placed in circuitry that uses "Motorola type MPU signals" or in circuitry that uses "Competitor type MPU signals". Per the datasheet, the "Competitor" is Intel.

DS12887

The datasheet for the DS12887 indicates that the DS12887 has no equivalent circuit (maybe Motorola patented it), and consequently, the DS12887 needs to be informed of which MPU signalling is being sent to it. It gets that information via pin 1 (which is an unused pin on the MC146818).

Pin 1 of the IBM AT's MC146818 socket is not connected to anything (I just now checked on my type 1 and type 3 motherboards).

However, a pin 1 related statement in the DS12887 datasheet is, "When connected to GND or left disconnected, Intel bus timing is selected. The pin has an internal pulldown resistance of approximately 20K." So the unconnected pin 1 of the IBM AT's MC146818 socket should not be an issue.

BTW: I see that the DS1287, the earlier chip, also uses pin 1 in the same way.

Chuck(G)
January 13th, 2017, 05:59 PM
I thought it more likely that the OP might have a dead or hacked DS1287 kicking around, which is why I suggested it as a test.

offensive_Jerk
January 13th, 2017, 06:19 PM
I ordered a spare chip just in case it was the RTC chip in here that was bad.
I just replaced it, and I think we're making progress!

The sockets I have lying around are 4 pins too big, so I ripped some of the pins out and soldered it to the board. I put the new chip in the socket. (picture 1)

I then ran Supersoft ROMS and I'm still getting some memory errors??? (picture 2)

However, after fiddling with the motherboard, I got it to boot with the IBM roms! I moved the jumper back to the 512k position, but the motherboard only counts 256k?? Do I have to set the base memory with the diagnostic disk? I would have thought the motherboard would know 512k when setting the jumper. (picture 3)


https://farm1.staticflickr.com/575/31452957694_40982b1570_k_d.jpg
https://farm1.staticflickr.com/711/32144465232_679caa3569_k_d.jpg
https://farm1.staticflickr.com/724/32145267892_0ecf1540aa_z_d.jpg

Chuck(G)
January 13th, 2017, 06:31 PM
Really, if you're going to replace sockets, just get some of this stuff:

http://i.ebayimg.com/images/g/OzsAAOSwBLlVAVYC/s-l500.jpg

Cheap, gold-plated, machine-pin. Just snap off what you need--and it leaves the space between the pin rows open. Search ebay for "round female pin header". That way, you ditch the dodgy sockets with no need to keep a bunch of different sizes.

offensive_Jerk
January 13th, 2017, 06:33 PM
Also I tried plugging in a Fixed Disk AT card, and the system wouldn't boot. Probably a short. I measured the resistance on the capacitor at C46 and it's measures 0. The other two caps to the left and right are the same 22uf and they don't measure 0 and make my meter beep.

Why is that cap like a hard black plastic? Could I just replace with an electrolytic axial 22uf capacitor?
https://farm1.staticflickr.com/281/31484308373_2f9909dec1_k_d.jpg

offensive_Jerk
January 13th, 2017, 06:34 PM
Really, if you're going to replace sockets, just get some of this stuff:

http://i.ebayimg.com/images/g/OzsAAOSwBLlVAVYC/s-l500.jpg

Cheap, gold-plated, machine-pin. Just snap off what you need--and it leaves the space between the pin rows open. Search ebay for "round female pin header". That way, you ditch the dodgy sockets with no need to keep a bunch of different sizes.

That's a great idea! Ordering now.

offensive_Jerk
January 13th, 2017, 07:41 PM
So the Fixed Disk/Floppy card that came with this must be shorted. I compared that suspect capacitor (C46, marked KT322) to another fixed disk card in my other AT and it does not read as 0 ohms.
I temporarily swapped the floppy cards out and got it to boot the diagnostic disk.

I set the system to 512k base memory. After rebooting, I'm getting a memory error:
040002 0004 201-Memory Error

I'll check minuszerodegrees and see what that's about.

Edit

0000 0000 0000 0100 > so bit 2 is failing on the 256k bank? I'll see if I can find out which chip that is.

modem7
January 13th, 2017, 08:51 PM
I ordered a spare chip just in case it was the RTC chip in here that was bad.
I just replaced it, and I think we're making progress!
So, faulty MC146818P. I have annotated that at [here (http://minuszerodegrees.net/vcf_motherboard_failure_history.htm)].


However, after fiddling with the motherboard, I got it to boot with the IBM roms! I moved the jumper back to the 512k position, but the motherboard only counts 256k?? Do I have to set the base memory with the diagnostic disk? I would have thought the motherboard would know 512k when setting the jumper
As you have discovered, the 5170 motherboard is not that 'smart'.


Could I just replace with an electrolytic axial 22uf capacitor?
Yes, or another tantalum.

BTW. Note that doing the measurement that way, you are measuring the resistance of the line - the short could be on a diffrent component on the same line. To prove that the capacitor is short, you need to remove one of its legs from the circuit. The particular capacitors that you refer to are a relatively common failure.


I set the system to 512k base memory. After rebooting, I'm getting a memory error:
040002 0004 201-Memory Error
I'll check minuszerodegrees and see what that's about.
Edit
0000 0000 0000 0100 > so bit 2 is failing on the 256k bank? I'll see if I can find out which chip that is.
40002 --> Address 256K
The first 256K bank is okay; failure encountered when testing the second.
0004 --> Yes, bit 2

offensive_Jerk
January 13th, 2017, 09:47 PM
Ok. The system now POSTs with ZERO errors!!!!!!!

Thanks guys!!!!!!!!!!!!

offensive_Jerk
January 14th, 2017, 03:23 AM
Ok. The system now POSTs with ZERO errors!!!!!!!

Thanks guys!!!!!!!!!!!!

Well, zero errors with the ibm Roms. Haven't tried with super soft again since it's booting.

offensive_Jerk
January 20th, 2017, 07:34 AM
Also I tried plugging in a Fixed Disk AT card, and the system wouldn't boot. Probably a short. I measured the resistance on the capacitor at C46 and it's measures 0. The other two caps to the left and right are the same 22uf and they don't measure 0 and make my meter beep.

Why is that cap like a hard black plastic? Could I just replace with an electrolytic axial 22uf capacitor?
https://farm1.staticflickr.com/281/31484308373_2f9909dec1_k_d.jpg

I just wanted to add that I replaced the three capacitors on the bottom of the Fixed Disk controller and the system now boots with the card inserted. No more short!

Anonymous Coward
January 28th, 2017, 09:56 PM
I have the same post card as the OP, and a bad XT-286 motherboard that shows "04 03" (bad PIT?), and a type 3 AT board that shows "11 32". Any ideas about what might be causing the problems?

modem7
January 28th, 2017, 10:45 PM
I have the same post card as the OP, and a bad XT-286 motherboard that shows "04 03" (bad PIT?), and a type 3 AT board that shows "11 32". Any ideas about what might be causing the problems?
How about you create two new threads. Otherwise, people like me get easily confused trying to sort out which post belongs to which problem.