XT clone motherboard - Diagnosis
Introduction
As a comparison, let us look at the IBM XT (IBM 5160). IBM released detailed technical information, including the source code for the motherboard (and other) BIOS. So, with the right test equipment, and the following knowledge, it is possible for me to fault find down to component level in an IBM XT.
- Good knowledge of electronics.
- Good knowledge of how computers of that era work.
- Knowledge of how to operate the test equipment.
- Knowledge of where to measure, and what to expect there.
- Knowledge of what the test equipment is telling me.
That changes if, say, long ago, someone swapped out the IBM CGA card for a Google CGA card. I do not have a circuit diagram for the Google CGA card. I cannot fault find that card down to component level. And I am not going to spend the many many hours required to work out what the circuit diagram is. Beyond some basics, such as a good visual inspection, and re-seating chips, I will probably throw the card away. But maybe I am lucky, and the problem on the Google CGA card is easy to find, such as a short-circuited capacitor, stopping the power supply from working. Maybe the card has an oscillator unit, and I can check its output. But without a circuit diagram, I am limited.
XT clone motherboards
The main problem with clones is a lack of detailed technical information, specifically, circuit diagrams.
With a faulty XT clone motherboard (the motherboard proven as being faulty) that appears 'dead', I can do basics, such as a very good physical inspection, re-seating socketed chips, and looking for the presence of +5V (and that +5V being within tolerance).
Above, I used, "motherboard proven as being faulty". That is important. Because a motherboard that appears to be dead might in fact be fully functional, with the symptom being caused by something else.
Examples:
• An ISA card that is faulty in a way that inteferes with motherboard operation. I have experienced that myself.
• Short-circuit in keyboard, resulting in the power supply shutting down.
• Short-circuit in a floppy or hard drive, resulting in the power supply shutting down.
• Short-circuit in an ISA card, resulting in the power supply shutting down.
So, let us pretend that someone has sent to me an XT clone motherboard that is described as 'appears dead'.
How will I deal with it ?
Important
In the following text, you will see the term 'known-good' appear often. It is important that everything you use to diagnose the motherboard is proven as working.
• For example, if you buy a replacement 8088 CPU online, maybe that CPU is faulty, or not even an 8088 (i.e. rebadged something else). So, what you could be doing is removing a working 8088 and putting a faulty/bad CPU in its place.
• For example, if you create for yourself, a Rudd's Diagnostic ROM, maybe you did that wrong.
1. Quick inspection.
At this very early stage, I do just a quick inspection, looking for obvious things (like the 8088/V20 CPU missing from its socket, like missing BIOS ROM).
2. Verification of 'appears dead'
I have a known-good set of: XT-class PSU, CGA card, CGA monitor, and speaker. <----------- VERY IMPORTANT: KNOWN GOOD
I connect those up to the motherboard, and set the motherboard's two video switches to the CGA setting. I turn on the PSU. I see nothing on the monitor. I hear no single beep from the speaker, either early on, or later, when one would be expected at the end of the POST. I hear no 'intelligent' series of beeps, the motherboard informing me of a problem.
At this time, I check the +5V line on the motherboard, verifying that I see +5V, and that the +5V is within tolerance (typically +/- 5%). If the +5V is missing, or out of tolerance, then because I have a known-good PSU, it must be an overload or underload situation, and so I investigate that.
I measure the +5V as present, and within tolerance, and so I continue as follows.
( INFO: I am using a known-good PSU, and so I see no point in verifying that the PSU is generating the POWER GOOD signal. )
3. Ruud's Diagnostic ROM
I possess a known-good Ruud's Diagnostic ROM (RDR). I pull out the motherboard's BIOS ROM, an in its place, fit RDR. It will auto-detect the CGA card.
Why am I using RDR now? If RDR displays on the CGA screen, or I hear the short-long-short pattern of beeps that RDR does, that tells me a lot about the state the motherboard, and I follow that diagnostic direction.
I see nothing on the CGA screen, and I do not hear the short-long-short pattern of beeps, and so I continue as follows.
4. Remove unnecessary stuff
I see that an 8087 NPU (maths coprocessor) is fitted. I remove it. It is not required, and it may be the problem.
Unlike most XT-class BIOS ROM's, RDR does not require motherboard RAM in order to show something on the MDA/CGA screen. I remove the RAM.
I turn on the PSU. RDR still does not show anything on the CGA screen, and I do not hear the short-long-short pattern of beeps, and so I continue as follows.
5. Re-seat socketed chips.
I see that the 8088 CPU and some other chips are in IC sockets. Just in case a poor connection has developed between those chips and their socket, I re-seat the chips.
I usually completely remove the chip. In that way, I will see if a leg had been accidentally bent up under the chip. Then I carefully put the chip back; carefully, because I do not want to create a bent-up leg.
I turn on the PSU. RDR still does not show anything on the CGA screen, and I do not hear the short-long-short pattern of beeps, and so I continue as follows.
6. Detailed inspection.
- Missing components.
- Damage to components.
- Damage to traces on the circuit board (e.g. from battery damage).
- etc.
The detailed inspection did not reveal a problem, and so I continue as follows.
7. Reset - part 1
Easy to check:
Is the CPU being held in a reset state? I turn on the PSU. I measure the reset pin (pin 21) on the 8088/V20 CPU. It is at a 5V TTL level of LOW, so I have eliminated that as a problem.
8. READY signal to CPU
Easy to check:
Is the CPU being told to pause/wait? I turn on the PSU. I measure the ready pin (pin 22) on the 8088/V20 CPU. It is at a 5V TTL level of HIGH, so I have eliminated that as a problem.
9. Swap in a known-good CPU
The 8088/V20 CPU has proven to be very reliable, but still, it needs to be ruled out.
I pull the existing CPU out of its socket, and in its place, fit a known-good CPU. In fitting the replacement, I am very careful that I do not bend a leg up under the chip.
I turn on the PSU. RDR still does not show anything on the CGA screen, and I do not hear the short-long-short pattern of beeps, and so I continue as follows.
10. Swap in other chips
Some other chips are in sockets, and I have known-good spares of them. I pull them out of their sockets, and carefully fit the spares.
I turn on the PSU. RDR still does not show anything on the CGA screen, and I do not hear the short-long-short pattern of beeps, and so I continue as follows.
11. Clocks
Are the various clocks being generated? This is the point where the lack of a circuit diagram for the XT clone motherboard starts to be a problem.
Critical to the startup of the motherboard is the clock signal that goes to the CLK pin on the 8088/V20 CPU and to the CLK pin on the 8288 chip, and so I can use my oscilloscope to see if that clock is present on the pins that receive the clock. Maybe the clone XT motherboard does not have an 8288 chip, instead, the 8288 functionality being in something else, such as an M1101 chip. If so, which pin on that chip receives the clock? Perhaps there a datasheet for the chip that will inform me of that?
Be aware that when measuring the 4.77 MHz clock signal, if your oscilloscope does not have adequate bandwidth for that, you will not see a 'squarish' waveform. More information about that is at here.
If the clock signal is missing, what circuit generates it? Is is an 8284A chip like what the IBM XT motherboard uses, or is it something else?
I see the expected clock signal on both the CPU and 8288 (or chip that contains 8288 functionality), and so I continue as follows.
12. Reset - part 2
At power-on time, the reset pulse on the reset pin of the CPU needs to be HIGH for a minimum period.
This XT clone motherboard runs at 4.77 MHz, and contains an 8088 CPU. Therefore, per note 1 at the bottom of here, the reset pulse on this XT clone motherboard needs to be 840 ns or longer.
I set my dual-channel digital storage oscilloscope (DSO) to monitor:
Channel 1 = The POWER GOOD signal from the power supply.
Channel 2 = Either pin 10 of the 8284A chip, or pin 21 of the 8088 CPU. Of those two, the CPU is the preferred location.
Trigger set to: Channel 1, rising edge, +2.5V level.
I turn on the PSU, and see my oscilloscope record a reset pulse of adequate width/length. An example recording/capture is shown at here, that example being for a particular {IBM 5160 motherboard + power supply} combination that I own.
Note that some XT clone motherboards do not use the POWER GOOD signal in the process of generating a RESET signal. An example of this alternate way of generating RESET is show at here.
13. xxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx