Bit stuck high

[zz28zz]

Hey everybody. Newbie here. I've been working around electronics for many years but just recently took a position that requires repairing circuit boards among other things. Struggling with the digital side of the boards so bare with me.

I'm working on a older tester that uses various circuits boards in a card cage. There's a DOS based controller connected to the cage.
When I run the stuck bit test on a particular board, it fails for having 2 bits stuck high.
There are 12 bus xcvr's and 1 inverter hex bus driver OC TTL connected to the bus.
Is there an easy was to tell which IC is causing the bit stuck high??

Thx in adv.

[Nusa]

Doesn't the test software give some specifics as to which pins it is complaining about? Or if the terminology is really "bits", it may be talking about something else on the board that has nothing to do with the bus interface itself. Depends on what that board DOES, doesn't it?

If you have a small enough pool of candidates, you could simply monitor suspect pins with a scope and see what you see.
If you have both working and failing examples of the board, you can verify suspect behavior on the good one.

[james_s]

A logic probe and datasheets for the ICs involved is a classic way of tracking down this sort of fault. An oscilloscope can be used too but a logic probe is often more convenient for quick go/no-go type checks like this.

[Tomorokoshi]

If a thermal camera is available and the fault causes contention that makes the part dissipate more power than usual, it may be possible to at least track down the part that's involved.

[Armadillo]

Normally, the other devices should be able to pull the line low. In this case, so it seems you are talking about hard stuck to VCC. So take the board out, don't apply power, use ohm meter to measure each device bus pins to VCC and identify the internally shorted device to VCC [low ohm, isolate the VCC pin if necessary].

[Samogon]

Normally, the other devices should be able to pull the line low. In this case, so it seems you are talking about hard stuck to VCC. So take the board out, don't apply power, use ohm meter to measure each device bus pins to VCC and identify the internally shorted device to VCC [low ohm, isolate the VCC pin if necessary].

Exactly that i did with MUX input stuck to Vcc

[zz28zz]

The test screen shows the stuck bits as BYTE 0/ BIT 0 and BYTE 0/ BIT1.

I spent a good deal of time connecting our digital scope to each data line (0-7) and running the stuck bit test. Found that data lines 0,1 and 4 had much more activity (like 27 downward spikes) than the other data lines on the bad board (is it re-trying??). On the good board there were only a couple of downward spikes for each data line when the stuck bit test was run.

When zoomed in on the spikes, each spike is actually 2 sets of 4 spikes. Not sure what that's all about.

The only single IC that's connected to data lines 0,1 and 4 is the "inverter hex bus driver OC TTL". I replaced that IC but the issue didn't change. The IC was harvested from a scrap board, so it serviceability is unknown.

In the attached pics I have one scope channel connected to data line 0 on the good board and the other scope channel is connected to data line 0 on the bad board. I set the trigger for falling slope and the threshold abt 0.5V lower than "high' and started the stuck bit test; first on the good board (#1), then on the bad one (#2). I then repeated the tests for each of the 8 data lines.

We don't have a logic tester available at this time to use. No thermal camera either. I did try touching each suspect IC but couldn't find one that stuck out thermally.
We have datasheets for the IC's, but like I said I'm still a bit green with digital circuits.

What else can I try??
Just saw armadillo's reply. That sounds good and it's an easy thing to check.
Still would like to understand what the scope is trying to tell me.

[Nusa]

What's the nature of the board? Does it contain any IC's that include a few bytes of non-volatile memory for configuration? EEPROM does have a finite cycle lifetime and you're clearly dealing with equipment old enough that it may be a concern.

[zz28zz]

No EEPROMs on this board. It's a s/w controlled P/S from the 80's. They usually don't have issues on the digital side unless it was hot-swapped, which is what I think happened to this one.

[Nusa]

So evaluate what's most vulnerable in a hot-swap situation and check those things first.

If you have a donor board or other spares, you can start swapping suspect components until the behavior changes. You already did the hex inverter. Save the stuff that made no difference as probably good.

[zz28zz]

In the past, the first IC connected to the bus is usually the one that gets zapped during a hot swap. In this case, data line 0 is connected to 4 bus xcvr's and the hex inverter. Problem with swapping multiple ICs is that the pads pull off these boards fairly easily.

On a prev repair (of the same type board), one of the bus xcvr's shorted data line 5 to ground after a hot swap. I was able to identify the bad IC by probing data line 5 at each IC with a 5 1/2 digit DMM while board was on the extender board and sys powered up. The bad IC's voltage on data line 5 was slightly lower (~8uV) since it was closer to ground.

Wondering if I can do something similar with this board. Like unplug the card edge connector and jumper 5V to data line 0 then jumper the 5V trace (thru a resistor maybe) to ground, then measure data line 0 at each suspect bus xcvr like I did above.

[zz28zz]

Well the mystery continues. Checked one of the bus xcvrs probing VCC and the data 0 pin. No short found. Since the 5v and data line 0 are common between the remaining suspect ICs, didn't bother checking the other bus xcvrs.

Spoke to a retired EE very familiar with these boards and he suggested cutting traces before trying to pull multiple ICs.

[james_s]

Cutting traces is brutal, I never do that unless it's an absolute last resort, I've seen a lot of butchered PCBs. Pulling ICs is easy with proper tools.