Data Lines stuck high - How to troubleshoot?

[Van Diesel]

Hi folks

I'd be interested in anyone's opinion and assistance with this fault, please.

I'm fault-finding a board that has a 6502 processor.
I don't have a schematic.
The data bus of the processor is buffered using a 74LS245.
Three of the lines on the buffered side of this IC appear to be stuck permanently high (d0, d2 and d3) . . . the others are buzzing with data (so my logic probe says).

Now . . . how do I find what's up . . . is there an easy way to figure out which IC on the data bus is causing the problem? Or might it be something else??
It is not the 74LS245 buffer itself . . . it's fine (the three data lines are still high with the IC removed).

On the buffered side, the data bus is connected to a 2 x 74LS273, 74LS374, 4 x 74LS244 (I think these buffer the button inputs) and another 74LS245.

Any thoughts, please?

Thank you
Van
Van Diesel2021-01-07 13:34:57

 

[jonhughes]

I have a few boards like this where a component is dragging address lines permanently to ground, with a large number of chips attached to the lines. Those boards also have no schematics which is the main stumbling point for me. One day though I may be angry or motivated enough to trace everything out (but not there yet).

I’m no expert but as far as I know the ls245 is bidirectional so the buffered side could be any side depending on the enable status of the chip.

I will follow this thread with interest.

 

[Van Diesel]

Hi Jon

You are right.
My statement 'The Buffered side' meant the side OPPOSITE the data bus lines of the processor . . . the ones going out from the 74LS245 to the rest of the board.

Van

 

[Monstermug]

sounds like a job for Slice. But like most pcb repair probs u just have to painstakingly eliminate with probe until u find culprit. datalines should normally toggle so if u get a low numbered dataline not toggling then u half way to finding solution. follow lines bit by bit along the databus. u find it or them. Monstermug2021-01-07 15:45:58

 

[ColinD]

Are the lines still pulled to 5v with the board unplugged ? (use a meter to check continuity to 5V)

 

[yoganuggy]

If pins on the buffer chips are stuck then compare the stuck pin values to it's truth table to see if the chip is faulty.

If the CPU R/W pin is stuck on read it could be a bad RAM. I had that problem on a data east board recently with garbage on the screen

 

[sukhbir]

Hi,on the ls245 pin 1 (dir) should be high and pin 19 (c/e) should be low for data to flow from input to output.if pin 19 is high then no data will flow as it will remain in isolated state.

 

[bones]

It sounds a nightmare without schematics, I did read somewhere once that Mark at retroclinic has had success with the old BBC computers which use the 6502 or 6512 processor. I reckon you've probably got an internally shorted ls chip somewhere but equally it could be a ram/rams too. Good luck but maybe it's worth checking some of the 6502 processor forums for info.

 

[John Bennett]

It sounds near impossible. If something's shorted on a bus then I don't know how you'd track it down other than sneakily clipping the same data line on each device until it's freed (then soldering them back up).

If a pin is shorted then only track resistance separates them, so you'd have to have an incredibly good resistance meter and lots of luck. As mentioned, worth seeing if it is a hard short to the rail using a multimeter.

 

[mlynn]

I once saw a video on YouTube I think it was a Chase HQ board where as a quick fix the tester piggybacked an identical chip on top to diagnose if the chip itself was faulty without de-soldering anything. Once he found the fault he made a permanent fix by de-soldering the old component and replacing it.

As the 74LS245 is a multi pack inverter chip it should be cheap to replace. If the input is constantly low and the output is constantly high then the traces leading to it or from another circuit would seem to be at fault.

 

[Muppz]

It sounds near impossible. If something's shorted on a bus then I don't know how you'd track it down other than sneakily clipping the same data line on each device until it's freed (then soldering them back up).

If a pin is shorted then only track resistance separates them, so you'd have to have an incredibly good resistance meter and lots of luck. As mentioned, worth seeing if it is a hard short to the rail using a multimeter.

Polar ToneOhm is the thing for these. Mine is the 550 which is a 1990s bit of kit. Like a continuity meter but it squeals higher pitched the closer you get to the source of the short, has a 0 - 200 milliOhm range.

The latest 950 series is more of an ATE system for serious money/

 

[ColinD]

There is a Fluke current probe that comes with the old logic trouble shooting kit that would probably be handy to try, but as above something like a tone ohm can be very very useful.

 

[Van Diesel]

Thank you for all replies.

sounds like a job for Slice. But like most pcb repair probs u just have to painstakingly eliminate with probe until u find culprit. datalines should normally toggle so if u get a low numbered dataline not toggling then u half way to finding solution. follow lines bit by bit along the databus. u find it or them.

Thanks - I'm watching that project with interest. It's out of my budget though unfortunately.

Are the lines still pulled to 5v with the board unplugged ? (use a meter to check continuity to 5V)

Yes - Good point - I already tested that. The bad data lines are not short-circuited to 5V.

If pins on the buffer chips are stuck then compare the stuck pin values to it's truth table to see if the chip is faulty.

Thanks, but that's not going to work in this case (hence my call for assistance).

For example - the 74LS245 with the enable pin high (i.e. NOT enabled) will have it's outputs set to high impedance in the truth table. However, all the inputs and outputs measured while the board is being tested have activity. There's no way to know if it's working or not, other than cutting the individual legs of the chip and testing either side (I did this a few moments ago to ensure that one chip was at least OK on that individual pin) . . .
. . . which I think is my answer. I guess I was hoping for a short-cut where there isn't one.
I need to go through each and every one.

Van
Van Diesel2021-01-08 12:54:50

 

[John Bennett]

Out of interest, if you put the multimeter to the diode voltage setting and measure each data pin to 5V and 0V (in both directions, so 4 readings per data line), do the failed bus lines read any different to the working ones?

 

[Macro]

It depends on the PCB layout, if you can cut a trace to divide the number of chips it connects to (preferably into 2) then you can at least eliminate some.