Kurtzweil P88 matrix problem?

[Sowhat]

I got a Kurtzweil P88 keyboard in for repair with the complaint that the battery was dead and the ">>>" key did not work. The battery was indeed dead so that got replaced and the ">>>" switch no longer clicked and was a permanent short. I dug up a Yamaha switch of the same height and hole pattern and replace that switch which now clicked and worked good... Job done, right? Well... a few hours later the customer brought it back and said the ">>>" key worked sporatically and then stopped working all together. He suggested I swap the switch with another known good but seldom used switch and although I could see nothing wrong with the replaced switch I followed his lead and did the swap. He took it back home because he forgot the power brick and checked it out there. The swapped switch made no difference and in fact the Yamaha switch which is now in the "F" key position worked fine there. The switch matrix is debounced by a 74HC541 octal buffer (U1) and the ">>>" key runs to A(3) of the buffer... A(2) handles the "<<<" key which works fine.

I think I have a faulty buffer chip. Any one run across this before? Could the permanently shorted ">>>" key have caused the buffer to weaken and finally go belly up?



Note: I keep saying key but I actually mean button, the 88 keys are just fine, this is all control panel buttons I'm taking here.

[Enzo]

MAybe, but probably not. Still the IC is cheap.

But my approach would be to turn the thing off, and go down the row of pins on the IC. Pins 2-9. As far as I can tell all those circuits are the same. Measure resistance to ground. If one of the transistors inside has tanked it sometimes makes the resistance reading funny. If all the pins measure say 10k but the one pin measures 2.6k, then that is a problem.

And while it is running, scope those lines, does that one look different? DO you see any contention on the bus?

Also I have to think those switch lines are strobed, so compare the IC pin served by suspect switch S8 and compare it to when you press S1. Do the waveforms look different? Then compare what the S9 input looks like compared to the S8 input. Also with it sitting idle, measure the resting voltage at the pins on the IC, does this input sit at about the same as the others? This is logic so we think voltage doesn't matter as long as it is past the thresholds, but if the row of pins are all sitting at 4.8v or at 0.1v, whatever, and that one input is hanging in at 3.5v or 0.4v, that is a clue the IC may be bad.

And I can only see the tiny part you clipped. Is ther anything on that return line other than those NO switches? Is there any other thing tied to that line, like a chip output or a selector switch? It could be possible something else is confusing that input. A hanging strobe on another return say.

[Sowhat]

And I can only see the tiny part you clipped. Is ther anything on that return line other than those NO switches? Is there any other thing tied to that line, like a chip output or a selector switch? It could be possible something else is confusing that input. A hanging strobe on another return say.

The schematic is in a huge pdf. service manual and I only clipped a screen capture of that schematic for use here. Here's the whole capture, hope this helps.



The nodes of this matrix are all plain old regular N.O. switches. Off the top are the group switching transistors and off to the right is V+. There are a lot of switches in this matrix that don't seem to correspond to physical switches on the control panel... it's a big clunky multipage schematic covering a few models of related Kurtzweil stuff, guess I should start hunting those things down, and today was my day off.

[Enzo]

There is the 10k pullup at the right on the end of each return line. The one for that return could be compromised. All it takes is one marginal voltage.

And just for grins, the strobes - the vertical lines through the matrix - run LEDs as well as switch lines, so scope that strobe and see if it looks different from the next one over. You can scope both strobe and return right at the switch legs if they are convenient. A bum LED could be holding the strobe marginal.

I am just thinking in my head, potential things.

I am assuming VCC is +5.


But we are dancing around the chip. It takes a lot longer to describe the tests I just discussed than it does to do them. A few quick checks like that, and I would probably be reaching for a new IC myself.

I have similar manuals, I used to be a Kurz service center. I know how large they are. I don't fault you from using the clip, I was just pointing out I did not know what lay beyond the page.

[Sowhat]

Sage advise there Enzo…

I had the IC in my hand ready to just replace it when I figured maybe I should check some signals first with the scope like you suggested. I found the switch strobe signal on a good button and then checked for it on the problematic button. It was not present, so, I traced it back to the sequencing drive transistor for the matrix column and the signal was present… I back tracked from trace to jumper to trace and I found a spot where the signal stopped. I slid it under my inspection microscope and this is what I saw…



A cold solder! There, since it was manufactured. I reflowed this solder joint and everything came back to life and worked reliably now. What probably happened is this connection got worse over time and the customer just pressed the key harder and harder to try and make it work just like we all would probably do until he broke the switch. They are tiny and it don't take much pressure to wreck them. It was already hit or miss before I replaced the switch and it was coincidentally just time for it to stop working all together… just my luck!