Debugging UVICS-3 (Uni Vibe in Crybaby Shell)

[Kevin Mitchell]

If you have read my previous thread then I should note that I have rebuilt the board only reusing resistors.

I have built the UVICS (project document -schematic on page 6) and am in the middle of debugging. There are two issue at the moment I would like to address but let's go at it slowly. I have included the charge pump which is pushing out ~17.5 volts. I'm using a dual speed pot and the board is pretty much loaded with the BOM parts.

Problem #1 - bulb is lit but does not pulsate
-Speed and depth control do nothing

Problem #2 - vibe/chorus toggle in one position (which position is unknown at the moment) kills audio signal
-We'll get to this later on

I would like to start with the lfo. I believe the issue must be a short or bad solder somewhere but I'm having troubling locating it. Here are the readings of the 3 transistors;
Q13 -MPSA13 (driver)
E- .75
B- 2.11
C- 1.47

Q11 -2N3906
E- 17.49
B- 16.92
C- 17.49

Q12 -2N3906
E- 12.83
B- 17.5
C- 17.5

And fyi the voltage regulator
E- 15.09
B- .01
C- 17.5

[PRR]

Short is around Q11 Q12.

Q12 appears to be blown (Vb-e is several volts, >1V is fatal).

Could instead be wrong parts or wrong pin-out.

> my previous thread

Might be better to keep it all together, or at least link to the previous chapter.

[Kevin Mitchell]

> my previous thread

Might be better to keep it all together, or at least link to the previous chapter.

I consider that thread irreverent to this debugging thread because the board was in very bad condition. Missing traces and was an aweful mess. Little embarrassed about that. This new board is a new build just reusing resistors -didn't want to buy new ones so I scrapped them.  That note was for anyone curious about the previous project because I said I was going to reuse all that I could which I ended up not doing -mostly for R.G.

Short is around Q11 Q12.

Q12 appears to be blown (Vb-e is several volts, >1V is fatal).

Could instead be wrong parts or wrong pin-out.

Q11 and 12 are 2N3906. I'm pretty sure all of the transistors were installed correctly. I'll swap out Q12 right now and post new voltages for the 2 transistors in just a minute. Thanks for the timely reply you're awesome!

[Kevin Mitchell]

Okay so I've replaced Q12 three times. E went down half the previous voltage (6.94) so I thought that transistor was bad then replaced it again with the same result. Still no pulsating.

Q11;
E- 16.91
B- 17.54
C- 17.54

Q12;
E- 6.94
B- 16.88
C- 17.54

I have no idea what's going on here.

[PRR]

Something is not wired right.

Q11 B should be around 9V-10V, Q11 E and Q12 B should be 0.6V lower, and Q12 E 0.6V lower than that.

Actually Q11 B should read much lower *while you put the meter to it*, due to the 22 Meg resistor. The emitter voltages should be a better clue. However you have solid full voltage at Q11 B, which aint right.

I see (in the numbers) a short from Q11 B to B+.

However that does not explain how Q12 E is sitting way-way low.

Until you find the wiring fault, throwing more transistors at it is likely to blow more transistors.

Check polarity of C22.

[Kevin Mitchell]

I noticed that c22 and 23 were bipolar caps so I switched them out thinking that would do it. But that wasn't the case. Still getting the same readings as previously listed.

I see (in the numbers) a short from Q11 B to B+.

I'm not sure what you mean by "B+" but I keep checking for a wiring mistake and I'm having trouble finding it. :/

[R.G.]

Paul is correct, as usual.

Here is some thinking about the circuit.
Your supply voltage is 17.54, on the collectors of Q11 and Q12. What should be happening is that R39 and R41 should be making a bias voltage for the base of Q11 of 4.7K/(3.3K + 4.7K)*17.54V = 10.31V. R40 is quite large, 2.2M, so you should be able to measure that 10.31V at the junction of R39, R40, and R41. If you cannot, something is wrong with the resistors or their soldering, or circuit board traces.

R40, 2.2M, lets a little current into the base of Q11 without shunting all the feedback signal from C19 that keeps the thing oscillating (when it oscillates!). The base of Q11 is a very high impedance point, and as Paul notes, a meter may load down the base so much that you don't measure Q11 base voltage accurately.

However, a meter load will make the base of Q11 read *lower* than it is without the meter loading, not higher. You're reading higher than normal. Paul's surmise that there is a short between Q11 base and the 17.54V supply, perhaps right at the Q11 collector, is probably right. To test this, turn off the power, set your meter to read ohms, and read the resistance between Q11 collector and Q11 base. If things are right, you'll read a low resistance with the meter probes in one direction, and a much higher resistance in the other direction. If it reads very low, in the few-ohms range and the same in both directions, you have a short between collector and base, and you have to go eyeball the board, perhaps under magnification, and find the short. If no short can be found, it is **possible**, but unlikely, that Q11 is internally shorted. This is the least likely thing to be the problem.

Back to how it should work. With the base of Q11 at 10.31 minus a volt or so dropped in the 2.2M resistor, the emitter of Q11 is much more measurable, and should be about 0.5 to 0.55V lower than the base unmeasurable voltage. The base of Q12 should be *exactly* the same voltage, as it's connected to Q11E by a solid trace of copper. The emitter of Q12 should be another 0.5 to 0.7V lower still, in the 7-8V range.

To simplify things, you might remove C22 and C19 just to make sure they're not clouding up the picture, but the problem, as Paul said, is probably a base-collector short on the PCB with Q11.

[Kevin Mitchell]

Thank guys. R.G. thank you very much for the breakdown. Very educating.

I cannot for the life of me find a short or a bad trace anywhere regarding the transistors or resistors accompanying them. I'll go at it again tomorrow night and do as you've suggested. It has to be a bad resistor out of the 3... Or as you said (but unlikely) Q11 has an internal short.

Here are the transistor voltages once again -now that I understand how everything works I can read these a little better. I can see around a bad read due to shaky hands or whatever.

Q11;
E- 16.92
B- 16.97
C- 17.54

Q12;
E- 6.96
B- 16.92
C- 17.54

[Kevin Mitchell]

Oh I forgot to ask to why after replacing Q12 the voltage of the emitter went down so much? Maybe the original transistor was bad? Something like this makes me question where the issue may actually be. But as I said I'll do everything you've suggested tomorrow night.

[Kevin Mitchell]

Sorry about the late update! Rarely have time to sit at my bench during the week. I measured everything and I think Q11 could be bad. I'll swap it out to find out.

So power source is pushing 17.32v today

Q11;
e- 16.71v
b- 17.32v
c- 17.32v

Q13;
e- 6.75
b- 16.71
c- 17.32


Reading at the junction before R40(2m2) is 10.16v as it should be.

Reading ohms as you've said c+ reads about 921 and nothing going the other way. q11 e to q12 b is small connection and a short should be easy to spot.

EDIT;
Misunderstood something for a second. So I guess the transistors are working fine and it has to be a short somewhere? Frustrating...

[Kevin Mitchell]

Maybe a broken wire with the speed control or a wiring error? I'll keep tinkering.

[Kevin Mitchell]

When the speed pot is down (max speed when in shell) and I turn it quick the light goes out for less than a second. Not sure if that's relevant but thought I should note. I know the taper itself isn't bad.

[PRR]

> power source is pushing 17.32v today
> Q11; b- 17.32v, c- 17.32v

Still shorting at Q11 B-C.

[Kevin Mitchell]

> power source is pushing 17.32v today
> Q11; b- 17.32v, c- 17.32v

Still shorting at Q11 B-C.

If I do as R.G. said and test resistance from b and c

Q11 reads;
934 with c+ and b-
1(nothing) c- and b+

testing a similar new transistor reads;
936 with c+ and b-
1(nothing) c- and b+

So if I am doing this correctly that would show resistance one way (c+ and b-) and not the other way. After testing Q11 that should imply that there is no short with b and c and the transistors are working properly from what I can tell.

Am I correct? By the looks of the voltages I understand why I am looking for a short but I promise I can't find a visual short on the board and the resistance test shows otherwise.

[R.G.]

We completely believe you that you can't find a short visually. The problem is that your voltage measurements are most consistent with a short, and that's all we have to work with.

Some PCB flaws are so small as to be invisible to the naked eye. Some people cut around all the pads and traces in the suspect area with a very sharp knife point, like an X-acto knife. Others use magnification. A meter is more reliable, as it tells you what the copper does electronically.

At this point, you've nearly exhausted the easy stuff. I would say to pull out Q11 and Q12, and C19, 20, 21, and 22.  This leaves only the resistors, pads, pads and traces and isolates this section from all the rest of the board excepting for power and ground.

Then go check every trace to every trace that could even possibly be shorted to the +V line, using your ohmmeter setting. Frankly, if it were me, I'd clip off the resistor leads on the component side of the board, too, and put in new resistors. Resistors are cheap. Only stop this when you're absolutely sure, as verified by your carefully-kept written table of every trace to every trace, that there is no copper short.

Then start rebuilding. Put in R39, R40, and R41 only and apply power. Use your voltmeter setting to test the voltage at the junction of the three resistors. This should be *exactly* as provided by the resistor-divider equation - V = R41/(R39+R41) times the power supply voltage.

Now check at the Q11 base end of R40. This will be the same voltage, but lower by the voltage divider rule of your meter's input resistance  and R40 from the voltage at R39/41.  Parenthetically, it also tells you the largest voltage you can ever measure at Q11 base if everything is all right, and also your meter's input resistance.

If all is well, put in R46 and Q12. Now measure the voltage across R46. It should be zero. There is no drive and no current flowing in Q12.

Now put in Q11 emitter should be a bit lower than (voltage at R39/40/41 minus about 0.5V). The voltage at Q12 emitter or across R46 should be yet another 0.5 to 0.7V lower still.

This is the procedure I would use if this were on my bench.

[Kevin Mitchell]

Do you think using pnp transistors instead of npn would explain the problem?   

[Kevin Mitchell]

Just replaced q11 and 12 with some 2n5088npn I happen to have. The lamp isn't pulsating but the voltages look a bit better from what can tell;
Q11;
E- 9.74
B-  ~6, drops down to 3.13 (meter load? If I recall 3.13 is about the voltage that should be coming off that 2M2r)
C- 17.5

Q12;
E- 17.5 (to note- goes down to ~16.16 with speed adjusted a full turn)
B- 9.74
C- 17.5

If this still doesn't look right I'll follow your procedure till I find the bug.

[R.G.]

Do you think using pnp transistors instead of npn would explain the problem?   

Boy, am I ever functionally fixated. You stated that right up front and I flat missed it. That should have been the first thing I noticed. Good catch. I am presently flogging myself for not noticing that. 

Your voltages look much, much better. Are all the caps back in now?

[Kevin Mitchell]

I haven't taken anything out. Only replaced those two transistors. I still don't know where the problem could be.

[R.G.]

OK, with the DC operating conditions set up, you at least have a chance, which you didn't before.

This little setup is a very strange phase shift oscillator. Q11/Q12 have a very large current gain. That current is fed into the emitter resistor of Q12, whence it's converted to a voltage by Georg Ohm and sent off to the light bulb driver through the depth control. The remainder of the current is fed through the two sections of the speed control to the stack of capacitors  and back to the base of Q11. The stack of capacitors and speed control are crafted so that a current just big enough to supply the input current needed to make that output current (yes, I know that's circular - in this case, that's the point) come out. It provides its own input and the speed/capcitors make that happen at only one frequency, so it oscillates there.

In the vibe clones, this may not work because of insufficient current gain, or wrong/too-small current fed back through the speed/caps, or it may be working fine, but the lamp driver may not be working, or the LFO may not be getting through the two caps and one pot of the depth control. The problem you now have is figuring out which it is.

A good start is to hook your meter across the emitter resistor of Q12, and note how steady the voltage is. If it doesn't change, it's ... well, steady. This is one of those few places that an analog meter is better than digital, but we can muddle through. Turn the speed control to middle. Did the voltage change?

Now use a clip lead or bit of wire and short across R41 momentarily. The voltage will go to zero, then pop back up to its former value, but does it wobble a time or two? Wobbling is an indication that it's almost oscillating but not quite. No wobbles indicate that it's not even close to oscillation.

While you're in there, check the voltages on all three pins of your lamp driver transistor and note what happens to them as you twiddle Tr1 and/or TR2 a bit. This is just to see if the lamp driver is functional and responsive.