Arbiter Trem Face—trem fades out as rate is turned up?

[eurekaiv]

I'm trying to repair this old Arbiter Trem Face and I'm having trouble with the oscillator.  I replaced a bad .68 cap, the rate pot (the original dual 47k was measuring over 38k/62k), and the trim pot which had become too dirty/finicky.  Overall the pedal sounds much better now and the trem effect is strong until you turn the rate up at which point the oscillations get faster and then quickly fade away as you play.  This happens at about 2/3rds up on the control and at best I get a moderate speed before it starts to exhibit the issue.  I didn't have much luck searching similar problems so I thought I'd ask for some help.  Thanks!

[R.G.]

I suspect that  your oscillator transistor has a marginal gain. That's usually what makes a phase shift oscillator like this fade out.

This may be the oscillator transistor, could be something about the circuit that's causing a bias shift or voltage shift as the frequency is turned up.

If you have not done so already, replace all the electrolytic capacitors with new, fresh ones of the same capacitance or as nearly as you can. Bum electros could be causing a change in bias or loss of gain as the oscillator is turned up.

I addressed the issue of getting a vintage effect running in my guitar effects repair FAQ at GEO.

[eurekaiv]

Thanks RG!

Since this was a vintage effect I didn't want to replace any parts I didn't have to and the original electrolytics tested ok but I'll go ahead and try swapping them just to see if that fixes things. It's easy enough to pull them and have a go at new ones.

[R.G.]

Beyond that, check it against the schematics at GEOFEX: http://geofex.com/Article_Folders/trem_face/tremface.htm and do the voltage-measurement thing when it stops oscillating. That will tell us what's going on. In fact, just do the voltage measurements before you replace caps. Turn the speed up til it quits, then measure all active device pins to ground.

It may be that the oscillator transistor has just lost some gain. A replacement will fix it in that case.

[eurekaiv]

So I checked against the schematic and double checked all the caps.  The two electrolytics are ok but they did read awfully high.  I went ahead and replaced them anyway just to get some fresh ones in for testing.  No difference in the effect.  I measured the voltages below...  all negative of course.  The one in blue goes way up and down when the effect is working but of course goes steady to the voltage noted when the rate is turned up high enough.

[R.G.]

Given that the electros are OK, the voltages look reasonable for a normal biased NPN. If the 330uF cap on the emitter of the BC184 is not in really good shape, that will lower the gain and cause this kind of problem directly. But if it's new, that should not be the issue.

I'd say replace the BC184 with a new high gain NPN. I'd use a 2N5088 because I keep lots of those handy. But any device with a gain over about 400 will do fine. What do you have?

[eurekaiv]

I have a ton of 5088s from building big muffs so I can try one of those. I also have bunch of BC109s and probably a half dozen other types that would suffice if it just needs a higher gain NPN.

[R.G.]

I have a ton of 5088s from building big muffs so I can try one of those. I also have bunch of BC109s and probably a half dozen other types that would suffice if it just needs a higher gain NPN.

Watch the pinouts. If that doesn't do it, start suspecting either that 330uF cap or components in the phase shift network.

[eurekaiv]

New transistor same result. 

[R.G.]

New transistor same result. 

Grak.

Let me recap a bit.

You replaced a bad 0.68uF cap - in the phase shift section of the oscillator, correct?
You replaced the electros.
You replaced the BC184.

Did you measure resistor values, including the speed pot?

Do you have another couple of 0.68uF caps to try in the speed section.

Here's what I think is going on. Phase shift oscillators need a voltage gain of at least 29 in most cases to make up for the losses in the phase shift network. As speed is turned up, the resistance to ground at the two capacitor junctions gets lower and loads the driving transistor. This lowers gain, and at some point, the oscillation will stop because the transistor can't support the necessary gain. The oscillations peter out. They return when the loading goes back down and the gain therefore comes back up.

If it's not a bum electro (first usual suspect!) and not an egregiously bad biasing situation, it's either enough gain from the transistor (tried that!) or the emitter cap isn't bypassing well (you replaced that), or something about the three 0.68uF and speed pot is putting too much load on.

While I'm thinking about it, the voltages look funny for a good battery. Are you using a battery, and if so, what's its voltage?

[eurekaiv]

Last I checked the battery was around 8.8V but I'll try a fresh one.  Right now the two electros are indeed replaced and the rate pot is a new 50k dual that has each side within a few k resistance wise.  The old rate pot was ridiculously far apart on each half.  The other .68 caps are original but I pulled and measured them all and only replaced the one that was obviously faulty.

[R.G.]

You might try measuring the DC voltage on the junctions of the 0.68uFs and the pot to ground. This should be 0.000000000V. Even a few millivolts indicates that something is leaking.

Out of curiousity, which 0.68 did you replace?

[eurekaiv]

Unfortunately I haven't had time the last couple of days to mess with this but I should be able to check those voltages and make sure the battery is still ok this evening.  As for the .68 cap, I'm pretty sure the one I replaced was the one farthest to the left in the schem.

[analogmike]

It's almost always the tantalum caps that are bad in those. Replace them all and that usually fixes it.

[eurekaiv]

My battery was down to about 8.5V so I stuck a fresh battery in that was giving about 9.3V. This improved the performance slightly at best.

I updated my image to reflect the additional measurements you suggested.  There is defenitely some voltage present at the two junctions. The cap I replaced was actually the farthest away on the board but the closest to the transistor in the schematic.  I marked it in green. I used a tantalum because the .68s I could find were way too big.  It's oriented with the positive as noted... the pedal didn't do anything at all with it the other way.

[theundeadelvis]

It's oriented with the positive as noted... the pedal didn't do anything at all with it the other way.

Did you have the tantalum cap oriented incorrectly at one point? I believe they are easily damaged if the polarity is reversed, so if you did, that cap may have been damaged.

[eurekaiv]

Really?  I've never heard that.  In any case, I wasn't sure which way it should I just soldered it in to see if it would work.  That would be really crummy if all along that's all the problem was.   

[theundeadelvis]

I've read this on a number of occasions, but I don't know enough on the matter to say for sure if it's true or not, or if it's less likely in a 9 volt application. If you have another, it may be well worth the trouble to give it a try.

From Wikipedia (for what that's worth): "All tantalum capacitors are polarized devices: correct polarity must be observed, otherwise the capacitor will depolarize and the dielectric oxide layer will break down, reducing the resistance of the device and causing it to fail."

[R.G.]

If you used tanatalum, replace them with film. Film caps up to 1uF are available, and if it's leakage in those caps, it's a long term fix.