EA Trem, long term use problem

[elevenoid]

I've got a problem with my EA Trem, I'll start by saying I built it back in 2009 and its been working like champ since then (probably pushing towards 500 shows) but has suddenly developed a strange problem:

It still works at all speed values except the very fastest, when the pot is dimed, or very close to full. Then it quite quickly stops pulsing at all, from eyeballing the LED which is still on at all times. The sound goes from effected to seemingly clean. I'm guessing some sort of weird feedback is going on in the circuit, because after I've maxed the pot bringing it back down doesn't restore the effect, it seems stuck. But a quick off/on on the bypass switch (or earthing  does restore. Obviously I use it mostly at full speed, d'oh!

Here's the circuit:


I have it with the LED wired into a 3PDT to act as on/off indicator

I don't really understand electronics at all, I tried quickly subbing out Q2 and Q3 for no change. Could a dirty or damaged Speed pot do this?

Here are the voltages,
when working:

Q1: D: 4.47, G: 1.25, S: 0.45
Q2: D: 0, S: 0, G: swings regularly between +/- 0.17
Q3: C: swings between 1.1-1.9, B: 0.57, E: 0

But in the error state Q2 reads all zeros, and Q3 C is stable at 1.53

Can anybody help with deciphering this? Please?!

Ben

[GGBB]

Mine does the same thing, but when I back off the speed it comes back - just not immediately.  This is normal and is just due to the oscillator caps not being able to cycle their charges fast enough.  In my case, the 1K resistor was replaced by a trimpot so I could dial in a faster top speed, so I have control over this problem.  In your case, I can only guess that aging has affected the caps and/or the speed resistor and pot so that you now have a problem you didn't have before.  I would also guess that because bypassing the effect disconnects the LED and therefore the power to the oscillator, the caps have a chance to discharge, thus restoring operation.

EDIT:
I suppose it's also possible that the speed pot or its connections are failing so that when you dime it there's no longer a path to ground.  Is lug 1 connected to lug 2 or is lug 2 going directly to ground (or to the 1K resistor if you reversed it)?

[PRR]

Yes, pot problem can do this.

And if you *only* run at high speed, you can increase the general speed range so you don't have to max the pot to get "your" speed. Change all three 1uFd caps to a smaller value. 0.5u makes it twice as fast throughout the pot range. That may be "too" fast, but 0.68u is a commonly available value which splits the difference.

[duck_arse]

and you can reduce the 15k between the 1uF's to increase the speed range.

[elevenoid]

GGBB,
I was being impatient, it will return after a few seconds under its own steam. I guess it is aged parts then...

PRR
Sadly not the pot, I swapped that out this afternoon for no improvement!

Everyone above:
It's a fairly messy perf board layout (and I made a hideous tangle of boxing it up!) I think it might be time to build up a veroboard with some of your suggestions incorporated. Thanks for your thoughts people.

[Davelectro]

I'd start by replacing the (22uF) cap connected between the jfet and the depth pot. Tantalum is recommended.

[GGBB]

GGBB,
I was being impatient, it will return after a few seconds under its own steam. I guess it is aged parts then...

PRR
Sadly not the pot, I swapped that out this afternoon for no improvement!

Everyone above:
It's a fairly messy perf board layout (and I made a hideous tangle of boxing it up!) I think it might be time to build up a veroboard with some of your suggestions incorporated. Thanks for your thoughts people.

Try Paul's suggestion of smaller caps.  If you want really fast as well as slow speeds, you can have a mode switch:


(C5, C6, and C7 are the three 1uF caps.)