Foxx tone machine debug

[punkaled]

Hi, just built the ftm from the JD Sleeps schem.
Slight prob with it, seems to work fine with the "octave" switch closed, but as soon as i open that switch i get a very low gated sort of signal.
Using an audio probe, i have found the signal seems to be fine up to and out of Q2 but stops at the diode, If i understand the diagram correctly, the octave is applied when the signal passes through both diodes and is full rectified, but there doesnt seem to sound an octave effect either. Is it possible one of the diodes may be broken (d2)? as i get a signal one side of it, but not the other side (surely the signal should be both sides but a little cleaner on the guitar side?)
i hope this question makes sense to someone and you can help!
cheers,
Aled.

[punkaled]

OK, so i tested it again this morning, and this is what im getting-
with the octave switch closed, i get a similar signal at both sides of both diodes.  With the octave switch opened i get a signal at the cathode of each diode, but nothing at the anode of either (the signal must be there but very very weak?)
I have treble checked the resistors in that part of the circuit, and all the solder joints seem fine, and no sign of any shorts. its really frustrating me now  :?

[punkaled]

Anyone? i dont understand this completely, do the diodes need to be at a certain voltage to be biased right (tike a transistor)?

[Mark Hammer]

The transistor just ahead of the diodes is functioning as a phase splitter.  That means that equal-but-opposite copies of the input signal are coming out of the emitter and collector pins of that transistor.  Those copies pass through the 4.7uf caps and a diode to a mixing point.just near the input of the "Drive" control.

The diodes serve to restrict the signal passing through to just one half-cycle.  Because the two copies of the signal are inverted, relative to each other, and because the didoes are oriented the same way, what you end up with is a "bump and flat bit" from each path, rather than a bump and a dip, and when you combine these anti-phase copies of the same input, you get two bumps where there might have originally been a bump and dip.  On a scope, it looks like the lower half-cycle has been "folded over" to become a positive-going half-cycle.

What does this imply for the diodes?  The diodes do two things: "chopping off" one half-cycle, and dropping the signal level by the amount of voltage-drop that they do.  GE diodes will drop the level somewhere in the neighbourhood of 250mv, but in my exoerience diodes of the same numerical type can show considerable variation over different lots/batches, from as little as 180mv up to 300mv or more.  Getting an obvious octave sound requires the "folded-over" part to be the same amplitude as the other part of the extracted waveform.  Consequently, the diodes should be roughly matched for voltage drop to get a more obvious octave up.  

If all you have is two such diodes and can't pick and choose, then another route might be to use a nulling arrangement like JC Maillet uses for his version of the Green Ringer.  Here, you'd stick in a small value trimpot in series with the 4.7uf cap, and before the 100k resistor to ground.  This would let the trimpot and 100k resistor form a voltage-divider to attenuate the signal on that one path.  Given that we're talking about inter-diode differences of at most around 100mv, fairly early in the signal path, not that much attenuation is needed to fine-tune the two chopped versions of the signal and match them for better octaving.  With a 100k resistor to ground, I can't see needing anything bigger than a 1k trimpot (set up as a variable resistor).  

Note that this trimpot would go in the path where the diode has a *smaller* voltage drop.  So, if your meter tells you that the diode after the emitter pin has a voltage drop of 267mv, and the one after the collector has a voltage drop of 192mv (and the voltage drop is what is taken away from the signal amplitude), the collector path will have the "hotter" output, hence need a wee bit of trimming back to match it to the already somewhat lower emitter output.

Bear in mind that this adjustment/mod is for improving the octaving effect.  The scenario you describe suggests to me that the 4.7uf caps may be oriented improperly.  Specifically, it seems that the one in the "no-octave" path is incorrectly oriented, or simply bad.  If there is a problem there of that type, then one would expect that:
a) the no-octave sound would sound gated and unable to "break through" unless you played real hard (sputtery)
b) you'd get a fuzz tone when the octave switch is closed (because the "octave" path is just an inverted copy of the other one, hence capable of sounding like a fuzz), but no octave effect, because you still have only one copy of the signal and not the complementary combination of both.

Make sense?

[punkaled]

Yes, thanks a lot man, that makes a lot of sense, especially what you said at the end about capacitors - thats pretty much exactly what im getting.
i'll check those capacitors again tonight.

[punkaled]

ok, i had another go at this thing with the audio probe. I dont think its the capacitors, theyre both the right way round, and there is a definate signal at the cathode of each diode.
With the octave switch open, i'm getting signal at the anode of d2, and all the way to the output, so i'm guessing that that section is OK. I dont get any signal at the anode of d1, as the anode is floating (not connected to anything as the switch is open) i dont know if i should hear a signal there or not?
when the octave switch is closed, the signals at the cathodes of each diode remain the same, but the signal at the anodes is very low and fuzzy sounding so it seems the problem is in the switched path somwhere i just cant work out what it would be as everything up to D1 seems fine and after D1 seems wrong. I have checked the drop on each diode, and both are within a few mV.
I'm using JD Sleeps schematic, and all the coupling caps are 10uF,
Also, the doc with the schem said that all diodes in the cct are Ge, but D1 and D2 could be anything so i used 1n4148s, should i change these for Ge's? could that be the problem?

[punkaled]

changed the diodes, no difference.
the signals are fine when taken from each diode seperatly, its when the two signals are mixed together that it seems to be screwed up. As i don't have a 'scope i cant tell exactly what is happening when the signals are mixed but they seem to be cancelling each most of other out or something.
I think im gonna give up on this now, im sick of the sight of it  :x