Tweet
Well, what fun and games today. The offender is SF Twin Reverb, the one with the master volume pull switch, described somewhere as a boost switch. I'll come back to the later.
The complaint is the vibrato channel breaking up. I looked at the settings that were used to make it happen - volume all the way, up treble all the way up, mid and bass half way, MV switch pulled. Sure enough it breaks up if you whack the guitar pretty good, especially in the middle range.
I reproduce the problem using a 300mVpp 1Khz sine wave and start probing. It's a very short oscillation burst. With this control setup there's a massive signal on the reverb driver (a gain of very roughly 50 x 50 x 1/10 =250 to this point) and it's getting driven all the way into hard cutoff and that is exactly where the burst happens. An interesting tidbit is the 2in wire than joins the 0.022uF cap to the 3.3M resistor that goes to V4. If I put my scope probe on the cap end of it, it makes no difference. If I put it on the 3.3M resistor end is kills the oscillation completely. A 2in bit of wire has an inductance of 50nH so this oscillation has to be VHF and given the rats nest layout and especially the wire to the MV switch from the reverb output trail right past the tone stack it's hardly surprising. I should add this amp is stock aside from the reverb transformer.
So my first question, has anyone else ever seen this problem?
So what to do about it? Well the whole business of taking a signal of the reverb output is rather bizarre. First, and I checked, the phase is such that it attenuates the signal to V4B, hardly a boost function. Second is that the phase shifts are significant and frequency dependant. If you wanted a gain switch it seems it would be a good deal easier to simply switch the 3.3M resistor value.
As one approach, I added an attenuator 4:1 on the reverb driver input, lowered the gain and added headroom by increasing the cathode(s) resistor to 1.2K. I also arrange the MV to switch a 1.5M in parallel with the 3.3M to actually give a boost. There is still plenty of reverb and you cannot overdrive the reverb driver any longer. I should be happy, but I'm not.
I suppose another strategy would be to leave the basic topology but add grid stoppers to V2B and V3 and also rewire the cable carrying the reverb signal to the switch with twin screened cable to cut out the feedback path but the sound when the reverb drive goes to cut off is pretty unmusical due to that inductive reverb tank load so I don't like this either.
Thoughts?
Bonus point area.
There's some other weird stuff about this design that I've never noticed before. Look where the low side of the MV pot goes: to the 100 ohm resistor in the tail of the LTPI. Why would they do that? Also I see a 0.01uf and 220k on the V6A plate, presumably to balance the signal levels but again returned to same 100 ohm. It all seems a little odd.
The complaint is the vibrato channel breaking up. I looked at the settings that were used to make it happen - volume all the way, up treble all the way up, mid and bass half way, MV switch pulled. Sure enough it breaks up if you whack the guitar pretty good, especially in the middle range.
I reproduce the problem using a 300mVpp 1Khz sine wave and start probing. It's a very short oscillation burst. With this control setup there's a massive signal on the reverb driver (a gain of very roughly 50 x 50 x 1/10 =250 to this point) and it's getting driven all the way into hard cutoff and that is exactly where the burst happens. An interesting tidbit is the 2in wire than joins the 0.022uF cap to the 3.3M resistor that goes to V4. If I put my scope probe on the cap end of it, it makes no difference. If I put it on the 3.3M resistor end is kills the oscillation completely. A 2in bit of wire has an inductance of 50nH so this oscillation has to be VHF and given the rats nest layout and especially the wire to the MV switch from the reverb output trail right past the tone stack it's hardly surprising. I should add this amp is stock aside from the reverb transformer.
So my first question, has anyone else ever seen this problem?
So what to do about it? Well the whole business of taking a signal of the reverb output is rather bizarre. First, and I checked, the phase is such that it attenuates the signal to V4B, hardly a boost function. Second is that the phase shifts are significant and frequency dependant. If you wanted a gain switch it seems it would be a good deal easier to simply switch the 3.3M resistor value.
As one approach, I added an attenuator 4:1 on the reverb driver input, lowered the gain and added headroom by increasing the cathode(s) resistor to 1.2K. I also arrange the MV to switch a 1.5M in parallel with the 3.3M to actually give a boost. There is still plenty of reverb and you cannot overdrive the reverb driver any longer. I should be happy, but I'm not.
I suppose another strategy would be to leave the basic topology but add grid stoppers to V2B and V3 and also rewire the cable carrying the reverb signal to the switch with twin screened cable to cut out the feedback path but the sound when the reverb drive goes to cut off is pretty unmusical due to that inductive reverb tank load so I don't like this either.
Thoughts?
Bonus point area.
There's some other weird stuff about this design that I've never noticed before. Look where the low side of the MV pot goes: to the 100 ohm resistor in the tail of the LTPI. Why would they do that? Also I see a 0.01uf and 220k on the V6A plate, presumably to balance the signal levels but again returned to same 100 ohm. It all seems a little odd.
Comment