Just got a Vox Supreme from the original owner

[Helmholtz]

I have the vague idea that the DCV between R19/R20 is used to make the clipper more or less inoperative (don't completely understand how).
When SW3 is closed and TR6 is fully conducting, the upper clipper pair is directly connected to ground and can do its job.

[Mick Bailey]

I took some direct measurements from the circuit with respect to 0v

SW3 open - TR4 collector 6.48v, TR4 emitter 6.77v.
SW3 closed - TR4 collector 0.1v, TR4 emitter 0.1v

Something subtle is going on with TR5 and maybe scoping the circuit would give more insight. There's more thought gone into the design than I'd expected and a much higher component count than merely switching a pair of clipping diodes across the signal.

[Helmholtz]

So with SW3 open none of the upper "diodes" will clip before the positive peak of the signal exceeds around 6.5V (which is unlikely).
Not sure about the negative going part of the signal.

[Mick Bailey]

It's puzzling that TR5 is held reverse-biased with SW3 open.

[Helmholtz]

It's puzzling that TR5 is held reverse-biased with SW3 open.

Not really, because it can only conduct from collector to base. That's what I meant with reverse conduction.
I should only conduct when the collector is more positive than the base.

[Mick Bailey]

I understand the junction operation, the puzzle for me is why is it hard reverse biased - it's function isn't clear at all, particularly when considering clean mode. The designer took the trouble to DC isolate the clipping circuit, add a basing network and two additional transistors for clean operation. With the distortion engaged the circuit is simplified to a straightforward pair of clipping diodes across the signal path, something that Vox had done with earlier models. In clean mode though I can't deduce what the circuit does - it appears to be inactive, but they would not have added those components to do nothing.

If I have chance today I'll scope the amp to see what's going on. It's interesting though that despite the abundance of information on this series of amps there doesn't appear to be any analysis at all of this part of the circuit.

[Mick Bailey]

The original factory schematic shows the key DC voltages and with this particular amp all are very close - nothing unusual. Caps all test good for ESR and DC leakage. The clipping characteristics and output levels on the scope are interesting. Within normal guitar signal levels clipping is asymmetric and only when the signal gets way higher (around 200mV) does it clip on both peaks.

Here's the clean signal output from the preamp with 100mV input (0.1V/div):



Distortion engaged with distortion control at minimum with 100mV input:



Distortion at maximum with 100mV input:



Distortion at maximum with 200mV input:



With the amp powered off I checked the operation of the clipping section by feeding a +/-sine wave across C10 from my signal generator and it shows good symmetrical clipping on both positive and negative peaks, indicating that the main TR4/TR7clipping pair are functioning properly. I'm thinking that at least some of the reduced signal level with the distortion engaged is down the the AC capacitive divider action of C9/C10 when TR5 is shunted by TR6. The output levels are not affected by frequency.

Apologies for the blurred shots - they looked OK on the phone and suffered when resized.

[R.G.]

In doing the PCBs I also did a fair amount of simulation of the circuitry. C10 and TR5 are a little mysterious, but I believe they're intended to prevent the voltage on the collector of TR6 from going negative when it's switched on. TR6 clamps the signal to ground when it's turned on, and TR5 clamps it to ground for negative excursions. The collector-base connection of a transistor makes an effectively more ideal semiconductor diode than an actual diode, as the current gain eliminates certain diverences from the ideal exponential curves. Maybe that's why they used diode connected transistors. Maybe. It simulates and works pretty much the same with diode models in those positions.

The lower volume when distortion is engaged is a bit odd. They even lowered it further with R21/R22. I didn't understand that one. Subbing in silicon devices for TR4/TR7 or changing R21/R22 brings the signal size up, as you'd expect.

[Helmholtz]

I don't think C10 has much effect at low to mid frequencies as it's shunted by rel. low resistance.
To find out, just lift it.
Considering that Ge diodes typically have a forward voltage below 200mV (@ 0.1mA and taking into account the voltage divider R21/R22 (-6dB), your clipped signal level looks about right.
Don't understand the asymmetrical clipping, though.
Does the TR3 collector signal stay clean?

How is VR1 wired? What resistance?
Looks as if it controls the NFB level.

Curious: What happens if you lift the collector of TR5?

Edit: Missed post above.

[R.G.]

I had to go look up the schemo for the UL700 series preamps, and my memory was right - the UL700 series used a very similar distortion setup, but used OA100 diodes instead of diode connected transistors for the clipping diodes and also used a hard switch for clamping the diodes to ground instead of an NPN transistor. If my guess about TR5 is right, TR6 would be unnecessary, and it's not there in the UL700s.

[Mick Bailey]

With taking a lot more measurements and scoping the circuit right through the signal path I've now concluded that the amp is operating as the designers intended. It's certainly an unusual circuit and I can't help but think they could have simply used a DPDT switch to operate the distortion and eliminate quite a few components. The earlier UL series used a pair of Ge diodes switched to ground, though the first two gain stages were boosted via transistor switches. They still used a similar voltage divider after the clipping circuit though. The designers must have considered the earlier design lacking to adopt this circuit for all of the subsequent range.

Rather unexpectedly, the MRB control modifies he clipped waveform as scoped at the output of the preamp. This being a Mk3 has six rather than the three ranges of the Mk1 and the effect is to progressively modify the top of the clipped peaks until they become more like a letter 'M'.

I find this to be a fascinating amp and the more I play through it wonder why they weren't more commonly used. I've been playing some Schenker and George Lynch tunes through it today - no other effects - and it's surprisingly good. I got lost in the playing rather than fiddling with stuff.

[Mick Bailey]

VR1 is wired as a 5k variable resistance. Whilst it would be interesting to lift connections there's the risk of lifting pads with the old and rather fragile board. I think it would be better to breadboard the circuit for further experiments.

[R.G.]

For whatever it's worth, the "architecture" of the Supreme is very reminiscent of the Thomas Vox Beatle. I think these two were being designed during the time Thomas and UK Vox were cooperating, so that might be reasonable.

As to why they didn't use hard-contact switches, I think it was because the very slowed-down transitions of the base of the switch transistor(s) - TR6 in this case - were possibly an attempt to minimize any clicks or pops when switching. The simulation work did show clicks when changing state that were much worse with hard contacts than slowed-down NPNs. The designers apparently didn't have the devices we have today to fall back on, notably P-channel JFETs available at a reasonable price.

[Helmholtz]

Does the TR3 collector signal stay clean?

Sorry, that was a stupid question.

[Mick Bailey]

It occurred to me today that the signal reduction with the distortion engaged could be to protect the power amp from an excessive square wave input, as the heatsink area and airflow are limited.