JCM 800 NFB & presence

[dtier]

I just meant that the 470r,10k,and 5k resistances should remain in their respective ranges unless redesigning/evaluating the whole the PI stage. That would include all of the presence pot configurations as they all retain roughly 5k as the feedback shunt resistance. Though technically the 470r bias resistor is not considered part of the tail.

 

[XTRXTR]

I think of a valve power amp negative feedback loop as being akin to an opamp. Almost invariably * they’re configured as non inverting voltage amplifiers.
As such, the main signal grid of the LTP can be considered as being the non inverting input.
Whilst the other grid, which is decoupled in an open loop amp, is the inverting input, used for the NFB signal in a non inverting opamp.
The signals at the inputs / grids are in the same polarity.

The Bassman AB165 and later models are the main exception, power amp is configured as an inverting opamp. It’s main and feedback signals are in opposing polarity, they’re mixed together at the main grid (inverting input), with the other grid (non inverting input) decoupled.

If you drive the grid the anode is 180°, If you drive the cathode the anode is 0°.

I like your analogy of the open loop opamp on the inverting side. As you say the DC is decoupled. If I follow you, without the NFB, the signal would have a certain runaway gain on the inverting side. Possibly becoming unstable with a honk or squeal at some resonant frequency - probably odd harmonic heavy distortion.

With the NFB you drive the inverting input grid pin 7 with enough of a 180° amplitude to partially disrupt/cancel the signal being driven by the shared cathode pin 3 and pin 8 at 0° that you have a reduced 0° output at the anode pin 6. Then of course the frequency of that NFB signal and the mixed anode output is shaped somewhat by the pin 7 grid coupling cap, the presence circuit, any other frequency changing component like an ice pick cap on the NFB resistor, resonance circuit etc.

I hope that is right because that makes sense to me as far as what the NFB is doing to the output on pin 6. The presence circuit makes more sense too.

 

[Pete Farrington]

without the NFB, the signal would have a certain runaway gain on the inverting side

Each section of a typical ECC83 LTP has a gain of about 30, ie about the same as an equivalent common cathode gain stage with unbypassed cathode.
It’s pretty much essential for valve guitar amps to operate correctly and stably both open and closed loop. As a presence control
An issue with using opamp terminology of inverting and non inverting inputs here is that unfortunately it runs counter to the operation of the LTP sections. ie the common cathode section used for tge main input inverts the signal, though from a opamp ‘system input to output’ perspective, it’s the non inverting input.
So whilst viewing the power amp as being a big opamp may help with getting a handle on what’s going on with the feedback loop (it certainly does with me), the terminology can make it a double edged sword when communicating those ideas.

 

[dtier]

An enigmatic circuit for sure. The input and the feedback are summed and present in both the outputs, hence Pete's op amp analogy.

The input at Va causes an in phase signal on the tail resistor. That is picked up by Vb which shares it. I think Vb is operating in common grid mode as far as Va is concerned. So Va is a common cathode with output inverted and Vb is common grid with output non-inverted resulting in complimentary outputs.

The feedback signal, as Pete stated, is applied to what would be the uncoupled input in an open loop amp, Vb's grid. I think the roles of Va and Vb are exactly reversed for the feedback signal. Vb is the input, operating as a common cathode, and Va, operating as common grid, picks it up on the tail resistor. So both outputs are the sum of the input signal and the feedback.

I stated "I think" because I am still learning and am not stating this as a fact.

 

[Pete Farrington]

The input at Va causes an in phase signal on the tail resistor. That is picked up by Vb which shares it. I think Vb is operating in common grid mode as far as Va is concerned. So Va is a common cathode with output inverted and Vb is common grid with output non-inverted resulting in complimentary outputs.

The feedback signal, as Pete stated, is applied to what would be the uncoupled input in an open loop amp, Vb's grid. I think the roles of Va and Vb are exactly reversed for the feedback signal. Vb is the input, operating as a common cathode, and Va, operating as common grid, picks it up on the tail resistor. So both outputs are the sum of the input signal and the feedback.

Yes, that’s how I see it too.

 

[XTRXTR]

@Pete Farrington
I get your point about the terminology differences. For clarity sake mixing those ideas could muddy the water when trying to understand how the PI, NFB, and Presence circuits are working together.

An ECC83 is a 9 electrode vacuum closed glass cylinder where as an Opamp is a cute robot that fell over, on a page.

 

[arthur.lowery]

Hi

I've just written this: https://www.marshallforum.com/threads/plexi-presence-options.135316/ after a morning of experimentation on a Plexi Black Flag.

Do the circuits in the pdf agree with your view?

Cheers
Arthur

 

[arthur.lowery]

Yes, that’s how I see it too.

Yes, I agree that for the signal input (from the tone stack), V3b acts as a grounded grid. The coupled cathodes mean that the input voltage is shared between V3a and V3b. If the tail resistor (470 + 10k + presence circuit) was infinitely large (or long, as in long-tailed pair), then the sharing would be perfect. As it's only 15.47k, then there's less of current swing on Vb3's anode. So to balance the circuit, V3a has a lower-value anode resistor.

Now, we could analyse from the other side - the effect of the negative feedback voltage. This is slightly more tricky as the feedback is fed to the bottom end of the 10k, and the grid of V3b. I did a LTspice simulation, and despite this asymmetry (and the different anode loads), the outputs at the anodes are quite well balanced. Probably I should do the maths!

In hi-fi amps I have designed, I try to keep things symmetrical, and use a constant current source instead of the 10 k resistor. The feedback goes into the grid of V3b: the signal into the grid of V3a.

Alternatively, many hi-fi amps use feedback to the cathode of the first valve. This works OK as the feedback network has a low impedance, so can drive sufficient current into this point. With a Marshall, there would be too many phase shifts between the first valve and the output stage - yes, the tone-control stack - so it would be unstable. Also, such a design would be stupid, as the negative feedback would mostly cancel out the effect of the tone stack!

 

[Pete Farrington]

Aiken’s analysis is good, he notes that applying the feedback to the base of the tail results in both negative and positive feedback, both of which help to balance the outputs.

The Long-Tail Pair

Guitar amplifiers

aikenamps.com

 

[dtier]

The fourth paragraph of Aiken's Long-Tail Pair article took awhile to sink in for me. Mixing the feedback into the tail, as opposed to just applying it to the vb grid alone, injects a signal into the tail that offsets the difference in the plate resistors, which were adjusted to balance the outputs for the input signal. All while keeping the tail resistance down as to not give up headroom in the stage. I have not put in the homework to determine if this is all necessary to get enough signal swing to drive a 6L6 or EL34 sufficiently , at the available plate supply voltage, as opposed to just using a larger tail resistance which Marshall did with their 18 watt EL84 series. But I am curious if this inverter circuit has been preserved traditionally or more for it's function or maybe it,s distortion character.

 

[arthur.lowery]

I guess one mod would be to use the negative bias supply to feed the long-tailed pair. Then the grids could be biased by the 0V (ground), cutting out the complexity and also providing more headroom to drive the output tubes. The early valve op-amps did this. https://hackaday.com/2018/08/09/op-amps-before-transistors-a-600v-vacuum-tube-monster/
I've used this trick on hi-fi amps. The "long tail" is the length of carbon resistor you need.

 

[Pete Farrington]

Also, such a design would be stupid, as the negative feedback would mostly cancel out the effect of the tone stack!

I think Fender kinda tried and realised that, with the short lived (3 months to be superseded!) 5E6 Bassman https://el34world.com/charts/Schematics/files/Fender/Fender_bassman_5f6.pdf

get enough signal swing to drive a 6L6 or EL34 sufficiently , at the available plate supply voltage, as opposed to just using a larger tail resistance

To achieve max AB1 power (typically V g1-k=0, requires the positive peak of the applied signal waveform to equal the bias voltage), KT66 require more signal swing than 6L6 than EL34.

With overdriven guitar amps, having much more signal swing available there than necessary is a drawback, as it’ll tend to cause bias shift, possibly leading to blocking distortion, as the degree of overdrive is increased.

Through the 60s, we can see Marshall realised this and managed it by adjusting the value of the LTP HT supply node voltage, using values 8k2 to up to 20k there.

 

[arthur.lowery]

H

Not to get off the original OP's subject but I suspect this phase shift is why some amps can get by with the global feedback reversed (positive) and not oscillate as it isn't perfectly in phase with the PI input.

ii, I suspect with reversed feedback the amp would find another (far lower) frequency to oscillate at. Have you experienced this? Cheers Arthur

 

[dtier]

ii, I suspect with reversed feedback the amp would find another (far lower) frequency to oscillate at. Have you experienced this? Cheers Arthur

My experience with a reversed feedback loop is limited to a 6V6 audio to guitar power amp conversion I did decades back. I knew something was up as it sounded different than the two I had done before. It had no audible oscillation but sounded interesting. A hotter overdriven sound that was reminiscent of EVH's Marshall but at the same time something about it was off. I suspect it was likely unstable. Once I realized what was up I corrected the phasing. I always wanted to revisit this. There seems to be a lot going on, tone wise, with phasing in the global feedback loop and the effects of a specific output transformer and circuit values on it.

 

[mickeydg5]

As brought up, I think the biggest thing was the 10k feedback resistor; very none Marshall.
Even Fender was not that far off.
You gotta get that negative feedback lowered.