The Marshall DSL "Resonance" circuit

[uneumann]

The prior discussion seems to assume this is a series resonance circuit producing a variable peak in the spkr output at the resonance freq.
At first glance I bought into that interpretation too.
Given some time to simulate this, I'm now seeing a different behavior. Here is my sim and the results.
Granted, this is just a sim of the NFB circuit - it doesn't include the PI or output stage - but it shows the general behavior of the circuit.

It turns out to be a parallel resonant circuit using the gyrator L value set by C139 and a parallel C value in the presence circuit C35.
The resonance is ~250 Hz and produces a peak in the NFB signal, meaning a dip or scoop in the spkr signal is to be expected - not a boost.
The effect peaks (~10 dB) when the Res and Pres controls are max (low resistance).
This isn't the whole circuit, and the PI tail connections will likely impact these figures.

Perceptually, the NFB increase at 250 Hz will reduce the resonance range content, which some people (me included) find a good thing. YMMV.
I find the placement of this control in the NFB loop (vs the tone stack or other preamp circuit) to be intriguing given the interactions with the Pres control.
I'd like to try one of these amps given an opportunity.

[GainFreak]

Considering the Resonance circuit is usually there to give a 2x12 or 4x12 cab some "help" in the 80-120Hz range 250Hz looks like too high.

[g1]

Considering the Resonance circuit is usually there to give a 2x12 or 4x12 cab some "help" in the 80-120Hz range 250Hz looks like too high.

The boost in the NFB means a dip at the output. So taking out some 250Hz range output might make it "seem like" there is more of the lower (than 250Hz) freq. content?

[Helmholtz]

Surprising.

So the resonant frequency varies with C5?
What happens if you add C89 and R40 from the Marshall schematic to complete the circuit?
Actually R40 "sees" around 11k in parallel from the upper tail.
What is the influence of C1 (C139)?
What do you see below 50Hz? The series resonant notch might show around 30Hz from gyrator formulas.

[uneumann]

Feel free to mod or edit my sim. I'm not that motivated to work out the last fraction of a dB or Hz. The sim tells me enough about what's going on.
As for bass boost - I don't think that's the Marshall intent. I can't read their minds, but I suspect their sense of "resonance" is borrowed from the music-mixing domain where the 200--300 Hz range is known by terms like " boxy" or "mud" or "boom" or "resonance". Freq's below that belong to the bass control anyway so bass can be boosted other ways.

In my subjective experience, I find that cutting that 200-300 Hz band (with a 1/3 octave EQ, for example) can help isolate the guitar from other instruments and drums.
Generally, less overlap with other instruments is desirable. Guitar amps often cut bass and boost treble to make a guitar stand out in a mix.
This is just another control to affect a specific band people seem sensitive to.
This is all subjective or course, but the circuit behavior aligns with this view.

[uneumann]

The boost in the NFB means a dip at the output. So taking out some 250Hz range output might make it "seem like" there is more of the lower (than 250Hz) freq. content?

Actually, you're right about the bass boost since the response chart shows that lower bass nfb freq's are attenuated.
So, while the spkr output at 250 Hz is cut, the lower bass freq''s are boosted. I hadn't notice that before.
That could be a desirable behavior.

[loudthud]

Speakers with a high Qts (>1) in a smallish (compared to Vas) closed box will tend to have a hump in the frequency response in the 200 to 300Hz range. This hump can be 2 or 3dB above the base efficiency.

[Greg Robinson]

I don't have the time right at the moment to go over uneumann's results, but they look fishy to me. I have previously simulated another Marshall's gyrator resonance circuit, this one is from the JVM410. And my simulations agree with what I've heard/measured with these circuits.
Here's the relevant schematics: Presence and Resonance control in centre right of page, & Power amp schematic

Here's the response with the resonance control being swept from 0/50/100 rotation:
Here's the value of C11 being stepped through different values, 22n, 10n (stock value), and 4n7:


And interestingly, here's the resonance control being swept with the NFB disconnected. You still get a degree of bass boost due to increasing impedance of the tail resistance at bass frequencies.

[uneumann]

I ....they look fishy to me.

OK - your circuit is similar, but your gyrator cap is 10n - the DSL uses 47n - which makes a difference. It moves the resonance freq down quite a bit.
Your comment got me to hook up the NFB circuit to the identical PI and an output stage. Here is the spkr out response (resistive load).

My transformer may be different, so scale factors may be off, but the 250 Hz dip is clearly there but not as impactful as I expected based on the NFB circuit alone.
The bass boost is even greater than the dip so that supports expectations that the control has a bass boost purpose.
The boost and dip are pretty localized - that may be the designers goal in using this circuit - and it also speaks to G1's (#18) and LTs (#22) comments.

[Greg Robinson]

OK - your circuit is similar, but your gyrator cap is 10n - the DSL uses 47n - which makes a difference. It moves the resonance freq down quite a bit.

Ah, I see now what tripped me up, you had the presence control set to 0.95 for your sweep, which caused that bump in the feedback signal around ~250Hz. I'd only taken a cursory look and expected the presence control to be set at 0.

[uneumann]

Here is a spkr signal plot with both controls varied from zero to mid to max. You can clearly see the damping effect of the controls.
The "L/C" dip is zero (flat response red line) when both controls are min since the control resistances are max and they both damp the resonance.
The dip is max when controls are set to min resistance (max res and pres green line).
The variation is smooth for settings in between.
The response is similar to many tone stacks, except that the dip is ~250 Hz.

Disclaimer: the degree of boost/dip depends on accurate OT specs which I don't have.

[Helmholtz]

I was curious to see the actual series resonance.
In fact simulation shows a distinct -12dB dip at 25Hz in the NFB circuit transfer response with resonance control at maximum and presence at minimum.

This corresponds to a gyrator inductance of 4H.
(This is a higher value than gyrator theory predicts (2.35H). But an emitter follower isn't an ideal buffer. Particularly output impedance isn't zero and gain is lower than 1.)

To crosscheck I simulated the response using a real inductor. I could exactly match the reponse of the gyrator circuit with an inductor of 4H in series with a 500R resistor.

[Helmholtz]

Disclaimer: the degree of boost/dip depends on accurate OT specs which I don't have.

The voltage step-down ratio from full primary to the 16 Ohm output of a typical Marshall 50W (2xEL34) OT is 1704 turns/ 119 turns = 1/14.32 = 0.07 (from a tear down of a Drake 784-139).
Corresponding to an Raa of 3.3k.

Full primary inductance @50Hz is
- around 5H at close to zero output
- rising to 100H at around 22W output
- dropping to around 13H at 60W output.

From this it follows that (open loop) LF gain varies with output power (more exactly with the primary side voltage-time integral).

[Helmholtz]

Full primary inductance @50Hz is...
- dropping to around 13H at 60W output.
.

Sorry I misread my notes. Full primary inductance at 60W is 50H.

[GoldtoneWorkshop]

Cool gyrator circuit breakdown from Aaron Lanterman from Georgia Tech. This is specific to it's use in guitar pedals, but might be helpful further reading for those like me who just learned what a gyrator circuit was from this thread.

https://www.youtube.com/watch?v=L4MGG3kt5pY

P.S. he does a whole series on the engineering behind guitar amplification and all of the math behind it starting at the pickup and going all the way through to the speaker.