Overcoming interaction between control settings in FMV tone stack

[Gnobuddy]

But it's worth noting that all of those amps (even the old Carvin) are designed for preamp gain, then EQ, then power amplification. So it's likely they are designed to avoid overdriving the EQ since that would nullify the intended design advantages.

Agreed, that is the only thing that makes sense to me. They probably had signal attenuation before the EQ circuits.

I'm not familiar with any of those amps, but if any have FX loops, they would already have a point in the circuit with pretty low signal levels (at the FX send, and FX return). It might have made sense to put the EQ bits in one of those locations.

-Gnobuddy

[Chuck H]

Agreed, that is the only thing that makes sense to me. They probably had signal attenuation before the EQ circuits.

I'm not familiar with any of those amps, but if any have FX loops, they would already have a point in the circuit with pretty low signal levels (at the FX send, and FX return). It might have made sense to put the EQ bits in one of those locations.

-Gnobuddy

In fact that IS the case. I almost mentioned it, but shot for brevity instead. Yep, the EQ and loop circuits are typically side by side in such amps. Usually EQ then loop.

[MarkusBass]

I couldn't find a trustworthy LTSpice model for a logarithmic pot, so I made my own.

There are several LTSpice models for a logarithmic pot and one of them is directly in LTSpice (but it's difficult to find it). I use it in my simulation (I will post it below). Please run it and tell me whether it runs OK without the symbol delivered additionally. I think it will run correctly.

This is a kludge to keep LTSpice happy. LTSpice has a hissy-fit if either resistance goes to zero in the pot. To avoid that problem, I just added one milli-ohm onto each resistor. It is negligibly small and won't affect accuracy at all, but it lets you use the "step" command without worrying about LTSpice throwing errors at the zero end of the range.
LTSpice is weird about the prefix "m". If you enter "1m", it means one milli-ohm. If you enter "1M", it still means one milli-ohm! You have to enter "1Meg" or "1meg" if you want 1,000,000 ohms.

I understand the equations but I was worried about the symbol "1 m" (with a blank in between). I think that if you write it with a blank, the "m" symbol will be ignored and you get 1 ohm instead of 1 milli-ohm as you wanted.

Can you tell us more about your LTSpice potentiometer model? Did you make it, did you find it? Have you checked the source impedance is correct at all pot settings?

As I wrote above, the model is directly from LTSpice. I used it a lot in my simulations and it's a shame but I didn't check it . Maybe you could verify it? The equation that I use is directly in the subcircuit definition. If you find it incorrect, you can modify the subcircuit (even with your equations).

You said also that you don't know how it sounds. But I hope that you know that in LTSpice you can specify a *.wav file as a source signal and save the output signal. In this way you can "hear" the circuit without building it.

Here is my version of the simulation (I also changed the extension of the file to TXT):

Voight_Tone_Control_V2.txt

Mark

[bob p]

Voigt wrote about it in the 1930s. It was designed for use with old record players. I could not find it with Google either -- I guess you have to be an old radio guy to know where to find it.

From: Wireless World, April 1940:

I've been corresponding with Duncan Munro by email, and I pointed him to this thread. It sounds like the Voigt tonestack will be included in the forthcoming update to DTSC. I think that's great news. I'd like to thank him for taking interest.

[benito_red]

You're very welcome!

Beware of what you hear from today's crop of mostly technically ignorant audiophiles. Many don't have a clue, and believe a lot of nonsense. The nonsense gets repeated over and over, and eventually, there is so much nonsense that it can be very hard to find the reality at all.

How do we know if delay in the feedback signal is a problem? Simple. If there was a delay in the feedback, it would immediately show up as either outright oscillation, or a big peak in the frequency response curve: delayed feedback is mathematically identical to phase lag, and too much lag causes oscillation. If there's not enough to cause oscillation, it causes a peak in the frequency response.

So if you measure a nice well-behaved frequency response, it's a guarantee that there is no mysterious feedback delay lurking in the background.

Measuring frequency response is quite easy to do, all you need is a sine wave signal and some way to measure the strength of the signal. You can even do this with the sound-card in your computer and a few external components.

In reality, today's semiconductors are very fast, and provide accurate feedback far, far beyond the highest end of the audio bandwidth. Of course the feedback is delayed - but the delay is so short that it doesn't matter until you get to frequencies far, far, far, far above audio.

For guitar, we only need things to behave well until maybe 10 KHz. This is trivially easy with the amazing transistors and MOSFETs we can buy cheaply today.

-Gnobuddy

phase delay is exactly what I meant, but IMO a "well behaved frequency response" is not a warranty that you do not have marginal transient instability even without oscillating, as the measurement of freq response is done by means sine wave signals, while the music signal is never a sine wave. Of course not every active tone control suffers of transient instability, but a passive tone control will never be affected for sure.

[Gnobuddy]

a "well behaved frequency response" is not a warranty that you do not have marginal transient instability even without oscillating, as the measurement of freq response is done by means sine wave signals, while the music signal is never a sine wave.

This is exactly the mathematical misunderstanding I was speaking about. Many such comments on the 'Web come from people who do not understand the relevant mathematics, and are suspicious of it simply because they don't understand it.

In fact, a well behaved frequency response IS a warranty that you do not have transient instability. The (sine wave) frequency response, and the (impulse or transient) time response are completely and 100% connected to each other by the Fourier transform. One completely describes the other, subject to certain minimal conditions that amplifiers certainly satisfy. Millions and millions of amplifiers have demonstrated this over the last hundred years or so. There is no mystery, there is no doubt, there are no transient-monsters hiding under the bed in the dark.

Perhaps you are thinking of the infamous "transient intermodulation distortion", or TIM. Some decades ago, Mattie Otala thought he had found something new, though in fact he had re-discovered what every good control systems engineer already knew: if you feed signals that are too fast for it to handle into a feedback amplifier, it will misbehave. (Peter Walker of QUAD fame pointed this out, very politely, shortly after Otala's paper appeared, but people enjoy fussing over something new and exciting, so Walker went unheard, and Otala was in the spotlight for a while.)

Otala got his fifteen minutes of fame by naming this "Transient Intermodulation Distortion" and publishing a paper on it. But it really was nothing new, and really described a failure of the engineer rather than a failure of the amplifier: you must not expect any amplifier, feedback or otherwise, to properly amplify signals that are outside its bandwidth. If you expect such signals to come along, you must bandpass-filter the input to your amp. And that is the end of the non-existent "problem" of TIM.

In the years since, TIM has faded to audio engineering obscurity as it deserves, and is only kept alive today in woo-woo audiophile circles, by faith rather than by logic and math.

Of course not every active tone control suffers of transient instability, but a passive tone control will never be affected for sure.

Agreed. However, to use a similar example: not every car will run out of petrol on the highway, but a horse will never run out of petrol. And yet, very few of us would choose to ride a horse rather than drive a car on the highway!

While running out of petrol is a possibility, it is a low-probability one, and therefore, not a high risk to take. And horses, like passive tone controls, have their own downsides! D

IMO engineering decisions should be made based on whatever design is most appropriate and best suited to the job. If a passive tone control is the appropriate solution, wonderful! And if an active tone control is the appropriate solution, wonderful, use that instead! (And don't worry about mysterious unknown transient effects that don't actually exist.)

-Gnobuddy

[J M Fahey]

In the years since, TIM has faded to audio engineering obscurity as it deserves, and is only kept alive today in woo-woo audiophile circles, by faith rather than by logic and math.

Its unrecognized illegitimate son is still very much alive, revered .... and fully misunderstood: "feedback is bad bad, and no feedback is Nirvana" , which of course is music in the ears for those who donīt understand Feedback and canīt design around it

[uneumann]

Is there really a community (woo-woo or otherwise) where "FB is Bad"? Am I missing out on something...?

[bob p]

Just ask someone who plays an AC 30.

[SoulFetish]

Agreed. As mentioned in my earlier post, the problem with an active tone control is even worse than just experiencing abrupt overload: the frequency response depends one the existence of negative feedback, and if you clip the output, the open loop gain goes to zero. Which means the feedback goes to zero as well. Which means the intended tone-control frequency response goes to he@@ in a handbasket! No more tone control, just some wildly unpredictable and quite unintended frequency response!

Also as mentioned in my earlier post, I think there is a simple solution: use an attenuator ahead of the tone control, set so that it is completely impossible for the tone control to ever clip, even when the stage driving it is at full clip. The idea here is that the preceding stage acts as a limiter, ensuring that the tone control stage itself cannot be clipped. Problem solved!

I have already used this approach successfully in a different situation: I have a valve guitar amp that uses a little MOSFET as a "sourceodyne" phase inverter. It works better than a triode phase inverter in every way, except it will probably sound nasty if clipped. So I put a fixed attenuator between the (valve) stage that drives it, and the MOSFET gate. Attenuation is set so that, with the valve at full clip, the MOSFET is still happily linear. (Easy to do because MOSFETs only need a couple of volts between source and drain to keep operating perfectly, unlike a 12AX7 that wants a hundred volts or more.)

This is an approach that Leo would not have liked at all: gain costs money! Who would throw away gain?

But of course, inter-stage attenuators to throw away gain are quite routine in high-gain amps of the sort Leo never imagined. The concept isn't new today, and has a long history.

-Gnobuddy

an attenuator is one way to ensure that the input of an active control does not clip. You would have to design it around the level at which the previous stage clips obviously. However another approach is to use a Cathode follower operating in it's linear range as the input of an active control. Even better would be to use a pentode cathode follower or mosfet source follower as the output voltage can swing almost rail to rail ensuring that it will not clip and act as a transparent stage.
I'm going to look into a post Phase Inverter active control for my next build. Since it's a balanced system anyway, I may use some sort of modified cross coupled tone circuit using the existing +/- supply and use the cathode followers here (below) as inputs (different output tubes this time. I love EL84s, but I'm tired of them at the moment and want to work on something new)

[Gnobuddy]

Is there really a community (woo-woo or otherwise) where "FB is Bad"? Am I missing out on something...?

Yes, absolutely, a huge community that doesn't understand feedback, has forgotten their high-school algebra, has never seen a differential equation, but is quite sure that feedback is evil, and more feedback is more evil. Exactly the opposite of what the actual feedback equations say!

You know what they say about the Internet, it allows people who share the same opinions to find each other and reinforce their beliefs. Sometimes those beliefs are extremely misguided, as in the case of the "flat earthers", and the "Feedback is evil!" people.

The next step after "Feedback is evil!" is to say that a well-designed solid state audio amplifier with 0.005% THD at all usable power levels and at all audible frequencies sounds bad, because, you know, "evil feedback!"

Keep in mind, these people are talking about low-distortion Hi-Fi audio, not guitar amps where we may actually want lots of distortion and little or no feedback, for perfectly good engineering reasons.

So what kind of Hi-Fi amp is best? A pre-WWII "directly heated triode" amplifier that costs a few thousand dollars, and puts out one watt of audio at 1% or 2% or maybe 5% THD. Now you get to also buy expensive speakers with 92 dB@1W@1m sensitivity - the only way to get enough listening volume. Never mind that these high-efficiency speakers usually suffer from a host of their own problems, including light, floppy cones suffering from massive cone-breakup at higher frequencies, limited bass extension because of the lightweight moving parts, etc, etc.

I'm more than willing to agree that some types of distortion sound good for musical instrument amplification, I just don't think Hi-Fi is the place for heavily distorting amplifiers with miniscule power outputs at exorbitant cost. Hi-Fi is about accurate audio reproduction, not intentional distortion.

When I first stumbled across the "feedback is bad!" school of belief, I used to try logic. For instance: try writing a paragraph in a lined notebook with your eyes closed, removing the negative feedback loop in which your eyes observe the line left by your pen and feed it back to your brain for correction. Does your handwriting get better without feedback, or worse? Did you do better at staying within the lines, or worse?

It didn't work, of course. I quickly found out that logic and the evidence of their own senses doesn't actually change the minds of most people, once they've formed an emotional attachment to a certain idea.

Which reminds me of a funny story. When I was quite young, and living in a small town, I borrowed a book on kites from the public library, and entertained myself one summer by building and flying several of them. On one occasion, I was out in a big open area right behind a bus-stop, flying a Delta Conyne kite I'd just made ( https://www.batchelors.net/kites-in-...ta-conyne.html ). An older boy waiting at the bus stop saw me and my kite, walked up to me, looked at my kite flying in the sky, looked at me contemptuously, and in a pitying tone said "How on earth do you expect that stupid kite to ever fly?"

I swear, this actually happened!

-Gnobuddy

[Gnobuddy]

Its unrecognized illegitimate son is still very much alive, revered .... and fully misunderstood: "feedback is bad bad, and no feedback is Nirvana" , which of course is music in the ears for those who donīt understand Feedback and canīt design around it

Excellent point! This is something that I find quite frustrating, so see the brilliant work of geniuses (like Harold Stephen Black) discarded in favour of superstitious nonsense. Any day now I expect to see amplifier designs optimized by reading tea-leaves at the bottom of a cup.

-Gnobuddy

[Gnobuddy]

Even better would be to use a pentode cathode follower or mosfet source follower as the output voltage can swing almost rail to rail ensuring that it will not clip and act as a transparent stage.

My worry was that the active tone control itself has signal gain (at settings where bass or treble is boosted), so its own output will clip, even if the signal coming into it isn't clipped.

To put some numbers to it: the one-knob tone control I'm experimenting with now has up to +12 dB of boost at 80 Hz and 5 kHz (i.e. a gain of about four times). Suppose you're using a 12AX7 with a 300 V B+, and its output can swing, say, 200 Vpp before clipping; the input will only need to swing 50 Vpp at 80 Hz or 5 kHz to cause the output to clip.

So I figure I have to do something to make sure the input to this particular tone control never gets larger than, say, 40 Vpp, to allow some safety margin.

If you drive it from a pentode cathode follower that can cleanly swing 250 Vpp without clipping, that's all well and good for the cathode follower, but that 250 Vpp will clip the tone control terribly.

The only way I can see to keep this from happening is to attenuate the output of the cathode follower down from 250 Vpp to 40 Vpp, i.e by a factor of 6.25 times. Now the cathode follower can be driven all the way to (hopefully good-sounding) clipping, but the active tone control still won't clip.

Does that make sense, or am I misunderstanding what you have in mind?

-Gnobuddy

[Gnobuddy]

Just ask someone who plays an AC 30.

An electric guitar amp may be the one place where the "Feedback is bad!" belief has at least some good justifications for existence.

For the last DIY guitar amp I built, I wanted a very smooth and progressive increase in distortion as the amp started to overdrive. If that's what you want, feedback IS bad!

But for Hi-Fi?

-Gnobuddy

[Enzo]

An active control has gain, as far as I know by definition. As to it clipping, EVERY stage in an amp has potential to clip. But it doesn't clip just because it is a gain stage. It clips because the signal level exceeds the ability of the stage.

You calculated some input level maximum. I won't quibble with the math, I will assume it is correct. The question then becomes : will you have that level at the stage input?

And like any control, just because the control goes to ten doesn't mean you have to turn it that far.

The passive tone stack has a MASSIVE insertion loss, the gains of other stages compensate. SO with an active EQ, we compensate in the design of the rest of the amp. Even on a clean amp, we can turn controls up to the point of clipping.