Lets discuss amp tone and how it changes w/volume and why

[Chuck H]

+1
Too much antagonism. Having the actual info is pretty boss, but in the end what works for daz is right for daz. Teemuks input has been spot on. But, to make it a personal note maybe his delivery could use a little polish Other than safe operation of the components I've never known of any rules when it comes to tone. And FWIW the bootstrapped MV circuit (in it's many guises) seems to be fairly popular. So your not alone in your preference. You don't need to be so thin skinned about it. (There! I've probably managed to avert this turning into a flame thread and offended two people at the same time!!!)

[daz]

No offense taken. I'm just tired of guys who just want to have a go at me under the guise of trying to be helpful. They always do the exact same thing....they pick thru my post till they see one thing they can argue over even tho they have to take it literally to even have an argument at all. If i said my car was silver he'd say no, you're wrong, it's metallic silver. I got past this cr@p in 3rd grade.

[jazbo8]

Just saw your post over at AX84, thought I would reply here to keep the thread together... from the schematic that you posted above, what you have done is a pre-PI MV, and as you found out, it messed up your GNFB. Now that you have isolated the problem to the MV, try this simple post-PI MV from Craig Anderson's 1982 Guitar Player article, which is a tried and true solution that many have been using for years. Good luck with the mod.

[Dave H]

Jazbo, how can a pre-PI master volume mess up the global negative feedback when it’s outside the loop? I think that any kind of post-PI master volume inside the loop will mess up the feedback because it reduces loop gain. The feedback will also be trying to negate the effect of a master inside the loop. It will be working more like a feedback control until the amp runs out of loop gain.

That BS (Bull Sh1t?) master looks like it’s outside the loop to me. Does it work by reducing the impedance looking into the LTP grid therefore loading down the driving stage? That would really mess up the frequency response if it’s being driven by a tone stack.

Thanks
Dave H.

[teemuk]

Rest assured it's in the loop alright.

Here's what the bootstrapped master volume does to response vs. conventional master volume (out_ref):



Bottom graph illustrates response when master volume is dialled down. Green curve is the response of an amplifier with conventional master volume control, blue graph is the response of an amp with bootstrapped master volume. You can clearly see the effects of decreased loop gain as decreased damping, which totally unlinearises the response. Same loss of feedback will also increase distortion, affect the "knee" of clipping, and possibly create some oscillation.

The graph also illustrates one of BS master volume's drawbacks: the potentiometer is referenced to cathode, which has signal in it. Effectively, the master volume can't be turned fully down. Blue curve of the conventional master volume portrays much more attenuation at same settings.

Upper graph illustrates response when master volume is dialled up. You can see that the blue curve (bootstrapped master volume) is now pretty much morphed into resembling the green one (conventional master volume). Feedback has increased and this has increased damping factor, levelling the response. Increased feedback also has effect on distortion and knee of clipping.

Thus, the conventional master volume setup provides the most consistent results throughout the dial range whereas the BS master volume essentially morphs between an amp with or without feedback. The differences between extremes of the dial are huge and every dial setting provides an unique blend "from between".

---

This graph illustrates the effect on feedback-based tone controls:



Blue curve is again the response of BS master volume and green the response of conventional master volume. Bottom graphs illustrate low volume settings, upper graphs high volume settings. As can be clearly seen, in BS master volume setup at low volumes there is very little negative feedback and the effect of the presence control is therefore quite small, almost non-existent in practice. Compare that to the effect of presence control in conventional master volume amp with constant loop gain. Upper graph again shows how characteristics of the amps turn similar towards higher volume settings and how the increased loop gain turns the feedback-based tone controls effective.

The BS master volume thus has feedback-based tone controls, which's effectiveness also "blends" in between low and high volume settings.

Yes, I know there are many people who prefer such variations, but concerning consistency of tone throughout the volume dial this BS master volume setup is an absolute catastrophy.

The plots were created with LTSPICE using a simulation of a standard Marshall -style power amp with 12AX7 LTP PI and 2xEL34 power stage, B+ 420V. Primary load was 8K with 8R load in the secondary and the feedback takeoff point was 8-ohm tap. BSMV setup used a dual 1Meg pot. Negative feedback was 82K shunted by 5K tail resistor in the PI. For presence simulation the tail resistor was turned to 5K potentiometer with wiper connected to 100nF cap. Secondary load is a circuit matching reactive characteristics of a real loudspeaker, thus in real life resonant frequency and inductive characteristics will vary from speaker to speaker.

[daz]

Just saw your post over at AX84, thought I would reply here to keep the thread together... from the schematic that you posted above, what you have done is a pre-PI MV, and as you found out, it messed up your GNFB. Now that you have isolated the problem to the MV, try this simple post-PI MV from Craig Anderson's 1982 Guitar Player article, which is a tried and true solution that many have been using for years. Good luck with the mod.

[ATTACH=CONFIG]26645[/ATTACH]

Thanks, I've tried those several times hoping each time that other changes i had made might make it work right each time, but it just never does. It drives the PI hard and the resulting sound is like the worse SS amp made in the 60's. Literally not usable. I read up on it and a lot of people say the same and it seems to be that the ones who like it are those with NON cascaded preamps which i suppose never drives the PI that hard and gets equal amounts of OD from every section. In fact, i'v tried every master i have seen and this BS master is by far the best so far. granted, it has issues due to no NFB and therefore it's too bright. But i'm working on that. I finally found a cap that makes the cut control work pretty well with it, and combined with a particular snubber at this point it works well. I'd rather have NFB because i like the way it smooths high end, but the tone is already a lot different and better than before. Plus the fact that made me post this thread in the first place, anything higher than bedroom volume was literally unusable. It really comes down to whether the amp is tuned for high or low volume. It was tuned for low and still is, and with the pre pi master that was a nightmare turned up because it wasn't JUST bright, it totally changed in every imaginable way to a sound that was unusable. Now it's basically the same sound when i turn it up, just much brighter. When i smooth that out with the cut and with a snubber at V1, the tone at stage volume is as close to the same as low volume as i've ever gotten from any amp. The only issue is it's a real pain to adjust the cut and tone controls just right every time i adjust the volume more than a tad. If it had NFB or some other form of high end smoothing that was more consistant thru the volume range it would make life a lot easier. In any case, i just found this tweak so I'm sure with time i'll find a better solution.

[teemuk]

More graphs.

These illustrate the effect of stage impedances and their interaction. Plots are frequency responses at the PI's input grid.



Graphs at bottom are once again illustrating response when master volume is turned down. Graphs at top are depicting responses when master volume is turned up.

Assuming there was an impedance bridging issue that would effect the frequency response according to MV's dial setting thse graphs would reveal it. The simulation again uses the same Marshall circuit as earlier. Tone stack is standard Marshall tonestack fed by stage with Zout=470R. Tone stack is followed by 1M master volume coupled to PI with 22nF cap.

First graph at left depicts the response of an unbuffered circuit versus the response of a circuit where a buffer stage (Zin=5M, Zout=1K) is inserted in between tonestack and master volume. As we can see from the graphs, the difference to response in both scenarios is practically nil. In addition, the top and bottom graph outline perfectly hinting that impedances really have no drastic effect to frequency response whether the MV is turned up or down. The shape of the response curve stays practically the same in each scenario.

This applies at least to the aforementioned standard Marshall -style PA circuit with given design specs. Other amplifiers with different circuit architectures and designs may naturally perform very differently.

Graph at the middle depicts the same reference circuit versus buffered circuit in which the buffer (having the same specs as earlier) is now placed in between master volume control and PI. Again both graphs outline perfectly. This setup even corrects the tiny 2dB skew BELOW AUDIBLE BANDWIDTH, which was the difference between buffered and unbuffered circuits in previous example.

Third graph depicts again the same response, now generic master volume vs. bootstrapped master volume. Again, the response of the ordinary master volume control, our reference circuit, outlines perfectly at both high and low dial settings revealing very little deviation in response across the potentiometer dial. We can't say the same thing about response of the bootstrapped master volume, which only is not worse in respect of attenuation but also introduces additional peaks to the response due to its feedback arrangement.

[Chuck H]

That's great work teemuk! (and I'm enjoying the graphs greatly) But in the end it doesn't matter that any master is making something different happen in the circuit because something different will ALWAYS be happening within the circuit. The only thing that matters other than safe operating parameters for the components is HOW IT SOUNDS! Daz reports that it sounds good to him and that's all that matters. Consistent responses and operating conditions DON'T guarantee the amp will sound/feel the same at lower volumes vs. higher volumes because some other things change too that are outside of the considerations you've plotted. If the different operating conditions make the amp sound/feel more consistent at all volumes to daz then that's what really matters.

Though I have to admit that after looking at the circuit and considering it, I probably would have canned the BSMV before even trying it! BTW, I'm enjoying the acronym "BS" for this circuit The fact that it so drastically affects the GNFB circuit is something I can't abide because it nullifies the use of control panel functions like presence, variable NFB circuits and resonance circuits. Pretty lame IMO. Just another circuit most people use because they can instead of because they should. But I'll reiterate... If daz likes it then it's right for daz. There's no point beating him up about what the circuit is ACTUALLY doing!?! He's listening to it. I think the validity of that has to be recognized. It's just insulting and antagonistic to imply that daz doesn't know what he should like or what sounds good to him because he doesn't have the technical prowess to understand the circuits or own bench test gear. That's kinda absurd actually.

EDIT: Now if you were to offer a circuit that DID accomplish your criteria for consistency that would change everything! Then it could be implemented and tested BY EAR! Since that is the only way to know if a circuit sounds the way you want it to.

[teemuk]

Consistent responses and operating conditions DON'T guarantee the amp will sound/feel the same at lower volumes vs. higher volumes because some other things change too that are outside of the considerations you've plotted.

The plots are of course useful only for analysis what happens in frequency domain.

There's no point beating him up about what the circuit is ACTUALLY doing!?!

I'm sorry if you feel that this is a personal attack All I'm trying to show is the actual function of the discussed circuits. If it disagrees with what Daz (or anyone else) hears then I can't help that.

The only thing that matters other than safe operating parameters for the components is HOW IT SOUNDS!

That's the ultimate goal. However, this thread is titled "Lets discuss amp tone and how it changes w/volume and why". I think the information in those plots and theory overall serves the purpose of this thread quite well: It explains what happens in the amplifier when you change its master volume dial and it explains why.

The theory behind and operation is usually pretty hard to argue against. Preferences about tone, on the other hand, are entirely subjective. For example, for every person preferring the bootstrapped master volume setup there is another person who dislikes it, ironically probably for the very same reasons.

[Chuck H]

Agreed!

So, to get back to the topic then...

I actually think the best solution to the PPIMV is a buffer stage!!! One little silicootie and the whole mess goes away (maybe, we haven't heard it ) Another consideration I thought about would be to use the standard type MV on a dual pot that simultaneously adds series resistance after the circuit to prevent loading the PI input. "Bright" circuits, like small caps or cap/resistor series or parallel networks could be employed to recover HF response. Which, with a circuit like this, should be the only significant losses.

[daz]

Consistent responses and operating conditions DON'T guarantee the amp will sound/feel the same at lower volumes vs. higher volumes because some other things change too that are outside of the considerations you've plotted.

Note the bold text. FEEL FEEL FEEL FEEL FEEL FEEL FEEL FEEL. Should i say it again so certain parties will listen? FEEL FEEL FEEL FEEL .....ah, nevermind. Won't get thru. Anyways, feel is the major difference with the pre PI MV. I could have EQ'd it, tho not nearly as well or easily as with the BS. But the main issue was it became so hard and brittle and un dynamic it was utter cr@p to play.

He's listening to it. I think the validity of that has to be recognized.

Maybe not by some, but the fact is i gigged out regularly w/o stopping for almost years, half of that as a 3 piece. I doubt i could have gotten gigs if i had no ear for tone. Not to mention i got lots of compliments over the years on my tone. But i'm 60 now and my hands no longer work so i play with amps for fun. But i'm sure some people look at a novice tech wannabe like myself as an idiot. I can't do anything about that any more than i can stick those people in front of my amp so they can hear what i hear, then have them play it so they can FEEL what i do, the latter by the way being far and away the most important detail to anyone who's ever learned how to really play. But that gets overlooked by those who never really learned to play that well because they never truly learned how to use an amp's dynamics.

[teemuk]

The plots are of course useful only for analysis what happens in frequency domain.

Oh, forgot to add:

Since plots like this can provide a good analysis what happens in frequency domain they DO help in deducing cause of issues taking place in other domains.

For example, let's say a plot shows no considerable increase in high frequency response throughout the dial. Yet, turning the volume up we do sense a boost at higher frequencies. With the results of the frequency domain analysis we KNOW the amp isn't really accentuating treble response so the perceived treble boost must come from some other source and from some other mechanism: e.g. it could be increased high order harmonic content in the signal due to increased clipping distortion at higher volume settings. We would easily HEAR it as accentuated treble response but frequency domain analysis and instrumentation gear such as an oscilloscope or spectrum analyzer would quickly show us the proper cause for the effect we perceived in the tone.

As important as they are, our ears can and will easily fool us too.

[Dave H]

Rest assured it's in the loop alright...

Teemuk, I think I must have the wrong circuit then for the BS master vol. Can you point me to a schematic of the BS master circuit you used for the simulations? I was working from the schematic Daz posted. There must be more to it than that because I tried 1M, 47k and 4k7 for the PI grid resistors in Ltspice with the feedback disconnected and the open loop gain hardly changed so I concluded that the BS master was outside the loop. Perhaps I’m not doing the simulation correctly? I’ll try it again when I have the correct schematic.

[teemuk]

Dave, "scope" the signal at PI cathodes in BSMV vs. generic MV. Now think what happens to the degenerative feedback at that node when you close the feedback loop. You'll see it.

The schematic I used is the "classic" Marshall circuit. LTP input, two EL34 power tubes. Zp=8K @ Zs=8R. 82K and 100K plate resistors, 1M bootstrapped grid resistors, 22nF coupling in input, 100nF bypass in the second grid input. 470R bias resistor, 10K tail resistor, 5K and 82K feedback resistors. Presence control optional. Basically, the standard circuit which covers about 99% of Marshall PAs.

For bootstrapped master volume simulation the 1M grid resistors were simply replaced with 1M pots (like in the schematic example posted previously). In fact, I tried various schemes of this: potentiometer only in one grid input other grid resistor being fixed value, different grid resistance/pot values e.g. 500K, 250K, etc. "asymmetric" values (e.g. 250K pot vs. 1M pot, 250k pot vs. 1M resistor), all kinds of additional grid resistors, etc. None seemed to have great impact on overall performance one way or the other. I even tried what would happen if one of the pots was reversed (as it seemed like a logical scheme to compensate the feedback ratio): it actually somewhat normalized the response through the dial but resulted into great asymmetry in PI side gains and consequently to considerable amount of distortion in the output.

[jazbo8]

It seems that too many variables are involved, making the amp hard to mange. So perhaps some compromises can be made, can you do without some of the controls such as the variable NFB and/or the presence control? e.g., can you get the tone/feel by setting the MV and the tone control on the guitar alone?