Inter-stage dividers in high gain circuits

[Gtr0]

G'Day MEF Folks.

I am never satisfied with just doing stuff by copying other schematics, I truly want to understand what's happening and why. I have read as much material as I could find (Merlin's books and site, Ampbooks books, Rob Robinette's site, Aiken's site, various posts by reputable users on MEF etc) to grasp a decent level of education as well as experimenting dozens of times on circuits, yet some of the very most basic fundamentals are a mystery to me.

So, today I would like to ask about inter stage dividers in high gain (4 stage) circuits.

Resistors are one of those things that remain a tad mysterious to me. I have a decent level of understanding when it comes to RC low/high pass filters etc, but I am mystified when it comes to values chosen on inter-stage dividers.

I understand, of course, that higher value resistors to ground (as they work in a voltage divider) will keep more signal pointing to the next stage... and conversely and smaller value will dump more signal to ground, but the way I am starting to see things is that these inter stage dividers are more or less chosen for their ratio more than anything... i.e. that a 220k/220k divider will pass half of the signal to the next stage. But the same can also be said about 100k/100k, or 1M/1M for that matter.

Are these values chosen based on how much resistance to ground is desired, then worked out as a ratio to pass on signal second? Or is grid stopping the initial desire and then a ratio to ground is chosen based on that? Or is there something more to it? I assume next stage grid leak is a factor etc.

I see some high gain amps use a 220k/470k divider while other amps use a 470k/1M... I realize there is a small difference in the ratio here, but I assume the "gist is gotten".

As always I appreciate your wisdom and guidance :-)

[Mick Bailey]

The divider can't be considered in isolation from the rest of the circuit and the impedance requirements of the circuit either side have to be factored into the resistor choice. If the values chosen are too low then there will be an excessive loading on the source, if to too high then noise can become an issue. As an extreme example, you can see that in isolation a 1R/1R divider would halve the input voltage just the same as a 100k/100k or a 100M/100M divider would, but they would work very differently in an audio circuit.

Another aspect is the resistances have to be considered along with source and especially load capacitance such as cable capacitance and that of the following stage. Then the voltage divider also behaves as a filter - usually a low-pass filter - the divider rolls off treble. This sometimes needs correcting with a treble bypass capacitor, like you would use on a volume control. I sometimes make this switchable to give a treble boost.

You are right in that resistors can have a dual purpose - with a tube stage one of the divider resistors also serves as the grid leak resistor for the following stage and this influences the choice of values.

[J M Fahey]

Agree and add: in very high gain circuit, plain gain does not tell the whole story, there are "hidden in plain sight" desirable secondary effects.

Say you have 2 preamps side by side, one has 3 stages full tilt, possible block gain is 50*50*50=125000X , maybe real world you get 50000X, in any case beastly gain, and the other has *four* full tilt stages, but you add dividers so end to end gain is also 50000X.

Both "should" sound the same, huh?

Well, they don´t, we all know it but not why.

The second one will have more sustain, have more liquid, elastic sound, will "sound better", will be "more responsive", etc.

Fact is that overdriven tubes have a "sweet spot" where they compress somewhat (say 10/12dB) more than plain clip (like a diode clipper or overdriven transistor stage would).

Careful attenuator selection guarantees when one tube is driven beyond its sweet spot the earlier one enters its own, an so on,one of the reasons Tube preamps are much preferred over SS.

Don´t think Designers did much Math and simulation, guess they added an extra stage, found it wild uncontrollable and added an attenuator to tame it ... and found a better sound "by ear" so they left it that way.

Rinse and repeat.

Now for comparison some thrash players don´t care about such small nuisances, and are happy with chainsaw sounding clipping, those are happy with SS amps (Randall, Ampeg, Crate, even Peavey Bandit).

[teemuk]

I don't buy the "sweet spot" -theory.

For one, typical tube gain stage has gain ranging from 50 to 60, and overdrives at input levels of maybe 1 to 3 volts peak-to-peak... yet, interstage dividers are often only 1/2th ratio and maybe 1 to 10th in extreme cases - and that not even including "treble peaking" bypass, which reduces the ratio even more. No 1 to 50th ratios in sight. If there's a "sweet spot" the division ratios vs. stage input sensitivity ratings do not correspond to it.

Secondly, even most "high gain" amplifiers generate the distortion in just select few of the preamp stages. Carvin's eleven cascaded stages is an isolated incident amongst typical designs and they intentionally introduce as much as DC offset shifting as possible instead of little bit of clipping in each stage. The DC shifting is really one of the key aspects of having several cascaded stages that distort.

I regard interstage attenuation as a relic and often there's more effect to tone from the aforementioned and associated "treble peaking" than from the attenuation of the divider. So, more important to attenuate select frequencies than to just attenuate everything.

[Mick Bailey]

With one of my amp builds I used three pots with limiting resistors instead of voltage dividers and it did give a much broader range of sounds than I had expected. Along with a master volume I really liked that arrangement but it only really came to life when I swapped the preamp cathode resistors for trim pots. It was pretty clear when each stage was biased just right and since then I've often thought of placing an externally adjustable trimmer below each level control.

[Chuck H]

teemuk and Juan,..

You're both right. Some things can be done to get better gain from fewer stages. I do it. It's been done a lot by many designers that, as teemuk mentioned, are clipping stages selectively and going for minimum tube count. And, it's not always easy. Sometimes more stages just makes it easier. And in many cases a 12ax7 stage is just as affordable as the extra components and other design considerations that it would take to get the sound you want from one less gain stage. Cascade stages need to correspond to each other to make desirable results. Whether you're shooting for symmetry from a previously cold clipped or hot clipped stage, trying to get controlled asymmetry or you're trying to mitigate the aforementioned treble peaking. Sometimes a voltage divider is needed to get the right amount of clip on the OTHER side of the wave form. Using NO dividers means accepting what you get without trying to control the shape of the waveform. It doesn't matter. Creating "good" sounding distortion is a black art . And also a reason many vintage designs are revered (sort of, see note *). Less so than it once was because of dedicated "guitar amp" designers. But still a bit undefined as Juan eluded.

FWIW I don't see how voltage dividers between stages is a relic'd artifact.?. Vintage designs for hifi always shot for the most gain in the fewest stages. Vintage guitar amps were always built with mind toward economy and wasting gain wasn't on the agenda. All kinds of gain stages have been used in vintage guitar amps and if a voltage divider was used there was always a good reason for it. So where are these relic'd, arbitrary dividers that are used now because they were common place then?

JM2C

* There was a time in the middle of the golden age of tubes when amplifier designers did try to make clipping "graceful" so it wouldn't sound discordant if the signal was pushed a little. It's these designs and models that spawned most of the classic guitar amp designs (happy accident). Later the hi fi guys just said screw it and went for clean performance only about the same time as the guitar amp guys were going the other way Also a happy accident. In the end our ears are trained for what we like and what we like is what our ears are trained for. So to speak.

[tedmich]

it only really came to life when I swapped the preamp cathode resistors for trim pots.

Hmm intriguing! I could have sworn Merlin said such pots were very noisy when moved, are they?
Maybe in series with bypass caps or the typical parallel arrangement?

[Enzo]

If I read him right, he doesn't adjust them while playing, he adjusts them to find the exact resistance for best performance. Yes, DC voltage on pots generally makes noise during turning.

[Gtr0]

The divider can't be considered in isolation from the rest of the circuit and the impedance requirements of the circuit either side have to be factored into the resistor choice. If the values chosen are too low then there will be an excessive loading on the source, if to too high then noise can become an issue. As an extreme example, you can see that in isolation a 1R/1R divider would halve the input voltage just the same as a 100k/100k or a 100M/100M divider would, but they would work very differently in an audio circuit.

Hmmm... I often forget about impedance... mostly for two reasons... first I am only at the stage that I can built off of an existing schematic and tweak to taste. Second, I find impedance to be the biggest mystery (at least at the moment) in terms of clear understanding. It would probably help me to understand it if I were to create a situation where I can clearly see (or rather hear) the negative effects of too high an impedance in the circuit... perhaps I already have and can not recognize it.

Say you have 2 preamps side by side, one has 3 stages full tilt, possible block gain is 50*50*50=125000X , maybe real world you get 50000X, in any case beastly gain, and the other has *four* full tilt stages, but you add dividers so end to end gain is also 50000X.

An older Splawn type design comes to mind, tons and tons of distortion from two 12ax7 valves.

While I am not entirely sure how to achieve it on my own, though trying to learn, I can only stand on the shoulders of the work of others that came before... try to learn why these particular components were chosen (whatever they may be) and subjectively improve (if possible)... subjectively of course being the key word.

I also feel like there is no "reinventing of the wheel" in a stage by stage design... without getting into insane tone shaping networks (5150 comes to mind) it seems to be that it is more about the collection of stages and how they work together. I guess what I mean to say is there is no real way to reinvent the wheel, but there are different purposes and performance levels in the style of the tires.

And with that, I look at stages like compression on a snare drum... not a good thing to smack the snare hard with a single compressor once and be done with it, but better to build up the tone/shape/attack of the snare by layering softer compression through several stages... the snare track, the multi-mic snare bus, the drum bus, the final mix bus. etc etc. Of course speaking of types of music where snare drum compression is a big part of it's sound.

[Mick Bailey]

Hmm intriguing! I could have sworn Merlin said such pots were very noisy when moved, are they?
Maybe in series with bypass caps or the typical parallel arrangement?

The pots are direct replacements for the resistors and they are noisy when moved, but are just for setting the bias and then leaving them. In one build I also used this approach to optimize the biasing and then changed them out for fixed resistors of the closest value. It did strike me though that different tubes of the same type do respond to slight adjustments and having trimmers is beneficial.

Hmmm... I often forget about impedance... mostly for two reasons... first I am only at the stage that I can built off of an existing schematic and tweak to taste. Second, I find impedance to be the biggest mystery (at least at the moment) in terms of clear understanding. It would probably help me to understand it if I were to create a situation where I can clearly see (or rather hear) the negative effects of too high an impedance in the circuit... perhaps I already have and can not recognize it.

Resistors generate noise, so unnecessarily high values contribute avoidable noise to the following stage, In addition high-impedance circuits are more sensitive to external noise that's coupled into the circuit (such as EMI).

[walkman]

I regard interstage attenuation as a relic and often there's more effect to tone from the aforementioned and associated "treble peaking" than from the attenuation of the divider. So, more important to attenuate select frequencies than to just attenuate everything.

Would you expand on this statement and demonstrate how you would achieve this with a single or two 12x7 preamp, either for high or low gain?

Having read Merlins preamp book he outlines some example circuits. These appear to use both interstate attenuation of all frequencies and frequency emphasis through use of different cathode capacitors and small coupling capacitors.

How would you address other areas to focus on, that often get overlooked, such as impedance or dc level shift ( Merlin also mentions these in his book, though I feel I have less experience understanding these effects when listening to an amp )

[jack briggs]

Have you possibly got "D" and "E" reversed? Looks like "E" values in the table are for the bias resistor and "D" values the bypass cap.

cheers,

[Chuck H]

I too would be interested in this distinction!
As a possibly aid to communication below is a gain stage worksheet that l like to use. It does nothing fancy (just voltage divider math) but I find it useful for comparing amps. Not much "high gain" at 2x 12 AX7s so I included some component values for the nice little gain box; a Peavey Bravo, which uses 3x 12AX7 and needs the final tube for its tone stack and PI.

I suspect the stage 3 values are off; my 2 examples seem to have a 50% splitter there and while there is definitely a trace for a cathode bypass cap, its not there. My guess is any bypass (with or without a 68% divider) made the troublesome reverb tank scream even more than it already does.

There's no component based voltage divider here. There IS voltage division, but it's between the previous circuit (not shown) output impedance and the load resistor (C). Resistors B and C are not forming a voltage divider. They are a load (C) and a grid stop (B) resistor. A voltage divider would place the load AFTER a resistance that is in series with the movement of voltage through the circuit.

[Chuck H]

Here's a modified diagram. C remains the load resistor, but I've added G. G would be the series resistor necessary to achieve a component based voltage division. B is a grid stop resistor and is not part of the voltage division. If you were to imagine G and C as a single potentiometer and the junction between them as the wiper it would represent how a typical gain control is wired as a voltage divider.

[Helmholtz]

Resistor B is not part of a voltage divider and does not drop voltage (as long as there's no grid current). Just a grid stopper, no influence on gain.

Edit: Missed Chuck's replies above - they only showed with updated page after sending my post.