where to cut gain?

[daz]

I know this is a question who's answer will vary depending on the circuit. But generally if you have a marshall style 3 cascaded stage preamp then cathode follower and you want to lower gain with a split load on either V1a or b, where do you think is generally the better place to do it for the best dynamics? Both stages have the same plate and cathode resistor and bypass values.

[Gingertube]

Daz,
You are better off with an interstage attenuator. That gives you the option of some tone shaping with cap(s) across one arm of the attenuator.
In general you want that after the second stage so you control when the third stage goes into and comes out of overdrive.

Attenuating too early in the chain has some non optimal effects:
1) It compromises noise performance
2) It makes it difficult to arrange for the 3rd stage to go into overdirve first followed by the second stage (and the reverse in comming out of overdrive).
3) If you attenuate the output of the first stage then you may not be able to get the 2nd stage to go into overdrive at all. If using an attenuator here it will need to be "mild", not more than say a divide by 2.

Cheers,
Ian

[NateS]

If dynamics is what you seek... just wondering out loud, but I wonder about some local negative feedback around the preamp stages. Trade away some of that gain dynamically, and up the touch sensitivity of the crunch perhaps?

[olddawg]

How about just trying different tubes. Not all 12aX7s are created equal.

[jpfamps]

I would agree with Gingertube, attenuating in the second stage would be best, for the reasons given.

I like using split anodes to attenuate as it is better for noise performance as you are not putting a high valve resistor into the signal path which you do with an interstage attenuator.

On the other hand using an interstage attenuator means that you can have quite a large "grid stop" resistor which will limit current if the following stage is driven into grid conduction.

[Mick Bailey]

I've just finished an amp where I've put gain controls after both V1a and v1b and I really like the variation in both dynamics and tone. It really opens up the voicing of the amp.

[NateS]

I like that. I was going to mention v2a is a colder clip than v2b. An option i thought about instead of individual volumes is a ratio control, but it would have to either be linear or log/antilog.

[daz]

I've just finished an amp where I've put gain controls after both V1a and v1b and I really like the variation in both dynamics and tone. It really opens up the voicing of the amp.

I've done that a number of times but i always come to find that i only like it set at a certain point. Plus i find that a resistor to ground after the coupling cap gives me a better tone every time than a voltage divider there, be it set, a pot, with a treble bleed cap over either one or not. In fact, i really cannot seem to find any way to cut the gain down that doesn't hurt the tone, and I've tried most everything including everything mentioned here. I guess it's just a matter of losing harmonic complexity when the OD is lessened. That makes sense since it's OD that generates complexity. But oddly i can lower my guitar's volume to get the same degree of OD as if i had cut the gain in the amp circuit and the harmonic complexity is retained much better.

[Chuck H]

But oddly i can lower my guitar's volume to get the same degree of OD as if i had cut the gain in the amp circuit and the harmonic complexity is retained much better.

What is the nature of your guitar? Does it use a bright cap circuit on the volume control?

I agree with Ian also. There's not a lot of OD possible from stage one to two. This is where you need gain to get the ball rolling without undue noise.

A simple load resistor forms a divider with the plate impedance of the driving stage. So even a simple load resistor is a voltage divider. As is the plate resistor since it terminates at a decoupling capacitor with a very low impedance. A post coupling cap voltage divider can be tuned with caps across the series resistor, load resistor or both. If you add capacitors to a loaded circuit you get phase shifts that may change the feel and responsiveness for better or worse I suppose. I notice the frequency change for the most part. The only way I know to add harmonic complexity is to add distortion. Clipping is the most common (but not the only) distortion we take advantage of with guitar amps. The effects of input and output impedances can have an affect on dynamics. A lower impedance output circuit will be less plagued by impedance rising with frequency as this applies to audio. In a guitar amp all these things are happening at the same time. And how each particular reacts to another is eminently variable. Also, many parameters change with operation. Things like sag and bias shifting alter operating conditions. Which in turn changes how all the above inter relate on the fly. Perhaps this is why there is a certain degree of black art to voicing guitar amps. Of course there are measurable parameters and many would argue that it's strictly a science. But the constant variability and sheer number of considerations do make comprehensive and quantifiable testing less than satisfactory as a sole means of guitar amp design. So we use our ears for a lot of design work WRT guitar amps. That it's actually a black art, bordering on alchemy is evident in the sheer number of designs. Compare to hi fi where there are only three paramounts that matter, wide frequency response, transparent dynamics and low distortion. Once you start to intentionally distort and clip the wave form, especially with tubes, many proper definitions of relative goodness or badness evaporate and we're left with only our instincts, intentions and perseverance to guide us.

All amps use inter stage attenuation. There may be other criteria altered by your loads and dividers that haven't been accounted for. It is absolutely true that if you intend to reduce gain, but want an exact replica of what you had, only less of it, you'll need to tune the rest of the circuit for whatever you did to alter the gain.

[Gingertube]

I agree with Ian also. There's not a lot of OD possible from stage one to two. This is where you need gain to get the ball rolling without undue noise.

....... Also, many parameters change with operation. Things like sag and bias shifting alter operating conditions. Which in turn changes how all the above inter relate on the fly. Perhaps this is why there is a certain degree of black art to voicing guitar amps. Of course there are measurable parameters and many would argue that it's strictly a science. But the constant variability and sheer number of considerations do make comprehensive and quantifiable testing less than satisfactory as a sole means of guitar amp design. So we use our ears for a lot of design work WRT guitar amps. That it's actually a black art, bordering on alchemy is evident in the sheer number of designs. Compare to hi fi where there are only three paramounts that matter, wide frequency response, transparent dynamics and low distortion. Once you start to intentionally distort and clip the wave form, especially with tubes, many proper definitions of relative goodness or badness evaporate and we're left with only our instincts, intentions and perseverance to guide us.

All amps use inter stage attenuation. There may be other criteria altered by your loads and dividers that haven't been accounted for. It is absolutely true that if you intend to reduce gain, but want an exact replica of what you had, only less of it, you'll need to tune the rest of the circuit for whatever you did to alter the gain.

The best (only really) text on this "black art" is Merlin Blencowe "Designing Tube Peamps for Guitar and Bass", the second last chapter (Chpt 14) of the book. It really is worth the US$39 price for this chapter alone.

Whether you call it a "black art", "magic" or "alchemy without the pointy hats" there is really a fair bit of art in addition to the science. In the final evaluation this is one place where the only valid bit of test equipment is one standard pair of ears and the processor between them.

Don't let the "voodoo" put you off, it's like learning to tune a guitar, the more you do it the better you get at it. If the amp is for you then the only opinion which counts is yours.

Cheers,
Ian

[Chuck H]

Don't let the "voodoo" put you off, it's like learning to tune a guitar, the more you do it the better you get at it.

Well put. I'll add that taking time to understand what's happening helps a lot. If something seems to be working, find out why. An example of failing to do this would be the .1uf presence cap value. Many low tech builders consider this the standard because it's what's been used in all the classic amps with a presence control circuit. The same builders (designers?) then alter NFB loop values (very popular now) with no real knowledge that to affect the same frequencies they need to change the presence cap value. I've seen boutique amps with the .1uf value on the presence control that, relative to other circuit values, is only affecting frequencies above 10kHz. Whereas in the Marshall/Bassman circuit the .1uf value has a -3dB at about 350Hz.

This mentality gets repeated all the time in the form of builders cutting and pasting circuits from different amps together attempting to make a hybrid.

The point is... Just because a 1uf cathode bypass cap for stage two on any given amp seems to be the best value DOESN'T mean the same relative goodness can be achieved in ANY amp by arbitrarily using that value. It's not enough to know part values for stages and circuits in a rote way unless you always mod or build the exact same amp. You do need to understand what's actually happening rather than just what's in there in order to get better and be able to implement discoveries that give good results.

[daz]

You do need to understand what's actually happening rather than just what's in there in order to get better and be able to implement discoveries that give good results.

I would say you don't need to, you just need to in order to get there faster with less trial and error. Might be a LOT faster too, but i think like myself you can get there only knowing a few basics. I know how certain components tend to sound when changed in value generally and with that and a lot (read: a HELLOFALOT) of trial and error i have a amp that sounds better than most production amps and quite a few boutique amps to my ear at least. But it took a long time. Actually i reached that point years ago and everything since has been a curiosity and a desire to find the ultimate, tho all that experimenting often leads to losing the tone or feel i had early on and i end up getting it back at some later point like i did with the voltages. But it still took me a few years to get it to where it's back to now. I still wanna perfect things tho. there are exactly 2 things i would like to improve that i think would make it perfect. But they are 2 things i'm not sure are possible w/o a major circuit overhaul.

[cbarrow7625]

Reduce gain and keep harmonic complexity is the goal? OK, sounds good. The way I interpret that is that you would probably like more definition on the lower strings while retaining the "singing" saturation for higher notes/strings. The simplest way to achieve that is, of course, use a smaller coupling cap so that the higher frequencies are "un-attenuated" while the lower frequencies have a reduced ability to over overdrive the stage at a rate decreasing at 6dB per octave (for our typical single component high pass inter-stage coupling cap). I think I've chatted with you enough on this forum to know that 1.) You have tried that and 2.) It doesn't work "right" because you lose all of the definition, chunk, attack (call it what you will) on the lower strings. Am I right? I have the same reaction.

So how do we overcome this? We probably need to try a more complex circuit to shape the signal (akin to tone control pre-emphasis but without the typical built in mid scoop). the way Isee it, the problem is the steady 6dB/Octave decrease in the signal at lower frequencies. I don't think we want it steady; I think we want it to drop to a certain level, shelve off and then drop off again at 6dB/Octave at the "normal" lowest note that we want to recreate. Hope that sounds similar to what you are thinking about, otherwise the rest of this post is useless!

At the simplest, we could just split the signal off of the Stage 2 plate and run a (pulling values out of thin air without actually calculating frequencies; bear with me) 330-470pF cap to a 470k-1M resistor to ground to create the initial high pass and 6dB /octave rolloff. In parallel, we could place a .022uF cap to a voltage divider (maybe a pot at first so that we can adjust the lower frequency shelf content) in the same 470k-1M range. The high pass cap & resistor would keep those crisp, saturated high frequency response that I think you are looking for. The larger value cap & voltage divider would allow you to set the exact shelf level and final roll off frequency, while the series resistor would create just a bit of a low pass, what typically "dulls" some of the top end on our "normal" single voltage divider stages. If you really want to get crazy, you could actually add a low pass cap and additional mixing resistor after the voltage divider to reduce the interaction with the 470k/1M high pass. Even crazier, you could add a passive mid/low mid-notch circuit as part of the voltage divider for further tone shaping.....hmmmm sounding an awful lot like a tone stack now.....probably way overkill; simple is better.

I have my doubts that implementing it in that way would get you exactly what you are looking for though. An idea I've always wanted to try but have not had the opportunity yet is to implement the above ide but instead of running both caps from the plate I would run the high pass (470pF/1M) from the plate as well as add a cathode follower to the same stage (stage 2) and run the low frequency portion from the output of the CF. That should allow for "tighter" lower frequencies with a larger cap and smaller resistor to ground for additional "control" of the signal, followed by a large-ish (470k-1M, or larger?) mixing resistor to attenuate the "shelf" signal by 3-6dB. If more attenuation is needed, a voltage divider can be used before the mixing resistor. Actually, if a 470k-1M mixing resistor is used, going to a 100k resistor to ground for the cap coming off of the cathode, that 470k-1M resistor can serve as both the mixing resistor for the low shelf and the grid leak / high pass resistor for the 470pF high pass circuit - that gets simple. Maybe I'm way off base here, but I always feel that lower frequencies following a cathode follower with smaller resistors feel much tighter and have better control but I just don't like how the HF ends up sounding "hard" - I don't have a better word to describe it. This arrangement may allow for the best of both worlds.

Any thoughts?

[daz]

Any thoughts?

Yeah, but they are lost in the spinning mess that was my mind till i read your post ! LOL! Yeah, i'm a bit lost and will have to read it a few times and try and pick some things up. My 2 big weakness' mentally are math and trying to understand a lot all at once. But i'll read it a few times and see what i can get from it. One thing i CAN say tho is yes, i've tried smaller couplers and have had good results to some defgree but never found it necassary to go too low. I have tried all the way down to fender specs in the pf range and then tried beefing up lows elsewhere like fender tends to do, IE: much smaller couplers but much bigger bypass caps on the pre stages, opposite of how marshall does it. I liked the results, mainly because the grass is always greener but in the end i prefer the way it is now. I only have one coupler thats smaller than the typical .022 marshall uses, and thats a .01 at V1B. I suppose i should try going back to smaller on V1A too and at the PI input as i had a while ago. It did seem good and reduces gain some and i don't recall what i went back to mostly .022.

On a side note, one of the things that i don't like about my gain structure is the gain pot itself. It's one of the reasons i wanted less gain. What i'm getting at is the treble bleed cap. As you know, a bleed cap on a pot causes more gain to pass, albeit only in the cap's range. But never the less there is a negative side effect of that.....you have to turn the pot a lot further down to get the gain to where you want, and that causes it to never clean up real well. But w/o the cap the tone goes too dark and loses clarity and focus. .

[Alan0354]

What is the nature of your guitar? Does it use a bright cap circuit on the volume control?

I agree with Ian also. There's not a lot of OD possible from stage one to two. This is where you need gain to get the ball rolling without undue noise.

A simple load resistor forms a divider with the plate impedance of the driving stage. So even a simple load resistor is a voltage divider. As is the plate resistor since it terminates at a decoupling capacitor with a very low impedance. A post coupling cap voltage divider can be tuned with caps across the series resistor, load resistor or both. If you add capacitors to a loaded circuit you get phase shifts that may change the feel and responsiveness for better or worse I suppose. I notice the frequency change for the most part. The only way I know to add harmonic complexity is to add distortion. Clipping is the most common (but not the only) distortion we take advantage of with guitar amps. The effects of input and output impedances can have an affect on dynamics. A lower impedance output circuit will be less plagued by impedance rising with frequency as this applies to audio. In a guitar amp all these things are happening at the same time. And how each particular reacts to another is eminently variable. Also, many parameters change with operation. Things like sag and bias shifting alter operating conditions. Which in turn changes how all the above inter relate on the fly. Perhaps this is why there is a certain degree of black art to voicing guitar amps. Of course there are measurable parameters and many would argue that it's strictly a science. But the constant variability and sheer number of considerations do make comprehensive and quantifiable testing less than satisfactory as a sole means of guitar amp design. So we use our ears for a lot of design work WRT guitar amps. That it's actually a black art, bordering on alchemy is evident in the sheer number of designs. Compare to hi fi where there are only three paramounts that matter, wide frequency response, transparent dynamics and low distortion. Once you start to intentionally distort and clip the wave form, especially with tubes, many proper definitions of relative goodness or badness evaporate and we're left with only our instincts, intentions and perseverance to guide us.

All amps use inter stage attenuation. There may be other criteria altered by your loads and dividers that haven't been accounted for. It is absolutely true that if you intend to reduce gain, but want an exact replica of what you had, only less of it, you'll need to tune the rest of the circuit for whatever you did to alter the gain.

You cannot say it any better. Every circuit you put in between change the sound. Even using higher value resistors for divider sounds different from lower value because of the 100pF input capacitance of the next stage.

It is a black art. You cannot use theory to do tuning. Believe me, I can theorize and calculate response until the cows come home, but that really does not help. In distortion, all the harmonics interact and create new side bands, too many to theorize. That's the reason I am quite disappointed lately because the need to test the sound in real life to fine tune. I got my amp very good testing in my own room, but if that's doesn't mean anything, then it's getting hard for me. That's the reason I am entertaining getting into audiophile power amp as it should be easier......my own room!!!