Please help in improving the sound of a gain channel.

[Alan0354]

Ok... New kettle of fish to fry. You've been busy. Don't go all daz on us (hope he sees this )

Here's what I'd try (with brief, non technical information why)

R3 150k You need gain here to start the clipping early. It's almost impossible to clip the second stage too much with the first. This get's the ball rolling.

R7 100k A common value and chosen mostly for familiarity with what it'll do with other changes. You don't need as much gain here because there is plenty of input signal for this high mu triode to pound the following stage. As you already know because you have a 50/50 divider on this stage.

R8 2.7k (second stage cathode resistor) This is a tad cold. Just enough to get a little asymmetry into the wave form. This helps promote 2nd harmonics and keeps things from getting too square and hollow sounding. You could try values up to 4.7k here. Higher than that and it the bypass cap shelf becomes really dominant and tricky to voice for IMHE.

C3 .022 Using two small caps in a row is something I've only seen on amps with four or more cascade gain stages. This is part of what's thinning your sound and forced you into other circuit oddities. A little lack of bottom end clarity is OK and a small part of the sound. It probably won't be objectionable because of the extra shelving effect of the raised value cathode resistor on this stage.

C12 jumper/remove You don't need the gain here. Cathode follower clipping is more of an effect than a quantified flatter/rounder waveform. It compresses a little and takes the sharp edges off the square wave. There's plenty of gain for that without the bypass cap. Besides, it's only bumping the top end and you had trouble balancing that in the tone stack. Removing this bypass cap should help.

R22 jumper/remove You shouldn't need it with the other changes. You'll be voiced too dark if you keep it.

VR5 25k This is a Marshall style circuit. They sound better with fuller mids and 10k keeps a lot of mid scoop in the tone.

That's my take. I'll bet six beer PNW that anyone following will back these changes. That doesn't guarantee you'll like it, but it should get you out of the box, so to speak. You may end up needing to change the R20/R21 divider values to restructure gain a little. Maybe not. If it sounds too dark at low to moderate gain settings add a bright cap of 220P to 1000p to the volume control.

I can't wait, I did the first three ones in RED and iT DOES MAKE QUITE A DIFFERENCE . Note that red does not mean offensive, just want to catch your attention with the color.

I definitely get more bass out of the mod.

I tried removing R22 ( well, I have a 1M trim pot). That does not work. I don't know what to make of it. It sounded very strange when R22=0 and crank up beyond half.

As For the mid pot, I cannot crank it pass 2/3 or it will be really dominant. So I am leaving it alone for now as the extra 15K can easily covered by the 1M Bass pot.

I think I still have a lot of experiment to do. If you have anymore advice, please let me know.

Man, you know your stuff!!!! Theory can never explain what you did. Now I just need to message the component all over again.

Thanks for your help.

[Alan0354]

I experiment a little more. I change R7 to 150K to increase the gain of the second stage a little. I change R20 and R21 from 220K to 470K to reduce loading to the second stage. All in all, I am loving it. I get back the gain I loss from removing the C12 that is the cathode bypass cap of the V2a.

Before, when I change from the Utah to the GB128 type speaker, the sound is so throaty it's like putting a big tube in front of the speaker and the sound came through the tube. Now, the sound is a lot more natural with the British sounding speaker. In fact, the amp sounds a lot closer between the two speaker, it can easily be adjust out by the tone controls.

Again, Thanks you Chuck of helping again.

[Chuck H]

You won't get good results removing R22 until the other changes are made.

You could put the 100k value on V1B plate and recoup gain by jostling the divider between stages two and three. Right now it burns half the AC. Using a 470k load and changing the 470k series resistor to 220k would have increased gain. For that matter you could have left the 220k load resistor and changed the series resistor to 100k for roughly the same VAC. It's a good idea to keep loads when possible. A stage where you're going to throw away some gain is a good chance to use a lower value load. Less noise. That's a good thing in a Marshall type amp.

And indeed there is theory behind these component changes. It's just that not all of it is the sort that's taught. There are no classes that demonstrate how to make good sounding distortion. Only how to eliminate the most distortion. And circuits like that usually make the worst distortion when clipped.

I implore you not to judge the changes one at a time. Since some are designed to work together it makes for very confusing results. Not to mention the possibility of doing one, judging it out of context and undoing it, then going on to the next and deciding you don't like it either. When in fact if you had kept the other change it would have all made sense in the end. I thought I was clear about that. But you'll do whatever you'll do I suppose. It's your amp after all.

[Alan0354]

You won't get good results removing R22 until the other changes are made.

You could put the 100k value on V1B plate and recoup gain by jostling the divider between stages two and three. Right now it burns half the AC. Using a 470k load and changing the 470k series resistor to 220k would have increased gain. For that matter you could have left the 220k load resistor and changed the series resistor to 100k for roughly the same VAC. It's a good idea to keep loads when possible. A stage where you're going to throw away some gain is a good chance to use a lower value load. Less noise. That's a good thing in a Marshall type amp.

And indeed there is theory behind these component changes. It's just that not all of it is the sort that's taught. There are no classes that demonstrate how to make good sounding distortion. Only how to eliminate the most distortion. And circuits like that usually make the worst distortion when clipped.

I implore you not to judge the changes one at a time. Since some are designed to work together it makes for very confusing results. Not to mention the possibility of doing one, judging it out of context and undoing it, then going on to the next and deciding you don't like it either. When in fact if you had kept the other change it would have all made sense in the end. I thought I was clear about that. But you'll do whatever you'll do I suppose. It's your amp after all.

I tried C3 to 0.022, it's a lot of bass. But I do like it better with something like 1500pF. I also try to jumper R22 to 0ohm with C3 = 0.022, still it does not work. The only think left is the 25K pot. But I am not using the mid at full, I think I use max at 3/4 way.

[Alan0354]

And indeed there is theory behind these component changes.

Can you tell me what is the theory? I know you try to get more gain from the first stage using 150K plate resistor. Even you said to look at the whole change, not individual, but I want to know where it all started.

1) Why lowering the current of the second stage by increasing the cathode resistor to 2.7K or larger? I know you increase the -ve grid bias to a little over 2V so you get more input dynamic range.

2) Do you want to lower the plate resistor back to 100K to get asymmetrical clipping?

3) Why do you want to lower the resistance of the voltage divider R20 and R21? Is it for better frequency response like you lower the grid stop resistor?

[Chuck H]

Can you tell me what is the theory? I know you try to get more gain from the first stage using 150K plate resistor. Even you said to look at the whole change, not individual, but I want to know where it all started.

Clipping a clipped signal creates the complex harmonics that more saturated distortion is known for. It's only possible to overdrive the second stage with the first a little bit. The extra gain helps.

1) Why lowering the current of the second stage by increasing the cathode resistor to 2.7K or larger? I know you increase the -ve grid bias to a little over 2V so you get more input dynamic range.

It only increases dynamic range on the saturation swing. The cutoff swing actually clips sooner. Remember that an amplifier, to some degree, has the most headroom at center bias. As if there is a finite box that the incoming signal fits in (not exact, but close enough). If you move off center you get more possible swing on one end but less on the other before clipping. This makes for asymmetrical clipping. Having at least one stage asymmetrical can sometimes improve the distortion character because a perfect square wave tends to sound hollow and dry.

2) Do you want to lower the plate resistor back to 100K to get asymmetrical clipping?

It's not necessary, but yes. This just happens to be the value I'm familiar with for the cathode circuit I suggested. Keeping the plate resistor at a familiar value makes the results more predictable. For all I know, and I think it's true, increasing the plate resistance somewhat moots the small change we made to the cathode resistance.

3) Why do you want to lower the resistance of the voltage divider R20 and R21? Is it for better frequency response like you lower the grid stop resistor?

There are many things happening at the same time. With the lower impedance for the divider there is the effect of a low pass filter because impedance rises with frequency. This can actually work to our advantage on two levels. The amount of loading, even with a 220k load, isn't enough to be objectionable for cleaner tones. With higher clipping levels to can help mellow things and keep the tone from getting too harsh. Because of the sheer gain at clipping there is no loss of HF fidelity, only less very high order harmonics that can sound hashy. The lower load also reduces input sensitivity. But since we have plenty of gain at that point to clip the stage it doesn't matter. The only notable effect of the lower sensitivity would be less hiss.

All of this is well known even for SS designs as far as I know. Not exclusive to tubes. It makes more sense when you consider that you're not shooting for fidelity, but rather pleasant sounding distortion. Even the clean tones for electric guitar are usually enhanced by designs that create distortions below audible clipping.

[Alan0354]

Thanks, I really appreciate your input. Now I can see things better. I have been experiment with different things keeping what you said in mind.

[teemuk]

Having at least one stage asymmetrical can sometimes improve the distortion character because a perfect square wave tends to sound hollow and dry.

Also, there's an interesting phenomenon that takes place in time domain: when you AC couple an asymmetric signal it results to a gradual DC offset bias shift. When you clip that signal again after the AC coupling the offset shift will dictate the asymmetry (or symmetry) of this clipping process. As a result the duty cycle of the output wave, and the harmonic pattern of the distortion will change in time domain and according to magnitude of overdrive, which is what happens in most tube amps that consist of more than two stages with more or less asymmetric clipping and an AC coupling in between them.

The input signal envelope will therefore have direct effect on harmonic patterns of the distortion, something that doesn't happen in symmetric clippers or with just a single clipping stage where harmonic pattern stays nearly identical but the magnitude of harmonics merely increases in relation to input signal's magnitude and level of overdrive.

Since you can change the harmonic structure of the distortion simply by varying the input signal envelope this, when carefully designed in, means a whole lot more touch sensitivity than you ever get with a single clipping stage.

If you take a look at most tube guitar preamps they rely highly on such process. The ultimate pinnacle of that is probably the Carvin's Quad-X which cascades nine AC coupled tube stages with carefully selected gains, DC bias offsets, and clipping characteristics together.

It's still just clipping, though. Huge importance also lies in how you EQ the signal pre and post the distortion stages. Distortion with signal at full audible bandwidth hitting the clipping stage sounds very different than hitting the same stage with a signal where lows are attenuated highly after 1 kHz. The first effect could be described as the buzzy "fuzz" while the second resembles lot more "modern", less "farty" distortion tones we have come familiar with.

If you look at those AC coupling caps between stages their essential job is also shaping the frequency response, not just AC coupling. Same thing with every other RC filter in the signal path, including those at cathodes. The cathode bypass also has a multi-purpose: with AC bypassed cathodes the grid-cathode voltage difference is usually higher and results to earlier grid conduction and clipping by that method. With higher cathode resistance values sans bypass you get an opposite effect. These will also affect the bias point of the stage and therefore overall symmetry and style of the clipping.

Like someone mentioned earlier, everything is interactive. Therefore one usually doesn't have great chances of landing to a good sounding design by taking one circuit bit from here and another from there, without fully understanding the whole process and the interactions.

All of this is well known even for SS designs as far as I know.

Those worth their salt, yes. If you think of it, why shouldn't it be? The distortion introduction and EQ'ing processes are universal and guys such as synthesizer designers have for ages been way ahead of what guitar amp designers are doing. Probably because they do not have obstacles in venturing to territories that for guitar amp designers would be sometimes considered taboos, like new technology, new circuitry or just new kind of "tone" in general. Which also brings us to notion that not even every tube amp designer seems to be familiar with the type of circuitry he is designing and its interactions. Although most of it still relies on principles outlined in the 50's.

[Alan0354]

Well guys, I have to button up the amp for now. I am going for capo tunnel surgery on my left hand tomorrow and I will be out for I don't know how long. I have not been able to play for more than just a few minutes workout having my finger all tingling and numb. I have not been able to play for years already so the surgery is necessary.

I'll get back to it whenever I can.

[Chuck H]

Their batting real high with that surgery nowadays. Be sure to do all the follow up physical therapy!!! Raising a beer to you with good wishes.

[Alan0354]

Their batting real high with that surgery nowadays. Be sure to do all the follow up physical therapy!!! Raising a beer to you with good wishes.

Thanks, I am looking forward to play a little bit more in the future.

[Alan0354]

Well, now that the Marshall is fixed, I have a constant reference for the distortion sound. By comparing with a constant distortion sound, I already hear some short coming of the gain channel already. I already put the Marshall together and on the side, and have the Bassman chassis out again really to experiment.

[Alan0354]

Also, there's an interesting phenomenon that takes place in time domain: when you AC couple an asymmetric signal it results to a gradual DC offset bias shift. When you clip that signal again after the AC coupling the offset shift will dictate the asymmetry (or symmetry) of this clipping process. As a result the duty cycle of the output wave, and the harmonic pattern of the distortion will change in time domain and according to magnitude of overdrive, which is what happens in most tube amps that consist of more than two stages with more or less asymmetric clipping and an AC coupling in between them.

The input signal envelope will therefore have direct effect on harmonic patterns of the distortion, something that doesn't happen in symmetric clippers or with just a single clipping stage where harmonic pattern stays nearly identical but the magnitude of harmonics merely increases in relation to input signal's magnitude and level of overdrive.

Since you can change the harmonic structure of the distortion simply by varying the input signal envelope this, when carefully designed in, means a whole lot more touch sensitivity than you ever get with a single clipping stage.

If you take a look at most tube guitar preamps they rely highly on such process. The ultimate pinnacle of that is probably the Carvin's Quad-X which cascades nine AC coupled tube stages with carefully selected gains, DC bias offsets, and clipping characteristics together.

It's still just clipping, though. Huge importance also lies in how you EQ the signal pre and post the distortion stages. Distortion with signal at full audible bandwidth hitting the clipping stage sounds very different than hitting the same stage with a signal where lows are attenuated highly after 1 kHz. The first effect could be described as the buzzy "fuzz" while the second resembles lot more "modern", less "farty" distortion tones we have come familiar with.

If you look at those AC coupling caps between stages their essential job is also shaping the frequency response, not just AC coupling. Same thing with every other RC filter in the signal path, including those at cathodes. The cathode bypass also has a multi-purpose: with AC bypassed cathodes the grid-cathode voltage difference is usually higher and results to earlier grid conduction and clipping by that method. With higher cathode resistance values sans bypass you get an opposite effect. These will also affect the bias point of the stage and therefore overall symmetry and style of the clipping.

Like someone mentioned earlier, everything is interactive. Therefore one usually doesn't have great chances of landing to a good sounding design by taking one circuit bit from here and another from there, without fully understanding the whole process and the interactions.



Those worth their salt, yes. If you think of it, why shouldn't it be? The distortion introduction and EQ'ing processes are universal and guys such as synthesizer designers have for ages been way ahead of what guitar amp designers are doing. Probably because they do not have obstacles in venturing to territories that for guitar amp designers would be sometimes considered taboos, like new technology, new circuitry or just new kind of "tone" in general. Which also brings us to notion that not even every tube amp designer seems to be familiar with the type of circuitry he is designing and its interactions. Although most of it still relies on principles outlined in the 50's.

This is a very good piece, really make me think more about the symmetry of the signal.

Let me see whether I have this right:

1) As Chuck suggested, I have the cathode resistor of the second stage set to 3K and plater resistor be 100K. The cathode is measured at 2.1V. This mean the input range is asymmetrical as the tube will go into cutoff upon a negative input signal faster than the tube get into 0V grid. Then on the plate, because the plate current is only about 0.66mA, only 66V drop across the plate resistor R7. So the plate can only swing up 66V maximum, but it can swing down a lot more. The effect of 3K cathode resistor has the same asymmetrical polarity. So both enhance each other. The result is the output of the second stage clips on the positive side a lot early than the negative side. this mean C3 is going to have a DC charged up when the second stage is clipping. The polarity of C3 of the side connects to the plate of the second stage is more negative than the side driving the third stage.

2) From the charging of C3, the plate current of the common cathode stage of the third stage increases as the second stage clips harder and harder. This will cause the average voltage of the plate of the third stage decrease when the second stage clips harder and harder.

[J M Fahey]

well guys, i have to button up the amp for now. I am going for capo tunnel surgery on my left hand tomorrow and i will be out for i don't know how long. I have not been able to play for more than just a few minutes workout having my finger all tingling and numb. I have not been able to play for years already so the surgery is necessary.

I'll get back to it whenever i can.

ok, good luck !!!!!!
[ATTACH=CONFIG]28071[/ATTACH]

[Alan0354]

Hi, I want to get an opinion. At this point, I improved the amp to the point that I like the sound at low volume. But when I cranked it up, it seemed to have less distortion and sounded harsh. I was even looking at the CF stage in the other thread. My question is whether it's my ears the room or it's real?

the reason I ask is because if you look at the schematic, this it typical high gain design. The master volume is after everything. This means all the harmonics, sound characteristics remains the same regardless of the master volume setting. I verified the power amp was not clipping as the volume is about half only. I matched the preamp/power amp so it only clip the power amp at the very last bit. So the sound should be totally from the preamp.

So why the sound change? Do other amps have the same thing?