ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Question about selecting bandwidth in diode clipping circuits

9/27/1999 4:44 PM
Mark Hammer	Question about selecting bandwidth in diode clipping circuits This forum has seen no end of threads concerning diode-based clipping circuits (indeed, the pre-occupation with distortion circuits is so rampant I sometimes wonder whether tboy should have a separate forum link simply entitled "Distortion"). The "Sweetest pedal" thread that David Manson started got me to thinking. When people talk about the "thickness" of distorted sound, they are generally talking about: 1) the amount of low end and preserved low-end dynamics, 2) the relative balance of even vs odd, and lower order vs higher order harmonics. The combinations of number of diodes, types of diodes, symmetry of diode combinations, diode thresholds, and all the other permutations and combinations of those little suckers has been chawed over endlessly, and many excellent posts have resulted concerning the benefits of asymmetrical clipping. We have also seen, over the years, here and on other forums, and in designs, the idea of multi-band (but band-separated) clipping circuits. What this enquiring mind wants to know is whether one should expect something useful from sticking caps in SERIES with clipping diodes, either separately for each diode- orientation-path, or between ground and the tail end of whatever diodes-to-ground combo one is employing? The TS-9 attempts to eliminate the potential for low-end to "overcrunch" by tailoring the pass-band at the front end. Is it possible to tailor the pass-band at point-of-clipping? I ask this because I think what a lot of folks would like is device that stays relatively clean for most notes, but only dirties up for notes you push. The difference in relative amplitude of lower notes and higher notes means that anything set to give crunch for higher notes over-responds to lower notes. If the threshold for clipping were different as a function of note fundamental, then it would be okay to let the entire spectrum of the guitar go through the entire circuit. Is this a pipe dream or what? And if it is not a pipe dream, can one use frequency-tailoring of the diode-clipping circuit to do it? If not, why?

9/27/1999 4:44 PM

Mark Hammer

Question about selecting bandwidth in diode clipping circuits

This forum has seen no end of threads concerning diode-based clipping circuits (indeed, the pre-occupation with distortion circuits is so rampant I sometimes wonder whether tboy should have a separate forum link simply entitled "Distortion").

The "Sweetest pedal" thread that David Manson started got me to thinking. When people talk about the "thickness" of distorted sound, they are generally talking about: 1) the amount of low end and preserved low-end dynamics, 2) the relative balance of even vs odd, and lower order vs higher order harmonics.

The combinations of number of diodes, types of diodes, symmetry of diode combinations, diode thresholds, and all the other permutations and combinations of those little suckers has been chawed over endlessly, and many excellent posts have resulted concerning the benefits of asymmetrical clipping. We have also seen, over the years, here and on other forums, and in designs, the idea of multi-band (but band-separated) clipping circuits.

What this enquiring mind wants to know is whether one should expect something useful from sticking caps in SERIES with clipping diodes, either separately for each diode- orientation-path, or between ground and the tail end of whatever diodes-to-ground combo one is employing? The TS-9 attempts to eliminate the potential for low-end to "overcrunch" by tailoring the pass-band at the front end. Is it possible to tailor the pass-band at point-of-clipping?

I ask this because I think what a lot of folks would like is device that stays relatively clean for most notes, but only dirties up for notes you push. The difference in relative amplitude of lower notes and higher notes means that anything set to give crunch for higher notes over-responds to lower notes. If the *threshold* for clipping were different as a function of note fundamental, then it would be okay to let the entire spectrum of the guitar go through the entire circuit.

Is this a pipe dream or what? And if it is not a pipe dream, can one use frequency-tailoring of the diode-clipping circuit to do it? If not, why?

9/28/1999 11:54 AM

GFR

quote:
"...should expect something useful from sticking caps in SERIES with clipping diodes,..."

I think LXH2's Marshall Simulator had a stage like this (unfortunately his site is gone). His idea is when the clipping begins, the frequency response changes, like in a power stage with NFB.

On the subject of bass cut on the TS-9. When you pass a complex signal into a clipper the bass frequencies modulate the high frequencies.

" target="_blank">http://www.geocities.com/SunsetStrip/Studio/2987/100p1k.gif">

(Blue wave is input, Red wave is after clipping). This makes chords sound bad. If you take some bass out:

" target="_blank">http://www.geocities.com/SunsetStrip/Studio/2987/100p1k2.gif">

Then it sounds better with chords. BUT... it will sound "thinner". So you have to trade thickness for clarity on chords. A Fuzz Face is very thick sounding, but you're kind of limited to single note soloing unless you're after some serious noise.

I think the best way to get out of this trap is to distribute the harmonics generation along several stages. Instead of having a sibgle clipping stage that does all the distortion, try a lot of "almost clean" stages in series. This should help with the intermodulation without the need to cut a lot of bass. RG has talked about this idea several times. And if you think about it... that's more or less what's going on inside a tube amp

9/28/1999 8:48 PM
Mark Hammer	I get your gist, but I was thinking of something a little different. In essence, what I was wondering was if the clipping threshold of a diode pair (or threesome) could be made to vary as a function of signal frequency, and whether that could be accomplished by sticking a capacitor in series with diodes going to ground. Referring to your scope pictures, I wondering whether if the time/distance between the last little "bump" and the current one is small/short, then the clipping threshold would be lower, but if the time/distance were longer/bigger, then the clipping threshold would be higher. As you rightly point out, the intermodulation distortion can be an unfortunate byproduct of too much bass and is something that people try to design away. The PAiA/Anderton "Quadrafuzz" was one attempt to do this, by splitting up the signal into 4 bands and distorting each separately before recombining again (and was not unlike how multi-band noise-reduction devices are used to provide time constants which are approximately appropriate to the particular frequency being compressed). In a sense one can accomplish what I'm talking about here by duplicating the same clipping circuit and feeding each with signals adjusted in level to be at, above, or below clipping threshold. They would then be mixed back together. The problem with this setup, of course, is that it not only uses a LOT of parts, but that you can't easily have a single "More" knob as on 2 or 3-knob stompboxes. To get a cleaner bass sound with more bite at higher frequencies, I would need to turn down the bass band going to its own clipping circuit, then turn it up post-clipping, safely sneaking the bass band past the clipping circuit, but acquiring more cumulative noise. Not good. So, I wonder, can the clipping threshold be made to "float" with the characteristics of the signal itself, specifically the frequency? I can accept kismet if there is no other way of achieving a low IM tone besides multi-stage or multi-path designs, but what IF...? What DO you get when the signal going through diodes to get to ground also has to go through a cap to get there? I know you can adjust clipping characteristics by sticking a small fixed resistor in that very same path, but what if it was a cap?

9/28/1999 8:48 PM

Mark Hammer

I get your gist, but I was thinking of something a little different. In essence, what I was wondering was if the clipping threshold of a diode pair (or threesome) could be made to vary as a function of signal frequency, and whether that could be accomplished by sticking a capacitor in series with diodes going to ground. Referring to your scope pictures, I wondering whether if the time/distance between the last little "bump" and the current one is small/short, then the clipping threshold would be lower, but if the time/distance were longer/bigger, then the clipping threshold would be higher.

As you rightly point out, the intermodulation distortion can be an unfortunate byproduct of too much bass and is something that people try to design away. The PAiA/Anderton "Quadrafuzz" was one attempt to do this, by splitting up the signal into 4 bands and distorting each separately before recombining again (and was not unlike how multi-band noise-reduction devices are used to provide time constants which are approximately appropriate to the particular frequency being compressed). In a sense one can accomplish what I'm talking about here by duplicating the same clipping circuit and feeding each with signals adjusted in level to be at, above, or below clipping threshold. They would then be mixed back together.

The problem with this setup, of course, is that it not only uses a LOT of parts, but that you can't easily have a single "More" knob as on 2 or 3-knob stompboxes. To get a cleaner bass sound with more bite at higher frequencies, I would need to turn down the bass band going to its own clipping circuit, then turn it up post-clipping, safely sneaking the bass band past the clipping circuit, but acquiring more cumulative noise. Not good.

So, I wonder, can the clipping threshold be made to "float" with the characteristics of the signal itself, specifically the frequency? I can accept kismet if there is no other way of achieving a low IM tone besides multi-stage or multi-path designs, but what IF...? What DO you get when the signal going through diodes to get to ground also has to go through a cap to get there? I know you can adjust clipping characteristics by sticking a small fixed resistor in that very same path, but what if it was a cap?

9/28/1999 9:59 PM
Petri Junno	>> What DO you get when the signal going through diodes to get to ground also has to go through a cap to get there? I know you can adjust clipping characteristics by sticking a small fixed resistor in that very same path, but what if it was a cap? I Have tried this cap in series with diodes thing with electra circuit and in the feedback loop of an opamp. It gives you a huge bass. So you might need to cut the bass after all to flatten it out or the bass will drown the highs.

9/28/1999 9:59 PM

Petri Junno

>> What DO you get when the signal going through diodes to get to ground also has to go through a cap to get there? I know you can adjust clipping characteristics by sticking a small fixed resistor in that very same path, but what if it was a cap?

I Have tried this cap in series with diodes thing with electra circuit and in the feedback loop of an opamp. It gives you a huge bass. So you might need to cut the bass after all to flatten it out or the bass will drown the highs.

9/28/1999 10:21 PM
Mark Hammer	Aha!!! So maybe it IS possible to pick a suitable value for the input cap in the first stage (setting low-end rolloff), a suitable value for the series cap connected to the clipping diodes, and a suitable gain level that will result in enough bass to sound "big" while not overwhelming the harmonics created by the clipping. For that matter, maybe it is possible to experiment with different cap values for each path of an asymmetrical diode configuration and get different frequency-sensity clipping thresholds for each half cycle. Would this assist in getting some magical combination of the right harmonics in the right balance and no thinning?

9/28/1999 10:21 PM

Mark Hammer

Aha!!!

So maybe it *IS* possible to pick a suitable value for the input cap in the first stage (setting low-end rolloff), a suitable value for the series cap connected to the clipping diodes, and a suitable gain level that will result in enough bass to sound "big" while not overwhelming the harmonics created by the clipping. For that matter, maybe it is possible to experiment with different cap values for each path of an asymmetrical diode configuration and get different frequency-sensity clipping thresholds for each half cycle. Would this assist in getting some magical combination of the right harmonics in the right balance and no thinning?

9/29/1999 12:05 PM
GFR	Mark If you put a cap in series with the clipping diodes: When the diodes are "off", no current is flowing through them. With no current to charge the cap, no voltage will develop across it. If there's no voltage across the cap then the voltage at the diodes will be the same as f the cap was not there. So they will start conducting at the same level, independently of frequency. When the diodes turn "on" you have a cap to ground, just like a guitar tone control. You will get a treble cut that depends on the input signal. That's what Petri heard as "bassier" - just less treble. LXH2's Marshall simulator used this to mimic the fx that the frequency response of a power amp changes when overdriven (because feedback can't compensate the response anymore). If you want a frequency content dependant clipper I think you could do something like this: 1) Split the signal in two paths, one will have flat response and the other will have bass cut like in a TS. 2) Use the bass-cut signal to turn on/off a pair of transistor B-E junctions. 3) Tie the collector of these transistors to the output of the flat response path. This way you are clipping the full spectrum signal based on the level of the filtered signal. Not so complicated, with not so many parts and you just need a single pot for "more" - you only need to ajust the gain of the filtered path to have different clipping thresholds. AND... you could use the output of the filtered path too if you want, or simply clip the output of flat signal. This would give you three fx in one box: full range distortion, full range freq dependant distortion and midrange hump distortion.

9/29/1999 12:05 PM

GFR

Mark

If you put a cap in series with the clipping diodes:

When the diodes are "off", no current is flowing through them. With no current to charge the cap, no voltage will develop across it. If there's no voltage across the cap then the voltage at the diodes will be the same as f the cap was not there. So they will start conducting at the same level, independently of frequency.

When the diodes turn "on" you have a cap to ground, just like a guitar tone control. You will get a treble cut that depends on the input signal. That's what Petri heard as "bassier" - just less treble.

LXH2's Marshall simulator used this to mimic the fx that the frequency response of a power amp changes when overdriven (because feedback can't compensate the response anymore).

If you want a frequency content dependant clipper I think you could do something like this:

1) Split the signal in two paths, one will have flat response and the other will have bass cut like in a TS.

2) Use the bass-cut signal to turn on/off a pair of transistor B-E junctions.

3) Tie the collector of these transistors to the output of the flat response path.

This way you are clipping the full spectrum signal based on the level of the filtered signal.

Not so complicated, with not so many parts and you just need a single pot for "more" - you only need to ajust the gain of the filtered path to have different clipping thresholds.

AND... you could use the output of the filtered path too if you want, or simply clip the output of flat signal. This would give you three fx in one box: full range distortion, full range freq dependant distortion and midrange hump distortion.

Page 1 of 1