Wierd note with unison bends

[jeff]

I've been noticing lately a wierd phenonomen when playing with a lot of gain. If you play a unison bend on the high frets like a G(15th fret E string) and bend an F up to a G(18th B string) It sounds like there is a 3rd note which lowers in pitch as the note is bent up. Almost sounds like a ring modulator.

For example listen to the end of rush's Working Man(~6:38)Rush-Working Man With Lyrics[/url]  where he's playing alone and bending notes. It sounds like the two notes are beating against each other and producing a wierd 3rd note.

I notice on my amp it's very pronounced. Is this just a natrual byproduct of high gain? Is this just the way it is with gain or can something be done so it's less noticable?

[tubeswell]

what's causing it on his amp and on your amp could be two different things.

My first thought is maybe its cone cry. What kind of speaker are you using?

[jeff]

I found a good deal on a Marshall 4X10 a long time ago. I'll look for model but they're Celestion 25W. I thought it was the amp but maybe it is the speakers. Have you ever noticed this sound while playing?

Now that I think about it this amp has no NFB loop could that have anything to do with it?

[HotBluePlates]

I've been noticing lately a wierd phenonomen when playing with a lot of gain. If you play a unison bend ... Almost sounds like a ring modulator.

It's intermodulation distortion. It is a byproduct of high gain in some circuits (like some high-gain distortion pedals).

What happens is extreme distortion causes not only multiples of the notes you're playing (harmonics), but if you play more than one tone, you get sum and difference frequencies of the notes you're playing (and their harmonics). So if the notes themselves are 250Hz and 400Hz, you get 650Hz and 150Hz, along with multiples of the fundamental notes and multiples of 650Hz and 150Hz.

Some designs have more intermodulation distortion than others. The difference frequency subharmonics stand out though.

I've seen references in engineering texts about intermodulation, but I can't recall anything suggesting how to reduce it (other than simply reducing distortion). That said, there are some designs that have gobs of distortion and don't have as-pronounced inter-mod. I don't know why; maybe PRR has heard an explanation.

My first thought is maybe its cone cry.

It could be, but the mention of high distortion is a tip-off. You can readily repeat this occurance with a fuzz-type pedal (or other high gain device with lots of harmonics) at volumes too low to cause speaker-related artifacts.

Is this just the way it is with gain or can something be done so it's less noticable?

To an extent, it's the way it is. But if you play with very high gain, you might have to roll off a LOT of bass to try to fight the inter-mod subharmonic. This is the real reason why mid-vintage and later Marshall amps seem so rediculously bright; they're stripping out the bass that's gonna get muddy, rolling and have weird subharmonics under heavy distortion.

[jeff]

That makes sense, to strip the bass a little. I have the bass knob kinda low to get the sound I like. But, the tone stack is after the preamp. Maybe if I strip the bass earlier in the preamp and then set the bass knob up I could reduce it some. That way, having the bass thinned out to begin with I'll be distorting a thinner sound instead of thinning out the distorted sound. Maybe it won't be so noticable or be accented as much, getting worse and worse, through each stage.

I have noticed it at my friends house with his rig but he uses much more distortion and it seems to be less pronounced.

[HotBluePlates]

Any pedals in the mix?

The catch is you might still notice it near/above the 12th fret. If you think about it, unless the notes are very close, the difference tone will still be higher than where you're likely to strip bass.

You'll hear this with Tubescreamers set high on the gain pot, with Fuzz Faces, etc. The more intense the distortion, the more you'll run into intermodulation.

[Ritchie200]

I've always wondered about this as well.  I think one of the best examples (in my mind) is early Metalica.  They had a "barking" almost blocking distortion-type squawk when they would chugga chugga on power chords.  A high gain pedal into the 14, cascaded 12AX7's in his Mesa....?  I always thought it was a combination they lucked into that fortunately happened to be somewhat musical.  I'm sure there are other examples that are not so pleasing to the ear!

Jim

[jeff]

Any pedals in the mix?

No, just straigt into amp. I though maybe I could "fix" it but I guess this is just a byproduce of distortion and will have to accept that's just the way it is.

[thelonious]

I guess this is just a byproduce of distortion and will have to accept that's just the way it is.

Probably, but there is an interesting thread here: http://forum.metroamp.com/viewtopic.php?f=28&t=31643 where a guy solved his particular ghost/intermodulation problem by playing with filter cap values. Whether there was also quite a lot of wishful thinking involved with his solution... I don't know. In any case, of course this wouldn't work with a tube rectifier.

[HotBluePlates]

His situation involved intermodulation of distorted notes played with hum in the power supply.

That's the same mechanism, but different reason, than the case described here.

Get a distortion pedal, the nastier the better. Make sure the bass is up somewhat and max out the gain. Play and hold a D (2nd string 15th fret) while bending an A (3rd string 14th fret) up to B. You'll hear a strange subharmonic tone going downward while you're bending up.

You'd probably only notice the extra tone during the bend; your attention isn't drawn to it while the musical intervals are static, because there's so many new harmonics added as a result of the distortion.

[jeff]

I tried thinning the sound using smaller caps, but do you think using snub caps may help?
Since the noise is more apperent on high frets, I'm thinking the higher frequencies are modulating to produce the noise in the audible range and the lower notes, when modulated, create the difference note too low to hear.

When a a tube starts to distort, does it creates higher and higher order harmonics?
Would using a snub cab limit these?

It seems that higher and higher frequencies are causing the low noise to reach the audible range so maybe instead of trying to filter out lows to filter out the effect I should be filtering out the highs causing the effect.

Does that make any sense? Whaddya think?

[HotBluePlates]

Try it, but it seems to address the opposite of what's happening. Ultimately, you're saying, "I want less bass, so let me cut the treble."

Intermodulation is simply a byproduct of a non-linear impedance causing distortion. Normally, this is done on purpose in the intermediate frequency (IF) stage of a radio, where a purposely-distorted stage causes a difference harmonic (subharmonic) to be created. The coupling network is tuned to pass the single exact frequency of this intentional subharmonic to pass to the IF amplifier stage. Generally, that "subharmonic" is 455kHz.

If reducing bass doesn't help, the only real cure is to reduce the distortion. Or live with the subharmonic effect of inter-modulation.

[jeff]

Ultimately, you're saying, "I want less bass, so let me cut the treble."

In a mannor of speaking, I was wondering if cutting the (ultra)highs would cure the disease and cutting the bass would simply be treating the symptoms. I don't want to filter out the bass, just the unwanted bass notes the high frequencies create.

This was my line of thinking:

1)If two low notes together(unison bend 3rd fret B string, open E) doesn't produce a sum note in the audio range but a high unison bend(15th fret B 12the E) does, then the higher the two notes are, the more pronounced IMD.
2) When a note is heavily distorted it produces high order harmonics.

 When a tube is distorted does it create high order harmonics that are frequencies beyond the speakers ability to reproduce?
If you don't hear those frequencies anyway can they be filtered out so it lessens the amount of IMD? Since the higher the two notes are the higher the sum note is.

So not really cutting the highs you can hear but are there ultra highs beyond what the speaker can produce causing notes, that when mixed, produce notes that the speaker can? I don't know much about harmonics.

Another reason I ask is because I've been noticing snub caps on high gain builds. I've always thought those were used to prevent oscillation but am now wondering if they are use to lessen the IMD.

Just a thought I may be way out of line here.

Thanks
  Jeff

[HotBluePlates]

... I don't want to filter out the bass, just the unwanted bass notes the high frequencies create. ...

The tube distorting creates the frequencies, not high frequencies creating new frequencies.

You play two notes of 100Hz and 120Hz through a heavily distorting tube. The output will consist of 100Hz, 120Hz, a sum frequency of 220Hz and a difference frequency of 20Hz. Also, there will be harmonics of 100Hz (200, 300, 400, 500Hz, ...) and 120Hz (240, 360, 480, 600Hz ...) as well as sum and difference tones of each of these new harmonics. Heavy distortion and complex chords often turn into a mess for this very reason.

But the sum/difference tones of the harmonics may not be too prominent, depending on the amplitude of these harmonics relative to the fundamental tones.

The sum tone of your two fundamentals will be prominent, because the two original tones were prominent, and the tube is distorting hard.

The difference tone would likely be inaudible, because your amp/speaker probably don't have good reproduction at 20Hz. Now if you play higher... well now the difference tone is in the range of your speaker.

Hopefully now you understand how sum/difference tones relate to the notes you play, and how these are separate from the other harmonics of the notes you play.

[genesys]

I've noticed and tried to diagnose this phenomenon for a long time.  It isn't just limited to tube amps, I can generated it easily in my digital modeling gear, through headphones and through transistor amps.  I think your first instinct is actually right.  I seem to be able to trace the effect mostly to the gain stage.  Backing off the gain seems to lower the effect.  Based on some other observations in the forum I have some other hunches about how to mitigate or use it.

I seem to experience that the subharmonic difference tone is the most pronounced/loudest when I'm doing unison bends in the middle of the neck on the middle/high strings (G/B/E).  It's been a while since I tried to diagnose it pseudo-scientifically, but I think I'm right to note that you can get a strong low subharmonic that bends a full fifth in unison down at the same time that I bend a fourth interval unison up a full step on the lower note of the two to yield a Third interval when the low note is bent and the high note stays the same.  It seems less noticeable (or less annoying to my ear) on unison bend that are lower in pitch.  This leads me to think that there might be something to the notion that the strength of the difference tone is particularly pronounced at a given range of pitches in the fundamental notes of he unison bend.  Perhaps the subharmonic effect intensifies within the gain stage to produce additional low frequencies.  That's just me speculating, because the subharmonics I experience are almost of equal volume as the bent tones.  It would be pretty ambitious to try to tame them, even when you know you can control them, because they just don't sound very musical layered with the unison bends, and that also don't have the "attractive" (I call it the motorcycle hum) clipping quality in the beats that they produce which can sometimes sound good on some amps.

[spacelabstudio]

I think what you're hearing is really just a product of how waveforms add.  You know how you can tune two guitar strings to another playing harmonics and listening for the beat frequency?  That beat frequency is the interfrequency modulation--it gets lower in frequency the closer the two notes get to each other and finally goes away when the notes are the same.  Now when you're tuning your guitar, you hear the effect of that frequency as beating, or a change in volume, like a tremolo effect, because the beat frequency itself is subsonic.  It doesn't produce a tone, but you can still hear it because of its effect on the volume of the other two notes.

When you're playing with a lot of distortion and your sine waves start to look more like sine waves, then you're adding a ton of higher order harmonics.  Each note in your bend has its own series of harmonics and they're going to interfere with each other and create beat frequencies.  The difference now is you have harmonics that are loud enough and high enough in frequency that their beat frequencies are no longer subsonic but can be heard as a tone in their own right.  Which is where you get your ghost notes from and why the ghost note seems to go down in pitch as the notes you're playing get closer to each other. 

My guess is amplifiers that let through more high frequency harmonics will have more prominent ghost notes, since there's more there to interfere.  If one note has a harmonic at 20k and another note has a harmonic at 22k, you can't hear their harmonics, but you can hear the beat frequency they make when summed together at 2k.

[genesys]

Spacelab -

You seem to be saying what I was trying to say, only more clearly and coherently.  I guess I'd echo your thoughts that the level of higher order harmonics being produce in the signal which then get amplified in the gain stage do seem to increase the "intensity" of the subharmonic/difference tone you hear.  Maybe a better analogy is to say there appears to be a multiplier effect where the incremental gain increase causes the higher order harmonics to gradually produce subharmonics that increase in kind.  Going back to middle school math, it also sounds to me that the relationship is probably a mildly geometric one (i.e., for each linear increment in gain, I think I hear a multiplied strength in the subharmonic).  It doesn't ever get as loud as the fundamentals of the 2 bent tones, but it's pretty close sometimes).  I also want to correct my earlier post a bit after trying to reproduce the interval of the descending subharmonic you get when doing the unison bends at given intervals.  Last night I think I noticed that the bending up the lower note of a unison bend with a Fourth interval (bent into a Third Interval) produces a Fourth interval when fully bent.  My earlier post suggested the resulting subharmonic was a Fifth interval change, but I've rethought that after trying it again.

Although the subharmonic tone produced seems to have two sonic elements (1-the 'fundamental' of the subharmonic or ghost tone, and 2-the beats produced by the intermodulation) to my ear the fundamental of the subharmonic is sonically stronger that the clicks/beats, in my opinion.  The effective tone of the subharmonic can be a little harsh, but not as annoying as really intense beating would be.

These thoughts seem to lend some credence to the idea that if the EQ can be rolled back or certain frequencies filtered from the signal in the amplifier, it's possible to reduce the  undesirable subharmonics (which is an alternative or complement to reducing the gain -- where I happen to think it all starts).

I think what you're hearing is really just a product of how waveforms add.  You know how you can tune two guitar strings to another playing harmonics and listening for the beat frequency?  That beat frequency is the interfrequency modulation--it gets lower in frequency the closer the two notes get to each other and finally goes away when the notes are the same.  Now when you're tuning your guitar, you hear the effect of that frequency as beating, or a change in volume, like a tremolo effect, because the beat frequency itself is subsonic.  It doesn't produce a tone, but you can still hear it because of its effect on the volume of the other two notes.

When you're playing with a lot of distortion and your sine waves start to look more like sine waves, then you're adding a ton of higher order harmonics.  Each note in your bend has its own series of harmonics and they're going to interfere with each other and create beat frequencies.  The difference now is you have harmonics that are loud enough and high enough in frequency that their beat frequencies are no longer subsonic but can be heard as a tone in their own right.  Which is where you get your ghost notes from and why the ghost note seems to go down in pitch as the notes you're playing get closer to each other. 

My guess is amplifiers that let through more high frequency harmonics will have more prominent ghost notes, since there's more there to interfere.  If one note has a harmonic at 20k and another note has a harmonic at 22k, you can't hear their harmonics, but you can hear the beat frequency they make when summed together at 2k.

[HotBluePlates]

I've noticed and tried to diagnose this phenomenon for a long time.  It isn't just limited to tube amps ...

I think what you're hearing is really just a product of how waveforms add.  ...

Yes to both, but add, "how waveforms add in a nonlinear impedance/amplifier."

I looked up all this stuff again. If you had a perfectly linear (0.0000000% distortion) mixer/amplifier, two input frequencies would be perfectly added and result in an output with the same two frequencies, only louder/softer as determined by the operation of the mixer/amplifier.

But if you have any non-linearity (distortion), you get harmonics of the original 2 frequencies plus sum/difference frequencies and their harmonics. How bad the intermodulation (IM) is depends on how much non-linearity there is.

There's an odd effect I re-read about, which lead me to originally think there was some old method to minimize it: If you apply negative feedback to the non-linear stage, depending on how the feedback is applied the total harmonic distortion may be reduced but the intermodulation left untouched. That results in the IM being more apparent and ugly by comparison.

So you will not get rid of IM unless you get rid of distortion. But it doesn't increase necessarily at the same rate as harmonic distortion. You might reach a balance where it doesn't matter or you don't notice it. If you have to cut the subharmonics, and can't cut gain, you'll have to strip out low end to reduce their effective volume.

If you think about it, this is why a single note sounds fuller/thicker with distortion than when played cleanly. There's harmonic distortion of the single note, and probably some small amount of IM subharmonics being created from the difference frequencies of the note and its natural harmonics.

This is also why octaves/fifths sound better with insane levels of distortion than 3rds/2nds. If you play a 1k and 2k octave, the main IM subharmonic is the difference, or 1k, which is the same as the lower note.