harmonic distortion

[Andy Le Blanc]

What is a good way to increase the levels of harmonic distortion
in an amplifier?....... even order?..... musical distortion?....

I noticed on the dumble spread a talk of affecting the balance of a chathode
coupled inverter to tune harmonic effects....
I been playing with paraphase inverters whos imbalance makes a very
characteristic distortion....
and wonder if there are more subtle ways to make more harmonic effects .....

[krash]

by design, a push-pull amplifier cancels even order distortion. Negative feedback further cancels it.

You can intentionally imbalance or unmatch the tubes in a p-p amp in order to get more even order distortion, since you will allow more distortion through without canceling.

Otherwise you can design in a stage to the amplifier that will produce 2nd order harmonic distortion and then you just amplify that in the amplifier as if it were regular signal. All of my amps employ an idea like this, and so do the 'wreck inspired amps.

[LOUDthud]

The idea that a push pull amplifier stage somehow (magically) cancels even order distortion is misleading. A push pull amplifier does not cancel even order distortion already present on its input. It's better to think of it this way. If you distort both halves of the waveform equally, you have odd order distortion. If you distort the wave differently on each half, you have even order distortion. Single ended stages like preamps produce mostly even order distortion until the waveform starts to clip on both sides. At that point, odd order distortion starts to rise but the even order doesn't go away. One thing that preserves even order distortion as the signal is increased is that grid current tends to cause a baseline shift. Too much grid current causes blocking distortion so the right grid stopper is one way to fine tune things. Another way to preserve even order distortion is to bias preamp stages slightly off center so that they don't clip equally on both sides.

Baseline shift and off center biasing prevent a preamp stage from producing a square wave with a 50% duty cycle. The 50% duty cycle has only odd harmonics and sounds hollow. The problem you run into with multistage preamps is that the stages invert so one stage distorts one half of the wave and the next stage distorts the other half. This can lead to excessive odd order distortion.

Theoretically, an RC network cannot create distortion. A pure sinewave input will not produce anything but a pure sinewave output, although the amplitude will usually change. But if there is some distortion present on the input, the harmonics can be emphasized and the output wave will look different. You can see this at the volume control on a blackface Fender. The wave looks pretty clean on the output of the first triode but you can see the harmonics at the volume control.

Consider the PI and power stage of the 5F6A and it's Marshall derivatives. If you drive the input to the PI with a large sine wave, it won't produce a square wave with a 50% duty cycle. There is a baseline shift on the input that skews the baseline of the input wave. There is something interesting that happends on some Marshall amps. If you increase the signal beyond a certain point, the bottom of the waveform starts to collapse after the initial positive to negative edge until it looks like an inverted shark fin. The size of the tail resistor seems to control this in some of my experiments.

I built an amp with a balance control like that seen in a Dumble. Looking at the output of a distortion analyzer, you could clearly see how adjusting the pot had an effect on the 2nd order distortion of small signals below clipping. It was a bass amp so I didn't look at what the effect on large overdrive was.

[Andy Le Blanc]

so maybe the right question is how do you NOT correct the responce of
an amp so it favors 2nd order and a regular decay of following harmonics
waveform imbalance..... unbalanced push pull
now I have experienced blocking distortion....
but what is the "right grid stopper" ?

[skyboltone]

You musta thought about that alot Mister Thud. That's about the clearest run through the stages I've heard in awhile. I'd like to repeat some of that study when I get a couple of amps out of the way. Especially the bit about the bias and tail of the PI. Did you take scope pictures when you did the study?

[angelodp]

Boy I sure would like to see the graphics on this. Can you guys post scope pics that show this distortion effect.

Ange

[Andy Le Blanc]

like wize....

Ive spent a lot of time on a scope but have had no good way to get a picture
it would be a good study...... musicality and harmonic content of guitar amps
ive seen the harmonic content affected by a good many things.....
in the HI-FI realm this is extremely significant... tells a lot about an amps subjective quality...

[LOUDthud]

I've got a Polaroid of the "Marshall Wave" somewhere. I actually made a photo album and it's in there with a lot of other scope photo's. This stuff was all done before I acquired a distortion analyzer.

The only other amp I ever saw that made the Marshall Wave was the Seymour Duncan Convertable. And it did it as you turned down the power with the power pot.

An interesting thing to look at is the X-Y plot of a preamp tube. Many scopes have an X-Y mode but on certain Tektronix scopes you put a vertical preamp in the Timebase slot and have control over the gain. I use X to look at the input and Y to look at the output. You like to think of it as a straight line but it's not. It has a curve that causes even harmonic distortion.

The right grid stopper is a tough question. When you start drawing grid current, the coupling capacitor starts to accumulate more charge. How fast it does this is affected by just about every resistor in the circuit as well as how much B+ sag there is. Then, when you cut a note off, it has to discharge back to where is was. So there are many things that come into play. As many of you know, the TrainWreck 3rd stage has no grid stopper. I think the magic there is the small coupling cap. On some amps you will see a diode that limits negative excursions on the input to a tube. This helps cancel out the charge accumulation on the coupling cap. At the end of the day you install a trim pot and try to find a value that works. You might not need one at all.

All these things fall under what I call the dynamic behavior of guitar amps. How an amp reacts when being overdriven. It is this area where tube amps react so differently from solid state amps. If you built a tube amp with regulated power supplys and lots of feedback stages such that the gain remained rock stable, you'ld end up with something nobody would want to play IMHO. Controlling the dynamic behavior (by design or by accident) is what distinguishes the best amps from all the rest.

[greiswig]

Really interesting discussion. I looked at a distortion analyzer (HP 331A?), but I couldn't figure out how it might work or whether it would be able to tell the difference between even order and odd order.

Any specific units you might recommend, LOUDthud? Or have you looked at any software/soundcard-based units?

[Ears]

The right grid stopper is a tough question. When you start drawing grid current, the coupling capacitor starts to accumulate more charge. How fast it does this is affected by just about every resistor in the circuit as well as how much B+ sag there is. Then, when you cut a note off, it has to discharge back to where is was. So there are many things that come into play.

Yes, and I think that it is easy to loose sight of the fact that any impedance introduced to the grid circuit will always increase distortion when positive grid current flows. Blame [edit: or thank] Mr Kirchoff.
I write this because I've read on many occasions (including Crowhurst) that gridstoppers are used to "reduce" the distortion from grid current. Such a claim, in my opinion, is misleading. Grid stoppers can reduce/eliminate blocking (and as Mr Thud points out the size of the cap is also relevant here) and mitigate Miller effects.

[LOUDthud]

First photo is the Marshall Wave. A time lapse shot. It squares up nicely at about a 40-60 duty cycle. As you push it harder the backside of the bottom half of the wave collapses.

Second photo is a shot of burst. The first few cycles are square but then the bottom starts to shark fin.

Third photo is a longer look at the burst. Note how the bottom half sags more than the top half.

[LOUDthud]

This is the X-Y plot of a 12AX7. IIRC 100K Rp, 1.5K Rk. Plus input to the right, minus to the left. Note how as the tube starts to cutoff on the left, the gain falls to near zero. Looking at these things is kind of tricky. At most frequencies you start to get loops caused by less than infinite cathode caps and the high frequency rolloff of the triode also causes a loop. The frequency of the test was probably between 500Hz and 2KHz. Those numbers on the screen are correct. Input 500mV per div, output 50V per div.

[LOUDthud]

I got one of those HP analyzers at a ham swap meet. Never could make any sense of the thing. I got a Tektronix AA501 and SG505 on ebay and thats what I like to use. Most distortion analyzers work like this. The incomming waveform is run through a twin T notch filter. This nulls one frequency. The resulting was is sent to an RMS voltmeter and then divided by the incomming wave to get percent. The Tek box auto nulls the fundamental, on the older HP you have to turn a dial to null it out. The usefull thing is to look at the distortion products at the scope output.

A friend of mine went out and bought a 24bit external sound card and the software to do FFT's. He can measure distortion down to something like .0002% the Tek stuff I have is good to .002%. It's nice to be able to digitize things and store them on your computer, a luxury I don't have at the moment. I've got a scope based spectrum analyzer but I don't think it works properly and it's really really slow.

[Andy Le Blanc]

being cheap I scrouged up a software thing on web called visual analyser 5.5
it works on old PCs and lame sound cards ... the basic function is ok
just having something that functions is a good first step.... free is better
I'll look around some more ... may we can find someting that can placed in the forums files for sharing....

[Andy Le Blanc]

VA - Visual Analyser 4.0
by Alfredo Accattatis

its free ware

at /www.hitsquad.com/smm/win95/SPECTRUM_ANALYZERS/

theres a lot of them maybe one can make it through the better judgement of the forum users... they are a usefull tool