Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 29, 2024, 06:08:33 AM

Login with username, password and session length

Recent Posts

 

Tube clipping causing crossover distortion and leading to compression

Started by skey, November 28, 2006, 10:37:00 AM

Previous topic - Next topic

skey

Here's an interesting Peavey patent. 

http://www.freepatentsonline.com/5524055.html

"... Thus, the output tubes 16 and 18 become over biased beyond class-B and at severe output clipping significant crossover distortion is generated as well. Consequently, at overload, the output signal of tube amplifier 10 will be clipped at the peaks. However, it will not be as "dirty" as a typical solid state power amplifier operating under the same conditions, because a large portion of the overloaded output waveform is forced or compressed into the severe crossover distortion region. To a musician, such a waveform is much more musical in nature and "cleaner" (i.e., less harsh) than a solid state amplifier at overload. Due to the compression (i.e., distortion near the zero crossover), the actual peak output clipping is reduced and is far more tolerable than that of the solid state amplifier. This phenomenon is thus, tube power amplifier compression."

There's circuit schematics that simulate this type of compression.

I commonly see diodes for clipping, but never anything for adding crossover distortion, and the effects it has on clipping.

teemuk

Interesting idea but I honestly doubt it's sonic qualities: The crossover distortion generates a lot of high order even and odd harmonics, which are not very ear pleasing. Of course, this is a distortion circuit (which aims to adding extra harmonics) and everyone has their taste about a distorted tone.

BTW, you can easily simulate the idea by putting two parallel diodes (in opposite orientations) to the signal path. Instead of shunting signal that exceeds their forward voltage to ground (diode clippers) just put them to the signal path where they will pass only the signal part exceeding their forward voltage. I doubt if you like the result. I don't.

Edit:

Just for curiosity I simulated the circuit in the patent and found out it has less to do with crossover distortion than advertized: The circuit controls the bias voltage of two opamp stages according to input signal amplitude. Then it individually "compresses" both halfwaves by clipping. The result is that due to varying DC offset the amount of clipping is dependant on the signal amplitude and not constant as in typical diode clipper circuit. Last, the two waves are summed, which further softens the knee of clipping.

Note that in no point of the process is the signal attenuated during crossing over the zero region - as happens in true class-AB/B amplifiers, in which each active device amplifies only its corresponding halfwave current. So, virtually only a very small amount of crossover distortion can exist (in this circuit). This amount is the result of summing the signals together - which is also the main reason why the circuit clips very softly! The figures 3A and 3F are misleading: This circuit does not output such a deviation of sine wave. (Circuit I described in paragraph two, however, does). Waveforms like these are not even the output of typical class-AB amplifier - they are output from true class-B amp (with no crossover distortion compensation whatsoever). True class-B amps are not used in audio applications because of the crossover distortion they introduce!

skey

Thanks for the feedback :)
Can you post the spice model?

Here's a spice model of an amp that closely matches the tube amp in the patent:
http://duncanamps.com/zips/el34_push_pull.zip
If you raise the input voltage to say 2v you will see something very close to diagram 3F.

At think the point is that class-ab tube amp overdriven migrates to class-b type output.

Interestingly, at http://tech.groups.yahoo.com/group/LTspice/ there is a schematic of a preamp that does crossover distortion as well as clipping:
Files/Examples/Apps/Audio Distortion Preamp for Electric Guitar



teemuk

Sure, I can post it... but not right now: My own computer is busted - again - so I can not access the files on its hard drive. I did some further simulations after posting and indeed found similar results as in one of the diagrams (so I was a bit haste to comment the waveform graphs in the patent). ...But the circuit had to be driven hard into clipping to get that crossover distortion (spice simulation can cope with this only to some extend - after that the simulation becomes just too slow). As far as I have understood the "crossover" section actually contains no audio data since its a product of two waveforms canceling each other out. Correct me if I'm wrong, though.

By the way, my opinion is that major part of the circuit's tone still comes from summing the two signals together (error correction): I built another model, where the "top" stage of the patent circuit was biased to half supply (0 V)and clipped symmetrically (like in ordinary diode clipper circuit). The "bottom" stage was just a buffer. The signals of both stages were summed together (I had to tweak the summing resistor value a bit) and (not so) surprisingly the result was almost similar to the patented circuit. This circuit did not have crossover distortion but I could not hear any difference in the tone though...

doug deeper

here is a fuzz i came up with that uses crossover distortion if you'd like to here what it sounds like at extremes!