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

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

Re: Crossover Notch Distortion

5/25/2000 10:46 PM
Ray Ivers
Re: Crossover Notch Distortion
I forgot about the instantaneous phase shift occurring within the OT itself - is this what you meant by 'change in volt-time'? Is there a certain amount of 'release time' associated with core saturation? If so, I can see how that effect might last long enough during each cycle to affect operation at the crossover region. I would think also that the voltage discontinuity you mentioned could also manifest itself as spikes at the crossover region, depending on the amount and frequency of saturation, and the amount of leakage inductance. I guess the best way to test would be to build a 15 watt EL84 circuit using a Hammond 1650W xfmr, to eliminate even the slightest possibility of core saturation. The ampere-turns that are 'eaten' - I guess they are just lost as heat, or maybe mechanical vibration.  
This was not meant to be a '20-questions' reply, just musings in question form, R.G. - but I'd love to hear anything you had to say on the subject.  
And now, a word from our sponsors:

5/27/2000 3:32 PM
Stephen Conner

Reid is right. I've seen the exact same effect on my 50 watt EL34 power amp. You can bias it as hot as you like, but if you drive it far enough into clipping, the crossover notch will reappear. I'm satisfied that this is caused by blocking due to self-biasing as Reid says. If you wanted to get rid of it, you would have to drive your power tube grids with cathode followers or MOSFETs - or play quieter. It doesn't bother me since it only happens at around volume 7, but my ears are already bleeding by 5.  
You can prove to yourself that it's not saturation by trying the same test with different frequencies of signal. Saturation gets worse as the frequency goes down (for tech heads: the core flux is proportional to the time integral of the voltage) but blocking distortion stays the same or gets better, depending on your coupling caps.  
Steve C.
5/26/2000 3:01 PM
Dave H.

"I have seen the notch emerge from a previously clean waveform at or near clipping. It strikes me that this is not 'crossover notch distortion' in the classical sense."
I think the effect is due to a shift in bias point not OT saturation. Iíve seen the notch on my 4 x EL84 amp when itís driven to clipping at 36w but the waveform looks the same when I disconnect 2 x EL84s and only push 16w through the transformer. If you set bias by viewing the crossover notch on a Ďscope with the amp clipping you probably will end up biasing it too hot. The crossover distortion you see emerge when the amp is driven hard into clipping is probably masked by the clipping distortion anyway. Classical crossover distortion tends to be more constant with signal and is therefore more objectionable at small signal levels because it is then a higher percentage of the output. When you looked at the crossover notch on the SS amp did you have to wind up the gain on the Ďscope and look at the cross over region to see the notch? You canít usually see the notch on the full sine wave because of the heavy -ve feedback used.  
If you were setting the bias on a hi-fi amp you would adjust it to minimise the "classical" crossover distortion on small signals. You donít care what happens after clipping because the amp wonít be operating in that region but what about a guitar amp which is normally overdriven? If you use the Ďscope/crossover notch biasing method do you set the bias to minimise the overdriven crossover distortion or the classical crossover distortion?  
5/26/2000 4:13 PM

I think Dave put it in perspective. The example of setting the bias for minimum distortion in a hifi amp is easy to visualize.  
We tend to take for granted that tubes are simple linear devices. That's true for some triodes when operated within strict small signal limits. But there really aren't any linear pentodes, and there are only certain places where their operating characteristic curves are flat and evenly spaced. Stray from that local region, whether it be from large input signal excursion, bias shift, or operating point shift due to the complex dynamics of an output transformer load, and the linearity behavior rules go out the window. The tube is no longer operating where you initially drew your simple resistive load line. The operating point is moving all around. Put a small pure sine wave signal on the grid and look at the output waveform across a resistive load. The output waveform looks just like the one in the textbook. Raise the signal level to a magnitude where it's competing with the control grid's negative bias and the grid no longer has complete control over plate current waveform.  
5/26/2000 4:15 PM
Peter S

I always set it to minimize the notch during heavy clipping near maximum volume, and check to make sure the tubes are not overdissipating at maximum volume. As long as the tubes are not overdissipating at maximum volume, I have not found the tube life to be noticably shortend by doing it this way. I always monitor the current draw on the tubes at idle when biasing the amp this way, it almost always falls in the 65% - 75% of max dissipation range on most amps. If I bias the amps colder, the distortion sound gets pretty raspy and harsh.  

<<First Page<PrevPage 2 of 2