• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

A question of bias...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
diyAudio Senior Member
Joined 2002
Hi,

Exactly the same techniques can be used in an output transformer, but if you don't ask, you don't get.

O.K., I exaggerated.

Still, what I wanted to point out is that many existing amps in the field do sport OPTs with inherently umbalanced windings.

While modding and biasing those amps I, and no doubt others, noticed that either input stages and/or outputstages aren't properly balanced.
No matter how well matched valves you pick, you can't obtain perfect balance.

But here's the kicker; some amps actually sound better when a slight umbalance is introduced at the output stage.

Why?

Is the tiny DC bias on the OPT keeping it out of hysteris or what is going on?
Is it we resurrect a little more H2 and bring back otherwise suppressed micro detail...?

So what's going on?

Cheers,;)
 
Ex-Moderator
Joined 2003
Bias

Agreed. Most commercial amplifiers have OPTs with poor balance.

Perhaps a bit of DC imbalance lifts us off the zero point of the hysterisis curve and increases the initial inductance, which, in turn, would reduce distortion at low frequencies and low amplitudes. I'm just thinking out loud.

As another "off-the-wall" idea, rather than DC bias, I've often wondered whether output transformer linearity would benefit from AC bias in the manner of magnetic tape. Perhaps one could deliberately make the output stage oscillate gently at a few hundred kilohertz.

I've just realised that in meandering, I've fallen squarely back onto the title of the thread, even if it is in a different context!
 
Re: Bias

EC8010 said:
As another "off-the-wall" idea, rather than DC bias, I've often wondered whether output transformer linearity would benefit from AC bias in the manner of magnetic tape. Perhaps one could deliberately make the output stage oscillate gently at a few hundred kilohertz.

David Berning did something similar with one of his amps ages ago. There was a tech article or two on it in Glass Audio about 2000.
 
diyAudio Senior Member
Joined 2002
Hi,

I've often wondered whether output transformer linearity would benefit from AC bias in the manner of magnetic tape.

Yes, yes, absolutely...that's exactly what I meant but expressed so badly.

Experiments have shown PP OPTs to benefit from this, it's delicate to implement but the effect is definitely there.

Once again. no generalisations made...just a small field to experiment with.

But it's a neat little trick nonetheless.

To be continued,;)
 
Ha! What a coincidence, there's a thread running on RAT about this very subject. :)

BTW Frank, hysteresis doesn't occur around zero crossing, but rather any time the waveform stops for a moment. Which is why a hysteresis loop (on the B-H curves) is open for the majority of the graph, rather than nonlinearity at the center. It occurs in SE amps just as much as in PP amps.

BTW, why hasn't anyone ever taken a PP amp and dropped like half the current to one side (using an OPT tolerant of DC of course)? I keep hearing chatter of SE amps with canceled DC (by a CCS opposite the driving tube), and full PP but it sucks as compared to an SE amp :rolleyes:, but why hasn't someone tried both!?...

Tim
 
Ex-Moderator
Joined 2003
Loops

Sch3mat1c said:
Hysteresis doesn't occur around zero crossing, but rather any time the waveform stops for a moment. Which is why a hysteresis loop (on the B-H curves) is open for the majority of the graph, rather than nonlinearity at the center. It occurs in SE amps just as much as in PP amps.

What do you have in mind here? My understanding is that I need to expend energy bounding the area of a hysteresis loop. On the other hand, if I draw a curve through a set of decaying hysterieis loops, I obtain a vaguely straight line with reduced gradient at either end, indicating reduced inductance. (I think this curve is called an ogive, but I wouldn't like to be adamant on this.) In this respect, a single-ended amplifier having DC bias at the centre of the curve should (ignoring saturation effects) avoid the reduced inductance at the ends of the curve.

Comments?

Oh, and another thing. Why is "hysteresis" so hard to spell?
 
Re: Loops

EC8010 said:
What do you have in mind here? My understanding is that I need to expend energy bounding the area of a hysteresis loop.

Yup, it consumes energy, and it should be taking it right off the peak (probably a clipping type distortion). Dissipation should also be proportional to power passed, since a small alternating range of V (primary @ freq.) > I (primary) > H (field) > B (field in core), whatever you want to call it, will have a lower velocity around the peak (voltage reverses quicker on a higher amplitude sinewave, does it not?). This is backed up by losses being in percentage, within saturation limits.

Come to think of it, I like this velocity analogy. Seems to me any time it accelerates, it produces a sort of pulling bias on the B-H curve, perhaps comparable to Zo of a signal source? But out of phase... So on the up or downslope it's pretty linear, because speed is relatively constant; but near the peaks it changes much faster.
Reminds me of the reciporical world of cap vs. inductor too.

On the other hand, if I draw a curve through a set of decaying hysterieis loops, I obtain a vaguely straight line with reduced gradient at either end, indicating reduced inductance. (I think this curve is called an ogive, but I wouldn't like to be adamant on this.)

For the most part it's linear, the flattening you see at the peak is where the waveform slows down - so the hysteresis reduces and catches up to the ideal curve, which would be a straight line extending out from the center, with slope = permeability.

In this respect, a single-ended amplifier having DC bias at the centre of the curve should (ignoring saturation effects) avoid the reduced inductance at the ends of the curve.

This would be true if the hysteresis effect were absolute, but it appears not; only saturation (which *does* reduce inductance towards the peaks).

Oh, and another thing. Why is "hysteresis" so hard to spell?

Indeed. :clown:

Tim
 
Oh, and I was going to attach this: a drawing from Radiotron Designer's Handbook 4th ed., showing a typical loop. Note that if my theory is correct, this loop occurs at any DC bias, positive or negative, and any signal level, both within limits of saturation. Of course, it's scaled to the signal strength, such that hysteresis loss is a percentage of the signal input.
The point O at center, and the line leading to the main loop, is how the iron responds to a change from zero flux.

Tim
 

Attachments

  • hysteresis.gif
    hysteresis.gif
    3.6 KB · Views: 100
...Ok I admit it, I've tried that...

Tim said...

BTW, why hasn't anyone ever taken a PP amp and dropped like half the current to one side (using an OPT tolerant of DC of course)? I keep hearing chatter of SE amps with canceled DC (by a CCS opposite the driving tube), and full PP but it sucks as compared to an SE amp , but why hasn't someone tried both!?...

I've tried this. I had a pair of push pull transformers made that were happy with up to 40mA of unbalanced dc through them. I tried the same valves with different amounts of imbalance up to 75%. I tried different valves with different currents and different valves with different current and different B+, I even tried 807 one side and 6B4G the other...

They all sound different and measure differently from each other but the best measuring and best sounding are matched PP output pair with a tweaked imbalance to give the smoothest harmonic distortion progression i.e. with 2nd higher than 3rd and a nice smooth roll off out to higher harmonics (actually after 5th or 6th they are in the noise...). This seems to be between a 5 and 20% imbalance depending on the circuit and individual valves...

I was inspired to try this by Stefano Perugini's website PAEng Design Site

ciao

James
 
diyAudio Senior Member
Joined 2002
Hi,

and best sounding are matched PP output pair with a tweaked imbalance to give the smoothest harmonic distortion progression i.e. with 2nd higher than 3rd and a nice smooth roll off out to higher harmonics (actually after 5th or 6th they are in the noise...). This seems to be between a 5 and 20% imbalance depending on the circuit and individual valves...

That's exactly what I mean, James.

A few friends of mine and myself came to this conclusion by experience.
It doesn't work on all amps but most class AB amps seem to benefit.

I also like the idea of using a pair of identical SE OPTs as an experiment.
It's likely to be an expensive project, though.

I vaguely recall Dave (Planet10) toying with the idea around this time last years.

Cheers,;)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.