Separate bias voltage

[Dave H]

I think he means this:

I don't understand it. If both pairs of tubes are in class AB then below the AB point all four tubes are on and above the AB point only two tubes are on. If one pair is now in class B then below the AB point (of the other pair) three tubes are on (both class AB and one class B) and above the AB point only two tubes are on as before so there's still an impedance loading change at the AB point. Is that correct?

[Mike Sulzer]

[ATTACH=CONFIG]41991[/ATTACH]

I don't understand it. If both pairs of tubes are in class AB then below the AB point all four tubes are on and above the AB point only two tubes are on. If one pair is now in class B then below the AB point (of the other pair) three tubes are on (both class AB and one class B) and above the AB point only two tubes are on as before so there's still an impedance loading change at the AB point. Is that correct?

Forget the class labels. Think of which tubes are biased on at idle and work out how the current through the tubes change as a ac voltage is applied. The so-called "B" tube on one side must be biased off enough so that it does not come on until the "AB" tube on the other side is turning off. Thus as the ac voltage increases, the operation changes from one tube on each side on to both tubes on one side on.

[Dave H]

Thanks, I've got it now. As the "class AB" tube turns off the "less than class B" tube is biased to come on.

[Chuck H]

'Course then you have the potential for two crossover notches with one in the audible portion of each wave. That could be fine tuned out, somewhat, but there would still be a two tube contribution, then only a one tube contribution in the audible portion of the waveform or the additional crossover notch. And even if perfection in duty transfer were achieved it couldn't be maintained because of the less than perfect impedance relationships. But that doesn't mean it would sound bad. Just thinking out loud. Not looking for reasons not to try it. I mean, if you have four separate bias supplies anyway!

[überfuzz]

I found it. It seems I messed up while doing the filter on the negative bias.

[Mike Sulzer]

'Course then you have the potential for two crossover notches with one in the audible portion of each wave. That could be fine tuned out, somewhat, but there would still be a two tube contribution, then only a one tube contribution in the audible portion of the waveform or the additional crossover notch. And even if perfection in duty transfer were achieved it couldn't be maintained because of the less than perfect impedance relationships. But that doesn't mean it would sound bad. Just thinking out loud. Not looking for reasons not to try it. I mean, if you have four separate bias supplies anyway!

In AB you have a tube going off while one stays on. I would think that another turning on at about the same place in the waveform could reduce crossover problems. Tubes shut off kind of gradually. I think you would do a bit of overlapping, and I personally would rather build this than try to analyze it in too great a detail. Of course you folks who do Spice...

[Chuck H]

Yup. If you're going to build it anyway all this head scratching is pale by comparison to having the circuit in front of you.

[überfuzz]

Of course you folks who do Spice...

... don't get to scratch our heads...

[KevinOConnor]

Hi Guys

Obviously the OP was referring to a matched set of tubes, which are of some nominal use when there is no means to hum-balance the output stage using bias pots. Ideally, there are cathode current-sense resistors and meter jacks added so that safe adjustment canbe made quickly and without removing the chassis from the cabinet.

The best solution was shown in post-8, where one bias supply feeds four bias pots as the bias-set network. This is the most versatile approach as it allows one to mix tubes, or to set the tubes at widely different idle points, as Mike suggested. The use of a high-biased pair and a low-biased pair was called "extended class-A" in a 1950s article which became the basis for mesa's 'simulclass' amps. In the latter, the high-bias tube pair were tied to the UL taps of the OT, so their contribution to total output is quite small. In my amps, each tube can be biased individually and can be independently operated in fixed or cathode bias to extend the tonal variety of the output stage.

As for the standard push-pull output with four tubes, there are two options regarding biasing: one adjustment per side , or one adjustment per tube. With one adjust per side, it is easy to set one side by meter so you know the tubes are idling safely, then set the other side by ear for minimum hum. This is important as the hum-balance point of the OT is slightly different from the DC balance point for the output stage. With individual adjusts, hum-balancing is a bit more difficult. You could DC-balance everything first, then skew one tube on one side for minimum hum. Or, you could set the first tube on one side by meter, then hum balance the second tube onthe other side, then DC balance the third tube on the first side and hum balance the last tube on the second side. A lot of back and forth...

As for the mixed biased tubes: I think the caveat for having listenable sound would be that the class-A region must be provided by at least one tube pair. Normal biasing to 30mA or so does this in typical amps and provides at least 1W of clean output. Additional tubes biased cold can be driven 'on' as more power is needed, but what happens to the 'hot' pair, which is undoubtedly receiving the same drive? Is it taking more than its share of the load? This depends on how hot and how cold the tube pairs are. In some amps I built years ago, it was possible to switch between PP and SE. In the SE mode, one tube would be cathode biased to provide the SE output, but the other tube would remain fixed-biased and its contribution could be dialled in/out. Because it was still idling at whatever the player had set it to, its current swing was asymmetric and the tonal contribution was quite fat.

Have fun

[Mike Sulzer]

Additional tubes biased cold can be driven 'on' as more power is needed, but what happens to the 'hot' pair, which is undoubtedly receiving the same drive?

The hot pair hit saturation sooner and the gain of the system drops; with further drive current through the cold tubes continues to increase. Thus, there is a sort of soft clipping followed by hard clipping when the cold tubes hit saturation.

[Gingertube]

For London Power method of separate bias for each tube and switchable cathode/fixed bias see the pdf attached to this post #48 here:
http://music-electronics-forum.com/t36969-2/#post351885

Some other of my notes on Power Scaling etc. in that thread too.

Cheers,
Ian

[Raybob]

I've done this with Twin Reverb circuit, but did it so that mismatched tubes can be 'matched' at least in idle bias current. It required parallel coupling caps (4 total) to separate signal going to each tube. The bias supply supplies all 4 bias pots with each bias feed going to it's own coupling cap signal.