I understand it is for SE operation. Just going off other specs I would assume 5-6Kpp to be around the primary needed so your 6.6K isn't too far off of what I would expect but I was unsure how much load resistance needed to change with huge changes in operation like you were up to. The 6V6, for example, goes from 10Kpp at 250V and 10W to 8.5Kpp at 280V and 14W.
Are you sure about that? The 6AQ5 datasheet, for example, says 5K for SE and 10K for PP. I always assumed you just doubled it. If that's not true I've got some recalculating to do!
"The question is often asked; "what load do the valves 'see' in Class AB?". The answer lies in the name 'Class AB' - a combination of Class A and Class B.
While both valves are conducting the amplifier operates in Class A, and both valves 'see' a load of ½ the anode-anode impedance of the transformer (1/2 Za-a). However, as soon as one valve cuts off, that half of the transformer's primary is no longer part of the circuit. Because the impedance ratio is the square of the turns ratio, the load presented to the remaining 'on' valve must be only ¼ Za-a. So the stage operates in Class B at higher signal levels. Drawing a load line to show this is simple:"
The Valve Wizard -Push-Pull
While both valves are conducting the amplifier operates in Class A, and both valves 'see' a load of ½ the anode-anode impedance of the transformer (1/2 Za-a). However, as soon as one valve cuts off, that half of the transformer's primary is no longer part of the circuit. Because the impedance ratio is the square of the turns ratio, the load presented to the remaining 'on' valve must be only ¼ Za-a. So the stage operates in Class B at higher signal levels. Drawing a load line to show this is simple:"
The Valve Wizard -Push-Pull
In practice, the tubes are developing their strongest gm (approx. gm = k Ip ^ 0.5) in the outer class B sections, so one might see an 8K OT used instead of the nominal 10K for class aB. A load line analysis can find an optimum. (trying to keep total sum gm across the full primary constant, so there is the question of optimum conduction overlap also, or use more N Fdbk to fix the discrepancy since it never comes out constant)
Dropping to 6.6K Zpri for class aB is generally running the 6AQ5 tubes harder to get more power. (and usually a little more distortion than an optimum Zpri)
Using tubes that are a little "over qualified" for the job will give you the best "design headroom".
Dropping to 6.6K Zpri for class aB is generally running the 6AQ5 tubes harder to get more power. (and usually a little more distortion than an optimum Zpri)
Using tubes that are a little "over qualified" for the job will give you the best "design headroom".
Last edited:
OH!!
I just looked at the 6AQ5 GE datasheet and they have a 5K OT for the class A SE case, which would put class AB P-P at a 20K OT, and they have already done a -HUGE- reduction to 10K for practicality. (so forget my 8K figure, that would be more like 16K)
Lesson:
Always use a TV Sweep type tube with high current capability. Later TV Sweeps have such efficient cathodes that they do not even incur a heater power penalty versus old low current tubes.
6AQ5: heater 0.45 Amps for 45 mA max DC spec, 10 Watts Pdiss
6BQ5/EL84: heater 0.76 Amps for 65 mA max DC spec, 12 Watts Pdiss
6LG6: heater 2 Amps for a 315 mA max DC spec, 28 Watts Pdiss
6KG6/EL509: heater 2 Amps for a 500 mA max DC spec, 34 Watt Pdiss
Small Sweep tubes: 12/6AV5 or 12/6BQ6 11 Watt Pdiss, 6GF5 9 Watt, 38/12HE7 10/15 Watt
Next size up: 12/6JN6, 12/6GE5, 12/6GT5, 21HB5, 6DQ6B, 6JR6, 6JG6, 6GB5 17.5 or 18 Watt Pdiss
I just looked at the 6AQ5 GE datasheet and they have a 5K OT for the class A SE case, which would put class AB P-P at a 20K OT, and they have already done a -HUGE- reduction to 10K for practicality. (so forget my 8K figure, that would be more like 16K)
Lesson:
Always use a TV Sweep type tube with high current capability. Later TV Sweeps have such efficient cathodes that they do not even incur a heater power penalty versus old low current tubes.
6AQ5: heater 0.45 Amps for 45 mA max DC spec, 10 Watts Pdiss
6BQ5/EL84: heater 0.76 Amps for 65 mA max DC spec, 12 Watts Pdiss
6LG6: heater 2 Amps for a 315 mA max DC spec, 28 Watts Pdiss
6KG6/EL509: heater 2 Amps for a 500 mA max DC spec, 34 Watt Pdiss
Small Sweep tubes: 12/6AV5 or 12/6BQ6 11 Watt Pdiss, 6GF5 9 Watt, 38/12HE7 10/15 Watt
Next size up: 12/6JN6, 12/6GE5, 12/6GT5, 21HB5, 6DQ6B, 6JR6, 6JG6, 6GB5 17.5 or 18 Watt Pdiss
Last edited:
...For practicality? What it really means is that the prior analysis is wrong. Each tube plate hooked to a PP primary half never sees more or any less the turns ratio for its own half. Doesn't matter if one tube is in cutoff. The windings are done in opposite direction for each half so that the idle currents can cancel each other. They are not like a single winding with a 50% screen tap. The CT creates a ground point for signal AC and the two windings are never aware of each other. The secondary is the only winding that sees the affect of current through the primary halves and the turns ratio is always fixed.
It happens, but not in this case. Don explained well what happens, but I would prefer to see it from the different side: assume load ipedance for both tubes P-P constant, but use the sum of Gms all the way, from peak to peak.
Huh?
Wait a second.
When both tubes are conducting, each tube is supporting half the load current as seen on the primary side (so effective turns Z is doubled). When one tube is conducting alone it has to support the full load current. (as seen on the primary side)
If a 5K OT load is optimum for the 6AQ5 in SE -Class A- pentode, then a 10K OT would be correct for a class A P-P case. Each tube would see the 2.5K half primary turns, but doubled to 5K since only half the AC current would need to be supplied.
For class B, the OT would need to be 20K P-P so that each tube would see the 5K half primary alone. Class AB falls somewhere in between due to the overlap in the region where the tubes have lower gm.
The 10K sited for P-P class AB on the GE datasheet would be half the nominal Zpri converted from the SE class A case. Likely some max V tube spec limitation, as well as the practical OT limitation, and the constant gm adjustment mentioned, came into play there.
Right?
Soooo, crush those 6AQ5 tubes and all those 6V6 look-a-likes! Fake wimpy tubes!
Or.... use a 20K Z OT, like I warned about earlier. (16K)
Or put up with more distortion with a lower Z OT and use some extra N FDbk.
Or bust some specs for higher power, which uses a lower Zpri.
Last edited:
My post was already covered above pretty well...
So is it correct to say that a 2.5K SE spec can be assumed to be a 5Kpp spec in PP?
6L6 datasheet isn't as straightforward so probably not. Looks like a lot of datasheets actually have similar numbers for load between class A and AB.
So is it correct to say that a 2.5K SE spec can be assumed to be a 5Kpp spec in PP?
6L6 datasheet isn't as straightforward so probably not. Looks like a lot of datasheets actually have similar numbers for load between class A and AB.
Last edited:
Yeah, when one tube is in cutoff the other one is carrying all the current load but it is going to produce it because it is close to the highest swing peak. But the winding ratio for its half doesnt change. The current through each winding is additive at any instantaneous second. One is falling while the other is rising and is the product of the volume control. Lower that and you come back into class A operation if no tube is ever in the cutoff region any longer.
The two windings of the primary are seperated by the filtering of the B+ and become a ground point to AC at the CT. No current from one tube ever travels through the other winding so it is basically non existant to each other. But the current through one winding half transfers magnetic energy to the secondary if the other is in cutoff.
A 2.5K SE (class A) spec would become a 5K P-P -class A- spec.
I hope we can all agree on that one.
Apparently more controversial: the class B P-P case would then become a 10K OT for equivalent loading on each tube.
And the class AB P-P would be in between 10K and 5K due to conduction overlap and tube gm variation.
------------------------------
[And in the case of the 5K SE 6AQ5, something like 16K OT for class AB.]
[But then modified further to remain within the tube Vmax, Imax, and Watts_max specs, including the screen grid rating, and probably also considering OT winding practicality. They end up with 10K for the 6AQ5 P-P class AB. They might also be heavily overlapping the tube conduction for the class AB case too (ie. bias setting running high idle current).]
I hope we can all agree on that one.
Apparently more controversial: the class B P-P case would then become a 10K OT for equivalent loading on each tube.
And the class AB P-P would be in between 10K and 5K due to conduction overlap and tube gm variation.
------------------------------
[And in the case of the 5K SE 6AQ5, something like 16K OT for class AB.]
[But then modified further to remain within the tube Vmax, Imax, and Watts_max specs, including the screen grid rating, and probably also considering OT winding practicality. They end up with 10K for the 6AQ5 P-P class AB. They might also be heavily overlapping the tube conduction for the class AB case too (ie. bias setting running high idle current).]
Last edited:
Unfortunately the best Z is not just one fixed for all voltages. Notice the differences in plate resistance for various plate voltages and screen voltage combinations. That will change the best load Z, and can be different for fixed bias or cathode bised too. I just use the data sheet as a guide.
Excellent point.
Screen V and B+ will alter the best OT Zprimary for pentodes.
The SE case given might be the best case for low distortion (and still reasonable power output) with all optimizations in effect.
P-P gets rid of even harmonic distortions (2nd harmonic being the largest), so the total design optimization can turn out differently (usually try minimizing the 3rd harmonic).
Then varying the screen V and B+ from that overall optimal 3rd H solution (by a modest amount) usually causes tolerable effects on the 3rd H distortion, if optimized correctly again.
The best OT Zpri (versus the class A SE derived figure) then is altered by the V versus I trade off made. Assuming there is sufficient headroom to do that. (use a tube with at least some extra qualifications, for design headroom)
And not exceeding 70% of tube Pdiss is usually sighted for longevity besides.
Clues:
(Beam) pentode tubes with nearly straight flat curves above the knees have the least sensitivity to OT primary Z as far as 3rd (and other odd) harmonics go. (they allow higher than normal Z primary OTs, without suffering screen grid current distortion taking off, which higher Z allows for lowering gm non-linearity, just like for triodes) Sharp low voltage knees give the best efficiency. But rounded knees give softer clipping. The 6LG6 GE datasheet gives a good example of a nice flat curve tube. (don't expect soft clipping though) (curved all over pentode curves are due to screen grid current pollution)
Screen V and B+ will alter the best OT Zprimary for pentodes.
The SE case given might be the best case for low distortion (and still reasonable power output) with all optimizations in effect.
P-P gets rid of even harmonic distortions (2nd harmonic being the largest), so the total design optimization can turn out differently (usually try minimizing the 3rd harmonic).
Then varying the screen V and B+ from that overall optimal 3rd H solution (by a modest amount) usually causes tolerable effects on the 3rd H distortion, if optimized correctly again.
The best OT Zpri (versus the class A SE derived figure) then is altered by the V versus I trade off made. Assuming there is sufficient headroom to do that. (use a tube with at least some extra qualifications, for design headroom)
And not exceeding 70% of tube Pdiss is usually sighted for longevity besides.
Clues:
(Beam) pentode tubes with nearly straight flat curves above the knees have the least sensitivity to OT primary Z as far as 3rd (and other odd) harmonics go. (they allow higher than normal Z primary OTs, without suffering screen grid current distortion taking off, which higher Z allows for lowering gm non-linearity, just like for triodes) Sharp low voltage knees give the best efficiency. But rounded knees give softer clipping. The 6LG6 GE datasheet gives a good example of a nice flat curve tube. (don't expect soft clipping though) (curved all over pentode curves are due to screen grid current pollution)
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.