Low output on JTM45 build

[Helmholtz]

It might be instructive to see scope pics of the grid signals at clipping (please specify tube type and bias voltage measured at the junction between the 220k bias feed resistors).
To see only power amp distortion, it's best to feed the input signal to the PI before the 0.02µ coupling cap, i.e. at the treble pot wiper.

[g1]

From what I can tell, post #32 indicates you did not try a mismatch into a lower load impedance?

I had set my load always at 8R and would change (reflected) output impedance with just flicking the amp's impedance selector between 8R and 4R.

[pdf64]

Not sure what you're referring to, as no one mentioned THD.

Fair point, I suppose that following on from your point about hotter idle increasing output power, I was thinking that for our application, how to assess output power (in a way that correspondings with the end result of moving speaker cones) is not a cut and dried thing. As the hotter idle seems to improve linearity and so reduce THD, and so increase the Vrms @ eg 5%THD. Whereas Vp, and how loud the amp can get in use, is unaffected.

Vp can be used for power calculation, but only if the signal is a sine. If not, power results will be wrong.
E.g. with increasing clipping Vp stays constant while real Vrms and power increase.

That's exactly my point
Using 0.71 of Vp (rather than Vrms @ xTHD) pretty much eliminates the effect of distortion / linearity, which to my thinking, makes it more useful to us, as an amp setting that results in a clean perceived tone may have a fair degree of THD from the amp. And then some more THD from the speakers.

There's no general relation between Vp and power.

My thinking is that Vp equates to the peak valve current minus transformer loss, and is what will cause the max cone excursion in the speakers. A problem with Vrms is that the waveform at the onset of clipping may be rather non sinusoidal, and with respect to 0.71 of Vp, Vrms might higher or lower.
So 0.71 of their Vp seems to give a better comparison between amp types, and helps to eliminate the variables of NFB and linearity.

BTW, what power output did you measure with your JTM45 build?
I plan to take some measurements on my JTM50 in a week or so.

The JTM45, with its high degree of NFB and highish OT primary impedance, gives a pretty good sine wave up to the onset of clipping, so the above issues don't apply
With the RS DeLuxe OT set to 6k6, it did about 33W with either 16 or 8 ohm secondary arrangements.
With the Ingo 8K OT, it managed about 30W.
The HT at OT CT during the above tests is about 420V.
FYI in either case, halving the load impedance (eg 8ohm load with OT secondary set up for 16ohms) reduced the power a bit in all cases.
Sorry for the delay.

My JTM50 type power amp (EL34, 50uF reservoir and g2 node caps,784-139 OT) seems somewhat more punchy. As I recall, it does about 40W with the GZ34, 50W with silicon rectification. Unless it's attenuated, I can't push it into overdrive with the band without drowning out everything else. Whereas the JTM45 sits nicely (if somewhat prominently) in the mix.

[Helmholtz]

With the RS DeLuxe OT set to 6k6, it did about 33W with either 16 or 8 ohm secondary arrangements.
With the Ingo 8K OT, it managed about 30W.
The HT at OT CT during the above tests is about 420V.
FYI in either case, halving the load impedance (eg 8ohm load with OT secondary set up for 16ohms) reduced the power a bit in all cases.
Sorry for the delay.

Thanks for your data/information.
What type KT66s did you use?

Whereas Vp, and how loud the amp can get in use, is unaffected.

I tend to disagree with your thoughts regarding the influence of Vp on speaker loudness.
A speaker is a loaded motor and converts electrical power into acoustic power with a constant efficiency (for a given frequency).
So more input power increases acoustic power and loudness.

There's a reliable relation between Vrms and power, not so for Vp.

As an experiment compare the loudness of sine signal to a square wave signal having the same Vp.
Square wave will be much louder.
This is at least partly due to the doubled power, i.e. the additional power contained in the harmonics.

Likewise a clipped signal is louder than a pure sine having the same Vp.

[pdf64]

Thanks for your data/information.
What type KT66s did you use?

Groove Tube, that I bought in the 90s, and have used a lot over the years, they're my favourite pair of beam type pentodes. I think they're Russian made, but I measured their heater current at 1.38A, so are a different valve than the modern Tung Sol, which only draw 0.9A, so may be more akin to 6L6.
With a TubeTown branded pair of KT66 that I got about a decade ago and have hardly used, the amp is less punchy, power output drops to about 28W. I think they're Chinese. They too draw 1.38A heater current.

I tend to disagree with your thoughts regarding the influence of Vp on speaker loudness.
A speaker is a loaded motor and converts electrical power into acoustic power with a constant efficiency (for a given frequency).

The issue may be that non linearty / distortion adds frequencies, but the phase relationship of the harmonics to the fundemental may result in the total Vrms being higher or lower than 0.71 of Vp of the fundemental.

Likewise a clipped signal is louder than a pure sine having the same Vp.

Absolutely, but if the amp can put out a ragged looking sinewave of 10Vp at the onset of clipping, we know that somewhere in there, there's a 7.1V sinewave. Whereas Vrms may be higher or lower than 7.1.
Below clipping, for a given Vp, depending on the amp design, non linearity can add or subtract to the Vrms, whereas Vp is a fixed metric.
eg with my AC30, the high power waveform fattens and rounds off the sinewave, such that it's difficult really to identify exactly the point at which clipping occurs. That tends to result in Vrms being higher than might be expected (ie >0.71 of Vp), as there's more 'area under the curve'. If NFB was applied, I suspect that Vrms @ xTHD would decrease whilst Vp would stay the same, once gain was normalised.
Whereas a cold biased amp with a low OT primary impedance and low degree of NFB might approximate a triangle waveform at high power output. That tends to result in Vrms being lower than might be expected (ie <0.71 of Vp), as there's less 'area under the curve'. If NFB was applied, the Vrms @ xTHD would increase whilst Vp would stay the same, once gain was normalised.

[Helmholtz]

Absolutely, but if the amp can put out a ragged looking sinewave of 10Vp at the onset of clipping, we know that somewhere in there, there's a 7.1V sinewave. Whereas Vrms may be higher or lower than 7.1.
Below clipping, for a given Vp, depending on the amp design, non linearity can add or subtract to the Vrms, whereas Vp is a fixed metric.
eg with my AC30, the high power waveform fattens and rounds off the sinewave, such that it's difficult really to identify exactly the point at which clipping occurs. That tends to result in Vrms being higher than might be expected (ie >0.71 of Vp), as there's more 'area under the curve'. If NFB was applied, I suspect that Vrms @ xTHD would decrease whilst Vp would stay the same, once gain was normalised.
Whereas a cold biased amp with a low OT primary impedance and low degree of NFB might approximate a triangle waveform at high power output. That tends to result in Vrms being lower than might be expected (ie <0.71 of Vp), as there's less 'area under the curve'. If NFB was applied, the Vrms @ xTHD would increase whilst Vp would stay the same, once gain was normalised.

My point was that higher Vrms typically increases speaker loudness, while higher Vp might not.

As Vp doesn't generally determine power nor loudness, I see no sense in focussing on it.
Calculating power from Vp is only permissible for sines, otherwise the result is not interpretable.
So it doesn't solve any of the problems you mentioned.

[pdf64]

My thinking is that 0.71 of their Vp helps to put a variety of amps we use, whose high power pre-clipping waveforms can range from linear to raggedly, on a more equal footing.
But I can see I'm not making progress with convincing you of that

[Helmholtz]

My thinking is that 0.71 of their Vp helps to put a variety of amps we use, whose high power pre-clipping waveforms can range from linear to raggedly, on a more equal footing.

Why not discuss a practical example, e.g. your AC30.

The AC30 exhibits considerable compression before eventually the signal tops flatten and show the typical ripple modulation.
What's the difference between the (real) power calculated from Vrms and the virtual "power" calculated from Vp at that point?

The latter should be lower, but what power does the amp sound like? (O.k., that's a rhetorical question ).

[tubby]

From what I can tell, post #32 indicates you did not try a mismatch into a lower load impedance?

I actually did, please see post #34. Post #32 was a typo

[tubby]

It might be instructive to see scope pics of the grid signals at clipping (please specify tube type and bias voltage measured at the junction between the 220k bias feed resistors).
To see only power amp distortion, it's best to feed the input signal to the PI before the 0.02µ coupling cap, i.e. at the treble pot wiper.

I did those measurements as instructed with a pair of matched Shuguang KT66. Parameters: Bias voltage between 220k bas feed resistor = -53V; 8R output into 8R load
Here are the screen grabs:
Signal at the PI input; 3,58 Vrms at onset of clipping at speaker output
Signal at power tube 1 control grid at pin of the tube socket; 17,45 Vrms at onset of clipping at speaker output
Signal at power tube 2 control grid at pin of the tube socket; 19,55 Vrms at onset of clipping at speaker output
Signal at speaker output jack; 13,22 Vrms at onset of clipping

[tubby]

Groove Tube, that I bought in the 90s, and have used a lot over the years, they're my favourite pair of beam type pentodes. I think they're Russian made, but I measured their heater current at 1.38A, so are a different valve than the modern Tung Sol, which only draw 0.9A, so may be more akin to 6L6.

Did the GT labeled KT66 maybe look something like this?


In that case they'd be of chinese origin (Shuguang, same as the Tube Town label).

[Helmholtz]

There's something wrong here.

At the onset of output clipping the grid voltages should be around 100Vpp.
Is your scope input in calibrated position?

Also I don't see any signs of clipping at the output.

[tubby]

There's something wrong here.

At the onset of output clipping the grid voltages should be around 100Vpp.
Is your scope input in calibrated position?

Also I don't see any signs of clipping at the output.

Ok, I think I misunderstood post #46. Just to make sure we're on the same page: With "grid signals at clipping" (post #46) you mean a fully clipped / square wave output waveform from the power amp?

[Helmholtz]

Ok, I think I misunderstood post #46. Just to make sure we're on the same page: With "grid signals at clipping" (post #46) you mean a fully clipped / square wave output waveform from the power amp?

With onset of clipping I mean just a tiny portion of flattened tops. Typically goes along with some ripple modulation at tops.
So we know we're just above max. clean output.

[Helmholtz]

FYI in either case, halving the load impedance (eg 8ohm load with OT secondary set up for 16ohms) reduced the power a bit in all cases.

Interesting.

Seems the counteracting effects ( increased saturation voltage, doubled OT losses and maybe lower plate and screen voltages) win.