PSE valve amplifiers are pretty rare, and I’ve read a lot of varying opinions as to why that is. Keeping the focus narrow, I’m wondering what the sonic difference would be between two output stages of equivalent gain, one of which using two smaller DHT tubes and the other using one larger tube? All other factors being equalized. If the idea of a single-ended design is to preserve even-order harmonics (vs. PP), what effect does using two tubes in parallel have on those even-order harmonics?
I did do a search on the topic beforehand. All the scope images I can find of multi-tube output are push-pull designs. Presumably, this isn’t done due to “smearing” between the two separate amplified signals, when recombined. Still, 2A3’s are a lot cheaper than 300’s!
I did do a search on the topic beforehand. All the scope images I can find of multi-tube output are push-pull designs. Presumably, this isn’t done due to “smearing” between the two separate amplified signals, when recombined. Still, 2A3’s are a lot cheaper than 300’s!
You cannot tell exactly a-priori, because it will depend on the characteristics of the two tubes you parallel, even if they are the same kind od tube.
They will become equivalent to a single tube with almost twice the input capacitance, twice the gm, half the rp. But that “almost” will depend on the characteristics of the two tubes you parallel.
Can I interject something, in the interest of clarity?
They do not preserve anything, they generate even-order harmonics. Signal components that are not in the original music.
Back to you.
Jan
Making well the PSE isn't easy job.
I use -now /still/ breadboarded- 801a PSE in my study, but construction was the challenge.
To find 2x2 -almost- identical (same parameters at given operating point, AND similar characteristic at even large input swing) is the nightmare.
From few dozens pair the tracking even possible.
BTW manufacturer selected pairs to guarantee only same parameters in given operating point, nothing else.
Using its as SE (tubes in different channel) is easy... mainly in desired op. point.
Using its as parallel tubes not trivial, especially then if you differs from manufacturer preferred op. point.
The harmonics does not deviate from SE.
801a PSE, 1W.
Red is the output, green is the driving (grids).
I use -now /still/ breadboarded- 801a PSE in my study, but construction was the challenge.
To find 2x2 -almost- identical (same parameters at given operating point, AND similar characteristic at even large input swing) is the nightmare.
From few dozens pair the tracking even possible.
BTW manufacturer selected pairs to guarantee only same parameters in given operating point, nothing else.
Using its as SE (tubes in different channel) is easy... mainly in desired op. point.
Using its as parallel tubes not trivial, especially then if you differs from manufacturer preferred op. point.
The harmonics does not deviate from SE.
801a PSE, 1W.
Red is the output, green is the driving (grids).
Last edited:
I missed the “preserve” point that @jan.didden highlighted, and indeed it is something often people on forums misunderstand:
the statement “push-pull cancels even harmonics” it doesn’t mean that reduces the even harmonics of the incoming signal, but that cancels most of the even part of the harmonics that are generated in the power amp.
Single ended amps (with one, two or more tubes in parallel) simply add even and odd harmonics to the input signal.
the statement “push-pull cancels even harmonics” it doesn’t mean that reduces the even harmonics of the incoming signal, but that cancels most of the even part of the harmonics that are generated in the power amp.
Single ended amps (with one, two or more tubes in parallel) simply add even and odd harmonics to the input signal.
Even order harmonics are the result of difference in amplitudes between the positive and negative parts of the sine wave. A single ended triode device will produce that to varying degrees. You just need to look at the load line on the characteristics curve graph.
Push-pull amps balance out the positive/negative amplitude difference because one tube's phase is inverted from the other so when superimposed, the amplitudes of the phases will be equal (if tubes are well matched). Of course, this will also depend on the accuracy of the phase inverter (note: not phase splitter), too, driving the output tubes.
Push-pull amps balance out the positive/negative amplitude difference because one tube's phase is inverted from the other so when superimposed, the amplitudes of the phases will be equal (if tubes are well matched). Of course, this will also depend on the accuracy of the phase inverter (note: not phase splitter), too, driving the output tubes.
The key is knowing which parameters are matched by the seller. "Matched" may mean different things to different vendors.
@jan.didden they are even harmonics. Odd harmonics keep the signal simmetric, even ones guve asimmetry to the signal (think about the different distribution of curves on the left va right of the working point).
In my book that is adding distortion.
The only way is the selection by tester as Sofia or uTracer or ROe tester
This for PSE and p-p
But in the same way for signal tubes if we want the THD or FFT same for both channels
No other way.
With Sofia I have the possibility to get the matching in auto mode with the closest tube as specs
But the number of stuff is, at the best, almost the triple to get a decent coupling.
Last, if I get the NOS/NIB tubes probably the job will be less heavy
PS: the time and money to build a good PSE is more than a very good p-p ( that I prefer I lot) mainly for the OT; and the selection of the tubes are the same. . For this reason I left PSE years ago.
PPSS: one of the best PSE is KIT III Audio Note, a old one
Walter
It's not a matter of adding distortion to the amp, the amp is adding distortion to the signal. I think every amp adds some distortion.
Actually though, some people do purposely add distortion to the signal. Check out the DIY H2 Harmonic Generator by Nelson Pass. The kit is available at the diyAudio Store.
The first term of the even order Fourier Series is Zero Frequency, the DC current in the SE tube.
And sure enough as predicted by the loadline the DC plate current changes as the signal is increased.
Why anyone would want to add to what the musician & mixer has crafted for us is a mystery in itself.🙄
And sure enough as predicted by the loadline the DC plate current changes as the signal is increased.
Why anyone would want to add to what the musician & mixer has crafted for us is a mystery in itself.🙄
This doesn't really happen. The sum of the output valves' currents is no different than the sum of push-pull output valves'. Transit time of similar valves is so equal as to be hard to measure. Electron valves work just fine at over 100 MHz, so .02MHz isn't much challenge. There's no smearing.
All good fortune,
Chris