Not in this thread, perhaps. But it was a hot topic a few years back. Search the Class-D forum.
That is the classic view, yes. In fact the opposite was also often proposed, that they may actually help by providing some sort of bias to the cable/crossover/voicecoil. Never proven, IIRC, but talked about.
Yes, I'm with you 100% on this. I said I was having trouble keeping a handle on it all. I came to my senses and stated the need to keep the two (objective & subjective) separate.
I agree with this. The revolution is occurring now. Maybe it isn't happening as anyone imagined it, or anyone desired it to be. That's beside the point.
No problems, except price for an exclusively built vacuum tube single ended stereo amp VS mass produced class D SMD rug. 😀
A phase shift isn't an issue providing the entire 20-20k is shifted by the same amount. With regards to phase shift however, I don't think a wide bandwidth amplifier is going to be of much concern when in a completed system, as there are other components that should dominate.
Interestingly though a wide bandwidth solid state affair should surely be better in this regard then class D amps with their low pass output filters and valves with their transformer limited bandwidths.
Depends on the filter...
Can't say what other class-D implementations are doing filter-wise, but many of the Tripath based amps incorporate a 2nd order rolloff in the vicinity of 100kHz, with a Q of 0.7 for 8 ohm loads... the phase shift remains quite small well past 20KHz
At 20KHz, the relative difference between a 100KHz lowpass and a 3.5Mhz lowpass function is so little as to not make any practical difference.
Of course, if some of the class-D amps are filtering at lower frequency, like at 30kHz, then the phase shift at 20Khz would be more of an issue. But at pretty much any lowpass filter starting beyond 75Khz, the phase shift at 20Khz should not be a point of concern.
Same applies to tubes, if the output transformer's frequency response extends beyond 75KHz with little or no peaking, it is likely that the phase shift would pose no problem there either.
Not necessarily. It'll depend on the low-pass filter setting.
Can't say what other class-D implementations are doing filter-wise, but many of the Tripath based amps incorporate a 2nd order rolloff in the vicinity of 100kHz, with a Q of 0.7 for 8 ohm loads... the phase shift remains quite small well past 20KHz
At 20KHz, the relative difference between a 100KHz lowpass and a 3.5Mhz lowpass function is so little as to not make any practical difference.
Of course, if some of the class-D amps are filtering at lower frequency, like at 30kHz, then the phase shift at 20Khz would be more of an issue. But at pretty much any lowpass filter starting beyond 75Khz, the phase shift at 20Khz should not be a point of concern.
Same applies to tubes, if the output transformer's frequency response extends beyond 75KHz with little or no peaking, it is likely that the phase shift would pose no problem there either.
Last edited:
Uniform group-delay
To keep things clear for those who may otherwise be confused by the terms, constant phase shift would create more time delay at low frequencies, less time delay at higher.
For example, a 2nd order Bessel low-pass filter exhibits increasing phase-shift with frequency such that it has uniform group delay, i.e. the time-lag from the filter is equal at 20Hz and 2KHz and 20KHz, and across the frequency range.
A 2nd order lowpass with Q of 0.7 will still have nearly uniform group delay below the cutoff frequency.
A 2nd order lowpass with a Q of 3 will not exhibit uniform group delay, i.e. some frequencies will arrive sooner, some later.
I get your point, but this may be clearer if it is stated as "shifted by the same amount of time, i.e. with uniform group delay".
To keep things clear for those who may otherwise be confused by the terms, constant phase shift would create more time delay at low frequencies, less time delay at higher.
For example, a 2nd order Bessel low-pass filter exhibits increasing phase-shift with frequency such that it has uniform group delay, i.e. the time-lag from the filter is equal at 20Hz and 2KHz and 20KHz, and across the frequency range.
A 2nd order lowpass with Q of 0.7 will still have nearly uniform group delay below the cutoff frequency.
A 2nd order lowpass with a Q of 3 will not exhibit uniform group delay, i.e. some frequencies will arrive sooner, some later.
Last edited:
i was talking about a class a 10-20w tube amp for driving four 17w tweeters 91db , nothing too expensive 400$ max 😱
In such case you need a time machine, to go back when tube amps were mass-manufactured. 😀
Today even Chinese SE amp with 20WPC can't cost $400.
They claim 12W per channel in ultralinear mode. It is no more than an ordinary amp that satisfies tube fashion, far from High-end. Most probably output transformers are too small, have cheap design. Even if small inductance is insignificant for tweeter amp, they may have too big leakage inductance and stray capacitance that affects reproduction of highs. EL34 is not the best choice for SE amp, for 20W the best is GM-70, but it is hard to cook properly.
I think it is better to find some Chinese 100W per channel push-pull amp, and re-use it's chassis (and probably power transformer?) for a 15-20 WPC SE amp. But minimally usable 20W SE output transformers will cost additional $240 plus shipment.
You may ask on Tube forum which Chinese amp may be used as a donor, I know people do that, and some of them are members of the Forum.
The prototype I am working on currently uses paralleled GU-50 tubes for output arranged as right-handed triodes and 5K primary 25W output transformers, with 600V stabilized anode voltage. Design target is 15WPC. The aim is to make a decent 15WPC High-end amp, less monstrous than with GM-70 tubes.
then you have the problem of the 'dynamic sandwich' or cluster-fudge of the output filter being dynamically shifted and loaded by the driver and the amp itself (as a pair). With any kind of real world high level signal, you find your instantaneous peak values of complex impedance IN the filter proper..this extreme level ends up modulating the filter in an untoward way. As in well outside of it's linear limits in the peak values. Thus the amps can sound very bad if the output filter is not over-specified, properly laid out, and tested to be dynamically micro linear at a 1000 different loading levels and polarities of load. Which an actual music signal will deliver to the filter at least a few thousand times a second.
Hi,
You are no doubt aware of Earl Geddes work on distortion audibility? And therefore you know the limits of distortion audibility?
And no doubt you are familiar with the work of Oohashi (et al.) on sound perception beyond 20KHz.
If not I suggest your knowlege is incomplete and needs upgrading.
If you knew the actual state of research on the subject you would be more careful with statements like the above.
Just because some CAN BE DONE, in principle, does not mean it actually IS done. Nor does it mean when something, anything, is done in a serious, scientific way or that the results hold any statistial power (meaning they can be generalised).
The problem is such dogmatic that are not rooted in actual scientific evidence views are present on both sides, you clearly illustrate one of these two positions, which holds "everything sounds the same, if certain minimum levels of performance in a certain minimal set of measurements are met." and this view is perpetuated in direct contradiction to established scientific facts that provide proof of the audibility of items outside this set of measurements and limits.
You have already illustrated the other side.
Has ever occurred to you that the truth is somewhere in the middle, and by it's entirely real and entirely dogmatic nor amenable to being dogmatised?
Ciao T
PS, my wife wanted a Dogma, one of these little Bug-eyed Chiwawa's, I said "NO"....
You are no doubt aware of Earl Geddes work on distortion audibility? And therefore you know the limits of distortion audibility?
And no doubt you are familiar with the work of Oohashi (et al.) on sound perception beyond 20KHz.
If not I suggest your knowlege is incomplete and needs upgrading.
If you knew the actual state of research on the subject you would be more careful with statements like the above.
Just because some CAN BE DONE, in principle, does not mean it actually IS done. Nor does it mean when something, anything, is done in a serious, scientific way or that the results hold any statistial power (meaning they can be generalised).
The problem is such dogmatic that are not rooted in actual scientific evidence views are present on both sides, you clearly illustrate one of these two positions, which holds "everything sounds the same, if certain minimum levels of performance in a certain minimal set of measurements are met." and this view is perpetuated in direct contradiction to established scientific facts that provide proof of the audibility of items outside this set of measurements and limits.
You have already illustrated the other side.
Has ever occurred to you that the truth is somewhere in the middle, and by it's entirely real and entirely dogmatic nor amenable to being dogmatised?
Ciao T
PS, my wife wanted a Dogma, one of these little Bug-eyed Chiwawa's, I said "NO"....
- Status
- Not open for further replies.