Suppose you fed a step waveform into the the circuit. If there is only phae delay but no time delay the output will respond immediately, albeit an exponential-shaped rise. If there is a time-delay the output will not respond immediately.
It is misleading to think of a filter as a form of time delay, It isn't. It sort of looks like it is in the special case of a continuous sinewave when you look at it on a scope. But it isn't a delay.
Real time delays cause problems, especially in control systems. Imagine driving a car if the steering column had a 5 second delay. You'd soon be in a ditch. Delays tend to make feedback loops unstable and lead to oscillation. There's a special set of Z functions for designing control loops for systems involving time delays.
I imagine the 1.5us delay quoted earlier is not entirely a time delay. I'm not sure what the time delay might be for a power transistor operating in its linear region...pretty darned small though.
It is misleading to think of a filter as a form of time delay, It isn't. It sort of looks like it is in the special case of a continuous sinewave when you look at it on a scope. But it isn't a delay.
Real time delays cause problems, especially in control systems. Imagine driving a car if the steering column had a 5 second delay. You'd soon be in a ditch. Delays tend to make feedback loops unstable and lead to oscillation. There's a special set of Z functions for designing control loops for systems involving time delays.
I imagine the 1.5us delay quoted earlier is not entirely a time delay. I'm not sure what the time delay might be for a power transistor operating in its linear region...pretty darned small though.
No
Yes it is !!! Never hear of group-delay (which can be quite frequency-independant for critical- and Bessel- lowpass-filters) ?
Agreed, but phase-delay causes problems TOO !!!! Otherwise it would be fairly easy to use NFB of whatever amount around any circuit topology one can imagine. This way we would end up with the perfect "wire with gain" at almost no cost. Which is unfortunately not the case, at least to my knowledge.
Regards
Charles
hi
Yes
How can I simulate it for instance with a cad
software?
I mean, does a module exist to introduce time
delay (with no shape change) in every component?
also in a cable? Where to put it?
How I have to think at the info propagation in a 3 leg
component?
I think (IMHO) this is very important to simulate
true feedback behavior
Bye
Federico
Yes
How can I simulate it for instance with a cad
software?
I mean, does a module exist to introduce time
delay (with no shape change) in every component?
also in a cable? Where to put it?
How I have to think at the info propagation in a 3 leg
component?
I think (IMHO) this is very important to simulate
true feedback behavior
Bye
Federico
To all interested
I noted that with a transfer like the above I
obtain spectra with only the fundamental and the third harmonic, irrespective of the level.
If I use this:
vo = a1*vi + a2*vi2
I obtain only the second harm.
But if I use any other exponent I have all the harmonics,
even or odd, below that exponent.
If I try, for example
vo = a1*vi + a5*vi5
I obtain 1th,3rd and 5th harmonic. If, at that level, I add
(or subtract) the right amount of 3rd order in this way:
vo = a1*vi + a3*vi3 + a5*vi5
I succeed to cancel the 3rd harm, but only at that particular level. Changing it the 3rd order come back.
So 2nd and 3rd components behave differently from the others.
bye
federico
So, there is time delay and phase shift. Is this makes contribution to distortion in sound? I imagine the feedback is coming too late than the input signal, so there must be some confusion in the differential.
Nelson Pass said that if the transistor gain stage is too long, there is some kind of "Hall of Mirror" effect. I assume this is due to time delay too. This effect makes the differential not effectively working to make correction signal for the next stage, because what had happened in the output stages comes too late to the differential, it makes the differential makes a wrong error signal. That is why he likes a 2 stages amp, maybe because the control is much better than many stages power amp.
I dont know about the fact. Is this delay good or bad to sound? There are 9 stages power amp out there in the market.
Nelson Pass said that if the transistor gain stage is too long, there is some kind of "Hall of Mirror" effect. I assume this is due to time delay too. This effect makes the differential not effectively working to make correction signal for the next stage, because what had happened in the output stages comes too late to the differential, it makes the differential makes a wrong error signal. That is why he likes a 2 stages amp, maybe because the control is much better than many stages power amp.
I dont know about the fact. Is this delay good or bad to sound? There are 9 stages power amp out there in the market.
lumanauw said:
Yes, time delay and phase shift. I believe a typical bandwidth measurement showing gain magnitude and phase, has the delay time effects added to the RLC caused phase shift. (Delay time)/(sine period)*360 = phase shift due to delays, for sinewave. If the delay times are constant, then they produce more phase shift as frequency increases, I think.
To me this effect is like the tail wagging the dog instead of the dog wagging the tail. I think heavy global feedback with phase shift (delays + RLC phase shifts) causes more TIM, IM, and THD. I think an amp that does the job with the least active devices with adequate individual bandwidths, will make a better amp. I think faster parts can have less delays. I think delays come from different sources. Delay related to the speed of current flow and delay related to digital gate propagation being different. Part of propagation delay is related to slew rates in digital circuits, I think. There are probably several physical effects that contribute to an overall delay in an active device with others in addition for multiple stages and the PCB. The physics of each active device type should dictate the additive delays in the individual part. I think we see the net effects as phase shift, usually.
I'm sure I've got this all wrong, and it will be presented correctly by the experts. I'm now ready for my intellectual whipping.
Re: There's Life Above 20 Kilohertz!
I just found the same survey
Comments on James Boyk's Course
"Projects in Music & Science" (EE/Mu 107)
California Institute of Technology
Curriculum Vitae — James Boyk
Current:
* Concert pianist and internationally-known recording artist
* Artist teacher of piano; coach of interpretation for all instruments
* Consultant on high-resolution audio (recording and playback)
* Consultant on choice and care of fine pianos
* Writer
Links:
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
http://www.its.caltech.edu/~musiclab/
http://www.cco.caltech.edu/~boyk/
http://www.performancerecordings.com/
dimitri said:
I just found the same survey
Comments on James Boyk's Course
"Projects in Music & Science" (EE/Mu 107)
California Institute of Technology
Curriculum Vitae — James Boyk
Current:
* Concert pianist and internationally-known recording artist
* Artist teacher of piano; coach of interpretation for all instruments
* Consultant on high-resolution audio (recording and playback)
* Consultant on choice and care of fine pianos
* Writer
Links:
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
http://www.its.caltech.edu/~musiclab/
http://www.cco.caltech.edu/~boyk/
http://www.performancerecordings.com/
Dear Lineup , maybe at a live show there is life above 20khz
but in the living room it should be considered undesirable.
(We can't hear it, why should we waste amp power..)
(0 - 50k log)
This FFT of vivadi-four seasons doesn't show much content
>20Khz. I took this off the output of a self amp at near full
output with a 47pf VAS miller cap.
I've found that the miller cap value is very dependant on the
use of the amp. For a sub amp a large value seems to give
more tight bass and less high harmonics at clipping but for
full range a small 22-47pf value gives "crispness" to
various instruments.
The loudspeaker that is being driven is often ignored for
what it does to even and odd order distortions. I've
seen the amp above show different characteristics driving
different loudspeakers. Isn' t the speaker just another L-C-R
network that is a part of the amp's circuit ?
As far as the "tube" sound goes.. I'd love to build a 12ax7
preamp to hear real vacuum tube compression as opposed
to my DSP 32bit compression algorithms emulated by my pc
sound card. OS
but in the living room it should be considered undesirable.
(We can't hear it, why should we waste amp power..)
An externally hosted image should be here but it was not working when we last tested it.
(0 - 50k log)
This FFT of vivadi-four seasons doesn't show much content
>20Khz. I took this off the output of a self amp at near full
output with a 47pf VAS miller cap.
I've found that the miller cap value is very dependant on the
use of the amp. For a sub amp a large value seems to give
more tight bass and less high harmonics at clipping but for
full range a small 22-47pf value gives "crispness" to
various instruments.
The loudspeaker that is being driven is often ignored for
what it does to even and odd order distortions. I've
seen the amp above show different characteristics driving
different loudspeakers. Isn' t the speaker just another L-C-R
network that is a part of the amp's circuit ?
As far as the "tube" sound goes.. I'd love to build a 12ax7
preamp to hear real vacuum tube compression as opposed
to my DSP 32bit compression algorithms emulated by my pc
sound card. OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.