Skip to main content

Topic: Time delay in audio amplifiers (Read 847 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • Nichttaub
  • [*][*]
Time delay in audio amplifiers
I just saw a known audiofool at another forum trying to espouse the idea that audio amplifiers can have extreme amounts of propagation delay.  He insisted that some of the higher end amps had delays on the order of 500 µS.  Further, he asserted that this needed to be taken into account when setting up bi-amplified speakers using different amplifiers.

This seems completely untenable; wouldn't an amplifier with inverting feedback and a delay of that magnitude turn into a high power oscillator at 2 KHz?

I did some research (both print literature and online) and it seems a much more reasonable value is on the order of a few nanoseconds, which of course would be completely undetectable to the ear and of no consequence to any setup or positioning of the equipment.

Do we have a new winner in the audiofool lottery?

  • saratoga
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #1
Quote
He insisted that some of the higher end amps had delays on the order of 500 µS. 

Plausible, especially if there is any EQ or other filtering.

Quote
Further, he asserted that this needed to be taken into account when setting up bi-amplified speakers using different amplifiers.

Take into account how? It is just a time delay, generally you don't care if it takes a few microseconds longer after you hit start for the sound to come out.

  • Nichttaub
  • [*][*]
Re: Time delay in audio amplifiers
Reply #2
Quote
He insisted that some of the higher end amps had delays on the order of 500 µS. 

Plausible, especially if there is any EQ or other filtering.

Not in a circuit with a feedback loop around it, though; he was talking about a bog-standard power amplifier, not an AV system with extra processing modules or even a DSP preprocessor section.  His contention was that the linear amplifier had an actual hundreds-of-µS propagation delay from input to output, and in a basic feedback circuit amp that would not be possible.

Quote
Quote
Further, he asserted that this needed to be taken into account when setting up bi-amplified speakers using different amplifiers.

Take into account how? It is just a time delay, generally you don't care if it takes a few microseconds longer after you hit start for the sound to come out.



His thought process seemed to be that it somehow affected positioning of speakers, choice of amplifiers, etc.  That just seems bizarre.

  • DVDdoug
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #3
Quote
His thought process seemed to be that it somehow affected positioning of speakers, choice of amplifiers, etc.  That just seems bizarre.
Yes, bizarre.  A difference  between left & right could be an issue but 1/2 of a millisecond is unlikely to be heard.  It's about the time it takes the sound to move from your left eat to your right ear or being off-center by a few inches.

Or, maybe this audiophool should calculate how long it takes for the sound to get to his ears after coming-out of the speakers.

Quote
His thought process seemed to be that it somehow affected positioning of speakers, choice of amplifiers, etc.  That just seems bizarre.
Yeah...  Computers usually have a few milliseconds of latency (or more).    I've got Beatles CDs with more than 50 years of "latency".  :D :D :D
  • Last Edit: 26 September, 2017, 05:04:35 PM by DVDdoug

  • saratoga
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #4
Quote
He insisted that some of the higher end amps had delays on the order of 500 µS. 

Plausible, especially if there is any EQ or other filtering.

Not in a circuit with a feedback loop around it, though; he was talking about a bog-standard power amplifier, not an AV system with extra processing modules or even a DSP preprocessor section.  His contention was that the linear amplifier had an actual hundreds-of-µS propagation delay from input to output, and in a basic feedback circuit amp that would not be possible.

You don't need DSP or anything like that.  Even a 40 degree phase delay at 1 KHz is over 100us.  The filtering built into the amp could conceivably add quite a lot of delay.  Not saying it will for any given amp, but having half a wavelength of delay at 1 KHz isn't unreasonable. 

Quote
Quote
Further, he asserted that this needed to be taken into account when setting up bi-amplified speakers using different amplifiers.

Take into account how? It is just a time delay, generally you don't care if it takes a few microseconds longer after you hit start for the sound to come out.



His thought process seemed to be that it somehow affected positioning of speakers, choice of amplifiers, etc.  That just seems bizarre.

It sounds like he is just confused about what delays actually mean. 

  • exponent
  • [*]
Re: Time delay in audio amplifiers
Reply #5
I dont know what he has been smoking but there is no delay that is significant. I did a real world measurement with an oscilloscope. I applied channel 1 on input and channel 2 on output. I had to drop the timebase way way down and saw nothing. Now that I think of that makes sense as most amplifiers use feedback and it would be useless if there was significant delay.There could be phase shifts but those would be a function of speaker load.

  • Nichttaub
  • [*][*]
Re: Time delay in audio amplifiers
Reply #6
I dont know what he has been smoking but there is no delay that is significant. I did a real world measurement with an oscilloscope. I applied channel 1 on input and channel 2 on output. I had to drop the timebase way way down and saw nothing. Now that I think of that makes sense as most amplifiers use feedback and it would be useless if there was significant delay.There could be phase shifts but those would be a function of speaker load.

Exactly my point - a delay of (as he asserted) 500 µS would be 1/2 wavelength at 1 KHz (not 2 as I mistakenly said above) and at that frequency would provide positive feedback instead of negative at the amplifier inverting input.  That would turn the amplifier immediately into an oscillator assuming its gain was >1 at that frequency.  Typical amps have a phase shift of less than a few degrees over the audio band - again, inside the stage which contains the feedback loop.  In a basic power amplifier like the ones he was considering, that is the only stage.  Any filtering present inside the stage would have to be dominant pole compensation.  Other filters (HPF/LPF, Zobel, etc.) either before or after the feedback stage would not affect the stability and would add phase shift.  But again, that wasn't the type of amp being discussed; it was a basic power amp, one stage, with no controls.

Re: Time delay in audio amplifiers
Reply #7
I just saw a known audiofool at another forum trying to espouse the idea that audio amplifiers can have extreme amounts of propagation delay.  He insisted that some of the higher end amps had delays on the order of 500 µS. 

Could be. Ordinary bandpass filters have measurable delay.   DACs have latency which is delay by a different name.

Quote
Further, he asserted that this needed to be taken into account when setting up bi-amplified speakers using different amplifiers.

Could be. That's why careful workers measure the FR of their complex systems. Or play pink noise and listen for peaks holes. Preferably the former. Of course we're talking golden ears and measurements are anathema.

Quote
This seems completely untenable; wouldn't an amplifier with inverting feedback and a delay of that magnitude turn into a high power oscillator at 2 KHz?


Only true if the delay is inside the feedback loop. Most amps have input and/or  output filters  that are outside the loop.

Particularly switchmode amps.


  • pdq
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #8
If we are talking about strictly analog circuitry, doesn't any time delay also imply a non-flat frequency response?

Re: Time delay in audio amplifiers
Reply #9
If we are talking about strictly analog circuitry, doesn't any time delay also imply a non-flat frequency response?

No. The filter that changes phase without changing gain is called an "All-Pass" filter.

You may be thinking about "Minimum Phase" filters. Their phase shift and gain are locked together by the Hilbert Transform. If you know one, you can calculate the other one from it using either the forward form of the transform, or its inverse.

  • saratoga
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #10
If we are talking about strictly analog circuitry, doesn't any time delay also imply a non-flat frequency response?

No.  Wires for example have time delay but hopefully flat frequency.  Group delay dispersion (frequency dependent time delay) can imply a non-flat frequency response, but for what its worth, amplifiers do have that as well since they are hopefully filtering out ultrasound and/or DC. 
  • Last Edit: 27 September, 2017, 03:01:07 PM by saratoga

Re: Time delay in audio amplifiers
Reply #11
I think that he mistook the power up delay with the signal delay, although this delay is important to take in consideration for the power up sequence. And a power amplifier can easily have a power up delay of 0.5 seconds or more.

  • pdq
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #12
So if an "All-Pass" filter produces a negative phase shift at some frequencies then this is considered to be a delay. What if it produces a positive phase shift?

  • pdq
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #13
Wires for example have time delay but hopefully flat frequency. 
Unless the OP has 150 km of transmission line in his amplifier, I don't think this is it.

  • saratoga
  • [*][*][*][*][*]
Re: Time delay in audio amplifiers
Reply #14
So if an "All-Pass" filter produces a negative phase shift at some frequencies then this is considered to be a delay. What if it produces a positive phase shift?

Assuming a negative phase shift means a time delay, a positive phase would means that a signal output preceded input (anticausal), which is not possible in a real world filter because information flows from past to future (causality).

You can simulate filters like that though in digital systems. They aren't constrained by the Kramers Koenig relationship, which in causal systems relates changes in magnitude to changes in phase.