What to look for in a power amp?

[microstrip]

Sure - the trick is to make a *high powered* amplifier that's also super fast.

What it is meant by "fast" in this topic? In electronics, it is usually referred as the slew rate of the output stage - one of the fastest I know of is the Atmasphere (600V/microsecond). But audiophiles use the term to define a type of sound - usually related with "slam" and "engaging", sometimes "inducing listening fatigue", but non-related with slew-rate.

 

[JackD201]

Within the context of this topic I believe it has been established that by fast it is the slew rate and rise times. Although I have observed that the published specifications do not have the same Vout or are just not specified which would require some conversion.

 

[DonH50]

^^^ +1

Also note that many (if not most) amplifiers have higher slew rates for smaller signals than for large.

 

[garylkoh]

About the only thing that knowing the slew rate of an amplifier reliably indicates is the frequency beyond which a sinusoidal waveform will begin to be distorted. That is why amplifiers with extremely high slew rates are needed for high frequency applications - like microwave amplification.

If a 300W into 8ohm amplifier (49V peak output) has a slew rate of 600V/us, then it can reliably do about 400kHz before slew-rate limiting distorts the waveform.

Unfortunately, with music, even a single rimshot needs multiple waveforms, and IMO slew rate is not a good indicator for speed.

A single rimshot is delivered over about 240ms. If the amplifier takes time to build up from the first waveform and it is the 3rd or 4th wave before it delivers full voltage, then the amplifier sounds slow. If the amplifier delivers the first wave, and then sags after that, it sounds "soft". You have good "slam" if the amplifier delivers all the waveforms, but is clipped with square tops - and then it sounds fatigueing.

Because of limiters used in a lot of digital recording, the perfect amp (even if it never reaches its limits) will sound fatigueing with CDs. So, it is a fine balance for the amplifier designer to design an amp that users won't say is "bright and fatigueing".

 

[DonH50]

Hey Gary,

I didn't quite follow that one. What do you consider the requirement for amplifier speed? I gather full-power slew-rate is not it, but am not sure what you feel is "it". It sounds like a combination of speed and sustained power output in your example?

Curious,
Don

 

[garylkoh]

I didn't quite follow that one. What do you consider the requirement for amplifier speed? I gather full-power slew-rate is not it, but am not sure what you feel is "it". It sounds like a combination of speed and sustained power output in your example?

Sorry I wasn't clear.

Full-power slew-rate I consider "speed" in terms of frequency power bandwidth. So, beyond a certain slew rate, it is only important in the design of microwave ovens. For a 300W power amplifier, you only need about 200V/us.

My requirement for amplifier speed is a combination of speed and sustained power output. Again, beyond a certain time it doesn't matter (your loudspeakers will melt down first). So, I use the time needed to deliver 1 Coulomb of charge.

 

[DonH50]

Got it, thanks! Microwave ovens aren't all that HF, though a lot higher than audio... - Don

p.s. I think it's 49 V rms for 300 W...

 

[ack]

What it is meant by "fast" in this topic? In electronics, it is usually referred as the slew rate of the output stage - one of the fastest I know of is the Atmasphere (600V/microsecond). But audiophiles use the term to define a type of sound - usually related with "slam" and "engaging", sometimes "inducing listening fatigue", but non-related with slew-rate.

My definition of fast is not slew rate necessarily, it's how closely (well) the electronic equipment follows the waveform. This includes slew rate but also settle time (due to thermal tails). It is claimed that faster circuits will follow the waveform more closely, thus presumably more accurately. But designing a fast circuit that's also very powerful, and above all, musical is allegedly extremely challenging. Perhaps a lot of negative feedback is required? I wouldn't know - just read this some time ago on a'gon I believe

 

[Orb]

Within the context of this topic I believe it has been established that by fast it is the slew rate and rise times. Although I have observed that the published specifications do not have the same Vout or are just not specified which would require some conversion.

Just to add anyone able to comment about the power-rail-output stage relating to the perception of amp speed.
I mention this because Chord Electronics is deemed to be very fast and possibly detailed/analytical like Krell but the slew rate of the Chord gear is not amazing, however they do have a very interesting power-rail-output solution.

This is oversimplifying it but quoting John Franks from awhile ago:

"The basic concept of the Chord power supply is to share current demand between the power supply rails, achieving this with a strong magnetic flux.
The only practical way of doing this is with a high frequency topology.
I came up with a design which worked well and it was demonstrated at a Brighton hi-ti show in 1982. At the time, though, the amplifier wasn't really a commercial possibility because the only suitable output devices were to military specification and necessarily very expensive."

And their products are acknowledged to be detailed and fast, while used by many professional studios and still in some of the mastering suites of Abbey Road (that is meant to be aligned to Classe or some other company these days).

Cheers
Orb

 

[microstrip]

Within the context of this topic I believe it has been established that by fast it is the slew rate and rise times. Although I have observed that the published specifications do not have the same Vout or are just not specified which would require some conversion.

It is more complicated than that. When analyzing the "time performance" of the amplifier you have to deal with small signal bandwidth and large signal bandwidth (this last one is usually mainly related with the slew rate), as they affect the signal differently. But as soon as you consider the frequency limitations imposed by the brick-wall filters in the digital recording process you will find that the differences in "time performance" for any reasonable amplifier can not be a justification for sounding different.

 

[JackD201]

It is more complicated than that. When analyzing the "time performance" of the amplifier you have to deal with small signal bandwidth and large signal bandwidth (this last one is usually mainly related with the slew rate), as they affect the signal differently. But as soon as you consider the frequency limitations imposed by the brick-wall filters in the digital recording process you will find that the differences in "time performance" for any reasonable amplifier can not be a justification for sounding different.

Just answering what was meant by "fast" in this thread microstrip, the engineer's or audiophile's "fast". Don't look at me man, I didn't say nuthin' about slew rate or rise time being anything. It was your question. Remember?

What it is meant by "fast" in this topic? In electronics, it is usually referred as the slew rate of the output stage - one of the fastest I know of is the Atmasphere (600V/microsecond). But audiophiles use the term to define a type of sound - usually related with "slam" and "engaging", sometimes "inducing listening fatigue", but non-related with slew-rate.

 

[Gregadd]

Energy storage is also a factor in amplifier speed.

 

[Gregadd]

Greg, All of the above is IYHO, is it not?

For the record my opinion is never humble. There are many contributors to this thread. As far as my contributions are cocerned, I prefer to call it adivce.

 

[microstrip]

Energy storage is also a factor in amplifier speed.

Yes, but curiously in both ways - some people claim that amplifiers with large capacitance in the power supply are slow and use small capacitance in the power supply, but others claim that due to the large capacitance their amplifiers are fast!

 

[microstrip]

Just answering what was meant by "fast" in this thread microstrip, the engineer's or audiophile's "fast". Don't look at me man, I didn't say nuthin' about slew rate or rise time being anything. It was your question. Remember?

Jack, my apologies, I should have quoted only your true and appropriate statement "Although I have observed that the published specifications do not have the same Vout or are just not specified which would require some conversion". My comment only addressed this part ot the sentence.

 

[Gregadd]

I hope you guys learned some things. I lnow I not only learned some things, I am better able to express the things I thought I already knew. More importantyl there is no reason to fear amplifier measurements. They can help you match the right amp to/your speaker/ room/music interface. As always the "proof is in the puddin' not the recipe. Feel free to add to the htread as you see fit. See you at the show.

 

[ack]

But designing a fast circuit that's also very powerful, and above all, musical is allegedly extremely challenging. Perhaps a lot of negative feedback is required? I wouldn't know - just read this some time ago on a'gon I believe

Just to close this argument, I just re-read the TAS BAlabo review, and it is claimed that:

BAlabo may not have gone about it in quite the same way as Soulution did with the Switzers’ very-high-speed, very-high-bandwidth, very-high-negative-feedback circuits...

and from the Soulution review:

the received wisdom about solid-state was that negative feedback was a bad thing—only to be applied sparingly and locally... Soulution have turned this conventional thinking on its ear. Together they decided that it wasn’t feedback itself, but the speed at which the feedback loop operated that was the problem.... to eliminate the time-related distortion, the graininess and edginess that feedback engenders, you have to make those feedback loops correct errors instantaneously. This means that circuits must operate at incredibly high speeds and with incredible precision. Forgetting about shorter signal paths and fewer parts, Schürmann apparently found ways to do this very thing, reducing propagation delay times (the amount of elapsed time it takes to correct a signal via feedback) to 1–10 nanoseconds...

 

[DonH50]

TIM was discovered to be related to loop delay decades ago. That is one of the reasons for all the ultra-wideband amps in the 80's and continued presence in the market today. However, 10 ns prop delay corresponds to about 11 MHz bandwidth; seems like overkill. I am sure it is not for someone... Not to say fast feedback does not matter, but too much bandwidth can cause other issues. among them higher noise and stability under reactive loads. One way around this is a fast circuit with limited bandwidth at key points to reduce noise and improve output stability.

 

[ack]

In the same review, the Soulution specs list a 2MHz bandwidth for the 700 - could it be for that reason? And the Spectral speaker cables have a general bandwidth limitation at around 800kHz (-3dB).

However, I just noticed something strange: a) the Soulution 700 is no longer listed on Axiss' web site (http://axissaudio.com/amplifiers/ampSoul.htm), as if it's about to be replaced already; and b) the 710 stereo amp is claimed to have zero feedback.

 

[Gregadd]

Specs for Spectral DMA260. http://www.moremusic.nl/spectral/dma-260.html

now that's fast!