What to look for in a power amp?

[FrantzM]

Interesting article (if a little vague, but it's not an IEEE paper for goodness' sake). Since audio circuits have very high gain and wide dynamic range, proper signal isolation and grounding is incredibly difficult. I am not surprised the article dwells a bit on that; most folk probably do not realize just how challenging it is to build a quiet, (relatively) broadband, high-gain amplifier chain.

A lot of audiophiles do not realize that some premium parts provide no audible benefit and in fact are worse in audio circuits. A prime example is metal-film resistors: they are noisier than other film resistors and have higher self-capacitance. Vishay (have not checked Caddock lately) has articles on this, but it is well-known to *ahem* techies in the design arena and has been for decades. They are great for stable bias circuits, properly decoupled of course. Certain types of film capacitors are better than others, etc. The differences are measurable and often audible. As usual, finding the right components requires a blend of science (engineering) and art (plus listening), with a deep understanding of not only the technical parameeters but how they translate to audibility.

And, of course, there are always cost considerations...

Don

You are making too much sense now.. You leave nothing for people to discuss about ... .. What's wrong with you ??..

Now for the cost consideration... Come on!.. You are being way too nice ... Parts represent a tiny fraction of the price of most High End products ... Even if one were to use the most expensive parts throughout the circuits ... Can I venture less than 10% and still be nice? By the way Spectral is that kind of company using the best parts possible parts and clean layout and topology.. yet they don't charge the proverbial arm and leg compared to several of their competitors ...

 

[Gregadd]

Here is what Richard Freyer of Spectral had to say on pricing in his TAS Interview p.98 February 2009 Issue 190:

It seems as though Spectral products are underpriced relative to the amount of R& D that goes into them.

RF-We've already been very blessed by economic success in some other fields and with other investments. To limit our compromise Spectral by the need to be financially successful doesn't suit our purposes.
When you consider what you're going to be known for -either when you are gone or what you are known for now- to say that it was more important to get to the bank first is just a cop out. We're better than that. With Spectral that is our opportunity as designers and music lovers to make a contribution, and we're stable enough financially that we can literally do whatever afford to do whatever we want.
It's very difficult today for a start- up company, particularly a specialty audio company, to survive at first. What we were able to do in the 70's and 80's , which was frankly very, very expensive research, today might almost be impossible for a young company. We did it while we could, and didn't ask the company to be profitable literally for years. We put tremendous financial resources into Spectral, and the payback for us was a set off technological building blocks and a knowledge base. That's what we bought for this tremendous investment in time and capital.
In our case, getting the technological breakthroughs, getting the performance advantage, and making the results is the company's goal , plain and simple. The financials have to be acceptable, but they're not the principal motivation. In addition, the Spectral customer to us is everything. They are the reason we exist. They are the reason we can do what we do. Our clients are the ones who believe in our mission and have partnered with us on the solutions. We owe to them our very best work. We never lose sight of that.

 

[Gregadd]

While I'm thinking about it, let's not forget Don's definition thread:

There are two posts at the beginning and end.

http://www.whatsbestforum.com/showthread.php?990-Units-Symbols-and-Terms-Oh-My!

 

[Orb]

Yeah clipping and what test-measurement used and presented are important especially spectrum/spectral tests or envelope.
Where this shows these 3; amplitude/fundamental and harmonic-partials/time.
Quite a few papers and presentations that focus primarily on sound theory (not audio playback associated) show the requirement of all 3 of these in an envelope.
The envelope can also apply showing overlay of the distortions as well, but I feel critically is seeing all of this in both frequency (or fundamental with harmonic-partials) and time domain (temporal).

For those who are interested or wanting to know more (appreciate a lot on these forums know and have experience in these subjects) here are some links:
The following link is good for getting an understanding, especially section 2.10 that shows how it applies in time varying spectral plot or envelope.
http://www.billbuxton.com/AudioUI02acoustics.pdf

For distortion I feel Nelson Pass presents it in an easy to read and see measurements:
http://www.firstwatt.com/pdf/art_dist_fdbk.pdf

Cheers
Orb

 

[Gregadd]

We have two knew amps claiming breakthroughs in speed. Spectrall and Dan Agostino. If only I could arrange a drag race.

 

[JackD201]

When folks say "speed" does that refer to the ultra-wide bandwidth approach?

 

[Gregadd]

When folks say "speed" does that refer to the ultra-wide bandwidth approach?

http://www.dagostinoinc.com/products/
http://SpectralAudio.com/bulletins/0410.pdf


Jack let Dan and Keith explain.

I would throw in the Atma-Sphere MA-2

220 watts/channel into 4, 8 or 16 ohm load before clipping.
*Power bandwidth: 2 Hz - 85 kHz within 1/2 db (open loop)
*Frequency response (1 watt, open loop): l Hz - 100 kHz within 3 dB
IM Distortion: less than .05% at full power
THD: less than 0.5%
*Square Wave Tilt: unmeasurable
Phase shift: less than 1 degree @ 20 kHz
Power supply current: 80 Amps
Input impedance: 100K single-ended, 200K balanced- selectable to 600 ohms
*Output section risetime: 600 V/micro-second
Output Impedance: ~1.75 Ohms (open loop)
*Power Supply Storage: 1200 Joules in output section, 100 Joules in Driver supply.

 

[JackD201]

Thanks Greg. I got it......I think.

 

[MylesBAstor]

Typically, high speed means how fast the amp can swing its output, ie from the most negative it can put out to the most positive. The faster it does that the higher the speed. This comes in handy in reproducing a square wave. A faster amp will have a more steep upramp of the lead edge of the square wave and also a faster downramp of the falling edge of a square wave.

Simply, the output changes faster. Faster means it takes less time to make the change.

Tom

But speed isn't everything. D'agostino once told me he could make an amp that was unbelievably fast but it sounded like crap. So there must be more to the puzzle than that.

 

[muralman1]

I have to say one more thing about class D. I already mentioned it might be the condition of a music signal that has been passed through oversampling and digital filters that conflict with the sampling of class D amps. I don't know if this has any real truth to it. All I know is, all oversampling players fail miserably here.

The other important point about great class D amps is they are darn hard to find compatible components for.

 

[JackD201]

Typically, high speed means how fast the amp can swing its output, ie from the most negative it can put out to the most positive. The faster it does that the higher the speed. This comes in handy in reproducing a square wave. A faster amp will have a more steep upramp of the lead edge of the square wave and also a faster downramp of the falling edge of a square wave.

Simply, the output changes faster. Faster means it takes less time to make the change.

Tom

Thanks Tom, now I get it fully.

 

[muralman1]

Can you please explain how the analog output from an oversampling DAC conflicts with the sampling of class D amp with an analog input?

What "great class D amps" are you talking about?

I wrote at some length in this thread and other topics about it all. In a nutshell, all oversampling DACs sound terrible with my system. Non oversampling DACs sound good, but rather bland. My Audio Note DAC has been modified to let all frequencies be equally expressed. This DAC performs all music venues as they are meant to be heard.

I really think that the reason oversampling DACs sound atrocious here is because my system is so spirited it can easily throw any component that doesn't match up. All cables that have but the scantiest insulation will produce white noise through the speakers. This system is that sensitive.

I am using a later edition of the H2O mono amps, the SE. I also have a very special preamp made by H2O, the Fire 2. They are pushing a close to dead short speaker, the Apogee Scintilla.

All equipment must be carefully matched. All equipment that does match will be very simple, causing the least amount of damage to the music signal.

When you get it just right, the real music takes the stage.

 

[muralman1]

"all oversampling DACs sound terrible with my system"

And you've tried every oversampling DAC on the market? Just because the oversampling DACs you've tried sound terrible on your system doesn't prove that their analog output conflicts in any way with the "sampling of class D amps."The other important point about great class D amps is they are darn hard to find compatible components for." Once again, what "great class D amps" are you talking about?

Please read what I posted. I told you what amps I am using. The best oversampler so far has been a Sony 999 ES Modright. The owner happened to have the same amps as I. My transport and DAC system outclassed his player so much, he set out to buy the best NOS player he could.

I was trying to give oversampling players the benefit of the doubt by saying there might be a technical mismatch. Really, I believe it is just because audio engineers have worsened the CD player with their over burdening the music. The 47 Labs Flatfish and Dumpty power supply transport I have just bought completely transformed the sound to the good from my former transport, a highly modified PSA. Everything must be just so. The less the music signal is touched, the better.

 

[JackD201]

Gee I never thought of my amps as "fast amps". FR of 5Hz - 155kHz, 2 Microsecond Rise Time peak to peak V out of almost 84V at 10kHz.

 

[Mobiusman]

I have owned many power amps during my 50 years playing with audio, and of course have listened to hundreds of "top contenders" over the years in many CES and audiophile systems, including a number at the fabled HP's Sea Cliff enclave. Although I no longer travel in the top ranks of audio due to financial limitation, I am still sufficiently fascinated that I go to CES, T.H.E. Show and almost always drop in on the local audiophile shop when traveling. Since a new power amp is on my agenda for the next 30 days and I am financially limited so I cannot consider the $20K+ stratum, recently I have been forced to ponder what makes a good amp for me more than my typical level of audio obsessing.

Previously I would have said that there is only one quality that matters after one establishes the real world limitations like cost, reliability, power, size, heat generation etc. That one thing is how much enjoyment it helps me derive my sound system, being realistic that what you hear is a combination of many components, listening position and environment. However, my friend Marty from this forum, always asks me pre-purchase what I think the resale will be since we both know that rarely do audiophiles keep things for years. So for me it comes down to the joy/dollar index and the dollar component includes resale.

 

[Gregadd]

Gee I never thought of my amps as "fast amps". FR of 5Hz - 155kHz, 2 Microsecond Rise Time peak to peak V out of almost 84V at 10kHz.

No doubt you enjoyed the results of a fast amp. Like excellent transient response.

 

[JackD201]

Yes I do. I guess I never looked at it from that angle though until I read this thread. So thanks to you, Tom and Ack for that.

 

[Gregadd]

...and Don.

 

[JackD201]

Definitely!

 

[muralman1]

I would just comment that "the less the music signal is touched, the better" is good for sure. However, in the world of digital reproduction, the more processing power brought to bear, the more "touch" the better. If you look at some of Don's tech threads there is clear evidence that more touch gets rid of more of the junk byproducts in the digital process.

And even in the analog world, it is proven that two stages of amplification will produce less harmonic distortion than one stage due to the cancelling effect of the second stage transfer curve verses the first stage. This is seen in tube circuits using a common cathode amp followed by a cathode follower, on in transistor circuits seen by following a common emitter with an emitter follower. Now, is the cancellation perfect, no, but it does happen so the simpler and less touch rule technically is not really valid in the Real World. In the theoretical world, yes indeed, you are 100%.

I do not however, deny that in your system, somehow, the older original cd circuits are presenting a better sound to your ears and some of your firends and that is what makes this hobby so much fun or so frustrating depending on your point of view!

Tom

If you could hear my system, I would cherish you as the latest convert. Here I have a transport that measures it's circuit paths in millimeters. A DAC that has no op-amps, just the necessary DAC and clock chips, drive tubes with coupling caps can be so pristine to the faintest highs.... Why on earth does anyone believe that wringing the music signal through successive chips has anything good to do at all?

I don't care what any of the industry folks have to say about what we need. I will do my best to get a top level OS player in here. My science background is not in circuit theory. It is in the natural sciences. I had to learn about quantum physics. It is the science that governs the very small, like electrons and their fields. Transistors owe their existence to designers using quantum mechanics. One cannot understand transistors, semi-conductors, and lasers without a thorough knowledge of quantum mechanics. You can't do it with classical physics. From what I hear here, the fellows who thought up upsampling, and oversampling did not apply quantum mechanics. From what I understanding they leaned on higher math for oversampling computations.

One of the basic tenants of quantum mechanics is you can't do so much as measure the very small without changing it's character. My system feeds on free circuits, including the speakers which are the closest thing to a dead short. My speaker cables are designed to keep that well cared for music signal intact to the point of work at the speaker.

This system is proof less is more, much more.