• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Stereophile's Jim Austin disagrees w Atkinson; says tubes have something that can't be measured

In Stereophile's review of the Mastersound 845 Compact integrated amplifier, which is a tube amp, editor in chief Jim Austin is once again publicly disagreeing with a reviewer; this time it's John Atkinson.

Basically, the TL;DR is that in the measurements section of the review, Atkinson (JA1) qualifies his comments by saying -- I paraphrase -- that he's commenting on its performance as a tube amp, with the recognition that tube amps don't perform as well as solid state. He adds the following explanation in the article's comments section: "When I write 'The amp performs along the lines of what one would expect for a tube amp with zero negative feedback,' the measured performance predicts departures from a neutral sonic character that will be audible."

To which Jim Austin (JA2) responds:

I just want to make it clear that the opinion expressed by JA1 here, though very well-supported, is not universally shared.

It's true--no one connected with reality can deny it--that certain features in old-school tube amps cause departures from neutrality, especially with loudspeakers with impedance curves that drop below, let us say, 4 ohms, which is most modern loudspeakers. No one can deny it because they are measurable at clearly audible levels. But there's another school of thought--embraced by certain other Stereophile writers--that believes that something less tangible is retained in some such amplifiers that is lost in demonstrably more accurate ones. Such opinions are based on subjective experience--self-perceived connection with the music. This makes them literally irrefutable-- they cannot be tested objectively, so they cannot be contradicted, which is annoying--yet (and this is my opinion, as the magazine's editor), in a magazine committed to subjective experience--to listening--above all else, such opinions must not be dismissed out of hand.

Edit: I thought I should add that the opinions/beliefs I'm referring to are held by many of the most experienced, devoted, passionate audiophiles. I do not take that lightly.

Jim Austin, Editor
Stereophile



Still sticking to the "some things can't be measured" trope.
Why were you reading that?
Why am I reading this?
Must have too much time on my hand!
 
If you want the truth about Stereophile, read their media kit, which is aimed at prospective advertisers:
http://www.avtechmediausa.com/mediakit.stph.pdf
Some stuff is omitted, such as graphs showing how median subscriber age and actual subscriber #s (versus Circulation) has been trending, but even if aging and declining, Stereophile's readers have shown a willingness to purchase products featured in the magazine!

OTOH, I imagine that an ASR media kit would reveal an audience of skeptical cheapskates. :p
 
I get paid significantly less than Stereophile writers, but not having to deal with those stunning levels of intellectual dishonesty is vastly more valuable than the monetary difference.
 
Yes, me too.

Concerning the Mastersound amp: Looking at the rectangle pass through, I would say that is sub-par in context.

Picture from the review: Small-signal 1kHz squarewave into 8 ohms.

523-Mas845fig02-600.jpg
Well...
Still, I am a supporter of hybrid amplifiers where everything is in its place.
Long time ago two unknown to the big world Ukrainians designed the beautiful unity gain buffer based on a diamond topology (it was published in Electronics World, November 1992).
Later this buffer became the main output solution in some extremely good Audio Research amplifiers.
Until now, many adhere to this buffer (Andrea Ciuffoli for example).
And here is an oscillogramm of the same test, this is an amplifier without global feedback at all, but the amplitude for this test is more difficult:

diamond_square (1).jpg


There is no panacea, and there can be no perfection either, but a tube voltage amplifier with such a buffer works perfectly without any global feedback and without artifacts associated with it.
And the best part is that such buffers once assembled (they are very simple) last a very long time.
I assembled them a long time ago.
The amp with them isn't cheap of course, but not because of the active components or tubes, how it always works out.
But you can change tube voltage amplifiers and each time get something new and still very good.
I still like it this way.
(It's just a pity that a truly beautiful case for such a thing is extremely difficult to make in amateur conditions)
 
If we consider whether it is possible to measure something, we must define what are we measuring.

Regarding distortion, which is a non linear phenomenon, the plot thickens a lot.

You can quote a distortion figure in percentage for an amplifier. You must quote the frequency of the signal being measured (ok, we assume a sinusoid), its level, and we obtain a figure. A figure is a single dimension. How much information can a single figure portray?

Two different pieces of equipment with the same measured frequency response and distortion can sound radically different.

What is the harmonic content of that distortion, it, its spectrum?

How does it vary with level?

How does it vary with signal frequency assuming a sinusoid?

Do other complex products such as intermodulation appear when we are not playing a humble sinusoid but real music?

So, you can measure it. But good luck trying to obtain a measurement that can really characterise it in a useful way.

What is a useful way? Well, let's imagine that we devise a way to measure and characterise that somehow yields a vector magnitude such that two different pieces of equipment with the same "distortion vector" sound "the same".

And that's the problem. When recording music (except classical) it's not uncommon (especially for voices) to choose microphones that have a valve preamplifier that distorts a bit when pushed hard. And some of the coveted mastering equipment, especialy passive equalisers and dynamic compressors, do distort in a pleasant way. Or a way that most people will consider pleasant.

For dynamics processors you could boldly say that they "macro distort" (applying compression or limiting to a signal is non linear distortion) and also "micro distort" (inherent distortion to the components and circuit path unrelated to the settings applied on the control panel such as ratio, make-up, attack and release times, etc).

There's nothing wrong about it. We use our audio equipment to play music, not to reproduce laboratory measurements.

Is there anything wrong about the measurement first approach in fhis forum? Of course not at all, I am not claiming that. There is some performance any good piece of equipment must satisfy not to be considered junk. But beyond that some designers can make some adjustments in order to obtain that "something else".

When photography was invented I guess some people thought it would be a guarantee of absolute truth. That said, no matter how faithful photography is, just by composing a frame you are deciding how you want reality to look like. When recording music you choose microphones and their placement.

When mixing and mastering you make tons of decisions about equalization (which can be corrective or somewhat creative) and surely you will tweak dynamics. You can do it in an obvious, blatant, creative way, or you can be subtle and do it in a way that people won't notice. But you are altering reality.
 
Jim Austin's promotion has been most unwelcome IMHO; not least because of this nonsense, but because he cannot resist dropping banal footnotes into virtually every article written by others. I (figuratively) feel myself getting stupider reading those.
 
If we consider whether it is possible to measure something, we must define what are we measuring.

Regarding distortion, which is a non linear phenomenon, the plot thickens a lot.

You can quote a distortion figure in percentage for an amplifier. You must quote the frequency of the signal being measured (ok, we assume a sinusoid), its level, and we obtain a figure. A figure is a single dimension. How much information can a single figure portray?

Two different pieces of equipment with the same measured frequency response and distortion can sound radically different.

What is the harmonic content of that distortion, it, its spectrum?

How does it vary with level?

How does it vary with signal frequency assuming a sinusoid?

Do other complex products such as intermodulation appear when we are not playing a humble sinusoid but real music?

So, you can measure it. But good luck trying to obtain a measurement that can really characterise it in a useful way.

What is a useful way? Well, let's imagine that we devise a way to measure and characterise that somehow yields a vector magnitude such that two different pieces of equipment with the same "distortion vector" sound "the same".

And that's the problem. When recording music (except classical) it's not uncommon (especially for voices) to choose microphones that have a valve preamplifier that distorts a bit when pushed hard. And some of the coveted mastering equipment, especialy passive equalisers and dynamic compressors, do distort in a pleasant way. Or a way that most people will consider pleasant.

For dynamics processors you could boldly say that they "macro distort" (applying compression or limiting to a signal is non linear distortion) and also "micro distort" (inherent distortion to the components and circuit path unrelated to the settings applied on the control panel such as ratio, make-up, attack and release times, etc).

There's nothing wrong about it. We use our audio equipment to play music, not to reproduce laboratory measurements.

Is there anything wrong about the measurement first approach in fhis forum? Of course not at all, I am not claiming that. There is some performance any good piece of equipment must satisfy not to be considered junk. But beyond that some designers can make some adjustments in order to obtain that "something else".

When photography was invented I guess some people thought it would be a guarantee of absolute truth. That said, no matter how faithful photography is, just by composing a frame you are deciding how you want reality to look like. When recording music you choose microphones and their placement.

When mixing and mastering you make tons of decisions about equalization (which can be corrective or somewhat creative) and surely you will tweak dynamics. You can do it in an obvious, blatant, creative way, or you can be subtle and do it in a way that people won't notice. But you are altering reality.

And I want my audio system to do as accurate as possible a job of getting the altered reality the engineers recorded to my ears.

"Two different pieces of equipment with the same measured frequency response and distortion can sound radically different" - my response is simply OK, I'll believe it when someone proves it.
 
Distortion is normally measured over frequency and not at a single point. It is also common to plot it over varying signal levels.
IMD is mathematically related to HD so you can predict multi-tone intermodulation distortion from a swept single-tone HD measurement.

If all frequency response and distortion parameters are the same under the same test conditions the amplifiers will sound the same under those conditions. If the distortion spectral content is different, it will show up in measurements. If the output impedance is different, it will show up in the measurements of different loads. If something sounds different and the measurements do not show it either the measurements are either incomplete or misinterpreted. Or perhaps just ignored.

Audio reproduction is not the same as production. If a microphone distorts the sound is a way desirable to the performer (or sound engineer), that is captured on the recording, then the goal of the playback system is to accurately repeat that sound with whatever distortion was included. It should not add additional distortion.

All of these measurements are useful and provide meaningful results. Yes, you will usually need to look at more than just one number, but to imply measurements are not meaningful is a disservice, a failure to recognize their benefit, and lack of understanding of what the measurements convey.

Measurement instruments provide far greater dynamic and frequency range than our ears, and are much better at determining distortion and noise, making the oft-cited claim that we can hear more than we can measure questionable.
 
Last edited:
And I want my audio system to do as accurate as possible a job of getting the altered reality the engineers recorded to my ears.

"Two different pieces of equipment with the same measured frequency response and distortion can sound radically different" - my response is simply OK, I'll believe it when someone proves it.
With "distortion" I mean "distortion figure" as it is quoted in specs. Ie, simple scalar value.
 
I think Jim Austin is missing the main point of his own argument:
- Everything can be measured.
- Tube amps don't measure as well as SOTA solid state amps.
- Jim likes the sound of some tube amps better than solid state amps.
- Therefore; he likes the sound of distortion and/or other artifacts created by tube amps better than the neutrality of SOTA solid state amps.

If he is saying there is something about tube amps that cannot be measured, he is wrong (I'm not sure he actually says that). He needs to determine and define what characteristic of the measurements he likes, whether its a certain type or level of distortion, or noise, or whatever.

Ultimately I suspect he's making this entire statement because he's afraid the manufacturer will view Atkinson's statement as negative and is worried they'll withdraw their advertising support.
Nail. Head.

Its perfectly fine that he is not really into HiFi, but there is no way he can state that because it will kill their advertising revenue.
 
OTOH, I imagine that an ASR media kit would reveal an audience of skeptical cheapskates. :p
Hey! It's not that I'm cheap. I go for value ;)
 
Coincidentally, related to this topic:

Just yesterday I was at my friend's place listening to his system. I'd brought my Brother-In-Law (a very ASR-oriented type audiophile) over to listen to some fun gear.

Among the gear my friend is reviewing is a new solid state pre-amp by the designer formerly of Blue Circle Audio.

Idiosyncratically enough, aside from volume and input selections, it also has 3 knobs for adjusting the sound to taste: Transistor, Tube, Warmth.

In Transistor mode it just operates like a standard transistor pre-amp. But you can dial up the "tube" mode knob to get a more "tube-like" sound. It was fascinating because it really did what it advertised. For instance on an orchestral track with some high piercing strings and horns, they were fairly icy and thin and piercing on that particular track, but dialing up the "tube" knob both thickened out, slightly rolled off, and softened the sound. Very much like what many have associated with the "classic tube" sound.
It was easily discernible to all of us, and I have no doubt everyone here would easily have heard the difference. (Interesting that my accuracy-focused brother in law preferred the more tubey-setting for some tracks).

Not sure exactly what the "warmth" knob does, but it did seem to thicken out the lower mids of the strings.

Apparently the alterations are all done with resistors/capacitors, if I understood correctly.
 
Coincidentally, related to this topic:

Just yesterday I was at my friend's place listening to his system. I'd brought my Brother-In-Law (a very ASR-oriented type audiophile) over to listen to some fun gear.

Among the gear my friend is reviewing is a new solid state pre-amp by the designer formerly of Blue Circle Audio.

Idiosyncratically enough, aside from volume and input selections, it also has 3 knobs for adjusting the sound to taste: Transistor, Tube, Warmth.

In Transistor mode it just operates like a standard transistor pre-amp. But you can dial up the "tube" mode knob to get a more "tube-like" sound. It was fascinating because it really did what it advertised. For instance on an orchestral track with some high piercing strings and horns, they were fairly icy and thin and piercing on that particular track, but dialing up the "tube" knob both thickened out, slightly rolled off, and softened the sound. Very much like what many have associated with the "classic tube" sound.
It was easily discernible to all of us, and I have no doubt everyone here would easily have heard the difference. (Interesting that my accuracy-focused brother in law preferred the more tubey-setting for some tracks).

Not sure exactly what the "warmth" knob does, but it did seem to thicken out the lower mids of the strings.

Apparently the alterations are all done with resistors/capacitors, if I understood correctly.
Or the 3 knobs do nothing and rely on your brain to do the real work. :)
 
Back
Top Bottom