Jon Risch distortion test signal

[skeptic]

"Which of course has absolutely no bearing on my intelligence or deductive reasoning abilities for the purposes of this thread."

True but it does bear on the knowledge you have to understand the crux of the debate. Were you to apply your intellegence and powers of deductive reasoning to study the applicable science and underlying mathematics, it is entirely possible that you could acquire that knowledge. Unfortunately that takes many years in an engineering school, not minutes or hours on the internet.

"I don't see that it does any such thing."

As I said, you don't understand the implications of the issues at the heart of this debate. If what Mr. Risch says is correct, either the entire mathematical analysis by Laplace and Fourier which completely characterizes both periodic and non periodic waveforms is wrong, or all of the work done by PHDs in electrical engineering and mathematics to develop the instrumentation to precisely make the measurements don't reflect the mathematics they are based on.

You can characterize my comments as hyperbole or posturing if you like but the sharpest criticisms I can level will be as nothing compared to what this theory will face by the huge corporations like HP who manufacture the worlds finest instrumentation and who will defend their technology and the underlying science behind it. Guess who I'm putting my money on.

"Jon clearly based his new test method on the already-existing methodology, which would make his version simply an evolution and refinement of the previous approach"

No! He claims the present method is inadequate and gives false results. He postulates that it disguises the true answers. At the very least, he will have to show that nonlinearities in the existing equipment or random noise disguise the actual results using the present method and that his doesn't. This is a very steep uphill climb. He will face the people who have designed and tested this equipment. Unless you've worked in industry, you have no concept of how hard manufacturers work to tear down their own products to find out exactly what the limits of their capabilities are.

"And history is replete with people who thought outside the box of conventional scientific wisdom, were reviled at the time for doing so, and were later proven to have been correct. "

It's also far more replete with crackpots who could impress the rabble in the streets but were proven no more than charlatans. I'll believe he's the second coming when I see in the paper that he walks on water.

[RobotCzar]

I really do try to put myself in your shoes when I read comments like that. You relate from a completely different, detached point of view. You know what you've read. I know what I have experienced.

rw

I will soon not post here at all. But, I will take the time to try to make a couple of points to you:

1. When you are cornered about some assertions you make (e.g., "simple" test signals are inadequte to judge amp performance) you resort to talking about what you experience. That is not fair. Are you always talking only about your experience or are you making wider claims (objective claims) about amp distortion and audibilitiy? You really shouldn't move back and forth between facts and experiences. The point I made above is that it is scientifically and mathematically valid to test amps with pure tone sine waves. That statement is contrary to what many high end audiophiles say and they are simply wrong about it because they are using their intuition and "common sense" which happens to be dead wrong. And, if they are wrong about something as basic as this (like Jon R. is) then why should they be given credibility about other electronic and audio issues?

2. If you really are only comming from the view of your experiences, then that is of very limited use to the rest of us, because we are not you. That is the essence of being "subjective". Many of us who support a more rational approach do not deny that you preceive what you claim to percieve, we just doubt the REASONS for that perception. For example, perception is known to be greatly influenced by expectation. Therefore you might really perceive a difference in the sound of a component you expect to sound better (perhaps because it is more expensive). The standard way to protect against such a bias in scientific testing is to prevent you from "knowing" which component is the more expensive (this is called blinding). In that case your expectations cannot influence your perception.

I can't help it that many many people have claimed to easily hear amp differences (like you do) but have failed to demonstrate that they can when they have nothing to hang their expectations on (i.e., they are blinded). But, that is pretty much what has always happended when audiophiles claiming to hear differences in low distortion amps have been tested (to my knowledge). it would be very illogical for me to assume that your experience is due to amp differences when you haven't taken the time to test yourself by taking the proper precautions to eliminate known reasons for people to 'experience" audible differences (e.g., level matching).

I have merely tried to point out that your explanations (which are for the most part straight "party-line" explanations offered by high enders) are contrary to fact, logic, and evidence. I don''t doubt your experience, only your explanations of your experience.

[FLZapped]

This I'm still not clear on. If it's necessary to obtain results out to the nth degree for the method to have any usefulness, how do you explain the results Jon did obtain, since he presumably did not do that? Certainly limiting the calculations to some established threshold would make them easier to interpret, but would doing that also automatically invalidate the results?

What Jon claims he did and what Jon can actually produce are two different things. If you had the history with him that many of us do, you would immediately know this without explaination.

I didn't say the results are invalid, just unusable. Like that old quip: "Too many cooks spoil the broth" - too many test tones spoil the test.

One other thing you have to understand is that these tests must be able to be performed outside the engineering lab. Make it too complicated makes it unusable anywhere but the engineering lab. Okay, so how is a tehnician out in the field going to be able to relate his findings to an engineer if they aren't speaking the same "test language" because he cannot perform this test and be able to interpret the results?

-Bruce

[J Risch]

No! He claims the present method is inadequate and gives false results. He postulates that it disguises the true answers. At the very least, he will have to show that nonlinearities in the existing equipment or random noise disguise the actual results using the present method and that his doesn't.

You are not being clear here. Which present method?

If you are talking about single tone HD measurements, this is wrong because a single frequency measurement DOES NOT FULLY CHARACTERIZE THE DUT AT ANY OTHER FREQUENCY. I know that this is not so much the case for RF amps, which operate over a very small % octave basis, but for an audio power amp, which operates over a three PLUS decade range, or over ten octaves, it is indeed the case. This information is in the literature, and has been provided by many a power amp or preamp review. A simple linear equation, such as is used to determine the X order intercept point FAILS with most audio devices.

If you are talking about the prior multitones that were being used by others before I came up with the Phi Spectral multitone, then they were guilty of either a lot of distortion product cover up, or were gulity of a HUGE amount of distortion proiduct cover up (ala the AP FASTTEST signal).

When I state that the traditional single tone HD measurement is not going to be relevant for most audio devices, I am not trying to repeal Fourier or rewrite Laplace, not at all, I am simply looking at the inherent limitations of the existing signals that are currently used.
A single tone HD measurement all by itself does not provide the correct HD information for other frequencies that are far removed from the measurement frequency. A single 1 kHz HD measurement will NOT tell you what the HD is at 100 Hz or 10 kHz, for audio devices, it just isn't so.
When I say that the classic two tone IM signals do not excercise the entire audio spectrum, this is a completely true statement. Measurements taken with the three major classic IM methods will all DIFFER in the levels of distortion they record when used with audio devices. Doesn't this tell you something? It DOES invalidate FLZapped's comments, and renders your sweeping claims that I am trying to repudiate Laplace and Fourier as over the top and irrelevant to the issues at hand.

Jon Risch

[J Risch]

What Jon claims he did and what Jon can actually produce are two different things. If you had the history with him that many of us do, you would immediately know this without explaination.

Totally irellevant to the technical discussion, and quite untrue as well.

I didn't say the results are invalid, just unusable. Like that old quip: "Too many cooks spoil the broth" - too many test tones spoil the test.

Two things. It is entirely one thing to say it is unusable, and another to say it is difficult to use, or that it is not the same as what I am used to doing. You are saying it is unusable, and this is not true. Is it not like what you are used to? Yes. Is it difficult for an inexperienced person to interpret? Yes. That does not invalidate the power it has, or the capability when used by competent people.

The second thing is, that by merely LOOKING at the FFT spectrum analysis result, one can SEE how high the levels of the distortion producs are. If you are concerned about the noise floor levels, then run an FFT analysis with a -110 dbFS signal level coming off the CD-R. This is the baseline for that particular combination of gear and measurement set-up. (This avoids the automatic muting that some CDPs have built in when they encounter enough samples of digital "zero input" in a row.)

In my case, I was able to regularly achieve a noise floor where none of the noise/distortion components was above approx. -105 dB from the nominal reference level. The multitone signal had pure tone components that were at -16.5, -21, or -22.5 (for 6, 10 or 12 tones respectively). So even the 12 tone Phi Spectral signal had a dynamic measurement range of over 82 dB. I could be fairly certain that ANYTHING that was registering at a level above -105 dB was a distortion product, and to check, all I had to do was cursor the frequency, and look at my spreadsheet.

An inspection of the measurement graphics from my paper would show that in most all cases, when stimulated with the Phi Spectral multitone signal, DUT's had SOME distortion products at levels well above -105 dB.

For the loudspeaker measurements, I even showed the noise floor due to the acoustic element as a greyed in set of data on the spectral plots, it can easily be seen that the distortion signals were well above this noise floor, and ALL of the spectral content above this floor was indeed distortion. So just by LOOKING at the plots of one DUT against another, it was immediately obvious that one had 6 or 10 dB of broadband low level distortion than another. As a concrete example that all can go look at, see Fig Y, at:
http://www.geocities.com/jonrisch/page10.htm

This shows one of the Phi Spectral test signals (not the only one) coming out of the reference CDP coming off a CD-R, and straight into the spectrum analyzer thru the usual signal path (included a known clean mixer to set levels).

Note the level of the primary tones, at a visible level of approx. -22 dB, while the noise floor (and/or residual CDP distortion) was at less than -105 dB, except for a little 60 Hz hum component at -100 dB. When I the tested another CDP, I found that I would see distortion products at -95 dB, a full 10 dB higher (this was Fig. 22 in my paper, vs Fig. 20). Of course, the noise level of the unit in Fig. 22 was checked, and it was approx. at the level of the spectral content in Fig. 20 (or comparable to my website Fig Y). This is a very simple way of looking at the overall audio band distortion content of one DUT vs. another.
Oh, BTW, more level in between the primary tones is BAD, less is GOOD.

One other thing you have to understand is that these tests must be able to be performed outside the engineering lab. Make it too complicated makes it unusable anywhere but the engineering lab. Okay, so how is a tehnician out in the field going to be able to relate his findings to an engineer if they aren't speaking the same "test language" because he cannot perform this test and be able to interpret the results?

This seems to say that if it is to difficult, it should not be tried or used. Why limit the cut-off to a tech? Why not say Joe Sixpak should be able to read the results? In point of fact, the modern test instruments we have can do this, push a button, and you get a THD read out. But we have found that this simple number is just not enough information, and does not directly correlate with what we hear.

Tube amps with a THD number of 2% sound MUCH better and more lifelike than a SS amp with a THD number of 0.008%. Or a more palatable comparison, the modern SS amp of 0.1% THD sounding better than the SS amp with 0.01% THD. These simple one dimensional metrics are highly flawed, and do not give us the ability to judge strictly by the numbers what amp will sound better, more lifelike and realistic to us.

In terms of what it takes to perform such a Phi Spectral multitone measurement, this does not exactly require a rocket scientist to do. As I outline in my paper, I used a digital based tone generation program (in my case, Cool Edit), created the test signals one pure tone at a time, mix pasted them together at the proper levels, burned the test signal to CD-R, and found out which CDP would play it back with a sufficiently low distortion level to use as the test signal source. What about the spectrum analyzer? With modern computer sound card based software and shareware, and a studio grade sound card costing less than $200, anyone can generate and spectrum analyze an audio device with a Phi Spectral multitone.

With a suitabkle full duplex soundcard, ne could simultaneously generate the test signal without the intervention of the CDP, and then take the signal in fromthe DUT and record it for later analysis using any one of a number of software programs. There exist a few such soundcard based software packages that can actually generate a multitone, and then analyze it nearly real-time.

These low cost and relatively simple test set-ups may be why some of the major names in measurement are not that interested in making the Phi Spectral available: it would literally wipe up their expensive hardwired dedicated equipment, and make it hard to for them to sell it at a huge profit. AP is not exactly going to shoot their FASTTEST and put it out to pasture, they have too much investment in it, even the TM and such make it their own special proprietary signal, which they have been marketing for years now. NIH is alive and well, and actively upheld at many companies.

I agree that it would be a good thing to make it easier to interpret and to share the data garnered from such a Phi Spectral multitone test, rather than be limited for conveniences sake to an A/B measurement comparison graphic (though this would still be a huge step forward in determining the performance of an audio device when stressed with more than just a single or pair of tones).
But that still does not mean that you can not calculate the THD, or the total IM distortion, or the total distortion by including all the relevant distortion products above a certain level, or even taking the distortion product data, and weighting the various products according to their order (7th order weighted higher than 2nd order by X amount, etc.). Given that this is a relatively trivial excercise to do on a modern computer based measurement system, it is not just a pipe dream or an impossibility as you seem to insist it is Bruce.

The Phi Spectral has already been used successfully to measure and compare various audio devices, and has turned up some interesting data in it's own right, such as the discovery of CDP digital filter clipping, and how much it differs from brand to brand, part number to part number. For you to continue to deny it has any usefulness or validity in the face of this, and of my explainations, defies reason and logic.

Jon Risch

[E-Stat]

I can't help it that many many people have claimed to easily hear amp differences (like you do) but have failed to demonstrate that they can when they have nothing to hang their expectations on (i.e., they are blinded). But, that is pretty much what has always happended when audiophiles claiming to hear differences in low distortion amps have been tested (to my knowledge).

I think it was Mtry who finally produced a lone story of some audio dealer in Florida who was unable to hear the difference between a Yamaha integrated and his Krell amp, if memory serves. That is the only report I've seen on this board to back your assertion. If such other proof exists, I'd be curious to read it. When pressed, he admitted that he could not find any citations that include the kind of equipment that is available today.

I have merely tried to point out that your explanations (which are for the most part straight "party-line" explanations offered by high enders) are contrary to fact, logic, and evidence. I don''t doubt your experience, only your explanations of your experience.

"Fact" has yet to be established from my standpoint. See above.

You are correct in saying that I cannot explain the reason behind audible differences among a range of components once you eliminate basic variables such as loading, major frequency response variations, etc. It is also true that when you ask ten audio engineers that same question, you will get ten different interpretations. I am not referring to guys who design two-way radios, power distribution systems, telephone systems, etc. These are engineers who live and breathe audio reproduction, many of whom having been doing so for decades. That "science" has irrefutably found the answer completely ignores the reality of the audio market. If such were true, why doesn't some well financed company simply produce these allegedly "perfect" inexpensive killer components and put everyone out of business?

I will also agree that any subjective opinion of musical reproduction naturally involves a set of priorities established by that listener. For that matter, my priorities have changed over time. My overall posture is to encourage folks to hear a range of components to get a feel for the variations found. While I have no conceptual problem with blind testing, I do object to blind acceptance of "truths" set forward by some engineers who claim to have found the holy grail. Just like when they did in the early solid state era chasing low THD numbers. Do you acknowledge that there are many companies whose products of today measure worse than their counterparts from twenty and thirty years ago? Who do companies ranging from Crown to Pass Labs (formerly Threshold) find that chasing the single THD metric was futile and not beneficial to providing the best musical reproduction? Just like when they did in the early 80s when CDs emerged. Perfect sound forever? It was those "wascally" audiophiles who immediately pointed out "hey guys, something isn't right here". Twenty some years later, we have a greatly improved product. On that topic, I hear conflicting defenses to that observation. On the one hand, you have those who claim that there have been absolutely no advances in basic circuit design and thus audible fidelity for decades. On the other hand, others posit that the early CD engineers knew of the limitations but simply didn't have the technology at the time. Go figure. RBCD is still not "perfect" IMHO. I do not accept the notion that "the only reason behind the high resolution formats is for strictly marketing reasons".

You continue to harp on the "if it's more expensive, it must be better" notion. Many times that is the case, many times it is not. On two different listening situations, I heard differences between CDPs and power cables for which I was rooting for the underdog. Objectively, however, I evaluated their performance for what it is - or what it was not. BTW, I ended up purchasing the "loser" in the CDP comparison and am quite pleased with that purchase of three years ago.

In the final evaluation, it is the continuous process of questioning our realm of known science that betters the technology. We all benefit. I'm sure glad that we're not stuck with those dreadful early sixties SS amplifiers today, despite their "better" measurable performance.

rw

[skeptic]

I'll grant you this much, at least you no longer present us with 10 page postings with countless hot link references to endless other papers, articles, and whatever the way you used to. Trying to get through even one of your old postings was only slightly more tedious than reading War and Peace from cover to cover. I admit I never actually got all the way through even one and eventually I gave up trying.

My problems with your theories will be as nothing compared to the cross-examination you will face from people who are actually expert in this field. I am merely a hobbyist with a long time ago bachelor degree level education in electrical engineering. That's nothing compared to what you will have to deal with at the highest professional levels. Statements like "The correlation between which unit sounded good, and which didn't followed the Phi Spectral measurements almost exactly." will get you laughed out of the arena assuming they ever even let you in. Considering you have lost every prior debate over cables you've had with "just lil ole me" through simple logic, you'll have to be a lot more convincing than you've been here. (right, in my dreams, your favorite expression to refute history.) If and when the day comes where the Jon Risch method for non linear distortion measurement and analysis replaces the existing and become de rigeur, you will earn my respect which I will gladly announce in public. But until that day, I'll just sit back and watch to see what happens in the real world.

[FLZapped]

You can characterize my comments as hyperbole or posturing if you like but the sharpest criticisms I can level will be as nothing compared to what this theory will face by the huge corporations like HP who manufacture the worlds finest instrumentation and who will defend their technology and the underlying science behind it. Guess who I'm putting my money on.

Except that they are now known as Agilent.

Unless you've worked in industry, you have no concept of how hard manufacturers work to tear down their own products to find out exactly what the limits of their capabilities are.

Indeed, most people would here would pass out if they ever sat through a real engineering design review.

As Skeptic has pointed out, Jon has no real evidence that his methodology is any better, nor any more elegant that what we already use. In fact, I have pointed out some of it's short comings, which Jon cannot adequately address, except to say you need a huge amount of computing horseppower to support it. Why, when what we currently have is elegant, transportable, accurate, and obvious to implement and interpret.

Quite frankly, his design of it is based soley on speaker systems. Which is obvious by the seperation of the tonal groups; One falls in the low frequency driver range, while the other in the high frequency driver range.. As I said earlier, I believe he dreamed this up to support his belief in bi-wiring. Even so, it is still unecessarily complex to measure what he is trying to achieve and has no useful application in electronic circuitry, much less speaker systems.

-Bruce

[skeptic]

"Indeed, most people would here would pass out if they ever sat through a real engineering design review."

Design review? That doesn't even begin to tell the story. They put them in thermal cycling chambers to see what happens if they are repeated exposed to extremes of temperature, in humidity chambers, in rf and acoustically shielded rooms to measure all types of emission. They may drop it off the roof of a building or run over one with an automobile. They'll test it at every conceivable voltage to see just where and how it will blow up. As for performance, the design and performance is checked over and over again and compared to the competition. Field surveys to see what customers like and don't like about them. Not just in industry but in colleges where post doctoral researchers are using them every day to perform leading edge research. Every reported field failure is documented to find out why it happend. And then when they have amassed as much data as they can, they go back to the drawing board to start all over again. Meanwhile, other scientists and engineers are working on the next generation of equipment to see if they can punch any hole they can find in the basic concepts and design as well as incorporate new features which will be useful to the end user and give them an edge over the competition. Mr. Risch will be chasing a 100 mile an hour train which is already far down the tracks and he will not only have to catch up to it, he will have to pull out ahead of it and beat them at their own game. They don't just dabble in this. It's their life's work. And they are a team with diverse skill sets including many most of us have never even heard of. Not just one company. Every big company in the business. As I said, it's a very steep uphill battle to overthrow the existing theory and practice. Can one individual do it alone? As I said, I'll watch and see. I'm betting on the establishment. The world of internet chat boards and consumer magazines is kindergarten compared to the world these guys play in.

[Resident Loser]

...re:"if it's more expensive, it must be better"

Check this out:

http://www.vmpsaudio.com/dc31.htm

---AND---

re:"...I'm sure glad that we're not stuck with those dreadful early sixties SS amplifiers today, despite their "better" measurable performance..."

Let me posit this thought...many early CDs got a bad rep due to the fact that most of the analog program material was mastered, mixed and otherwise hocus-pocused on equipment contemporaneous with the performance, for home playback on similar vintage stuff...consequently the transfer to digital was less than stellar...soooo...back to the masters...here a tweek, there a tweek, everywhere a tweek-tweek...new learning curve for the new medium...

Since most of the stuff played back thru those "dreadful" amps was probably produced on tubed-gear, for playback on similar gear(and let's not discount the fact that true "audiophiles" of the era were lstening predominantly listening to classical, using Mac and Marantz and Fisher and HHScott et al) IMHO you have a similar problem...particularly since the recording companies are loathe to retool and/or retrofit for any reason, particularly for the "pop" crowd...

So, I don't think things reached an "equilibrium" of sorts, until the record producers sprung for the SS stuff and used updated techniques as required in the recording process.

Perhaps the SS amplifiers weren't the problem and it was simply the software "incompatiblities"...

Early SS guitar amps did did just-plain-$uck, however...

jimHJJ(...although my vintage "Pignose" is a hoot and a half...)

[FLZapped]

"Indeed, most people would here would pass out if they ever sat through a real engineering design review."

Design review? That doesn't even begin to tell the story. They put them in thermal cycling chambers to see what happens if they are repeated exposed to extremes of temperature, in humidity chambers, in rf and acoustically shielded rooms to measure all types of emission.....

Oh, I was talking looooong before product certification....I was talking about the point where the ink is still drying on the velum and all the enginners gather in a room to "review" the new design. Which is more applicable to JR's gaining acceptance of his supposedly better mouse trap. What we are doing here pales by comparison.

Certainly, product verification and certification is also rough - but on the equipment, not the engineer, well, at least not directly.

A very small fraction of the tests we run become published specifications.

-Bruce

[E-Stat]

Check this out:

http://www.vmpsaudio.com/dc31.htm

No revelations made there by Mr. Cheney. Questions of mid-fi vs. high end aside, there are simply more entertainment choices and reduced fidelity, yet popular formats for music such as MP3 available today that have drastically reduced the number of folks who simply sit down to listen to music. Before you cry elitist regarding my MP3 comment, consider that I am listening at this very music to MP3s stored on my computer. I ripped all my CDs. There's no denying the convenience of having one's entire musical library immediately accessible. It works great for background listening which is really what most folks do.

re:"...I'm sure glad that we're not stuck with those dreadful early sixties SS amplifiers today, despite their "better" measurable performance..."

Let me posit this thought...many early CDs got a bad rep due to the fact that most of the analog program material was mastered, mixed and otherwise hocus-pocused on equipment contemporaneous ...

We're talking about different animals here. I am referring to the earliest players, not recordings. Perhaps you never suffered with a first gen unit. I had an early Magnavox unit (really Philips) that I gave to my Mom when I moved to a Pioneer unit. I listened again to that unit a few years back when she passed away. Nope - it still sucks with new recordings. My $69 Toshiba 3950 DVD player of today is significantly smoother in the high frequencies. BTW, the earliest Telarc recordings mastered on the 16/50 Soundstream system remain pretty darn good today. Conversely, I have some remastered Astrid Gilberto recordings from the early sixties that sound quite clear on my current rig, albeit replete with ping-pong stereo effects, limited bandwidth, and in a couple of cases, gross analog distortion.

Since most of the stuff played back thru those "dreadful" amps was probably produced on tubed-gear, for playback on similar gear(and let's not discount the fact that true "audiophiles" of the era were lstening predominantly listening to classical, using Mac and Marantz and Fisher and HHScott et al) IMHO you have a similar problem...particularly since the recording companies are loathe to retool and/or retrofit for any reason, particularly for the "pop" crowd...

Really don't know where you're going here. The AR integrated amplifier (my first step up from an Electrophonic 8-track unit) remains a hard sounding amp. The Crown IC-150 preamp still drags fingernails across the chaulkboard with your choice of recording.

Perhaps the SS amplifiers weren't the problem and it was simply the software "incompatiblities"...

Then again !

rw

[J Risch]

In fact, I have pointed out some of it's short comings, which Jon cannot adequately address, except to say you need a huge amount of computing horseppower to support it. Why, when what we currently have is elegant, transportable, accurate, and obvious to implement and interpret.

As I had feared, you fail to fully comprehend what the new tests signal is all about, and what it accomplishes.

I have more than adequately addressed the majority of your comments and concerns, leaving only the one about ease of use/transport of the results. I have posted a reply to that aspect as well, and pointed out some simple expediants that would allow one to compare DUT's visually if nothing else.

Quite frankly, his design of it is based soley on speaker systems. Which is obvious by the seperation of the tonal groups; One falls in the low frequency driver range, while the other in the high frequency driver range.. As I said earlier, I believe he dreamed this up to support his belief in bi-wiring. Even so, it is still unecessarily complex to measure what he is trying to achieve and has no useful application in electronic circuitry, much less speaker systems.

If you had bothered to actually read the paper, instead of only looking at the pictures at my web site, you would find that I have proposed a number of variations on the Phi Spectral multitone, rather than just one single set of tones.

The single graphic example at my web site is what I call a Hi-Lo Split Band Phi Spectral. This is one out of approx. 5 different major variations I came up with.
After further research, I believe that testing with just 3 or 4 of them would suffice to give a pretty good picture of what was going on with a given DUT.

For the paper, See:
http://www.geocities.com/jonrisch/PhiSpectral1.htm
and the next two pages. The text is all there, as are the specifc frequencies I recommend.

[ Even so, it is still unecessarily complex to measure what he is trying to achieve and has no useful application in electronic circuitry, much less speaker systems. ]

This is not true, and I have posted sufficient replies for folks with the knowledge to understand why. I think that your reaction and comments have bee a knee-jerk reaction to me, rather than the actual content and idea's, and that you have made all of your judgements based on the one graphic at my website.

The Phi Spectral multitone has a wide application to electronic circuits, and has already detected a heretofore unknown aspect of CD player behavior with regard to digital filter clipping differences between brands and models of digital filters.

As for it's use with loudspeakers, this could quite possibly be the most powerful and useful distortion test signal ever applied to loudspeakers in the history of measurement. It should be considered indispensable for any loudspeaker system or transducer designer.

Jon Risch

[FLZapped]

As I had feared, you fail to fully comprehend what the new tests signal is all about, and what it accomplishes.

Sure I have and that's not much.

If you had bothered to actually read the paper, instead of only looking at the pictures at my web site, you would find that I have proposed a number of variations on the Phi Spectral multitone, rather than just one single set of tones.

I did. I wasn't impressed.

This is not true, and I have posted sufficient replies for folks with the knowledge to understand why. I think that your reaction and comments have bee a knee-jerk reaction to me, rather than the actual content and idea's, and that you have made all of your judgements based on the one graphic at my website.

Sure it is, it's easy to show why mathematically. I read your paper, remember.

The Phi Spectral multitone has a wide application to electronic circuits, and has already detected a heretofore unknown aspect of CD player behavior with regard to digital filter clipping differences between brands and models of digital filters.

So say you. Using your signal to characterize clipping would be about as useful as using pink noise for the same.

Sorry Jon, but you just haven't created anything new that is likely to displace the much more simple and elegant methods already in existance.

-Bruce

[RobotCzar]

"I think it was Mtry who finally produced a lone story of some audio dealer in Florida who was unable to hear the difference between a Yamaha integrated and his Krell amp, ....."

What you really mean is a "story" that meets all your requirements. Because you think that every person in the world would have to be tested on every piece of equipment, your "story" needs will never be satisfied. Instead, I'll ask you to provide any kind of story of a person that can distinguish level-matched, properly performing amps. You know what, there ain't any--not even one ole' story of a guy in Florida. There are plenty of references to blinded, level-matched amp tests over the years, from Stereo Review, High Fidelity, Audio, The Audio Critic,etc. I'm sure you will find something wrong with all these reports because they didn't use "today's" stuff. So go ahead an silence all us irrationals who are out to spoil your fun and show that you can tell your amp from a low end Pioneer reciever. I don't think you can. I'd bet big money on it.



"Fact" has yet to be established from my standpoint. See above".

"It is also true that when you ask ten audio engineers that same question, you will get ten different interpretations."

How would you know? You haven't asked them. Be aware that the term "audio engineer" often means a "recording engineer" who knows how to work a mixer and learned on the job. Why not ask a scientist or an EE? I have often begged people of your beliefs to do so. They generally refuse. You know, "science can't be trusted and all..."


"I will also agree that any subjective opinion of musical reproduction naturally involves a set of priorities established by that listener."

I said more than that, I said that a purely subjective opinion will only be valid for the person who holds it. Why bother sharing your subjective opinions?

"You continue to harp on the "if it's more expensive, it must be better" notion. "

No, I attribute that opinion to you. You have said many things like "it is ridiculous to think that better components won't sound better". I suppose one must think that if one has spent a lot of money on stuff. I prefer evidence.


"In the final evaluation, it is the continuous process of questioning our realm of known science that betters the technology. We all benefit."

There is a big difference in reasonable questioning of valid issues supported by logic, theory, and evidence and simply making up any excuse for believing what you want to believe.

I am more than a little disappointed that you did not respond about your belief that "simple test tones" are inadequte to evaluate the distortion and frequency response of audio electronics. If you really have an open mind and really want the facts, you will find out about this issue. (Please, NOT from Jon Risch, who is not a scientist, it is doubful he is the EE is claims to be.)

I have agreed with you that a single 1KHz spec is not sufficient to characterize the electrical performance for a audiophile, one must test severa freqs across the audible band. But, there is no reason not to use test tones. A 20 Khz test tone at full volume is going to test how "fast" your amp is better than all the goofy spew put out by non-technical audio crackpots.

Having said that, a single 1KHz distortion spec is probably a good idea for most consumers (including most so-called audiophiles). I very much doubt you would distinguish amps with identical distortion at 1 KHz.

No need to respond. This is my liast post in this forum.

Good luck.

Robot Czar

[E-Stat]

What you really mean is a "story" that meets all your requirements.

Nope. I mean simply results > 1 with equivalent gear

How would you know? You haven't asked them.

That assumption would be wrong.

Why not ask a scientist or an EE?

You're batting zero on the wild guesses. I have spoken with about half a dozen EEs who design and build audio equipment.

You have said many things like "it is ridiculous to think that better components won't sound better".

You mistake me for someone else.

I am more than a little disappointed that you did not respond about your belief that "simple test tones" are inadequte to evaluate the distortion and frequency response of audio electronics.

I use deductive reasoning. If something measures great using one criteria and yet sounds bad, then that criteria is insufficient. My evidence, however, is based upon experience thus I do not post it here.

This is my liast post in this forum.

Good luck to you as well and happy listening.

rw

[J Risch]

RE this post's comment by you, your replies have far from shown what you have claimed (the following are quotes from FLZapped's previous posts):

"1) You can ONLY measure harmonic distortion with a single test signal."
"2) You ONLY need 2 test signals to measure IM."
and
"It is exactly what I said; Two tones and you get Intermodulation products of a non-linear device."
" Exactly, but why bother to sweep, either you are going to see harmonic distortion generated by the non-linearities of the device, or you aren't."
" Once again, you are going to generate intermodulation products if your device is non-linear, period. Two tones is all that is required to predict and see all products."
"The physics don't change with the operating frequency."
" Your multi-tone test signal is completely unusable to measure intermodulation distortion and is a violation of the recognized definition in measuring harmonic distortion."
"The problem is, your test signal will generate SO many products it would be impossible to distinguish noise from distortion products making it unusable as a viable test signal."
"It is completely unusable to make meaningful measurements."

I have responded to, and answered all of these claims.

You CAN measure HD with more than one tone present, you just have to do it properly, using a multitone signal like the Phi Spectral, which avoids the cover-up and stepping on the various distortion products by the other distortion products. That is the new idea behind the Phi Spectral, which a lot of folks never catch on to.

If you measure IM ONLY using two tones, you will not have characterized the DUT for ANY other combination of tones, ONLY for those two specific tones used.
All of the classic two tone IM test signals tend to measure out differently on a great many audio components. This proves that usng ONLY two tones, you will NOT see what the actual non-linearity of the DUT is.

Audio devices have a bandwidth in terms of octaves FAR in excess of any other type of amplifier or electronic device, including RF. This is what makes the simple linear equations used to generate the 2nd order intercept and the 3rd order intercept for RF amps WRONG for most audio devices. HD at 1 kHz is not going to measure the same as HD at very low or very high audio frequencies, nor is a single pair of IM tones going to provide the full bandwidth behavior of the DUT. The physics DO change with frequency!
(and level, and signal history, etc.)

The Phi Spectral multitone can and has been used to measure HD, THD, IM, Total IM, and Total Distortion, all with one test signal applied. The classic single sine wave for single frequency HD can not do this for the whole audio band when only one frequency is used, nor can the classic two tone IM test signals, nor can the heretofore available multitone signals.

ALL distortion products can be measured, can be distinguished from the noise, and can be calculated as noted above.

So say you. Using your signal to characterize clipping would be about as useful as using pink noise for the same.

Sorry Jon, but you just haven't created anything new that is likely to displace the much more simple and elegant methods already in existance.

As a matter of fact, when a DUT clips the Phi Spectral multitone, a host of distortion products arise in the clear area inbetween the primary test signal tones, and clearly indicate clipping, and whether or not it is asymmetrical, or symmetrical, or if it has a poorly behaved recovery, etc. Far more can be determined using the Phi Spectral multitone than with a single pure tone, or two tones.
(An aside: I suppose that you think that all audio devices clip at the same point at all frequencies too, eh? Once again, this would be false.)

As I said earlier, some folks "get it", and some don't. I think we can clearly see what category you fall into.

Jon Risch

[J Risch]

(Please, NOT from Jon Risch, who is not a scientist, it is doubful he is the EE is claims to be.)

You have taken several pot shots at me, without bothering to respond directly, and your comments seem based on others perceptions and comments, rather than on your own understanding (or lack thereof). Reminds me of mtrycrafts.

Since you have not provided any proof of your asssertions or claims regarding me or the new test signal, I can only assume that you have no such evidence or proof, just personal opinion formed from reading other posts. The points in those other posts have been effectively rebutted. If you have any evidence or proof, please cite it, otherwise, I wish you would indeed make good on your promise, and cease posting here.

RE my credentials, I am a Senior Project Engineer at Peavey Electronics, a member of the Audio Engineering Society for 24 years who has presented 3 papers, and have 3 US patents (and 2 more pending) to my name. I post under my real name, as I have nothing to hide. What is your real name, and what are your credentials?

Jon Risch

[J Risch]

Jon, I'd appreciate it if you could elaborate on what you meant about the "past signal history" portion of this statement:
That's a new one for me.

Some audio devices, most notably power amps (but not ONLY power amps) can vary how they respond in terms of distortion based on the past history of the signal.

A trivial example would be the thermal heating of a power amp (output transistors, emitter resistors, various portions of the PS) with a sustained LF signal such as an organ note, and then a cymbal crash right afterward compared to just a naked cymbal crash without the prior sustained LF signal. The amp will react differently, and the distortion will be different.

A more esoteric example would be when electrolytic caps are used to couple the AC signal, or in feedback loops, or in portions of the circuity that provide a bias point or operating voltage reference, the DA of the electrolytic cap can become a significant factor, and cause the operating points to vary with signal content and level. It helps to realize that the DA of many electrolytics can be as high as several percent, and some of the modern super miniature electrolytic caps are the worst, with DA in the realm of 10% or higher. This is enough to actually shift the voltages/currents that they are providing a reference for.

Some audio circuits do not behave well when faced with strong high level asymmetrical transients, they can shift their internal operating points, or a DC servo can saturate, causing a DC output offset shift that varies with transients or signal content, etc.

Usually, examination with a simple single test tone will not reveal these kinds of behavior, but a multitone has enough crest factor and enough slew to push such devices into non-linear operation, similar to the difficult musical portions, and show the distortion occuring. The first example would not be covered as well, but some of it would manifest to some degree.

I have developed several new variations and additions to the original Phi Spectral multitones, some of which explore LF envelope behavior, and others which focus on low level loss of detail (signal information). I really need to do a new paper, and present these to the engineering community as well, before nature or my health put a premature end to my time on this planet. :-( However, I really want to get my book on crossover design done first.

Jon Risch

[FLZapped]

A trivial example would be the thermal heating of a power amp (output transistors, emitter resistors, various portions of the PS) with a sustained LF signal such as an organ note,

For how long? 30 minutes? Of course, you're ignoring that most who wish to test this will pre-condition the amp by running pink noise through it for a period of time, like 30 minutes, first.

and then a cymbal crash right afterward compared to just a naked cymbal crash without the prior sustained LF signal. The amp will react differently, and the distortion will be different.

Sounds like an impluse test to me.....

A more esoteric example would be when electrolytic caps are used to couple the AC signal, or in feedback loops, or in portions of the circuity that provide a bias point or operating voltage reference, the DA of the electrolytic cap can become a significant factor, and cause the operating points to vary with signal content and level. It helps to realize that the DA of many electrolytics can be as high as several percent, and some of the modern super miniature electrolytic caps are the worst, with DA in the realm of 10% or higher. This is enough to actually shift the voltages/currents that they are providing a reference for.

Still sounds like an impulse test.....

Usually, examination with a simple single test tone will not reveal these kinds of behavior, but a multitone has enough crest factor and enough slew to push such devices into non-linear operation, similar to the difficult musical portions, and show the distortion occuring. The first example would not be covered as well, but some of it would manifest to some degree.

Hmmmm....still seems like a job for a simple impulse test to me.

Once again, much more simple and elegant means are availble to get the job done that are self-evident in their outcome.

-Bruce

[skeptic]

"A trivial example would be the thermal heating of a power amp (output transistors, emitter resistors, various portions of the PS) with a sustained LF signal such as an organ note, and then a cymbal crash right afterward compared to just a naked cymbal crash without the prior sustained LF signal. The amp will react differently, and the distortion will be different."

Finally something we can agree on. Although the FTC measurement method requires a preconditioning of operating the amplifier with all channels driven to 1/3 rated power for twenty minutes to heat them up, the changes by sustained high power demands can alter the performance of amplifiers. For one thing, heating is a very real problem with vacuum tubes and bipolar transistors because their characteristic curves change. This shows the enormous advantage properly stablized amplifiers using negative feedback have over amplifiers with no negative feedback since the gain is far less dependent on variables such as power supply voltage and temperature. Non negative feedback amplifiers performance drift all over the place. They may never reach a stable operating condition. It also shows why emitter resistors to prevent thermal runaway in bipolar transistors are so important. It demonstrates that in today's technology, the field effect transistor which is inherently thermally stabalized is the device of choice for audio amplifiers. It also shows the effect of poor power supplies whose bias voltage can be pulled down significantly by sustained high power demands and which may not recover without oscillation or other transient instability when overloaded. The effects of all of these aspects of amplifier performance howerver can be quantitatively meausured using standard methods for IM and Harmonic distortion. What is really required is a change in the test procedure to demonstrate the differences between amplifiers rather than the similarities as we have today. They really have to be pushed to their limit and beyond. No standard test that I am familiar with even remotely simulates the real life operational conditions of driving a loudspeaker and some loudspeaker systems with their high back emf and highly reactive low impedence loads present amplifiers with terrible challenges they are not designed to meet. The caution not to use the current crop of low end A/V receivers with speakers of less than 8 ohms is only the most grudging concession to this limitation of their real life performance. Not that they won't merely perform adequately, but that they are prone to failure under such conditions.

"I really need to do a new paper, and present these to the engineering community as well, before nature or my health put a premature end to my time on this planet. :-( "

I cannot believe there will ever come a time when this board and Cable Asylum will not be graced with your "unique" observations and opinions. After all, only the good die young. Just watch out for a little girl with a dog, three weird friends, and a bucket of water.

[J Risch]

For how long? 30 minutes? Of course, you're ignoring that most who wish to test this will pre-condition the amp by running pink noise through it for a period of time, like 30 minutes, first.

Tsk, Tsk. For the IHF/FTC style testing, you do the pre-conditioning only for the power rating portion of the test (which has distortion limits), but this is NOT the segment of the test regime where you actually run the distortion test portion for the specs.

Sounds like an impulse test to me.....

A simple single impulse would not evoke the same kind of reaction from the amp as what I described. Not enough sustained energy is demanded, just a quick blip, and it's over.

Besides, how does this three times repeated call for impulse testing jibe with your mantra:
"You only need one tone, no, make that two tones, no, make that one tone, wait a minute.... "

You just shot your self down Bruce. Your stance was that a single frequency HD measurement and/or a single IM measurement, and you now knew what the DUT distortion behavior was. Why would you need to use an impulse if these other two were all that was required?


BTW, there is an interesting paper at AP web site about Nonlinear Distortion, see:

http://www.audioprecision.com/bin/Co...nt_Methods.pdf

"Comparison of Non-linear Distortion Measurement Methods" by Richard Cabot.

His list of 6 things that can increase HF distortion, and 5 things that can increase LF distortion (over the levels present in the midband), pretty much shoots down your contention that a simple linear equation describes the distortion for an audio device.

So does the listing of distortion for the various test circuits he shows the measurement results for.
Note very carefully that NONE of the IM tests measure the same, nor do they correlate entirely with the THD measurements. See Figures 8 thru 17.

We see in this paper that HD levels vary with frequency, and with level, as do IM distortion levels. This pretty much puts to rest your ridiculous contentions as outlined in my post #92 at:
Jon Risch distortion test signal

I fully expect an attempt by you to repeat some of the same tired old claims that have already been thoroughly addressed and rebutted, but that won't change the facts of the matter.
Pretty much every thing you have said about the Phi Spectral multitone test signal is wrong, in error, or irrelevant.

Jon Risch

[FLZapped]

Tsk, Tsk. For the IHF/FTC style testing, you do the pre-conditioning only for the power rating portion of the test (which has distortion limits), but this is NOT the segment of the test regime where you actually run the distortion test portion for the specs.

Just applying a similar methodology to your posting, Jon, never mentioned IHF/FTC. I guess that makes your test invalid too.

A simple single impulse would not evoke the same kind of reaction from the amp as what I described. Not enough sustained energy is demanded, just a quick blip, and it's over.

What you described was an impulse following a sustained envelope, not the two superimposed on them.

[quote]
Besides, how does this three times repeated call for impulse testing jibe with your mantra:
"You only need one tone, no, make that two tones, no, make that one tone, wait a minute.... "

Oh lookie here, no argument, so we're back to the ol' ad hominem attack. I never said that and you know it, Think I'm goign to call you MM the second, or did he learn from you?

You just shot your self down Bruce. Your stance was that a single frequency HD measurement and/or a single IM measurement, and you now knew what the DUT distortion behavior was. Why would you need to use an impulse if these other two were all that was required?

Never said that either, Jon. You're trying to reach outside the scope of the argument here in a vain effort to try and make you test signal viable. I just countered with a much more simple and elegant way of dealing with the problem you presented.

-Bruce

[FLZapped]

BTW, there is an interesting paper at AP web site about Nonlinear Distortion, see:

http://www.audioprecision.com/bin/Co...nt_Methods.pdf

"Comparison of Non-linear Distortion Measurement Methods" by Richard Cabot.

What a waste of time. This paper was nothing more than regurgitation of what is already well known and well documented. I didn't see a single bit of new ground broken. Much like Jung's paper on capacitor distortion. And much like your test signal, this one too takes a large amount of computing power to even hope to get results, nor does it point to being any more accurate than what is already in existance and much easier to implement and interpret.

His list of 6 things that can increase HF distortion, and 5 things that can increase LF distortion (over the levels present in the midband),

And nothing new in the lot of them. Yawn....

pretty much shoots down your contention that a simple linear equation describes the distortion for an audio device.

Never said that, Jon.

The problem with the methodology described in this paper and your test signal is that there is no manual way to verify that you are getting reasonably correct results as there are with the current more simple methodologies that have made it through the standards process and withstood the test of time.

-Bruce

[J Risch]

You really don't get it, or are refusing to see the light.

Your comments and claims are on record, and to try and deny them is the silliest thing to attempt. Your claims have been rebutted, and your comments are shown to be a slanted and personal view that has no relationship to reality.

End of any further attempts to explain things to you Bruce, it is futile and not worth the effort.

Jon Risch