Cable Distortion Measurements: Part Deux

Status
Not open for further replies.
john curl said:
Steve Eddy has little or no idea of what he is talking about in this case...

I don't know, it strikes me as an honest effort here. Though it may not come across that way, I don't think he ultimately has an axe to grind (a rather petty victory if there was one). I think Steve truely would like to understand this. Personally, I like'd to know one way or another, where is ground zero.


JF
 
JF, there are more threads than just this one where SE attacks my input.
I have the equipment, the experience with the equipment, and the tech manual of the equipment. SE can't be bothered to have any real evidence that contradicts my statements, but he does so anyway.
 
john curl said:
JF, there are more threads than just this one where SE attacks my input.
I have the equipment, the experience with the equipment, and the tech manual of the equipment. SE can't be bothered to have any real evidence that contradicts my statements, but he does so anyway.

Understand. I don't have the details, but I understand you are an expert at amplifier design as well (to say the least). I'm glad (as many are) that you contribute to a public forum (the World Wide Web is a great tool for international communication--Hi again Frank in Belgium). I'm also glad that Steve shakes you guys up a bit. As you know, there are a lot of aspects to technology--nobody knows it all. If Steve was a crackpot, people would ignore him. He certainly is persistant at this, though. Reading the manual and calling the mfg is square one and more that some users do. He doesn't have equipment, but then many of us don't have the equipment (my plan is to see how much a local stereo shop charges to run measurements--my guess is about $30-$50 for fully certified measurements).


JF
 
john curl said:
It is strange when people discuss test equipment that they have no experience with.

Hey, I'm just relaying what I was told by the gentleman at Sound Technology. If you don't like it, take the matter up with them.

First, the 1700B is spec'd at: .0018% harmonic distortion at mid frequencies. So far as my experience is concerned, this includes the oscillator as well as the analyzer section.

Great. But when I specifically asked what the 1700B was spec'd at, I was told it was between 0.0018% to 0.0025% for the generator, and the same for the analyzer.

Again, I'm only relating what I was told by the gentleman at Sound Technology. If you've got a problem with that, call them up and set them straight.

The FUNDAMENTAL limitation of this system's measurement capability is the NOTCH. It is the fundamental limit and is almost all the meter reads, except for noise at low levels. After all, a -95dB notch is slightly less than .002% You can't measure with the meter alone, lower than the notch itself.
This can be seen with ANY spectrum analyzer, FFT, or wave analyzer. The added distortion components from the oscillator or the input stages are lower in level by a significant amount. I have further improved the performance by upgrading the input and oscillator op amps.

Looking at your Mac The Scope plot of the van den Hul cable you use for reference, if we assume that you're notching the fundamental by 95 dB, we're looking at the third harmonic being down about -115 dB relative to the fundamental.

It doesn't matter what you did with the opamps. Your plots speak for themselves. Just as the AP plots speak for themselves. And it's abundantly clear that your system isn't measuring as far down as the System Two Cascade. Period.

That being the case, if the high order distortion products you're measuring are indeed being produced by the cables themselves, they would be even more obvious when measured using the System Two Cascade.

Yet there's no sign of them.

Steve Eddy is not an 'objective' arbitrator here. He minimizes the quality of the equipment, and its cost to duplicate it.

How can I not be objective John when we're comparing actual measurements? The plots speak for themselves. Your system's measuring down to about 125 dB below the fundamental, the System Two Cascade's measuring down to about 145 dB below the fundamental.

These facts have absolutely nothing to do with me. They are what they are. So why do you continually attempt to personalize this issue?

In fact, if it is so cheap to obtain a ST1700B, then why doesn't he buy one, if only to pick it apart?

Why? I've never said you're not measuring down to the level that you claim you are. Like I said, your plots speak for themselves. So why would I need to go buy a 1700B?

Steve Eddy has little or no idea of what he is talking about in this case, but that won't stop him from confusing the rest of you.

Again, John, the plots speak for themselves. How much of an idiot does one have to be to not realize that a system that's resolving down to -125 dB relative to the fundamental isn't as resolving as one that's resolving down to -145 dB relative to the fundamental?

se
 
john curl said:
JF, there are more threads than just this one where SE attacks my input.
I have the equipment, the experience with the equipment, and the tech manual of the equipment. SE can't be bothered to have any real evidence that contradicts my statements, but he does so anyway.

Again, John, I have never said that you're not measuring to the levels you claim to be measuring. I've fully accepted that you are.

I have only said that your system's not resolving to the same levels as the System Two Cascade is. And by your own claims, it's not. Do we need to go through the math again?

Your system's third harmonic is about -115 dB relative to the fundamental. Your noise floor using FFT averaging is is about 125 dB relative to the fundamental.

The System Two Cascade's thrid harmonic is about -130 dB relative to the fundamental. It's noise floor using FFT averaging is about -145 dB relative to the fundamental.

So tell me, John, is -130 dB further down than -115 dB? Is -145 dB further down than -125 dB?

se
 
Crackpot or Despot?

"If he was a crackpot, people would ignore him. But he certainly is persistant at this. Reading the manual and calling the mfg is square one and more that some users do."


Ah yes.............. I remember the last time he called the manufacturer, Jensen Transformers. They ended up trying to distance themselves from comments in confidentiality to him that did not stay confidential. They also asked to be left out of any involvement in discussion of the debate at hand. Steve Eddy later decided that one of technical statements made by the president of Jensen, Bill Whitlock in one of his articles was false and Jensen Transformers was quite the reference he needed. As for ignoring crackpots, try it some time when he spends a week trying to defend some erroneous position with methods that bring to mind the "it depends of what the definition of 'is' is" defense. At some point, the people worth reading will stop bothering, if being second guessed by people with no technical qualifications and a clear agenda, is the only reward for ones efforts. Ask yourself what Mr. Eddy has added to your knowledge of audio design and what exactly he hopes to accomplish with this week long harangues. For my part I can think of nothing except his desire for notoriety by his constant harassment of one the leading audio designers working. I am sure the targets of his obsessive fixation with would love to ignore him, if he would just shut up once in a while.
 
Actually, your last statements, attributed to me are off a little. The real noise floor is -120dB and therefore everything is moved up 5dB. This makes an even greater departure from Bruno's measurement, so be it. Please do not speculate on my very early input.
For the record, and the rest of you, my measurements were at first considered by me to be RELATIVE rather than absolute. Some cables have more distortion than others. When I first used my HP3580 linear spectrum analyzer and later the MAC the Scope with my MAC computer, I was unsure of the absolute accuracy of the distortion levels. Now, because I use a calibrated FFT in the HP3563 and the voltage levels are on the screen, it is much easier to pin down the absolute distortion level. Actually, it is still the changes in the harmonic levels between different wires, that is still the most important.
 
johnferrier said:
Reading the manual and calling the mfg is square one and more that some users do.

I did that to satisfy my own curiousity as much as anything else.

I couldn't imagine that AP would offer a multi-thousand dollar measurement system that couldn't do the same type of measurement as John, but I couldn't say for certain that it could. so I checked it out.

And while I didn't think that Sound Technology was making current models of their 1700 series, looking at how it was presented on their web page, it looked as if they were. So I called 'em. And while I was on the phone with them, I asked about that 0.0009% figure that Charles was ascribing to the 1700B.

John likes to think that I'm just out to get him. I don't know if he truly beleives this or if he intentionally uses this as a smokescreen.

The only thing I'm out to get at is the truth. And it doesn't matter to me where the truth lay. If I'd have found out that not even the System Two Cascade could do the same type of test as John, I'd have reported that, just as I reported that my suspecting that ST was producing current model 1700 series equipment was incorrect.

se
 
Re: Crackpot or Despot?

Fred Dieckmann said:
Ask yourself what Mr. Eddy has added to your knowledge of audio design...

Okay, I don't have time to polish a reply...

Maybe indirectly learned (averted learning):
AP System II is the best equipment to use to test an amplifier.
From the AP datasheet, 25mVrms (which as John Curl indicates is a typical signal and I concur) is more challenging to measure than a 2Vrms signal--the noise/distortion floor moves up (okay, everyone knows this but the graphs in the datasheet made it clear).
Silver is the lowest electrical and thermally conductive metal (useful for pulling heat out of parts...yes sorry, learned it because of these threads).
I also, like Eric's idea of making a "directionless" cable--if nothing else the parallel conductors lowers the resistance.
Solid wire may be better than stranded.
And, if it's important enough, by asking questions, one can get answers.
And other things that don't come to mind right now. Of course, other people have contributed equally, but Steve did start these threads.

I will add that the rest of my circuit has been narrowed down. Information on conductors and interconnects is what I'm finalizing.

The subler phenomena in conductors is still forumulating and is not concrete yet. Though, it doesn't seem to harm to care about the details.

Again, it's a lot of work even to make a small amplifier. I just want to get the details right...

I could learn this other ways alone, of course. But why do we bother with a public forum then?


JF
 
Re: Crackpot or Despot?

Fred Dieckmann said:
Ask yourself what Mr. Eddy has added to your knowledge of audio design


to give you one data point: Steve has contributed to this forum infinitely more than you have.

other than your superficially convoluted answers, you, Mr. Dickmann, have added nothing, zipo, positive to this forum.

If you really really want to sound sophisticated by writing in such a convoluted way, you can at least go through a law program. even one at a local community college will help you a long long way in that regard.

Again, please respect Steve's wish and let this discussion be a technical one, not about how over-inflated your ego is.

For that, you can open another thread of yours and I am happy to take it out with you there.
 
john curl said:
Actually, your last statements, attributed to me are off a little. The real noise floor is -120dB and therefore everything is moved up 5dB. This makes an even greater departure from Bruno's measurement, so be it. Please do not speculate on my very early input.

Fine. But whether -120 dB or -125 dB, when resolving down to -145 dB, there's still no sign of high order harmonics.

For the record, and the rest of you, my measurements were at first considered by me to be RELATIVE rather than absolute. Some cables have more distortion than others.

But you also say that after a time, a given cable will measure considerably less distortion than it did orginally. Which means that its level of distortion might end up being the same as that of another cable which originally measured considerably better.

You also say that distortion levels can change depending on whether you've cleaned the contacts or not.

You seem to have quite a lot of moving targets.

Actually, it is still the changes in the harmonic levels between different wires, that is still the most important.

But you also get changes in harmonic levels with the same wire.

Perhaps SY can help pin down some of these moving targets and maybe we can make better sense of this as a result.

se
 
Oh, for crying out loud ......

Can we please stop with the personnal b.u.l.l.s.h.i.t.

For the record, I don't know SE, I neither like nor dislike him. He reminds me of a "fixated terrier" with a rag. This is not to be insulting ..... it is actually a characteristic we look for in trainees. "Obsession is a virtue."

The concept, that members of the general community here are going to form their opinion of a given member based upon what another member tells them they should think, is pathetic.

SE's fundamental question (as it appears to me .... and a few others with no axe to grind), seems to have scientific validity.

1. JC has report a finding which is outside of mainstream "knowledge". No-one has argued that JC has not seen this, fabricated this, or in any way been dishonest

Certain people have suggested, well JC is an industry leader, this is therefore real ...... CRAP!

This is not meant to be offensive in any way to JC, but if we are going to accept he is infallable we should be referring to him as "The JC". Any scientist can err, be tricked by erroneous findings etc. (I am NOT saying his findings are erroneous, just that this is a possibility)

2. SE has said, hang-on, this may be real or alternatively a measurment artefact. Now, I acknowledge he can do this a little abrassively at times, but I don't think I would hang him for this just yet.

Science would suggest that we:

1. Find someone with the same equipment and see if this is repeatable.

2. Find someone with "better" equipment and see if this is repeatable.

OK now, for those of us who have an interest in reality here. JC has delineated the performance of his equipment adequately (I think).

JC and Charles, do you agree that Bruno's system is able to measure at least to the same level as John's system?
If not, why not?
If not, what specifically would need to be changed so that it was?

Also, there must be more people "out there" with this kind of equipment. Can we find them and get them to make measurements.

OK, they will not be measuring the same cables as JC/Bruno, but the first step is to see an "effect" ..... we can work on exact categorisation etc later.

mark

PS: SE, how "friendly" were the folks at Sound Technology? Do you think we could interest them in doing this as it is "at the edge" of their systems performance????
 
I find that 'moving targets' make sense. Let me explain to the rest of you.
You have heard of 'dirty contacts' haven't you? Well, I seem to be able to measure them. I start with some adaptors that I must use to make the test possible. I clean them at first, BUT over time they seem to get 'dirty' I clean them again, and the extra distortion goes away. Is this an impossible concept?
You also have heard of 'break-in' of cables. It seems that when I find a particularly 'bad' cable, I tend to use it for testing more often, sometimes accidently leaving it in the machine running with a test signal for days. You might ask, am I not paying attention? Yes and no. Sometimes I turn off everything EXCEPT for my ST analyzer. It likes to be on all the time, and with something connected to it. If I forget to replace the test cable with a reference, the test signal will continue to flow through the cable until I go back, which might be days later. Just last week, I accidently left SE's steel leaded cable in the analyzer, overnight. Well, overnight was not enough to change it much, so I can still use it for testing. However, many of my original RS cables that much is still made of, have changed for the better. Darn, but I still have one RS cable that measures pretty poorly, and I have yet other examples of cables.
 
Hello -

I was wrong about the ability of the AP to perform an FFT spectrum analysis on the residual output coming out of the analog notch filter, and Steve Eddy was right.

While I was verifiying that this was indeed possible, I decided to try and run a quick "Curl-type" test. Now the AP only has a choice of XLRs and bananas for I/O. Rather than complicate things by adding adapters, I chose to measure an XLR cable. Since I don't have a large selection of XLR cables, I decided to compare the signal through the cable against the direct output of the distortion analyzer (an internal bypass, presumably through relays). Here are my results of a single test run, done at 2.0 Vrms @ 1.00 kHz:

Frequency L Cable R Cable L Gen R Gen
1.00k -120.75 -120.72 -120.56 -120.55
2.00k -120.11 -120.12 -120.09 -120.55
3.00k -134.07 -134.04 -133.62 -133.55
4.00k -142.18 -142.42 -145.54 -145.60
5.00k -139.69 -139.88 -142.14 -142.23
6.00k -149.55 -149.71 -151.12 -150.73
7.00k -146.34 -146.22 -144.13 -144.11
8.00k -145.75 -145.90 -147.24 -147.16
9.00k -145.46 -145.43 -150.57 -150.79

(Please excuse the difficulty of reading this table. I don't know how to insert a tab, and the HTML apparently parses multiple spaces down to just one.)

(The number reported at 1.00 kHz shows the depth of the analog notch filter.)

There are a couple of interesting points to be seen. The first is that the two channels track each other quite closely. This would tend to support the idea that the measurements are relatively repeatable. The next thing to note is that there are significant differences to be seen. For example, the 9th harmonic is roughly 5 dB higher when going through the cables.

At the same time, this is hardly a definitive test. One odd thing that pops up is that some of the harmonics are *lower* when the signal is going through the cables. So perhaps there is a random component to the measured values, although the comparative consistency of the two channels would tend to negate that possibility. Also, it would presumably be more fair to have two different sets of cables rather than comparing a cable to a bypass.

At any rate, this is a tantalizing beginning. There are a lot of other things that I could do (such as try to figure out if I can average multiple runs to lower the noise floor), but currently I am suffering from the flu. I am heading home for the day to try and recover.

Charles Hansen

PS -- In my test setup it was important to turn the computer's CRT monitor completely off when running the test. Leaving it on raised the noise floor significantly, obscuring the low-level harmonics.
 
Re: Oh, for crying out loud ......

mefinnis said:
For the record, I don't know SE, I neither like nor dislike him. He reminds me of a "fixated terrier" with a rag. This is not to be insulting ..... it is actually a characteristic we look for in trainees. "Obsession is a virtue."

Not taken as an insult.

Though I am a bit curious... trainees for what exactly? 🙂

By the way, with regard to your previous request:

You may well be absolutely correct in what you say to Charles ...... for the sake of the rest of us, could you verify Bruno's set-up for Charles.

I just received Bruno's reply a moment ago.

I'd sent him the following commentary from Charles:

Actually I think Curl's method is rather simple and easily reproduced.

1) Use a high-quality, low-distortion analog oscillator.

2) Run the signal through the DUT.

3) Notch out the fundmental tone with an analog notch filter.

4) Make multiple time-windowed digital readings of the resultant residual output.

5) Perform an FFT on the averaged signal to generate a spectrum of the residual distortion.

However as near as I can tell, Putzeys has not done this. Based on what I've read on this forum and in the AP manual, I think what Putzeys has done is to:

1) Use a high-quality, low-distortion analog oscillator.

2) Run the signal through the DUT.

3) Convert the analog signal to digital.

4) Notch out the fundmental tone with a digital notch filter.

4) Make multiple time-windowed digital readings of the resultant residual output.

5) Perform an FFT on the averaged signal to generate a spectrum of the residual distortion.

If this is true, the problem is in step #3 where the analog signal is converted to digital. The ADCs used in the AP have insufficient resolving power to replicate Curl's findings.

Of course, I could be wrong. I am waiting for Eddy to confirm in an unambiguous way Putzeys' setup.


Here is Bruno's reply:

The AP2 certainly notches out the fundamental in the analog domain before sending it over to the digital analyser. That's the big deal about Audio Precision products ever since the first Dual Domain came out. It's been their main selling argument since ever. Of course, you *can* tell it not to - I mean if you really want to cut your finger no good knife will stop you (actually this option is only used to measure the onboard AD itself). Even more interestingly, while the AP has a digital notch filter capability (to directly read THD+N figures from A/D converters), it has no way to feed the signal from that notch filter into the FFT analyser. It's one or the other, but both DSP programs can't be run at the same time and you certainly can't pipe data between them. It means the experiment the poster suggests can't even be performed.
It is quite clear that this poster has *no* idea what he was talking about. Suppose the signal did get converted fundamental and all. Once you have the signal in the digital domain, there is no point in removing the fundamental. You don't improve the resolution of an FFT by removing the fundamental. An FFT of a data set representing a measurement and an FFT of the same data set with the fundamental removed are perfectly identical - apart from the fundamental. There are no more distortion mechanisms to worry about once digitisation is done (discounting numerical effects at -190dB).
This is why FFT plots showing the distortion performance of D/A converters always (should) have the fundamental removed: to prevent the analyser's AD from dominating the measurement. Plots showing distortion performance of A/D converters always have the fundamental still in.


So, that's out of the way.

se
 
Hi,

John F,

I also, like Eric's idea of making a "directionless" cable--if nothing else the parallel conductors lowers the resistance.

It doesn't lower the resistance any more than twice the gauge would, if there were to be more resistance one way or the other, it would be easy: measure and arrange accordingly.

What I assume Eric does is determine the directionality of a piece of wire and from that decides the direction of the send and return wires in his star quad configuration.

He's happy with that, so I'm happy for him...

Just running a pair of wires in parallel will give you more problems than you're trying to solve.
A star quad doesn't measure the same as a twisted pair on all parameters concerned and directionallity is not going to alter those facts.

As far as I'm concerned, I don't care all that much about the directioanlity of a piece of wire, once it's run in I don't hear a difference untill I reverse the direction. Why would I want to do that in the first place?
What I do think happens is a conditioning of the wires according to the way they're run in.
More later, perhaps.

Millwood,

to give you one data point: Steve has contributed to this forum infinitely more than you have.

Do you have proof of that? I seriously doubt it for I know Fred has contributed in a big way and still does...With Fred you have to know what he's on about though...

It may take some effort but let me assure you he knows more than he's letting on lately.
In short, I know where's coming from, understand what he means but don't expect too many other members do.

This is not a personnal attack against anyone, just my observation of what I consider facts.
Which, in anyone's book is a subjective one....

So it was Al Gore inventing the internet or was that a subjective observation as well?

Cheers,😉
 
Try this ...
 

Attachments

  • charles_table.gif
    charles_table.gif
    3.9 KB · Views: 207
SE asked ...Though I am a bit curious... trainees for what exactly?
Sorry, I am an examiner for the Australian+NZ College of Anaesthetists. I work fulltime in Intensive Care and spend a lot of time in teaching/training.

While the finer engineering points here pass me by, I have a deep-seated interest in scientific method.

cheers
mark
 
Charles Hansen said:
I was wrong about the ability of the AP to perform an FFT spectrum analysis on the residual output coming out of the analog notch filter, and Steve Eddy was right.

Thank you, but you could have left out the "I was wrong" and "Steve Eddy was right." That just feeds this notion that it's all about scoring points for me. I'm no more interested in your saying "I was wrong" than I am your saying "Steve Eddy was right."

When it comes to getting at the truth of something, let's keep things impersonal. Something either is the way it is or it isn't. And whichever way it ends up being, it doesn't end up being that way because someone said it is or it isn't. Or because someone was right or someone was wrong.

se
 
fdegrove said:
It doesn't lower the resistance any more than twice the gauge would, if there were to be more resistance one way or the other, it would be easy: measure and arrange accordingly.

What I assume Eric does is determine the directionality of a piece of wire and from that decides the direction of the send and return wires in his star quad configuration.

He's happy with that, so I'm happy for him...

Just running a pair of wires in parallel will give you more problems than you're trying to solve.


Frank,

I understood it as simply running a wire A to B and then a second wire B to A. And yes, at minimum, parallel resistance, etc.

I'm waiting for the results of the cable directionality test before going any further...

However, why would parallel wires give more problems?

Thanks.



Millwood,

How many threads have you ended, or triggered the end?


JF
 
Status
Not open for further replies.