Charles Hansen said:
Bruno's not using a System One Dual Domain. He's using a System Two Cascade.
Sure.
An externally hosted image should be here but it was not working when we last tested it.
This is of the cheap A/V cable. Generator set at 1kHz, 30mV, just as John used when he made his Mac The Scope plots. The dB scale is relative to the 30mV fundamental. The FFT analyzer is set to synchronous averaging which allows measurement significantly below simple power averaging, which gives this result:
An externally hosted image should be here but it was not working when we last tested it.
Even this is measuring below where John says he's measuring. If you take where John says he's measuring and overlay it on the first plot, it would look like this:
An externally hosted image should be here but it was not working when we last tested it.
se
Gee ...... I hope Bruno isn't listening!
Have we invited Bruno to actually join the discussion directly, or are we simply going to have a relayed report .... ???
I cannot profess to even remotely understand the finer technical points in this thread, or it's predecessor, but it is nice to see the thread come back to a straight technical discussion.
mark
Steve Eddy said:
Steve,
You have conveniently included the phrase "where John says he's measuring". Now I haven't been following this thread, so I don't know what Curl claims in this regard. But if you stop for a moment and look at the situation you will realize that the graph you posted is incorrect.
First of all, please consider the residual distortion of the ST versus the AP. If you go to the AP website you can see that the distortion spec for the System Two Cascade is:
Residual THD+N
20 Hz-20 kHz </=(0.0004% + 1 mV), 22 kHz BW [-108 dB]
</=(0.0006% + 2 mV), 80 kHz BW [-104 dB]
</=(0.0015% + 6 mV), 500 kHz BW [-96.5 dB]
10 Hz-100 kHz </=(0.0040% + 6 mV), 500 kHz BW [-88 dB]
Now go to the ST website, where it is claimed that the 1700 series measures distortion down to 0.0009%. Unfortunately, there is no additional information regarding the measurement conditions. Therefore we cannot directly compare the exact performance of the two machines.
However, can see that they are in the same ballpark. Depending on the test conditions of the AP, in some circumstances it is slightly *better* than the ST's spec, and in some circumstances it is slightly *worse* than the ST's spec.
Like I said before, the AP is a wonderful instrument. So is the ST. Audio Precision has *not* made some tremendous breakthrough in analog distortion measurements since the ST was designed. (In fact, the last breakthrough I can remember in analog distortion measurements was when HP developed the 200 series oscillators. That was a *long* time ago.)
And just to complicate things further, Curl has modified his ST to reduce the residual distortion. So it is quite likely that Curl's setup has a *lower* residual than your highly vaunted AP. However, we cannot ascertain this at the current time. Suffice it to say that the distortion residual of the generators is roughly comparable.
Next we turn to the analyzer side. While it is true that the System Two has some additional capabilities over the System One, I still don't believe that it can (by itself) duplicate the measurements that Curl has made. I have downloaded the user manual and failed to find any evidence that it can. In fact, everything I have read points to the opposite, that it *cannot* duplicate Curl's measurements. Now I may be wrong about that, as I haven't read the full 720 (!) pages in great detail, nor have I any experience with that particular machine.
But the graphs you have posted from Putzeys' work leads me to the conclusion that he is absolutely *not* resolving as low as Curl is in his measurements.
Simply put, it is clear (based on the above discussion) that your red overlay of Curl's measurements should be moved down at least 20 dB relative to Putzeys' measurements. When that is done, you will see that Putzeys' measurements are *not* adequate to resolve the effects that Curl is seeing.
Charles Hansen
Steve Eddy said:
"Speculation in the absence of data is a capital mistake." When you're talking about signal levels 6 orders of magnitude down from 30mV, there's no shortage of possible sources for error or unaccounted for variables. Including the source that John claims, the wire itself.
John is making some extraordinary claims, especially the one about being able to distinguish wires which have carried signal from those which haven't. I haven't seen any extraordinary proof, just some informal measurements. In the scientific world where I grew up, the first thing one does when getting results like this is to consult with peers, bring other experts into the picture to try to eliminate error sources or unaccounted for variables, publish full and complete procedures for people to replicate, and in general, tighten up the protocols. It's certainly the least that I would do, but I'm not an engineer, so what the heck would I know.
Originally posted by SY
"Speculation in the absence of data is a capital mistake." When you're talking about signal levels 6 orders of magnitude down from 30mV, there's no shortage of possible sources for error or unaccounted for variables. Including the source that John claims, the wire itself.
Sure, we are talking about micro levels and extreme care is needed when making these kinds of measurements as I am sure John (and Bruno) have.
John is making some extraordinary claims, especially the one about being able to distinguish wires which have carried signal from those which haven't.
Yes that one caught my eye too - if Johns measurements are proven correct then this is interesting indeed.
I haven't seen any extraordinary proof, just some informal measurements.
Sure, but we are working on establishing a proof (or experimental errors).
In the scientific world where I grew up, the first thing one does when getting results like this is to consult with peers, bring other experts into the picture to try to eliminate error sources or unaccounted for variables, publish full and complete procedures for people to replicate, and in general, tighten up the protocols.
We are getting there.
It's certainly the least that I would do, but I'm not an engineer, so what the heck would I know.
You mean you know every thing about bugger all, and bugger all about everything ? .........
I have some comments/questions .............
In the graph of JC's measurements there is no 4kHz residual ???.
The graphs contain both even and odd order harmonic content.
This suggests both symmetrical distortion (peaks compression)and non symmetrical distortion - ie directionality - http://www.dwelle.de/rtc/infotheque/qual_parameter/qualpar_04.html
Connectors have reputation for sounding different, and according to what I have read it seems that variable connectors (Rca ?) are being included in the measurements.
For this reason (elimination of one of the variables), when I get the direction test cables from Stuart, I will be listening to bare ended cables/wires connected to a fixed connection setup - ie solder on test or some kind of quick connect so that the only variables are the wire and absolute time.
In the name of correctness, I reckon that this kind of approach needs to be used/stated in further testing of extreme low levels of distortions in cables due to wires and/or dielectrics.
I understand that John has gone to these extreme measures in order to more clearly quantify the sonic differences in cables that he is hearing/heard.
If these kinds of measurements can be proven, then all of this discussion is indeed interesting.
That these measurements have not been proven conclusively previously only indicates that properly sensitive measurements have not been performed and collaborated yet by independant parties.
I look forward to further scientific collaboration and resoloution.
Eric.
"Speculation in the absence of data is a capital mistake." When you're talking about signal levels 6 orders of magnitude down from 30mV, there's no shortage of possible sources for error or unaccounted for variables. Including the source that John claims, the wire itself.
Sure, we are talking about micro levels and extreme care is needed when making these kinds of measurements as I am sure John (and Bruno) have.
John is making some extraordinary claims, especially the one about being able to distinguish wires which have carried signal from those which haven't.
Yes that one caught my eye too - if Johns measurements are proven correct then this is interesting indeed.
I haven't seen any extraordinary proof, just some informal measurements.
Sure, but we are working on establishing a proof (or experimental errors).
In the scientific world where I grew up, the first thing one does when getting results like this is to consult with peers, bring other experts into the picture to try to eliminate error sources or unaccounted for variables, publish full and complete procedures for people to replicate, and in general, tighten up the protocols.
We are getting there.
It's certainly the least that I would do, but I'm not an engineer, so what the heck would I know.
You mean you know every thing about bugger all, and bugger all about everything ? .........
I have some comments/questions .............
In the graph of JC's measurements there is no 4kHz residual ???.
The graphs contain both even and odd order harmonic content.
This suggests both symmetrical distortion (peaks compression)and non symmetrical distortion - ie directionality - http://www.dwelle.de/rtc/infotheque/qual_parameter/qualpar_04.html
Connectors have reputation for sounding different, and according to what I have read it seems that variable connectors (Rca ?) are being included in the measurements.
For this reason (elimination of one of the variables), when I get the direction test cables from Stuart, I will be listening to bare ended cables/wires connected to a fixed connection setup - ie solder on test or some kind of quick connect so that the only variables are the wire and absolute time.
In the name of correctness, I reckon that this kind of approach needs to be used/stated in further testing of extreme low levels of distortions in cables due to wires and/or dielectrics.
I understand that John has gone to these extreme measures in order to more clearly quantify the sonic differences in cables that he is hearing/heard.
If these kinds of measurements can be proven, then all of this discussion is indeed interesting.
That these measurements have not been proven conclusively previously only indicates that properly sensitive measurements have not been performed and collaborated yet by independant parties.
I look forward to further scientific collaboration and resoloution.
Eric.
mrfeedback said:
That these measurements have not been proven conclusively previously only indicates that some of us can still make baseless and anecdote-driven claims,
Seriously, I thought one only makes a claim when backed with data, facts and measurement.
Ever scince Al Gore invented the Internet 10 years ago, anyone can make claims unless they are proven false.
Charles Hansen said:
It's not incorrect regarding where John claims he's measuring. Here's what John himself said:
Steve, the notch filter in the 1700B notches out to somewhere between -94 and -100dB This can be shown by the RESIDUAL of the THD at all levels is a little less than .002% or -94dB. Of course at very low levels, NOISE makes the residual appear to increase on the meter. This residual includes the unnotched fundamental ( the highest peak on the left side at 1KHz), the averaged distortion residual, and NOISE. At low levels, noise is all important, and this is why I must use signal averaging or very narrow bandwidth filtering to remove enough noise to look at the distortion residual. If you start with my graphs with the IKHz (notch) then you can see that the noise boundary is about -125dB. The distortion occurs in this case about -115 to -120dB, or even as much as -110dB if the notch is not complete.
John says that his noise boundary is about -125dB relative to the 30mV fundamental. And that the high order distortion products are between -115 to -120 dB relative to the 30mV fundamental.
Since the dB scale in Bruno's plots are also relative to the 30mV fundamental, what I've overlaid corresponds to what John claims to be measuring.
The overlay by the way was made based on this Mac The Scope plot of his of a Radio Shack Gold interconnect.
An externally hosted image should be here but it was not working when we last tested it.
Um, that bit on the ST website is for their current model 1700 series. John's using a 25 year old 1700B which he says he's upgraded a few of the opamps in.
Are you saying that ST's current model 1700 series has in no way changed or otherwise improved over what they were building 25 years ago?
And they have three different models. The 1700 "Distortion Measurement System," the 1701 "Precision Distortion Measurement System," and the 1710 "Professional Distortion Measurement System."
They say of these:
Fully automatic nulling circuits measure distortion as low as 0.0009% in five seconds.
How do you know which of these three models that that 0.0009% refers to? Does it refer to the budget 1700, which "provides excellent performance at an attractive price" or does it refer to the 1701 which "features extremely low distortion and noise at high frequencies"?
So how on earth can you conclude that that ambiguous "0.0009%" figure has ANYTHING to do with a 25 year old 1700B?
Great. Except that John's not using any of the current production models from ST.
How can you say that when ST gives no conditions whatsoever for that 0.0009%" spec?
But what we do have are measurements made using a 25 year old ST 1700B and we have measurements made using a more recent AP System Two Cascade. And going by what John himself claims as to where he's measuring, the System Two Cascade is resolving significantly lower.
And even if one assumes that the System Two Cascade is only measuring as far down as where John's measuring, there's still no evidence of the high order distortion products which are plainly visible in John's plots.
He says he's replaced a few of the opamps. That's it.
But if John's claims are correct, it's not even close.
AP specs individual residual harmonics to be no greater than -130dB. Which is where it's at in Bruno's plots. John says in his system, noise is bottoming out at about -125dB. Which puts his third harmonic at about -115dB.
We have the FFT measurements from the System Two Cascade. We have John's claims as to what the levels are relative to the fundamental in his plots. So unless John has absolutely no clue just where he's measuring, it's rather clear that the System Two Cascade is resolving levels well below John's measurements.
It's really quite simple. The output of the oscillator feeds the DUT, the output of the DUT goes into the filter module where the fundamental is notched out, and the output of the filter module gets fed into the FFT analyzer.
It's clearly able to do it or it couldn't produce the plots that have been provided.
If that's the case, then John has absolutely no clue what he's doing or where he's measuring.
Why is it clear that the overlay should be moved 20dB down below the System Two Cascade's measurements? Based on where John claims he's measuring using a piece of equipment that he's been using for over 20 years, the overlay is right where it should be. If the overlay is wrong, then you need to tell John to learn a bit more about his 1700B because he obviously has no idea what its level of performance actually is.
se
Ok John I will open another bottle of Barolo and go back to the Xmas duties.
I have a AP portable one it may be Ok I will blow the dust off this bastard and try it together with my Card Deluxe soundcard later this week, maybe this can measure interconnect wires wich I am now very interested in maybe this can make me a fortune my new business interconnect cable maker with technical specifications on every handmade cable.
I have a AP portable one it may be Ok I will blow the dust off this bastard and try it together with my Card Deluxe soundcard later this week, maybe this can measure interconnect wires wich I am now very interested in maybe this can make me a fortune my new business interconnect cable maker with technical specifications on every handmade cable.
Steve,
Your combination of ignorance and willingness to argue over *anything* is too much for me.
1) Actually all of the ST 1700 series instruments were discontinued many years ago. The don't have a "current model" 1700 series.
2) There have been no fundamental breakthroughs in the performance of low distortion audio oscillators since the HP 200 series over 50 years ago.
3) At the risk of being repetitive, Curl's ST oscillator has roughly (say within 5 dB or so) the same residual distortion as the AP.
4) Nowhere that I can find in the AP manual does it say that unit is capable of the test you claim is being made.
5) I'd be willing to bet that Putzeys test setup is not what you claim it is. My strong suspicion is that he is instead using the digital analyzer section of his AP to perform both the notch filter and the FFT. This approach will not work for the very simple reason that the ADCs used in that device are incapable of performing at the required level of resolution.
Please confirm that Putzeys' measurement is being made in the manner you specified. You could do this in a few different ways.
a) Find the specific passages in the AP manual that even show that this is possible. (This of course wouldn't be conclusive, but it would be a start.)
b) Post a detailed description of Putzeys' setup *in his own words*.
c) The test setups for the AP can be stored as test protocols. Post Putzeys' test protocol here and I will take it to a colleague's that has a System Two and will confirm if he has in fact replicated Curl's method in at least a general way.
Until such time, I reject your claim that Putzeys' test in any refutes Curl's findings. While it is certainly possible, at this point I do not believe that the method Putzeys used has the resolving power of Curl's test.
Charles Hansen
Your combination of ignorance and willingness to argue over *anything* is too much for me.
1) Actually all of the ST 1700 series instruments were discontinued many years ago. The don't have a "current model" 1700 series.
2) There have been no fundamental breakthroughs in the performance of low distortion audio oscillators since the HP 200 series over 50 years ago.
3) At the risk of being repetitive, Curl's ST oscillator has roughly (say within 5 dB or so) the same residual distortion as the AP.
4) Nowhere that I can find in the AP manual does it say that unit is capable of the test you claim is being made.
5) I'd be willing to bet that Putzeys test setup is not what you claim it is. My strong suspicion is that he is instead using the digital analyzer section of his AP to perform both the notch filter and the FFT. This approach will not work for the very simple reason that the ADCs used in that device are incapable of performing at the required level of resolution.
Please confirm that Putzeys' measurement is being made in the manner you specified. You could do this in a few different ways.
a) Find the specific passages in the AP manual that even show that this is possible. (This of course wouldn't be conclusive, but it would be a start.)
b) Post a detailed description of Putzeys' setup *in his own words*.
c) The test setups for the AP can be stored as test protocols. Post Putzeys' test protocol here and I will take it to a colleague's that has a System Two and will confirm if he has in fact replicated Curl's method in at least a general way.
Until such time, I reject your claim that Putzeys' test in any refutes Curl's findings. While it is certainly possible, at this point I do not believe that the method Putzeys used has the resolving power of Curl's test.
Charles Hansen
Hello SY,
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.
Charles Hansen
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.
Charles Hansen
I can't comment about what Putzeys has done- I'll be interested in what he has to say. But when doing measurements of this low level, even a good procedure can still end up giving one unexpected errors. There are a lot of subtleties which don't matter when you're doing the sort of normal, gross measurement typical of audio, but can bite you in the butt when trying to push the capabilities of your equipment. That's why I keep harping on John to open up his lab to me (or someone else of similar qualification) to double check everything and try to eliminate any other error sources that he may have missed.
If we were talking about anything less than extraordinary claims that fly in the face of known physics, I wouldn't bother- I've got a life and a business to run. But either there's something funny in the test setup or there's something interesting here, and I can't tell which is which from a simple laundry list and the rather incomplete analysis so far presented. So, the half-hour drive and an afternoon playing with wires is worth it to me to try to resolve this question.
If we were talking about anything less than extraordinary claims that fly in the face of known physics, I wouldn't bother- I've got a life and a business to run. But either there's something funny in the test setup or there's something interesting here, and I can't tell which is which from a simple laundry list and the rather incomplete analysis so far presented. So, the half-hour drive and an afternoon playing with wires is worth it to me to try to resolve this question.
Hello SY,
Have you asked John to visit his lab? I think that such an undertaking would resolve a lot of questions. Thank you for making such a generous offer of your time. If John has declined your offer, please let me know and I'll see if I can persuade him to change his mind.
Charles Hansen
PS -- John already did this with another engineer (I think his name was Peter) to try an resolve some questions on the Audio Asylum forum. Unfortunately, it didn't diminish in any way Eddy's penchant for arguing.
Have you asked John to visit his lab? I think that such an undertaking would resolve a lot of questions. Thank you for making such a generous offer of your time. If John has declined your offer, please let me know and I'll see if I can persuade him to change his mind.
Charles Hansen
PS -- John already did this with another engineer (I think his name was Peter) to try an resolve some questions on the Audio Asylum forum. Unfortunately, it didn't diminish in any way Eddy's penchant for arguing.
Regarding ST do they sell the ST 1700 series analyzers? I have visit their site wich is a little bit obscure, why do they have this dark look? BTW I own a ST1000 wich is a nice instrument.
For a couple of years ago I talked to a guy who constructed the AMBER analyzers wich I think was even better then ST 1700 series.
For a couple of years ago I talked to a guy who constructed the AMBER analyzers wich I think was even better then ST 1700 series.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.