-
I came across a guy called Ethan Winer lately and found one of his video on cable tester https://youtu.be/ZyWt3kANA3Q?t=1 interesting. This guy has a huge fan base in "science review" circle on audio equipment.
He claimed that he designed a perfect method to show audio cables are nonsense, with a device he called null tester. Well, let's set cable part aside and focus on his tester.
This is his signal chain: Sandisk sansa clip's Headphone out --> Signal splitter that duplicate one channel to two output --> Null tester (which would take the signal and reverse phase of one channel, then stack up with the other) --> 80/100dB gain --> speaker.
So, in brief, this mimics active noise cancelling. Seems just fine and he showed it in video.
I think there is an obvious problem in his practice. He used a source that only does ~65dB in SNR. Despite his equipment is 110dB in SNR, I don't think he could ever achieved a detection resolution higher than his source is ever capable of. However nobody ever questioned that.
I'd like to know what you guys think of it. Cheers!
Now we closely examinine his circuit (he didn't post all). Fig. 1 is what he show to illustrate his idea on this null tester. Fig.2 is from his actual circuit https://audioxpress.com/article/you-can-diy-building-a-null-tester-device on his "null control".
His null control is rather... funny. I actually did a spice simulation on it (fig.3). It seems his null control can kill a 0.005V difference in the inputs, which is -43dBu before his 80dB gain. I didn't go higher because of time. That will be damn loud after 80dB gain!
My amateur mind leads me to think that his circuit doesn't work as intended.
-
What he really needs is a "differential amplifier" or "instrumentation amplifier".
Two easy tests I can think of for audio cables are frequency response and interference immunity. The former is trivial to do with a VNA and I would not expect much difference in the audio band. The latter would involve a spectrum analyzer or scope and some kind of repeatable noise source including a repeatable way to couple it to the cable under test, and there is very likely to be significant differences in cable quality. That said, for line level audio (which is becoming less and less common in consumer use thanks to the increasing use of digital audio), it's hard to go wrong with quality coax or STP. -
That's exactly what I am thinking. I still haven't figured out why he proposed his current design in the first place.
-
...
Tunning range 22/ 10422 = 0.002110919, it's simply means that his circuits would not be capable to "null out" any difference higher than this value. There is no constrain for lowest value. Classic bridge configuration, so I'd replace 100 OHm res for exact same 200 OHm pot, adjusted to the middle and left untouched.
His null control is rather... funny. I actually did a spice simulation on it (fig.3). It seems his null control can kill a 0.005V difference in the inputs, which is -43dBu before his 80dB gain. I didn't go higher because of time. That will be damn loud after 80dB gain! -
Thanks for the input, I wasn't trying to get the limit here. As he claimed his device has a noise floor of -110dB, which is the minimum detection limit of his experiment.
-
Immense overkill, when all you have to do is compare the input and output of the cable with an oscilloscope and a spectrum analyzer (both 2-channel of course).
Not in everybody's kit, but apparently this Winer guy is so long in the business that he should have those.
-
I agree. I am going to try out using a scope next week.
In addition, I think there is an obvious problem in his practice. He used a source that only does ~65dB in SNR. Despite his equipment is 110dB in SNR, I don't think he could ever achieved a detection resolution higher than his source is ever capable of. However nobody ever questioned that. -
In addition, I think there is an obvious problem in his practice. He used a source that only does ~65dB in SNR. Despite his equipment is 110dB in SNR, I don't think he could ever achieved a detection resolution higher than his source is ever capable of. However nobody ever questioned that.
Yes, you wrote that before, but I think you are mistaken. It does not matter what signal/noise that source has, since it gets split into equal parts for the two channels under test, and only the difference between those parts is what matters.
You could start with a white noise source, i.e. S/N ratio of zero, and the test would work nicely. If the two signal paths are identical, the two noise signals cancel out in the end, and you have demonstrated that the paths are identical (which is the point of this test).
Where you do need good quality components is in the parts of your test jig which are not identical between the two channels. A good differential amplifier, or inverter and adder, is important: You want to make sure that any measured differences between the two channels are indeed down to the transmission channels, and not due to your way of taking the difference. You can easily verify this by testing with two know-to-be-identical signal paths (e.g. two identical shielded cables). -
In addition, I think there is an obvious problem in his practice. He used a source that only does ~65dB in SNR. Despite his equipment is 110dB in SNR, I don't think he could ever achieved a detection resolution higher than his source is ever capable of. However nobody ever questioned that.
Yes, you wrote that before, but I think you are mistaken. It does not matter what signal/noise that source has, since it gets split into equal parts for the two channels under test, and only the difference between those parts is what matters.
You could start with a white noise source, i.e. S/N ratio of zero, and the test would work nicely. If the two signal paths are identical, the two noise signals cancel out in the end, and you have demonstrated that the paths are identical (which is the point of this test).
Where you do need good quality components is in the parts of your test jig which are not identical between the two channels. A good differential amplifier, or inverter and adder, is important: You want to make sure that any measured differences between the two channels are indeed down to the transmission channels, and not due to your way of taking the difference. You can easily verify this by testing with two know-to-be-identical signal paths (e.g. two identical shielded cables).
I don't get it. Say if we put a terrible source that only output noise into the test jig. We will still get noise out of the jig, even if we assume that cables make a difference. Does that make sense?
So when would you call there is a real difference? What's your boundary condition? -
I don't get it. Say if we put a terrible source that only output noise into the test jig. We will still get noisy out of the jig, even if we assume that cables make a difference. Does that make sense?
If it doesn't matter, it says that you are able to make a valid measurement of 0.0001V with a device that displays only 0.01V
You take a signal (any signal -- audio from your MP3 player, white noise, whatever) and split it into two equal parts. You send one part through cable A. You send the other part through cable B. At the outout ends af cables A and B, you let a differential amplifier take the difference between the two signals which have passed through A and B, respectively.
If the signal propagation through A and B is identical, that difference will be zero. That is true no matter what test signal you sent in.
Your difference amplifier (or output bridge) must be decent, and you must have a decent way of measuring low-level difference signals. But your signal source does not matter. -
I don't get it. Say if we put a terrible source that only output noise into the test jig. We will still get noisy out of the jig, even if we assume that cables make a difference. Does that make sense?
If it doesn't matter, it says that you are able to make a valid measurement of 0.0001V with a device that displays only 0.01V
You take a signal (any signal -- audio from your MP3 player, white noise, whatever) and split it into two equal parts. You send one part through cable A. You send the other part through cable B. At the outout ends af cables A and B, you let a differential amplifier take the difference between the two signals which have passed through A and B, respectively.
If the signal propagation through A and B is identical, that difference will be zero. That is true no matter what test signal you sent in.
Your difference amplifier (or output bridge) must be decent, and you must have a decent way of measuring low-level difference signals. But your signal source does not matter.
What's your boundary condition? Like I said, if you use noise as source, whatever you put into the jig will only output noise as well. So you got noise out isn't a valid evidence that you have equal signal.
Your current source is a terrible source since it has way lower resolution than the testing jig. We don't really know what caused the noise in your output.
I agree that a valid metric would be the output amplitude after nulling. If you got a significant increase in amplitude versus idle, you are seeing a difference. Your bondary condition would determine how much increase is valid as a real difference.
But for now, your conclusion was drawn when you got noise, which is not a proper judgement.
I am not going to argue that cables are equal or not. I am interested in your methodology.
-
kris2014, either you are missing the point here, or maybe I am.
In my understanding, the whole approach here is a differential measurement, i.e. you compare one cable with another one and look at the difference of the signal only. If there is no difference when comparing some stupid "high end" cable with a standard one, you have proven that the "high end" claim is bollocks.
As I have tried twice to explain in the above posts, you can run this comparison with any old signal, and hence any old signal source. If there is noise in the signal, the two signal paths will both faithfully transmit the same noise, and it will cancel out at the output (with high precision).
Anway, I have tried to expain this threee times now. If this still makes no sense, I better let someone else have a try. -
kris2014, either you are missing the point here, or maybe I am.
In my understanding, the whole approach here is a differential measurement, i.e. you compare one cable with another one and look at the difference of the signal only. If there is no difference when comparing some stupid "high end" cable with a standard one, you have proven that the "high end" claim is bollocks.
As I have tried twice to explain in the above posts, you can run this comparison with any old signal, and hence any old signal source. If there is noise in the signal, the two signal paths will both faithfully transmit the same noise, and it will cancel out at the output (with high precision).
Anway, I have tried to expain this threee times now. If this still makes no sense, I better let someone else have a try.
In your video, after nulling you still have a strong signal in terms of amplitude versus your background. That said, you have not null out the difference after all. If they cancelled perfectly, why would your output signal several times stronger than your background?
In your video, you never showed minimum level of output amplitude you could trim down to. -
2 observations:
- around 18min. of the video he highlights the quality and the importance of the quality of the cables he built, to intend to demonstrate in the end that cable differences are placebo. I'm I getting this right?
- "puff"? Never heard that. Is that a short for "Hufflepuff"? To much Harry Potter I guess. Sorry.
1 question:
If we take some sellers voodoo in one input, and this guy voodoo on the other input, of a differential amplifier... will it null? -
In your video, after nulling you still have a strong signal in terms of amplitude versus your background. That said, you have not null out the difference after all. If they cancelled perfectly, why would your output signal several times stronger than your background?
Which part of the video are you referring to? I see him demonstrating and explaining, e.g. at 25:00, that the residual noise of the difference signal is at -20dB. That's after 80dB of amplification, so -100dB on the scale of the original input signal.2 observations:
- around 18min. of the video he highlights the quality and the importance of the quality of the cables he built, to intend to demonstrate in the end that cable differences are placebo. I'm I getting this right?
- "puff"? Never heard that. Is that a short for "Hufflepuff"? To much Harry Potter I guess. Sorry.
"Handling noise", i.e. microphonic effects due to capacitance variations as you move the cable around, is a real thing; and he explains why he wants to minimize it for his demonstration.
And while I dislike the "puff" pronounciation for pF, it is a common colloquialism -- I am sure I have heard Dave use it multiple times in his videos. This video's author realizes that it warrants a "translation", and has hence put a footnote into the video. What's wrong with that?
Move on, guys, there's nothing to "debunk" here. -
Please read this IEEE article https://ieeexplore.ieee.org/document/7546717. This particle is on thermal noise over power splitter. It has pointed out that on the outputs of a Y shaped 2-R splitter, thermal noise is anti-correlated, meaning they will have different phase.
In the case of white noise and other types of random noise, due to their random nature, they are uncorrelated with anything. Therefore, after they passed through a splitter, they won't be cancelled in a phase reversal or in a phase detector.
I bet if you hook up a good white noise generator to your tester, you will never be able to null it out. -
Please read this IEEE article https://ieeexplore.ieee.org/document/7546717. This particle is on thermal noise over power splitter. It has pointed out that on the outputs of a Y shaped 2-R splitter, thermal noise is anti-correlated, meaning they will have different phase.
In the case of white noise and other types of random noise, due to their random nature, they are uncorrelated with anything. Therefore, after they passed through a splitter, they won't be cancelled in a phase reversal or in a phase detector.
I bet if you hook up a good white noise generator to your tester, you will never be able to null it out.
Sorry, mate, I think you are still confused about the input signal on one hand, vs. what happens to the two parts of the signal along the split path and difference amplifier on the other hand.
Yes, there will always be independent noise contributions affecting the two separate paths, and that will limit the ultimate perfection of your cancellation. In the case of this video author's jig, that's the -100dB to -110dB limit he is talking about.
No, the noise level of the input signal does not matter. That's the -65dB red herring you are chasing here.
-
I honestly never heard or had conscience of the term "puff".
Never read it in a book, but now I realize that it's a thing. There's even puff converters online: https://www.convertunits.com/from/puff/to/picofarad
Always learning...
About "Sir Null", to me he's not much different from the sellers he is trying to ridicularize, just in the opposite direction.
If he is trying to sell something to prove that the cables doesn't matter, his concern about his cables is a contradiction. If any sort of effect (capacitance, EMF, microphonics...) his acknowledge to be a concern, then one must understand that some with lot's of cash to spare, are in they're own right to be concern about microphonics induced in his speaker cables, by the wife walking on heals around the house, and so, place some stands along the cables.
Listening, is everything.
If one can't tell the difference, go for the cheapest or for the ones that goes best with rest of the house decoration.
Seriously. I have Klotz speaker cables in blue, and painted the wall behind the speakers in petrol blue. I like it
-
Kris, I think what ebastler is trying to suggest is this kind of setup:
The input signal is connected to the two different cables being compared, and fed through them to the inputs of a differential amplifier. If the cables have the same transmission characteristics and don't change the signal, then the differential map will have equal signals on both its inverting and non-inverting inputs, and there will be no difference to amplify. It really doesn't matter what the input signal is, as long as it can be passed through the cables without undue effect on it, because all you're going to be measuring is the difference between what emerges from the cables. If one of them changes the signal, then the outputs will no longer be the same, and the amplifier will magnify whatever difference is there, showing that the cable(s) ARE affecting the signal.
Hope this helps clarify.
-Pat