Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
They do see the same signal? Are you sure about that?? :)
I think you'll find that only at one single frequency do the woofer and tweeter receive exactly the same signal.
Yet another person who doesn't understand the concept. My goodness.
They never receive the same signal at any frequency. At the frequency where the amplitudes are equal, the phase is different.
And I don't know why you insist on pushing the crossover analogy in this thread. It really is irrelevant. Your point seems to be that two interconnects hanging off a switch box makes two parallel LCR networks, and hey, a crossover has two parallel LCR networks too! Therefore they must be comparable. Well, no they are not. The crossover is designed to be a frequency dividing network with cutoffs in the middle of the audio band. The interconnects are not. Neither is the switch box. And the source impedances are vastly different, and the loads are vastly different too. There is literally nothing comparable in your analogy.I suggest to read up on voltage sources.
Also read up on the specific requirements for this test called out by the engineers who have advocated it. You do need to satisfy some requirements for it to be valid.
The primary requirement is: for this "separation" of networks to occur and a valid A/B comparison to result, you do need to have a real-world source that is pretty close to a theoretical voltage source. For interconnect testing something like a zero ohm headphone amplifier would work well. For power amplifier testing, a solid-state amp with very low output impedance. If you satisfy that requirement in the conventional two-way speaker system example I mentioned, you'll find that disconnecting either the woofer/network or the tweeter/network individually will not alter the operation of the other in a significant way. When both are connected, the two legs will have minimal/no interaction with each other.
In either case, the characteristics will lump.....but only as seen by the source....which should be a voltage source. In that case, regards the loads, the networks will operate as if the other were not connected.
This is basic stuff fellas.
That's totally irrelevant because the output stage of a source component or pre-amp is far, far from an ideal voltage source. Typical output impedance values for components with solid state output stages vary from a few 10s of ohms to a few hundred ohms. For tube gear it's typically a few hundred up to a few thousand. And these impedances are reactive too, with output transformers and/or caps around the output for isolation, DC blocking, and filtering.
Let's go back to basics. An ABX test is a statistical hypothesis test. The hypothesis is that two different line level interconnects produce audibly different results during music playback from an audio system. A valid ABX test of the hypothesis can only be performed if the design of the experiment preserves or duplicates the conditions under which the hypothesis could be true. If your experiment depends on using a source that behaves as an ideal voltage source and isn't representative of a typical audio source component or pre-amp, then you aren't testing the hypothesis.
Using a typical audio pre-amp driving a typical audio power amp, you can't hand wave away the fact that this particular test setup makes the parallel capacitance of A the same as the parallel capacitance of B which equals the sum of the two interconnects. The open-circuited interconnect hanging off the switchbox may also be conducting RFI into the circuit under test. Even the switchbox itself is a problem. The mere presence of the switchbox in the circuit could invalidate the test unless its impedance is a couple orders of magnitude less than the cables under test. And you would have to be sure that the switchbox doesn't introduce a path for RFI or EMI to enter the circuit, and the switchbox has to preserve the relationship between the grounds of the source and sink components.
The QSC and AVA ABX comparators, for example, are essentially passive pre-amps. When you put one of those things in the circuit, you are not testing the original hypothesis.
You've highlighted some valid qualifiers of the Sanders test in question. (I noted those as well three years ago when this was mentioned by "E-Stat" on the other forum.)
Yes indeed there are some issues with the A/B testing as Sanders as outlined it, but what I was trying to do then (as now) is give Roger the benefit of the doubt regarding his thinking on the basic validity of the concept. I'm not the originator of this, but I do understand what Roger is getting at.
I was then elaborating on his scheme with things that would make it much more valid. But by this time, the heels were well dug in by other posters. :)
The speaker crossover analogy was not the best, but I was trying to highlight a similar scheme that might be more understandable for E-Stat. He had suggested awhile back that for any common source split path network configuration all the characteristics would lump and yield identical signals at both destinations. (This is the basic premise of his whole rejection of the Sanders scheme.)
In the speaker analogy, yes indeed, even at the one frequency where the low-pass and high-pass filters are exactly the same amplitude there 'may' be a phase difference. But the "they do see the same signal" statement posited by Bob Rex is clearly incorrect.
I will stipulate this whole concept is tricky for some to get their heads wrapped around. Circuit analysis here and how some parameters lump is more complicated than it might seem.
The rest of your post regarding ABX testing clouds the issue, a bit. Typical audio sources not being theoretical voltage sources...non-used open-circuited interconnects...switchbox concerns, etc, etc, are all qualifiers that may invalidate the scheme. All of these (and others) I already noted three years ago.
Let me reiterate again....I have NEVER said there aren't some possible issues with the Sanders interconnect testing scheme. :) But unlike others, I don't completely reject it out of hand because of a few issues.
If you haven't already, check out the original paper from Roger Sanders and you decide how misguided his thinking is on this topic.
Thank you for the reasonable response.
My point of view on this topic is that when it comes to ABX'ing audio cables, the only way to make sure you're not introducing other variables into the audio system is to swap the cables, which is tedious and makes it impossible for the listener to utilize short term auditory memory to make the comparison. I can't think of any easy way to allow quick switching without introducing something into the circuit whose effect on the circuit performance is on the order of the cables themselves, if not greater.
Some time ago, I recall reading the comments of Frank van Alstine who insisted his second generation ABX comparator was completely transparent and suitable for cable comparisons because it had no active components or capacitors in the signal path, just a network of relays for switching and attenuators for volume control (he had gotten rid of the cap coupling in the second gen). The box contained a power supply, and digital control and display circuitry, and his single ended circuitry introduced a whole new ground path. It was essentially a digitally controlled passive pre-amp.
The underlying basis of the hypothesis is that differences in the relatively small impedances of audio cables and connectors, possibly combined with differences in shielding, or other construction details, is enough to make them audibly different to certain individuals in certain systems. Regardless of whether or not you believe this hypothesis to be true or false, in order to test it you have to preserve the details under which the hypothesis could be true. That is something that Frank and many other ABX proponents seem to have a mental block about.
In the case of Sanders, it could be that he just didn't think through the details of his proposal. Or it could be that he assumed a-priori that the cable capacitance couldn't matter, and therefore he didn't have to control for it.
Good points. Obviously you are correct that to 100%, absolutely, fully, maximize, etc, etc, control an A/B test like this, you would not parallel the interconnects onto a single source....without some facility to completely switch out the unused one.
I have talked with Roger a few times, but not regarding this interconnect cable test scheme. So, I'm not really sure what his specific thinking was/is.
Regards lumping of characteristics: If we really want to get down into the weeds on this, the capacitances of two, simultaneously connected interconnect cables do not strictly appear in parallel with each other. Even a simplified equivalent circuit would illustrate this. There are multiple series and shunting elements to consider.
Regards possible ABX testing and gadgets to achieve it, it really is a can of worms with numerous potential pitfalls. I'm well aware. :)
> I have NEVER said there aren't some possible issues with the Sanders interconnect testing scheme. :) But unlike others, I don't completely reject it out of hand because of a few issues.>
Not possible issues - ISSUES.
You don't reject it, even though it is flawed, because it fits your biases.
So if you don't control ALL of the variables in a scientific test, how is it a valid scientific test?
Do you understand the scientific method?
Your test is pseudoscience - a parlor trick to make audible differences disappear.
So it's MY test now?? :)
I'm not the one that came up with this scheme.....let's not lose sight of that.
AND you've identified by biases too. Excellent! :)
This post is made possible by the generous support of people like you and our sponsors: