The AP2 gets a few mentions around here, even by people who swear by vinyl and tubes and not digital (no, don't worry I'm not starting those debates again!). What I would like to know is, if you don't like digital on principle (maybe you don't like the way it takes 'infinite' resolution and turns it into discrete levels or whatever), or you think you just don't like the sound of digital, whatever the technical arguments, do you think that the AP2 can capture the 'essence' of 'sound quality', even if the actual interpretation of the measurements is debatable? In other words, could we take a snapshot of a test signal acquired by the AP2 and put it into a PC and devise an as yet unknown algorithm that could measure 'sound quality', or does the digital nature of the AP2 mean it simply can't be done, end of story?
I have no problem with digital as a technology. Like everything else, it has it's own issues to understand. I don't know the unit in question, but it still comes down to to knowing what to measure, if the system has sufficient resolution, and if the algorithms are appropriate. It is not easy. The level of precision that has been brought to even mid level consumer analog is phenomenal. If we are talking 50Khz amps and worrying about IM distortion from high order harmonics, we are looking at a couple of megahertz sample rates.
Obviously, why bother with a mod which does nothing worth mentioning otherwise?
But then, none of us knows how many small steps it will take for the equipemtn to sonically disappear, do we? So we work on, step by step, hoping to reach it relatively soon.
But admit it - it's FUN!Every little step brings its own rewards.
Right. That's why it sounds weird when people demand to give them list of medical tests in order to check if the system can win all Olympic competitions.
System Twos are hybrid instruments of course, not purely digital. They have the outstanding differential analog front ends, analog fast-settling low-distortion signal generators and analog fast-settling notch filters, as well as good ADCs and DSP for extracting spectra from sets of data in the "symbol domain". They are superb instruments for determining frequency response and steady-state harmonic and intermodulation distortions, and pretty good at measuring noise considering they are general-purpose insturments. What this all has to do with how things sound is another story, but it's certainly not the digital portion of the instruments that make that determination impossible.
More likely it's that we're still not sure what to measure, after dealing with the easy and fairly non-controversial stuff.
In this interview, Mike Albinson of Quad amplifier fame alludes to a "bridge test"
MikeAlbinson
Does anyone know the details of this test? It sounds like the sort of thing that could be more easily done in software with a generic 'AP2' using real music signals - if everyone agreed that an 'AP2' can acquire with sufficient fidelity.
Edit: I know he's talking about subtracting two signals - input from amp output most probably - but what were the test signals, pass criteria etc.?
MikeAlbinson
Does anyone know the details of this test? It sounds like the sort of thing that could be more easily done in software with a generic 'AP2' using real music signals - if everyone agreed that an 'AP2' can acquire with sufficient fidelity.
Edit: I know he's talking about subtracting two signals - input from amp output most probably - but what were the test signals, pass criteria etc.?
Last edited:
CopperTop, it is about comparing the amp with a reference one. If you have such an amp the method is priceless, as priceless as a reference amp. If to combine my Tower-VII with AP2 you can test as close are say class D, or class AB amps, to the room heater that is the best that I ever heard. But it does not mean that the reference amp better than my Tower-VII can't be made.
"More likely it's that we're still not sure what to measure, after dealing with the easy and fairly non-controversial stuff."
Seems to me that is were we started 2364 posts ago.
"We shall not cease from exploration, and the end of all our exploring will be to arrive where we started and know the place for the first time."
T. S. Eliot
Hi,
I explained previously why. The AP2 is the de-facto industry standard for performing certain types of measurements. Several highly influential HiFi Magazines or reviewers use long term loaners (nice marketing ploy AP). So for many of "us" having our own AP2 is necessary self defence, as even experienced operators make mistakes and if you cannot point them out you end up with more criticism of measured performance than one would need to accept.
No.
The AP2 was designed to answer certain specific questions with a high degree of precision, repeatable and quickly.
In principle, the AP2 cannot measure any more than we could in the 1980's using needle meters, filter banks and so on, what it does do, it makes a measurement that would take many hours in seconds and it eliminates many possible experimental errors.
It is not as such the digital nature that precludes what you suggests. It rather is the nature of AP2 itself that precludes it. It is a special purpose device which does certain specific things very well, but nothing else.
Asking the AP2 to do something it was not designed for is like asking a top of the range Fluke DMM to give us an FFT of the AC on it's probles.
If the actual measurements needed are already present in the AP2 but only a new metric is needed (e.g. we substitute THD & N with the GedLee distortion metric), it may be possible to just re-program the software (either PC Software or Firmware).
Ciao T
I explained previously why. The AP2 is the de-facto industry standard for performing certain types of measurements. Several highly influential HiFi Magazines or reviewers use long term loaners (nice marketing ploy AP). So for many of "us" having our own AP2 is necessary self defence, as even experienced operators make mistakes and if you cannot point them out you end up with more criticism of measured performance than one would need to accept.
No.
The AP2 was designed to answer certain specific questions with a high degree of precision, repeatable and quickly.
In principle, the AP2 cannot measure any more than we could in the 1980's using needle meters, filter banks and so on, what it does do, it makes a measurement that would take many hours in seconds and it eliminates many possible experimental errors.
It is not as such the digital nature that precludes what you suggests. It rather is the nature of AP2 itself that precludes it. It is a special purpose device which does certain specific things very well, but nothing else.
Asking the AP2 to do something it was not designed for is like asking a top of the range Fluke DMM to give us an FFT of the AC on it's probles.
If the actual measurements needed are already present in the AP2 but only a new metric is needed (e.g. we substitute THD & N with the GedLee distortion metric), it may be possible to just re-program the software (either PC Software or Firmware).
Ciao T
Hi,
The test signals are whatever you like. It is about comparing the output of the DUT to a known device (it is not really possible to compare directly to the input).
As they allude to, even minor response differences cause a phase difference that will reduce cancellation, but if you compensate these and null loads of other stuff out, is the test still fair and reasonable?
Ciao T
The test signals are whatever you like. It is about comparing the output of the DUT to a known device (it is not really possible to compare directly to the input).
As they allude to, even minor response differences cause a phase difference that will reduce cancellation, but if you compensate these and null loads of other stuff out, is the test still fair and reasonable?
Ciao T
The nature of reality tells us that qualities can't be reduced to quantities. End of story.
<edit> @dvv : Metal film resistors in SMT aren't particularly expensive. I bought a couple of reels recently, I think they were around $60 apiece. 4,000 on a reel. The ones I measured (LCR meter) were I think better than 0.05%. That was until I overheated them with my soldering iron set to too high a temperature
Last edited:
Poking around with Google, I find an apparent description here:
http://www.angelfire.com/ab3/mjramp/distortiontest.pdf
It looks to be just what I thought, circuitry to take both the input and output and "cancel" them, leaving only the difference. A couple of R/C networks compensate for the high-frequency and low-frequency rolloffs of the amp. As one circuit shows, it's easy enough to do this with all passive components for an inverting amplifier, though apparently all amps thesedays are noninverting.
Of course, the AP2 uses the same or similar A/D converter chips as many computer audio interfaces, though presumably the AP's analog input and grounding design is as good as or better than most such interfaces.
There's that MSWindows .wav file diff program that does (almost?) the same thing digitally, mentioned earlier in this thread.
The nature of reality tells us that qualities can't be reduced to quantities. End of story.
<edit> @dvv : Metal film resistors in SMT aren't particularly expensive. I bought a couple of reels recently, I think they were around $60 apiece. 4,000 on a reel. The ones I measured (LCR meter) were I think better than 0.05%. That was until I overheated them with my soldering iron set to too high a temperature
Point taken, but please bear in mind that I am the low in the "low volume", meaning that I make and sell 50...60 units per annum of the filter, which is the core of my operation. Headphone amps, to which SMD would be pertinant, will not sell much, if at all, simply because they came about 5 years too late.
They are definitely not portable at 20 lbs and in the 19" rack format, they are not WiFi, they are neither Apple nor Android compatible, you can't use them to chat with your friends, they do not deliver 3D pictures, and you can't use them to live your life on the Internet. Hence, they were practically stillborn.
I don't mind at all, I did them mostly for myself anyway, but I'll be lucky if I sell a total of 10.
Going the SMD way, with such low volumes, is simply counterproductive.
Also, with kind of bum luck I have, I would need like 50 of those reels of 4,000, and at $60 you mentioned, that works out at $3,000, to which I would need to add shipping costs, customs duties and VAT, making the final price more something like $4,260.
So you see, while your advice is good such as it is, it's just not applicable in my case.
Last edited:
Good point, but also I think rather relative.
Remember, we cannot measure everything that is relevant, but not everything we can measure is relevant.
Increments in rising quality are, in my view, even more subjective than the basic design. Not many users will know the device as well as its designer does, so in most cases, the improvements would be actually known only to the designer. The designer is, objectively speaking, the man with the best historical perspective of the design, he knows best where he started from and where he is now with it.
Some improvements may be rather obvious to one and all, such as, for example, suddenly doubling your filtering capacity, but then this is hardly a minor change.
On a general note, while I am not a regulation freak, it seems to me that much of the debate about measurements does not address some of what I see as key issues.
A good example is the work load for amps. We are still using, at least offcially, simple lab resistors, when we all know only too well just how sadly inadequate that is.
An official standards (say, IHF) measurements should use a rough approximation of what might be called an "average" speaker, both as 8 and 4 Ohms. By this, I mean a simulation of an actual speaker crossover, say 3 way, with resistors, inductors and capacitors all mixed in so they give us a say -45 degree phase shift as a realistic, real world case.
Something along those lines, but surely a little more real life like, because a simple resistor sure as hell is NOT real life like at all. Letting things go on as they do simply prolongs our misery, these people determining the standards seem to have stopped evolving them something like 30 years ago.
Not that the Chinese care, mind you, they will still claim that a 12 Volt power supply for your PC speakers will deliver "1,200 Watts of S.M.P.S." power, whatever that means.
How can we expect our mesurements to more faithfully describe our products when our benchmark standards and values are so far removed from anything barely resembling real life?
A good example is the work load for amps. We are still using, at least offcially, simple lab resistors, when we all know only too well just how sadly inadequate that is.
An official standards (say, IHF) measurements should use a rough approximation of what might be called an "average" speaker, both as 8 and 4 Ohms. By this, I mean a simulation of an actual speaker crossover, say 3 way, with resistors, inductors and capacitors all mixed in so they give us a say -45 degree phase shift as a realistic, real world case.
Something along those lines, but surely a little more real life like, because a simple resistor sure as hell is NOT real life like at all. Letting things go on as they do simply prolongs our misery, these people determining the standards seem to have stopped evolving them something like 30 years ago.
Not that the Chinese care, mind you, they will still claim that a 12 Volt power supply for your PC speakers will deliver "1,200 Watts of S.M.P.S." power, whatever that means.
How can we expect our mesurements to more faithfully describe our products when our benchmark standards and values are so far removed from anything barely resembling real life?
If the input and output of a hypothetical amp were compared (with any real speaker load on the amp), and found to give a difference of zero with any and all test signals, including real music, to the limit of an AP2's resolution, could we declare that amplifier to be 'perfect', for all practical intents and purposes?
Well I did say "hypothetical". And yes, the question is precisely that: could a null at -110dB, as measured by a state of the art piece of equipment be considered 'perfect', or would there still be room to quibble?A null <-110dB? Good luck with that
My main concern is that I suspect there are people who would not trust an amplifier that had op amps and digital elements in it, on principle. (I'm not completely immune to those thoughts myself!). If the test equipment itself is built like that, could they accept its findings?
And another question I'm trying to tease out, is that some people think that an audio system's job is to actually add something that makes the final sound at the speakers more like listening to a live performance than the recording itself - many threads contain such suggestions e.g. the measured distortions of vinyl and tubes serendipitously enhance the sound. A true difference of zero would mean no added 'musicality'. Is this 'perfection' in such people's eyes (ears...)?
Hi,
The first pitfall is the comparison. How will you compare? Can you make sure you comparison system does not add or subtract something?
And of course you must use the actual system, mains cables etc. included, testing the Amplifier in isolation removes too many variables that can influence the outcome.
You may as well leave an AP2 out of this, as this is not a function it provides. It may be useful to inform yourself at AP's site as to how these devices work and what they can do and what they cannot.
If you instead suggested to use a system involving common DS A2D and D2A converters with normal (<= 200KHz) sample rates I would suggest that they are incapable in aiding such determination as you wish to make (and are unsuited to use in high quality audio).
On to perfection.
Now if an Amplifier (or a piece of wire) reliably does not alter any signal under any "designed for" conditions the Amplifier or Wire would be indeed 'perfect'.
Such a device does not exist and cannot exist, we would require to establish in what ways the amplifier (or wire) may alter the signal to remain:
1) Inaudible
2) Audible but not objectionable
3) Audible and objectionable
Then we could agree that an amplifier (or wire) may be regarded:
1) sonically transparent
2) sonically not transparent but offering good sound
3) sonically not transparent and offering bad sound
Having determined from the above which technical parameters actually matter and how (e.g. not THD but instead the best available distortion metric in terms of audibility research) we can measure how far devices deviate from 'audible perfection' and if the deviations are objectionable or not.
This would be what is called in other disciplines an "evidence based" approached. As we can observe no such approach is undertaken anywhere in audio.
Ciao T
The first pitfall is the comparison. How will you compare? Can you make sure you comparison system does not add or subtract something?
And of course you must use the actual system, mains cables etc. included, testing the Amplifier in isolation removes too many variables that can influence the outcome.
You may as well leave an AP2 out of this, as this is not a function it provides. It may be useful to inform yourself at AP's site as to how these devices work and what they can do and what they cannot.
If you instead suggested to use a system involving common DS A2D and D2A converters with normal (<= 200KHz) sample rates I would suggest that they are incapable in aiding such determination as you wish to make (and are unsuited to use in high quality audio).
On to perfection.
Now if an Amplifier (or a piece of wire) reliably does not alter any signal under any "designed for" conditions the Amplifier or Wire would be indeed 'perfect'.
Such a device does not exist and cannot exist, we would require to establish in what ways the amplifier (or wire) may alter the signal to remain:
1) Inaudible
2) Audible but not objectionable
3) Audible and objectionable
Then we could agree that an amplifier (or wire) may be regarded:
1) sonically transparent
2) sonically not transparent but offering good sound
3) sonically not transparent and offering bad sound
Having determined from the above which technical parameters actually matter and how (e.g. not THD but instead the best available distortion metric in terms of audibility research) we can measure how far devices deviate from 'audible perfection' and if the deviations are objectionable or not.
This would be what is called in other disciplines an "evidence based" approached. As we can observe no such approach is undertaken anywhere in audio.
Ciao T
- Status
- Not open for further replies.