G.Kleinschmidt said:
It depends on the field of improvement. My point was, NPN and PNP technologies are still different... Unfortunately...
lumanauw said:
I prefer when one shoulder is a voltage source, another shoulder is a current source. In such case it is possible to get the best result in terms of sound quality.
Bob Cordell said:
Beats me, even mediocre spectrum analysers these days operate between 20 Hz and a few GHz. I cannot see any reason for not being able to measure the spectral power density of a audio amplifier at the nth harmonic.
I would be very dissapointed in paying thousands of dollars for a high end amplifier knowing that the equipment was designed and tested using only a cheap 3.5 multimeter.
How do you achieve your distortion figures from pure theoretical calculation?
Regards
Nico
Nico Ras said:
There is a big difference between audio spectrum analyzers and RF spectrum analyzers, and any spectrum analyzer that is not PC-based is generally very expensive. You are describing an RF spectrum analyzer if you are talking about going out to 1 GHz. Extremely few of those go down into the audio band with 100 dB dynamic range. We are talking $20,000 and up pieces of laboratory test equipment here.
I personally use an HP 3580A, a very fine analog audio spectrum analyzer designed in the 70's. At the time it was about $10,000 when that was a lot of money. I bought it used for $700 about 9 years ago. It goes to 50 kHz and has 90 dB dynamic range. I enhance its dynamic range by 20 or 40 dB in amplifier tests by using my Distortion Magnifier described on my web site at www.cordellaudio.com.
More significantly, PC-based spectrum analyzers using PC sound cards are now available that put spectral analysis in the hands of just about anybody. Check out Visual Analyzer at www.hacca.altervista.org. If you use it with a 24-bit, 96 kHz sound card, you can get in excess of 100 dB dymanic range and get out to frequencies close to 50 kHz. It supposedly can also be used with 24-bit, 192 kHz sound cards for greater frequency range, but I have not yet located a reasonably priced sound card with 192 kHz A2D sampling rate.
For THD-20, my THD analyzer has a bandwidth out to 200 kHz, and can see 0.001% THD. But as pointed out earlier, spectral analysis of the distortion output out to only 50 kHz doesn't tell the whole story on THD-20.
If you find a spectrum analyzer that goes from 20 Hz to 1 GHz with 100 dB dynamic range for less than $1000, please tell us all.
Cheers,
Bob
Sorry this is not a beef directed at you, Bob.
You rightly mention PC spectrum analysers, scopes and the like, Besides the world is becoming a cut and past society. Moreover, the fact that if it cannot be found on the internet does not mean it does not exist.
I trust that you do not suggest that the "High End" audio industry considers a $1000 piece of test equipment expensive.
I believe in the fact that if it does not measure right then it is unlikely to perform correctly.
Test and verification equipment should be at the very least an order of magnitude better than the equipment you produce, even if you have to develop test equipment for the purpose at sometimes hundreds of thousands of dollars.
Not only that, the equipment must be calibrated regulary against know standards to ensure that the products designed and manufacture remains within specified limits.
How would one feel knowing that Goodyear Tyre company only had a short paved driveway at the back of Mr Goodyear's home where they develop and test all their tyres but claim to comply with all weather conditions and safety standards.
Kindest regards
You rightly mention PC spectrum analysers, scopes and the like, Besides the world is becoming a cut and past society. Moreover, the fact that if it cannot be found on the internet does not mean it does not exist.
I trust that you do not suggest that the "High End" audio industry considers a $1000 piece of test equipment expensive.
I believe in the fact that if it does not measure right then it is unlikely to perform correctly.
Test and verification equipment should be at the very least an order of magnitude better than the equipment you produce, even if you have to develop test equipment for the purpose at sometimes hundreds of thousands of dollars.
Not only that, the equipment must be calibrated regulary against know standards to ensure that the products designed and manufacture remains within specified limits.
How would one feel knowing that Goodyear Tyre company only had a short paved driveway at the back of Mr Goodyear's home where they develop and test all their tyres but claim to comply with all weather conditions and safety standards.
Kindest regards
Bob Cordell said:
us$140 ok?
http://www.esi-pro.com/viewProduct.php?pid=43
Dac not quite up to adc distortion performance, a few spurious lines in the noise floor but highest price/performance ratio I know of
I have the juli@ because I'm cheap but while looking I considered
EMU-1212m, -1616m, 1812m
lower end Lynx Studio cards (may not do 192K) but have very good #
you can search for RMAA loopback plots of all of these to compare
unfortunantly this list is quickly growing out of date:
http://www.pcavtech.com/soundcards/compare/index.htm
Nico Ras said:Sorry this is not a beef directed at you, Bob.
You rightly mention PC spectrum analysers, scopes and the like, Besides the world is becoming a cut and past society. Moreover, the fact that if it cannot be found on the internet does not mean it does not exist.
I trust that you do not suggest that the "High End" audio industry considers a $1000 piece of test equipment expensive.
I believe in the fact that if it does not measure right then it is unlikely to perform correctly.
Test and verification equipment should be at the very least an order of magnitude better than the equipment you produce, even if you have to develop test equipment for the purpose at sometimes hundreds of thousands of dollars.
Not only that, the equipment must be calibrated regulary against know standards to ensure that the products designed and manufacture remains within specified limits.
How would one feel knowing that Goodyear Tyre company only had a short paved driveway at the back of Mr Goodyear's home where they develop and test all their tyres but claim to comply with all weather conditions and safety standards.
Kindest regards
I agree with just about everything you have said here. You're right - a couple of thousand dollars for a high-end manufacturer should not be a problem. Let's face it, they should at least have an Audio Precision analyzer or the equivalent. The story is admittedly a little different for the DIYer, even the very serious ones like myself. There, the capabilities of PC-based instruments can really come into play well at a low cost.
Your comment about purpose-built instrumentation is also a good one. This is an area where the DIYer is not necessarily at a disadvantage, since he has the time to put together some of these types of things. Even though some do not consider the sonic quality of op amps the equal of discrete high-end audio designs, there is no doubt that they excel in terms of objective measurement capabilities. So it is not difficult to build a test instrument, like my Distortion Magnifier, that is easily better than the amplifiers it is being used to measure. I can see twin-tone IM products down to the -120 to -140 dB range with these techniques.
Finally, I suspect that you might be very disappointed in how some (maybe only a few) high-end manufacturers test their designs before they foist them on the public. This may only be a few, and I certainly do not intend to paint with a broad brush. While there is certainly no substitute for subjective listening tests, I suspect that some high-end manufacturers are so far into the subjectivist camp that they do little or no objective testing, and that, in my opinion, is a bad thing. My motto is "listening and measurement - beautiful together."
Bob
Bob Cordell said:
Bob,
Using a 96k or 192k sound card does not automatically give you range that is extended close tot he expected Fs/2.
These sound cards use DAC/ADC chips from Wolfson, AD, BB, Crystal - and all of these concentrate primairly on the audio band. This means that while close to Fs/2 performance is achieved at sampling rates up to 48k, for sampling rates over this (64k, 96k etc) the phase/group delay response close to the top of the audio band is taken as a priority.
This means that a 'slower roll off' filter is used, which in turn means that roll off may well start close to the top of the audio band, and be quite signifficant at say 30-40kHz for a 96k DAC. It is very unlikely that it will be the typical near-ruler flat up to close to Fs/2. It is worth downloading a few datasheets to check out the DAC performance.
Of course, this does not mean that a higher Fs card is not useful for exploring the spectral content of the signal above 20kHz, it's just that one has to account for the loss of resolution due tot he filter characteristic.
Finally, I have found in my own experiments that most low cost cards perform poorely mainly because close to zero attention was payed when designing the analog and power sections. What is the use of using a good 96k DAC when the output analog filtering and buffering is done by 4558 type OPAMPs and ceramic cap types that are not suited for the task by any stretch of immagination... building a small extra output board can change this.
BTW I forgot to mention that some 'venerable' instruments work marvelously well extended by a PC spectrum analyser. For instance, i use the output of a HP334 distortion analyser as the input to my PC sound card, extending it's dynamic range quite considerably.
Bob Cordell said:
I had a few run-ins with high end equipment with my role being the one trying to ressurect them after a failure.
Never mind objective testing, some have nothing to do with sensible (and I mean common sense) design and construction techniques. I do not advocate that only certified engineers are capable of good construction, no (seen enough blunders done by those, thank you
) - but certainly the constructors have to be aware of good engineering practises, if only to avoid liability. Given those examples, I am not surprised nothing is given objective measurement, I doubt the designers would know how. But the real questionis, what use is superbaudio performance, if there is no real relaiability, and in some cases if there is real danger involved to the consumer? And we are not talking cheap stuff either...
ilimzn said:
I had a few run-ins with high end equipment with my role being the one trying to ressurect them after a failure.
Never mind objective testing, some have nothing to do with sensible (and I mean common sense) design and construction techniques. I do not advocate that only certified engineers are capable of good construction, no (seen enough blunders done by those, thank you
This leads me to a little story to support the point.
My boss, CEO and good friend (an accountant), purchased his very high end system from the local exclusive agent after reading a review in Whatever Hi-Fi. He learnt all the jargon from these magazines and if one did not know better would think he is a connoisseur.
Before taking it home he asked one of the engineers to fire up the system and then explain to him where what goes.
When passing one of our tea stations, I overheard the chaps talking about so-and-so system, and I decided to drop into R&D to see what all the fuss was.
After speaking to the youngster, he told me that he always dreamt about owning a system ABC but this cost almost the same as the car he was driving.
Anyway, the youngster (okay so 35 is not young) though he would like to see what an amp with these specifications really looks like on real equipment and told me he at first though that our test equipment was faulty. It was not. He even asked me to verify his test set-up, it was correct.
I fail to see why this equipment was so highly acclaimed. Of course, we could not break this to the boss because he would only take a defensive stand supporting his favourite hi-fi reviewer who is obviously far more qualified in this area.
Although I was very temped, we did not dare open the equipment as there were warranty void if seal is broken stickers all over it.
Some things that was quite notable. Power output into a pure 8.2 Ohm resistive load and sine wave @ 1 kHz was some 30% short from the claimed continuous power and severe clipping occurred by the time the volume reached 12 o’clock.
The third order product alone was a few orders of magnitude higher than the claimed THD and two tone intermodulation products only 10 dB down from the fundamentals. Higher order products were very prominent throughout the audio spectrum and far beyond.
There was a no load spurious around 600 kHz and at every overtone to 5 Mhz but dropped by about 30 dB when a load was connected.
The intro to the owners manual was something like this:
Thank you for purchasing ABC. We have been designing and manufacturing audio equipment to the highest standard for over 35 years.
This equipment was designed by us to offer lifelike sound and is a natural extension of our love for music. We are committed to make products of the highest possible quality to enhance your life. We urge you to partner this equipment only with other equipment designed and made to the same impeccable standards.....
The list of specifications covers almost two A-4 pages as well as having mean time between failure calculations.
I had better say no more... except that the sticker revealed that it was manufactured in China (not even Taiwan)
Contrary to this, mechanically and aesthetically this piece of equipment was beautiful, if one could describe audio equipment as beautiful. The connectors where knurled and gold plated with solid nickel-plated machined knobs, buttons and panels. Everything was done on a massive scale and the shipping box claimed the weight as 47 kg. This specification I tend to believe was accurate.
Hi end audio is a 'cottage' industry. Many small manufacturers don't have the resources to know 'everything' about design and construction, but they give it a try, anyway.
The alternative is to buy from the likes of Bose and Sony. Is that what you want?
I think that complete analysis of a 20KHz sine wave is rather extreme, and we are only discussing it in order to accomodate Bob Cordell and his fixation on that frequency.
As I said before, I use 5K, because I think that it is a more appropriate test frequency, overall, but even 2 or 3K would be OK, if absolutely necessary. I measure TIM separately. The most important factor is to look at the types of harmonics generated by the amplifier. Is it significantly 2nd, 3rd, 5th, 7th, 9th? It is more important to look at the 7th and 9 harmonic components, than to get another decimal point on the distortion measurement.
For the most part, standard measurements do not tell everything about how a piece of equipment will sound. This is unfortunate, but that is why many people do not bother as much with measurements.
I agree with Bob Cordell, that we should measure carefully, as well as listen to audio components. I use a modified Sound Technology 1700B and then do an FFT with an HP 3563 or sometimes a spectrum analysis with an HP3580. FFT analysis has the important advantage of signal averaging to reduce the noise floor significantly.
I have also used virtual spectrum analysers with my computer. This works OK as well, especially if you can use a 96K sound card. Still, you can get plenty of info with a standard 44K sound card, but you must operate at 2K or so, in order to see the higher order harmonics, and you still need to get your dynamic ranged enhanced by some sort of pre-null of the test sinewave. Usually, a THD analyser works best, but a passive null network would work as well.
The alternative is to buy from the likes of Bose and Sony. Is that what you want?
I think that complete analysis of a 20KHz sine wave is rather extreme, and we are only discussing it in order to accomodate Bob Cordell and his fixation on that frequency.
As I said before, I use 5K, because I think that it is a more appropriate test frequency, overall, but even 2 or 3K would be OK, if absolutely necessary. I measure TIM separately. The most important factor is to look at the types of harmonics generated by the amplifier. Is it significantly 2nd, 3rd, 5th, 7th, 9th? It is more important to look at the 7th and 9 harmonic components, than to get another decimal point on the distortion measurement.
For the most part, standard measurements do not tell everything about how a piece of equipment will sound. This is unfortunate, but that is why many people do not bother as much with measurements.
I agree with Bob Cordell, that we should measure carefully, as well as listen to audio components. I use a modified Sound Technology 1700B and then do an FFT with an HP 3563 or sometimes a spectrum analysis with an HP3580. FFT analysis has the important advantage of signal averaging to reduce the noise floor significantly.
I have also used virtual spectrum analysers with my computer. This works OK as well, especially if you can use a 96K sound card. Still, you can get plenty of info with a standard 44K sound card, but you must operate at 2K or so, in order to see the higher order harmonics, and you still need to get your dynamic ranged enhanced by some sort of pre-null of the test sinewave. Usually, a THD analyser works best, but a passive null network would work as well.
ilimzn said:
I agree, these are all very good points. One 192 kHz sampling card I just looked at was flat to within about +/- 0.1 dB out to about 65 kHz, then pretty much rolled off very steeply. This aspect will vary a lot from card to card. Of course, for purposes of distortion spectral analysis, +/- 1 dB, or in some cases even +/- 3 dB is not a show-stopper.
All of this just underscores the value of IM tests whose distortion products lie in-band.
Bob
Nico Ras said:
I can guess what brands may fall into this category. I always get a laugh when a unit gets measured for a Stereophile review and misses its spec, and the manufacturer claims it must have been a defective unit. Ha!! When you pay that much money for a high-end unit, there should be extraordinary QA. Moreover, it is unbelievable that a manufacturer would provide a review sample that was not known to be absolutely solid.
This is all such a terrible shame, because once the big investment is made in the thick chassis and front panels and big heatsinks and beautiful machined knobs, it costs very little more to make something that at least passes a reasonable set of specifications. It just takes a little bit of diligent engineering and measurement verification; and there is no reason in the world that such diligent engineering and measurement should subtract from the subjective sonic performance.
Bob
Nico Ras said:
john curl said:
Yeah. You plunk down your 10k or 20k, and you pick a high end system. You can be lucky and get an amp worth its money. Or you get crap. Surely owners of both equipment will praise 'their' hifi into heaven. Of course a little measurement would quickly separate the corn from the chaff, but, hey, if it ain't a listening tests it doesn't count, right.
What a deploreable state this audio business is in.
Jan Didden
Bob Cordell said:
Couldn't agree more. Chalk this for the undeserved black eye negative feedback has earned in certain circles.
And we must live with it.
Rodolfo
PS. 20 KHz THD test has much to do with this, here we get a stressing combination of slew rate and low correction margin, the stage where mediocre designs falter.