I did understand that . I just wondered could it be relevant ? As said I never found evidence .
My re-digitizing idea . I guess even vinyl could be used ? We can digitize the original and the output ?
Nordost gave a demonstration of this saying it would prove some debates could be answered . They seemed to think all errors showed up in the time domain rather than amplitude domain . I have heard nothing since so imagine they came unstuck ? The were using some RAF facilities which are used for aircraft analysis . They also had the ex RAF scientists to help .
http://www.nordost.com/default/downloads/NewApproachesToAudioMeasurement.pdf
My re-digitizing idea . I guess even vinyl could be used ? We can digitize the original and the output ?
Nordost gave a demonstration of this saying it would prove some debates could be answered . They seemed to think all errors showed up in the time domain rather than amplitude domain . I have heard nothing since so imagine they came unstuck ? The were using some RAF facilities which are used for aircraft analysis . They also had the ex RAF scientists to help .
http://www.nordost.com/default/downloads/NewApproachesToAudioMeasurement.pdf
As I understand the article, they took the analogue output from a CD player, converted it to (hopefully uncompressed) WAV format in a PC, then compared it with the original digital data. They believe this somehow measures the quality of CD playback. I hope I have misunderstood them, as all this does is measure the quality of the PC ADC.
The comparison achieves data alignment via some software they have written (for analysing sonar?).
They find that there are differences in the two data streams, especially when the slew rate is high. The difference can be up to 10% of data MSB, or a timing error of up to two samples. Adding 'magic' ingredients (e.g. fancy mains cables, equipment stands) seems to improve things, and correlates with audible perceptions. Have I got it right?
Assuming their write-up was not a total fabrication, or written by a marketeer, then two things are glaringly obvious:
1. As mentioned above, they are actually testing their ADC as much as anything else.
2. A 2 sample data error is simply not possible, unless one of the replay methods has a seriously wonky clock. I suspect that their time-alignment software was fiddling with the time delays as the test was running, and it gets confused by high slew rates in the data.
To do the test properly, you would need synchronised clocks (not mentioned) and an ADC which significantly betters (not mentioned) the CD standard (not to mention near perfect anti-aliasing filters). As usual, the MoD is lurking in the background. I hope not too much of my taxes have been paid to this outfit for their sonar work.
It is reassuring that the 'evidence' for magic potions continues to be so incompetent, as it helps confirm that the doubters like me are right.
PS Extra thought: some timing errors actually guaranteed unless clocks synchonised, so I think this confirms that their software would have to track the data dynamically (like a PLL) so almost certainly they are seeing software limitations.
The comparison achieves data alignment via some software they have written (for analysing sonar?).
They find that there are differences in the two data streams, especially when the slew rate is high. The difference can be up to 10% of data MSB, or a timing error of up to two samples. Adding 'magic' ingredients (e.g. fancy mains cables, equipment stands) seems to improve things, and correlates with audible perceptions. Have I got it right?
Assuming their write-up was not a total fabrication, or written by a marketeer, then two things are glaringly obvious:
1. As mentioned above, they are actually testing their ADC as much as anything else.
2. A 2 sample data error is simply not possible, unless one of the replay methods has a seriously wonky clock. I suspect that their time-alignment software was fiddling with the time delays as the test was running, and it gets confused by high slew rates in the data.
To do the test properly, you would need synchronised clocks (not mentioned) and an ADC which significantly betters (not mentioned) the CD standard (not to mention near perfect anti-aliasing filters). As usual, the MoD is lurking in the background. I hope not too much of my taxes have been paid to this outfit for their sonar work.
It is reassuring that the 'evidence' for magic potions continues to be so incompetent, as it helps confirm that the doubters like me are right.
PS Extra thought: some timing errors actually guaranteed unless clocks synchonised, so I think this confirms that their software would have to track the data dynamically (like a PLL) so almost certainly they are seeing software limitations.
Last edited:
This is why I thought a Vinyl source a better choice . It is a complex signal which could be digitized simultaneously by the same ADC ( in / out ) . It could be as many bits as one required . One might say an already digital signal might make life too simple ( pre-digested ) . Someone said it was the slight wrongness of a Garrard 401 which was great for testing digital recording devices . Mature Cheddar so to speak . Mild Cheddar should not result .
I suspect the Audio community will not join the bandwagon with this as it costs money and some will be a bit frightened . I talked to Richard Black and Tim de Paravacini with similar observations . The latter said approximately " tried it and it doesn't work " . The problem seems to be like null tests , is the null reliable . I suspect it doesn't require the last word in ADC ( 24 bit seems wise ) , just that it truly measures what is says it will and is repeatable . I would even include a heater in my test box as it would make one simple variant assured ( 45 C ) .
If we here could devise a simple test using cheap parts it would be great . It would be nice to find it did measure something we hadn't imagined possible . Perhaps the effect of vibration on DAC's or the like ( pulled from thin air as an example , Naim told me Philips have measured it when 16 bit DACs , Bleutak on top is all it requires ) . Nordost running the show is not my idea of objectivity .
I suspect the Audio community will not join the bandwagon with this as it costs money and some will be a bit frightened . I talked to Richard Black and Tim de Paravacini with similar observations . The latter said approximately " tried it and it doesn't work " . The problem seems to be like null tests , is the null reliable . I suspect it doesn't require the last word in ADC ( 24 bit seems wise ) , just that it truly measures what is says it will and is repeatable . I would even include a heater in my test box as it would make one simple variant assured ( 45 C ) .
If we here could devise a simple test using cheap parts it would be great . It would be nice to find it did measure something we hadn't imagined possible . Perhaps the effect of vibration on DAC's or the like ( pulled from thin air as an example , Naim told me Philips have measured it when 16 bit DACs , Bleutak on top is all it requires ) . Nordost running the show is not my idea of objectivity .
He was just the mouth of the company . Shame I never had the e-mail of the designer of the system . He seemed very aware of the questions we might ask .
I think if these ideas are reliable they will be seen and used . Noel Keywood is a journalist and is able to debate such things . Richard Black is no fool by a long way . Martin Collloms is sensible as is Barry Fox .
I think if these ideas are reliable they will be seen and used . Noel Keywood is a journalist and is able to debate such things . Richard Black is no fool by a long way . Martin Collloms is sensible as is Barry Fox .
Even I can see so many holes in that paper to dismiss it as Madison Ave. Engineering. They are comparing the outputs of two totally different DAC's. Nothing more than one cap could make the difference in the measurement, let alone the clocking of the "high quality sound card" 10 pages of BS. But then again, they are selling power cords.
Yes, the more I think about that paper the more silly it seems. The main thing they will be seeing is phase shifts caused by the reconstruction filter in the CD player and anti-alias filter in the sound card, together with changes from asynchronous resampling. Then on top of that they either see some crystal clock wandering (probably inaudible anyway) or their software trying to compensate for this. If there is any tiny bit of real problem in there it can be solved by using better clocks: better crystals, better PSUs, not magic cables.
The brains behind this work might be brilliant on sonar, but perhaps not quite up to speed on digital audio?
The brains behind this work might be brilliant on sonar, but perhaps not quite up to speed on digital audio?
Acuity About Us
I would ask him to join us . I may have cooked my goose by criticizing his journalist ( Nordost paper ) ?
The test has to stand examination if it is to succeed .
I would ask him to join us . I may have cooked my goose by criticizing his journalist ( Nordost paper ) ?
The test has to stand examination if it is to succeed .
That depends on how you define success. For a marketeer, success means increased sales rather than peer-confirmed truth.nigel pearson said:
A standard technique is to put out an article which looks vaguely technical (with impressive graphs of 'before' and 'after'), but doesn't say exactly what was done and hides technical things like numbers. The marketeer may genuinely believe that he is making techy stuff accessible to the masses, but what he actually does (accidentally?) is make it harder for technical criticism to take place.
One way to get an impressive graph is to (accidentally?) exploit numerical coincidence. A sharp downward spike on an FFT (below the noise floor) must be an artifact of the calculation (e.g. windowing, rounding errors). It can look impressive when this 'detail' is washed out by the test. The skirt response on an FFT tells you very little about the signal being analysed; it is set largely by the windowing function used and the time period for the sample. People who buy (and sell) cables probably won't know this.
I have asked Dr Gareth Humphries-Jones to give his views and a link to our forum . I imagine he might see it on Monday ?
I am not sure if it is the same gentleman I met ? If so I expect it to be a proper discussion far removed from hi fi hyperbole .
I remember null tests were rejected by many . I never understood that . In that case it was perhaps the subjective people who didn't like them .
I am sure penicillin seemed to be snake oil at first , it wasn't easy to produce . Certainly it was ignored for years .
I am not sure if it is the same gentleman I met ? If so I expect it to be a proper discussion far removed from hi fi hyperbole .
I remember null tests were rejected by many . I never understood that . In that case it was perhaps the subjective people who didn't like them .
I am sure penicillin seemed to be snake oil at first , it wasn't easy to produce . Certainly it was ignored for years .
Perhaps what they should have done is find a suitable frequency standard and use that with a phase detector to measure the medium-term (secs->mins) stability of the CD player crystal clock. Essentially they have done this, but in the wrong way, so their results are almost meaningless.
Having got their results, they could then do some double-blind trials to see to what extent, if any, they correlate with sound perception. Note however, that the likely effect of wobble in a decent crystal osc could be the same as the listener moving his head a few mm forwards or backwards. Quartz crystals have much higher Q and better frequency stability than musical instruments, human voice, turntables etc. so even a measurable wobble is IMHO unlikely to be audible.
Having got their results, they could then do some double-blind trials to see to what extent, if any, they correlate with sound perception. Note however, that the likely effect of wobble in a decent crystal osc could be the same as the listener moving his head a few mm forwards or backwards. Quartz crystals have much higher Q and better frequency stability than musical instruments, human voice, turntables etc. so even a measurable wobble is IMHO unlikely to be audible.
I think the double blind test very useful . Proper double blinds should always include playing the same thing twice . It was found many years ago that an identical thing sounds brighter the second time around . Doubtless the brain has learnt and is focusing differently . By the third try the effect has past . It is subtle .
Some people can not do blind tests . They switch off for whatever reason ( fear ? ) . Other people are not gifted even though they might think they are . I would imagine a professional conductor not a bad choice ( Andre Previn even though now 83 would be my choice ) . I would recommended Bill Low of Audioquest if he would ( cables ) ? From analogue Hi Fi Frank Schroeder if he would ( sorry all my German friends , Frank seems the most trained spokesman of us all ) . 20 total ?
Some people can not do blind tests . They switch off for whatever reason ( fear ? ) . Other people are not gifted even though they might think they are . I would imagine a professional conductor not a bad choice ( Andre Previn even though now 83 would be my choice ) . I would recommended Bill Low of Audioquest if he would ( cables ) ? From analogue Hi Fi Frank Schroeder if he would ( sorry all my German friends , Frank seems the most trained spokesman of us all ) . 20 total ?
Pics, took me longer to be able too!
SteamPunk pictures by mrqueue - Photobucket
First is the speaker that I built with my Grandfather in high school, That is half of one, I cut it in half and made two, I gave the other two to a friend. The original drivers where 12" coax from Lafayette Radio.
What's in there now is from Pioneer and Audio Concepts with some U-Do-It from Needham MA
I originally planed on using the cabinets to test drivers but wound up with what is there now due to 'weighing anchor' so to speak, and she doesn't like the look of them btw! Perhaps some wood veneer and a grill would fix that?
I have these situated in a nook up stairs, I'll post pics of that, but the area is not ideal for listening, so I have to make it work with the new fan-geld digital realm, that I had sitting on my desk for quit some years when I noticed it said, crossover! Light bulb picture goes off!
The amp is an old Crown I bought off of craig(s) list, total water damage but after bias adjustments, sounds fine, it was repaired before I got it and shows. I now know why the older guys like analog Simpsons multimeters, these digital ones throw you off when the battery is low!
What I've had fun doing is going to the source of what music I like and listening to it played by the Artists, then trying to find recordings that are good and recreate the performance the best I can, to the way I like to listen!
So, over all, just have fun!
SteamPunk pictures by mrqueue - Photobucket
First is the speaker that I built with my Grandfather in high school, That is half of one, I cut it in half and made two, I gave the other two to a friend. The original drivers where 12" coax from Lafayette Radio.
What's in there now is from Pioneer and Audio Concepts with some U-Do-It from Needham MA
I originally planed on using the cabinets to test drivers but wound up with what is there now due to 'weighing anchor' so to speak, and she doesn't like the look of them btw! Perhaps some wood veneer and a grill would fix that?
I have these situated in a nook up stairs, I'll post pics of that, but the area is not ideal for listening, so I have to make it work with the new fan-geld digital realm, that I had sitting on my desk for quit some years when I noticed it said, crossover! Light bulb picture goes off!
The amp is an old Crown I bought off of craig(s) list, total water damage but after bias adjustments, sounds fine, it was repaired before I got it and shows. I now know why the older guys like analog Simpsons multimeters, these digital ones throw you off when the battery is low!
What I've had fun doing is going to the source of what music I like and listening to it played by the Artists, then trying to find recordings that are good and recreate the performance the best I can, to the way I like to listen!
So, over all, just have fun!
The speakers look great . Some Soviet ones I took apart recently had a similar look and sounded great .
Many who looked at Motional feedback ( Philips , circa 1975 ) also looked at negative output impedance . Some 1950's RCA amps had it . Professor Korn of Belgium had an amplifier called Servo Sound slightly before Philips . Meridian M series speakers might have it if a breif conversation with their rep was correct .
At college we had a Servo Sound where incredibly the germanium transistors survived and the cones didn't . It was said better quality speakers did not suit it and TV types were used as the servo liked them . No idea if true .
From what I understood the box might halve in volume for a given output if using Motional feedabck . One suspects it would be partly due to active crossover which helps .
The Linn Isobarik although not a servo speaker looked also to get more out of a smaller box . It has a pressure sealed extra drive unit inside the box . It seems to work . The latter Linns had external crossovers . I saw a pair for £350 recently . I was very tempted . I would love to build them a nicer crossover ( active ) . Also to try a bit of negative output impedance just to know if it is a no no .
I can see an argument for critical under damping ( 8 R resistor to speaker reducing to zero ) . One could split that into 4R + 4R and get the error signal . I dare say even 0R1 + 0R1 as an op amp will amplify it nicely . If one finds the error disappears at 0R1 then output impedance is doing the job instead ( I am sure it does ) . In the 1950, Wireless world estimated by experiment a damping factor of 3 to be critical . Speaker cones/surrounds were stiffer then . If we said 2R + 2R we might be in interesting territory . It could be switched off to have pseudo valve sound . Sid Smith of Marantz was convinced much of it was that ( the output resistance ) . The Marantz had low distortion yet still was valve in it's virtues . Another has said that divorcing the amplifier from the speaker a little bit is no bad thing . I could well believe some bandwidth limiting of the feedback could offer many opportunities . One could argue that some of the virtues of current drive might be tested and a sweet spot found .
Variable Amplifier Impedance
DVV you use servo's . Why not do both jobs in one ? It could be called variable damping . It will fine tune a speaker to a room ? We all need a unique selling proposition . Some rooms are so dry , a bit of bloom might do magic . It needs very fine control is my guess . Oh if I could dial my suspension of my car ( some can ) . Lets be fair VW it is not bad ..
Many who looked at Motional feedback ( Philips , circa 1975 ) also looked at negative output impedance . Some 1950's RCA amps had it . Professor Korn of Belgium had an amplifier called Servo Sound slightly before Philips . Meridian M series speakers might have it if a breif conversation with their rep was correct .
At college we had a Servo Sound where incredibly the germanium transistors survived and the cones didn't . It was said better quality speakers did not suit it and TV types were used as the servo liked them . No idea if true .
From what I understood the box might halve in volume for a given output if using Motional feedabck . One suspects it would be partly due to active crossover which helps .
The Linn Isobarik although not a servo speaker looked also to get more out of a smaller box . It has a pressure sealed extra drive unit inside the box . It seems to work . The latter Linns had external crossovers . I saw a pair for £350 recently . I was very tempted . I would love to build them a nicer crossover ( active ) . Also to try a bit of negative output impedance just to know if it is a no no .
I can see an argument for critical under damping ( 8 R resistor to speaker reducing to zero ) . One could split that into 4R + 4R and get the error signal . I dare say even 0R1 + 0R1 as an op amp will amplify it nicely . If one finds the error disappears at 0R1 then output impedance is doing the job instead ( I am sure it does ) . In the 1950, Wireless world estimated by experiment a damping factor of 3 to be critical . Speaker cones/surrounds were stiffer then . If we said 2R + 2R we might be in interesting territory . It could be switched off to have pseudo valve sound . Sid Smith of Marantz was convinced much of it was that ( the output resistance ) . The Marantz had low distortion yet still was valve in it's virtues . Another has said that divorcing the amplifier from the speaker a little bit is no bad thing . I could well believe some bandwidth limiting of the feedback could offer many opportunities . One could argue that some of the virtues of current drive might be tested and a sweet spot found .
Variable Amplifier Impedance
DVV you use servo's . Why not do both jobs in one ? It could be called variable damping . It will fine tune a speaker to a room ? We all need a unique selling proposition . Some rooms are so dry , a bit of bloom might do magic . It needs very fine control is my guess . Oh if I could dial my suspension of my car ( some can ) . Lets be fair VW it is not bad ..
I think this is a modern version of Motional feedback . They look interesting , especially the tweeter .
https://www.google.co.uk/search?q=P...EC6OA0AWFkYDQBA&ved=0CB8QsAQ&biw=1280&bih=885
https://www.google.co.uk/search?q=P...EC6OA0AWFkYDQBA&ved=0CB8QsAQ&biw=1280&bih=885
Quite simply, Nige, because I don't believe in it. Not saying it's all wrong, just that I don't believe in it. I believe in naturally (i.e. without global NFB) obtained good damping factors, one of the reasons why I prefer to use 4 pairs of smaller power trannies rather than two pairs of higher power trannies. That's why I like to use parallelled capacitors of decreasing values, and why I prefer to have no inductors at all in my signal path.
This DF control switch was built into some integrated amps in the early-mid 70ies (e.g. Accuphase), and I tried it then and wasn't impressed with the results.
Philips' Motional Feedback is quite another matter. It's not much on lower cost versions, but on the biggie model of the time (12" bass driver), I did hear some rather amazing bass notes, in clarity comparable with any of the time for a whole lot less money than the passive speakers, given that their speakers were fully active and thus required only a preamp.
In addition to which I happen to have some oddball ideas which are out of synch with the times, which naturally limits me to a series of 1, all the more so since I do not wish to get into the general electornics market, feeling that it's already so crowded that it's bursting at its seams.
While I really love to read comments posted here by various members, and am grateful for it, I nevertheless do have my own experiences and views, all of which boil down to a single word - moderation. Zero global NFB amps sound loose and not quite precise to, while large global NFB amps all too often sound cold and lifeless to me, too "mechanical". Experience teaches me that a balance between zero and "much" global NFB is usually the best all around solution. So, I will have neither 0 dB nor 60 dB of global NFB, but will aim for say 26 +/- 1 dB as what I feel is a reasonable compromise.
At that value, no matter what else I do, and given that I hardly ever use less than 3 series/parallel pairs of output devices, often 4 pairs, I will have what I feel is a high damping factor anyway, it's not likely to be less than 30:1 into 4 Ohms and at 20 kHz. My feeling is that I simply don't need any more.
My favorite NFB loop consists of a simple resistor voltage divider, I do my very best to keep it as simple as possible. I try to avoid parallel capacitors as much as I can, but I will not sacrifice overall stability for anything.
In fact, to put it simply, I agree with John on all points except one, he once wrote that he tries as hard as he can under the circumstances to avoid any global NFB - I do not, as I said, I feel a reasonable amount of global NFB actually improves the overall results soundwise.
- Status
- Not open for further replies.