Very little to do with the reproduction in my case. I don't get the repeat appearance of this comment. My wife and daughters could care less about the quality of the sound and have never EVER made one of those comments from the kitchen.
Last edited:
Hi,
You don't? Maybe worth looking at the schematic.
What I found with almost any circuits that use cnacellation to reduce even order HD is that they can arranged (balanced) for lowes THD with 3rd HD dominant or for a higher over all THD in which 2nd HD is dominant and 3rd HD is appreciably lower than in the "lowest THD" case and there is less "grass" growing with high order HD.
I find the 2nd case always sounds better, as I would expect, from basic circuit theory and five decades of distortion audibility research.
While the second case may have higher THD (THD being commonly used but meaningless), if we apply weighting to the HD as per Olson, D.E.L. Shorter or Lee/Geddes we find the second case has in ALL these metrics a lower level of distortion audibility...
So it is not "masking", not not "unmasking", it is simply less audible distortion despite greater THD.
As for the means by which JH achieved this in his 20W Amp, they should be blindingly obvious, but for the benefit of those who cannot "see it", the input stage deliberately introduces a small amount of imbalance (around 15%) which I suspect was carefully tuned...
I note some copies removed this imbalance in an attempt to "improve" the design.
BTW, the Pass F5 is an all FET version of the same circuit principle and it also is 2nd HD dominant, unless explicitly trimmed for lowest THD, because the P and N-Channel Fet's are not very good complements, so even order HD is not cancelled very well and higher order HD is not produced with levels as high...
In any amplifier using Mosfets, especially laterals, where the output stage is the dominant source of distortion this "2nd HD dominant" behaviour is also observable. This can be trimmed out with asymmetric source resistors but the results are again better THD but higher audibility of distortion.
I personally have long given up on low THD, I nowadays always tune for lowest 3rd HD and lowest higher odd order products. The even stuff then so far has always fitted in to produce a nice row of organ pipes on the FFT.
Ooops, now I am giving away the good secrets.
BTW, Impulse/Cepstral response of a given Amp is a different story, it has to do with other factors...
Ciao T
You don't? Maybe worth looking at the schematic.
What I found with almost any circuits that use cnacellation to reduce even order HD is that they can arranged (balanced) for lowes THD with 3rd HD dominant or for a higher over all THD in which 2nd HD is dominant and 3rd HD is appreciably lower than in the "lowest THD" case and there is less "grass" growing with high order HD.
I find the 2nd case always sounds better, as I would expect, from basic circuit theory and five decades of distortion audibility research.
While the second case may have higher THD (THD being commonly used but meaningless), if we apply weighting to the HD as per Olson, D.E.L. Shorter or Lee/Geddes we find the second case has in ALL these metrics a lower level of distortion audibility...
So it is not "masking", not not "unmasking", it is simply less audible distortion despite greater THD.
As for the means by which JH achieved this in his 20W Amp, they should be blindingly obvious, but for the benefit of those who cannot "see it", the input stage deliberately introduces a small amount of imbalance (around 15%) which I suspect was carefully tuned...
I note some copies removed this imbalance in an attempt to "improve" the design.
BTW, the Pass F5 is an all FET version of the same circuit principle and it also is 2nd HD dominant, unless explicitly trimmed for lowest THD, because the P and N-Channel Fet's are not very good complements, so even order HD is not cancelled very well and higher order HD is not produced with levels as high...
In any amplifier using Mosfets, especially laterals, where the output stage is the dominant source of distortion this "2nd HD dominant" behaviour is also observable. This can be trimmed out with asymmetric source resistors but the results are again better THD but higher audibility of distortion.
I personally have long given up on low THD, I nowadays always tune for lowest 3rd HD and lowest higher odd order products. The even stuff then so far has always fitted in to produce a nice row of organ pipes on the FFT.
Ooops, now I am giving away the good secrets.
BTW, Impulse/Cepstral response of a given Amp is a different story, it has to do with other factors...
Ciao T
Last edited:
John,
I suggest you also the reading of the paper TEMDECAY.pdf that can be retrieved from Temporal Coherence web site: I could have written nearly the same things!! I completely agree with the author.
If you're referring to the asymmetry of the two resistors coming off the emitters of the input devices (originally 200 and 300 ohms, later 200 and 240 ohms), according to Hiraga, this was done to improve symmetry due to NPN and PNP devices not being fully complimentary.
Perfect complementary pairs, which implies all parameters together and not just the precise condition of equivalence of Hfe at one current, do not exist in practice. We have already indicated this difference between PNP and NPN previously, a difference which had led us to slightly unsymmetrical values of resistance between the first and the second stages. In the first diagram, the values were respectively 200 and 300 ohm. In the final version, they change after adjus*tment to the more precise 200 (210 ohm) and 240 ohm. Saturation is thus quite symmetrical.
se
Steve,
But what does it really do this?
Drive is increased to the NPN input gain transistor compared to the PNP and it's bias is lowered, so the null offset correctly adjusted means the NPN transistor will produce more gain than the PNP even degenerated (and there is a lot of degeneration). Given that NPN normally already have higher Gm to start with this must create a 2nd HD dominant spectrum.
Maybe the translation that you read was flawed, or maybe JH simply did not quite express his aims right. Who knows, maybe JH also tuned for lowest 3rd HD? Or his tuning for symmetrical clipping produced the 2nd HD dominant result by accident? You can ask JH when you next see him what really went on.
It should really not take more a one second look at this schematic to see how things are done and what effects they produce, it is "tres primitive" and very obvious, no need to look at transistor datasheets or such...
Ciao T
But what does it really do this?
Drive is increased to the NPN input gain transistor compared to the PNP and it's bias is lowered, so the null offset correctly adjusted means the NPN transistor will produce more gain than the PNP even degenerated (and there is a lot of degeneration). Given that NPN normally already have higher Gm to start with this must create a 2nd HD dominant spectrum.
Maybe the translation that you read was flawed, or maybe JH simply did not quite express his aims right. Who knows, maybe JH also tuned for lowest 3rd HD? Or his tuning for symmetrical clipping produced the 2nd HD dominant result by accident? You can ask JH when you next see him what really went on.
It should really not take more a one second look at this schematic to see how things are done and what effects they produce, it is "tres primitive" and very obvious, no need to look at transistor datasheets or such...
Ciao T
Hans van Maanen's system sounds very, very good. If you look at the pictures you see that he has consistently made his speaker omnidirectional in the full audio band. But I believe that also makes them sensitive to unequal reflections in rooms. Hans' living room is rather large and he has the space to place his speakers away from room boundaries. Actually, his speakers remind me also of Siegfried Linkwitz' ORIONs. And they also perform best when place not too close to reflecting surfaces.
jan didden
jan didden
Thanks, Luigi, I had forgotten about that paper & that website. Yes, indeed, the information contained in that paper & website seem pretty much to match your own independent research. I doubt too many here bothered to read the paper when I referenced it initially, based on their responses to it - i.e. none.
It goes to the heart of everything that you are talking about:
- The human ear distorts sound by 25% (I've seen 30% mentioned elsewhere)
- So why do we bother designing audio systems of <1% distortion?
- could it be that the ear uses this distortion characteristic in an evolutionary advantageous way?
- could it be that we use this distortion to get more information about the sound than would be possible using a linear system?
- could this have led us to be able to detect the envelope of a sound (attack, decay, sustain, release)
- should we be analysing our audio systems in terms of non-linear acoustics to get an understanding of how they are perceived by the ear (i.e some measurement that finally correlates with what we hear)?
- is the current use of sine waves & FFT spectrum analysis flawed & leading to incorrect results?
A read of that paper could be a useful start in answering/analysing these questions
Any particular paper - there are a lot.
What about my questions above to you - any comments?
Yes I have heard his system, I know his work. It is all very well worked out and I can't find anything wrong in his papers (but then again, some of it is above my head).
His system sound very, very good. Have I heard traditional systems that sound as good? Yes I have. Does that prove anything? No
As I said, Siegfried Linkwitz tries to control off-axis directivity to avoid variations in reflected sound with frequency. Hans tries to avoid delay differences with frequency, and that also takes care of reflection differences *provided* you don't get too close to room boundaries. I have listened to both systems and they sound extremely clear, detailed, dynamic and balanced. I believe they reach similar heights with different means.
I am convinced that for great sound you can't dismiss room reflections and directivity changes and you have to somehow take care of it.
But don't forget, these guys have been working on their concepts for decades. It's allright to dream up something on a rainy afternoon, but getting a first rate piece of equipment realised is something else entirely.
jan didden
If your goal is to design an audio system that is transparent, that gives out the same as you put in, then whatever the ear does isn't relevant.
You only need to know the ear thresholds to know how far you need to take this 'transparent' thing.
jan didden
If the ear is what you are using to listen with then it is VERY relevant. I don't get your logic?
And the ear distorts 25 -30% so this is it's threshold - your point is?
His system sound very, very good. Have I heard traditional systems that sound as good? Yes I have. Does that prove anything? No
.......
jan didden
Of course it doesn't prove anything - it's purely subjective & should be completely discounted as of no importance whatsoever! In fact the reader should assume that you were coached in this listening & furthermore that you had strong expectation bias in both. So your observation may in fact be misleading!
How relevant was "what the ear does" when you listened to these systems?
Or how relevant is "what the ear does" in DBTs?
Last edited:
- Status
- Not open for further replies.