Sound Quality Vs. Measurements

Status
Not open for further replies.
diyAudio Member RIP
Joined 2005
I doubt it. I would expect that a cascade of stages, each with a smooth gain curve around zero-crossing, has a total gain curve which is also smooth although perhaps slightly less smooth. This assumes that there are no mathematicians' pathological functions lurking around.

I would not want to do a proof, but I expect that the product of a few rapidly-convergent smooth infinite series is itself a rapidly-convergent smooth infinite series. Distortion will then be monotonic with signal level, for sufficiently small signals.

If the nonlinearity is truly time invariant I'd tend to agree (and that's one of our tacit assumptions).
 
well you can expand to include slowly time variant nonlinearity too where the time constants are less than the nominal circuit speed - then the slow time variance is separable in the perturbation theory sense - one reason I really can't get excited over Lavardin/"memory distortion" from thermal device modulation - just too slow relative to feedback amp dynamics to have effects worth worrying over in a high feedback amp

of course low local feedback only amps could have thermal drift transfer function modulation of operating points give possibly audible response changes - clear win for global high feedback

the only class of time variant/memory distortion I see to worry about is hysteresis
 
Last edited:
well you can expand to include slowly time variant nonlinearity too where the time constants are less than the nominal circuit speed - then the slow time variance is separable in the perturbation theory sense - one reason I really can't get excited over Lavardin/"memory distortion" from thermal device modulation - just too slow relative to feedback amp dynamics to have effects worth worrying over in a high feedback amp

of course low local feedback only amps could have thermal drift transfer function modulation of operating points give possibly audible response changes - clear win for global high feedback

the only class of time variant/memory distortion I see to worry about is hysteresis


Unfortunately some small geometry SOI processes have thermal time constants ~3 usec as well as theta-j's of up to 2000 degrees/W so lots of attention to this is required.
 
Fuzzy distiortion

http://www.essex.ac.uk/csee/research/audio_lab/malcolmspubdocs/J7 Fuzzy distortion.pdf

Page 749 ( don't panic this is an exacted section of a larger text ) seems to be the discussion I remember .

Bob Carver . Wikipedia .

In 1985, Stereophile magazine challenged Bob to copy a Conrad-Johnson Premier Five (the make and model was not named then, but revealed later) amplifier at their offices in New Mexico within 48 hours. The Conrad Johnson amplifier was one of the most highly regarded amplifiers of its day, costing in excess of $12,000.
 
Last edited:
chip amps are different, I've pointed to the TPA6120 IMD plots with kHz bump as likely consequence of thermal effects, looked at your AD815 plots

but for "conventional" thru hole discrete audio power amps which I believe is the main subject the thermals have much longer time constants

maintaining Self's "optimum Class B" bias point could be a source of problems - he certainly has gone to some effort to stabilize the bias with output Q thermal transisents - even so he doesn't show distortion from "transient un-biasing" - and claims to have looked for it

but most here I think would go for the heavy AB bias of ~100 mA per output device - mostly avoiding the issue


I think we've had the discussion on Hawkford's fuzzy thinking in that article before - I rank it with the Cheever thesis as internet garbage to circular file
 
100 kHz bias could be said to be 198 kHz sampling if you look at it .
No, it is like the dither. Meant to mask the crossover magnetic distortion.
The bias mostly disipates as heat in the magnetic support at recording process (twisting the magnetic dipoles on spot). Some of it remains in the top layer, but just a little, exactly like dither.
 
I think we've had the discussion on Hawkford's fuzzy thinking in that article before - I rank it with the Cheever thesis as internet garbage to circular file

Unlike Cheever, Hawksford did some excellent work as well. He's an enigma: I always find it difficult to understand how the same person can produce fine analyses and then turn around and crank out junk without blinking. But "fuzzy distortion" has a nice memetic slogan feel to it, so it's useful for browbeating purposes.:D
 
Off to bed now .

I side this recently .

Smiley it is said Audio designers are aging with the customers . At the pub I was being universally blamed for the joke that Audio has become . I was horrified and said that it is customer led . The problem is it is putting off new people . I went to a Rock gig the other day and nearly cried . It was a time warp and everything was alright in the world ( girlfriends son's band ) . Marshall guitar amps and every expensive flavour of guitar . The kids still love music . So why don't they love hi fi ? We put them off . That's why I want to be doing this . They were on Radio Marlow ( we all have to start somewhere ) and were said to be a mix of the Arctic Monkey's and the Beatles , not bad I would say . .
Fuzzy thinking ? Guilty m'lord .
 
diyAudio Member RIP
Joined 2005
Unfortunately some small geometry SOI processes have thermal time constants ~3 usec as well as theta-j's of up to 2000 degrees/W so lots of attention to this is required.

As I may have mentioned I didn't pay much attention to audio in my early days, but did have some complex instrumentation projects for astronomy, and dug out a lot of things that had no direct references in that field. One of the tasks was to generate very fast and precise pulses, as well as measuring the settling time of fast amplifiers, and I studied how to do this before finding out that the scope amp people were on related techniques, and then some, although many of the details were proprietary. I was hampered by having a pretty limited suite of test equipment so in many cases built my own. Schedules were vague and I had indulgent chairpersons, which, despite exasperating some and making a few enemies, saved me from being put out to pasture, barely :rolleyes:

One of the trade magazines ran a story by Tektronix folk (iirc) about thermal distortions, and already in those days they mentioned that, despite thermal things being thought of as quite slow, they could see waveform distortion effects at frequencies as high as about a megahertz. Of course for vertical amplifiers, before sampling techniques came to dominate, there is the desire not to slow down the system with global feedback.
 
Status
Not open for further replies.