Bill,
Data was 1, 1, 2, 2, 9 mean is 3. Standard deviation is Sqrt((4+4+1+1+36)/5) = 3.033 TL072 deviates by 1.978 SD
Yes the source of errors is large, but 2 standard deviations is not statistically insignificant.
Scott,
My parable about foaming tubing was be sure you are considering the right issue.
The TL072 has 18 nV/sqrt Hz at its best fed from a 20 ohm resistor. If we consider the human hearing mechanism is most sensitive at 3,500 hertz and the critical bandwidth at 3,500 hertz is about 560 hertz, so the noise level masking a tone at 3,500 hertz from a TL072 would be 425 nV.
As the maximum level from a CD should be 2 volts that would mask any distortion signal 74 dB below maximum. Research hints strongly (Like a 2x4 up the side of your head) that 2nd and 3rd harmonic distortion are not really an issue at significantly higher levels.
However Higher order harmonics are much more of an issue. As the most energy in music is around 150 hertz and down by 3 dB or so at 300 hertz or in practice may be as high as 15-20 dB above the 11 harmonic critical band the noise would mask the harmonic from a distortion of .1% or less.
So it seems to follow the research that a noisy final buffer may indeed improve the perception of a sound reproduction chain.
The AD797 is .9 nV/RtHz. and the AD825 is 12 nV/RtHz.
So one could indeed test to see if folks prefer enough noise in the system to mask any high order distortions! (Or are we measuring the right thing to look at distortion?)
As to why the noise is not noticed, take a typical listening level of 75 dBa and even my best room is not quiet enough to not mask a noise floor of 1 dBa.
Last edited:
You obviously have not known many people involved in multiple startups that failed. Actually most of my contemporary senior technologists in the IC industry are millionaires several times over, collect exotic cars, airplanes, etc. all working as a hired hand.
Same here out west. SEATAC, PDX, SFO, SJC, LAX, SAN, AUS, TUS are overflowing with colleagues that banked 1E7 simoleons working as hired hands in the IC industry.
Last edited:
You have to look at it from a multivariate analysis, and B without any sort of exp procedure that we know people follow (neither existed), the whole endeavor is folly.
Yes, even going back to Harry Olsen. Then it was shown folks preferred limited bandwidth reproduction over full range! H.O. Showed it was the distortion products that influenced the results.
Of course the research does still go on. In some applications noise is used to improve measurements.
A quick google showed lots of papers. Here is one that covers quite a bit more.
http://mcdermottlab.mit.edu/papers/NormanHaignere_McDermott_2016_distortion_products.pdf
Did you really miss the papers or did you just not bother to look?
Last edited:
Obviously... a straw man.... average millionaire goes broke 3.5 times.
MY comment was about formal education of wealthy individuals and those who make the Most money are not from salaries. Have their own business. same relative education stats are for millionaiirs as for Billionaires. Only 18% of millionaires have a Masters. and only 6% have a PHd.
THx-RNMarsh
Last edited:
Again why are we obsessing about $$? There's tons of stats out there to point out that, for the most part, a Ph.D isn't worth it financially. It may suggest something very quick about the career trajectories we propeller head types take and the motivations therein. Very very few people I know from my (God knows how many) degrees are primarily financially motivated -- usually it's "good enough income" and go do something interesting.
It's all good, I really really just want to discourage people from extrapolating anything from that test, beyond running it for yourself and finding out what *you* like.
And I find folks want to dismiss anything that differs from their expectations. Interesting results often are worth further investigations.
The Lawrence Livermore Labs once were only able to confirm a result found by others because the way the set up their equipment everything was turned off by a single switch. A slight difference in design and they would have seen the new discovery well before any one else. But they weren't looking for that data. Otherwise they were well ahead of the rest of the world.
Doctors and lawyers don't have either, please Dick at least make an effort to construct some statistics that make sense. I'm only a BSEE so I fall in your bucket.
21% of NFL players (mostly millionaires) are felons I suspect virtually none have Phd's (a few might have Bentleys though more likely RR's).
Last edited:
Sorry Ed I was thinking of actual listening to music on audio components, the famous >100KHz content Japanese paper has been much abused. Maybe you could provide a test like PMA did, I have no source material with a noise floor low enough.
Scott,
You are drifting from the topic. Most audio perception research is based on the fundamentals. The Olsen experiment was done with live musicians. Recording folks often add reverb to sweeten the tracks.
Non-harmonically reared noise is well known as a making agent. It has been used in some pieces of pro gear to mask unwanted artifacts particularly in digital audio products.
Harmonic distortion is a reasonably well understood issue. The point is there are others. Looking at the AD797 versus the AD825 there are not simply noise or distortion issues, some folks look at phase and other issues.
My criteria for my systems is I have to think they are OK. Instruments certainly are a useful part of the process. But I can have no distortion and flat frequency response and still have a terrible system.
Fine with me, are you saying a phono pre-amp should have a reverb option or other effects box features? Once you open these doors, do what you want I have no interest.
That seems to completely ignore all the people who said they couldn't detect a difference? Surely that is cherry picking the data to fit the point you want to make? A bit like the pet food adverts that say '8/10 owners said their cats preferred it*' with the star pointing to a microdot that said 'for those who expressed a preference'.
Certainly I would love to find a test everyone would agree on that can solve the audibility question, but given we can never even get people on here to agree what music is best that'll never happen
. And it's just a bit fun after all...Finally something I may agree with. If you’re working too hard you don’t have time to make money.
The only issue is that not everybody, at least on this side of the pond, have the success/happiness/achievement/etc... metric measured in coins. It's a little more complicated than a number in a bank account.
Shifting again. In the recording studio a few folks decide on what they want to present. As to at home you certainly can buy reverbs, 4 channel simulators or whatever you like.
The issue was how noise masks distortion. Lower distortion and lower noise is better, but noise masking distortion has been known for a long time to improve perceived quality compared to the same distortion unmasked.
Bill,
Your choice to count nulls. Include them and you are still over 1 SD.
- Status
- Not open for further replies.