Stationary? A model thats order of magnitudes more accurate than anyones hearing. And there are also ways to model those non linearities, using, wait for it, ohms law.
Some reality. Other than power amps the voltages are a few volts.
A V divider with a shunt 100 times lower R than the input resistor drops the V coeff caused distortion by that 100 times, down to sub ppm. You can't hear that.
Are standard 1/4W 1% metal film resistors sufficiently linear for Hall Notch filters for looking at distortion at -160 dBc?
Or for use in the PCM1794 I/V & filter?
Or the ES9038Q2M I/V & filter?
Or for use in Samuel Groner's Super Op Amp? Samuel Groner's super opamp
Or for use in the PCM1794 I/V & filter?
Or the ES9038Q2M I/V & filter?
Or for use in Samuel Groner's Super Op Amp? Samuel Groner's super opamp
How do you know voltage coefficient is the only possible resistor imperfection? No thermocouples, no excess noise phenomena, etc? As said before, not everybody agrees on that.
Not everybody agrees on what is possible to hear either. The simplified calculations seen in some audio forums based on a misunderstanding of 'limits of audibility' and assumptions of hearing linearity are incorrect.
Last edited:
Maybe not all of them. If it really matters, how about buying Ed Simon's article then measure any resistors that are especially critical. Also, you might only need to measure parameters that are critical for your application.
Buy 1% resistors in lots or 10 or 25. Then measure them until you get a closer match.
1% resistors from the same batch will typically match much closer than 1%. The tolerance specification refers to how close they are to their rated value, not now closely they match.
Amazing that someone finally(?) mentioned that a shunt resistor between a low output resistance DAC buffer and a preamp can't add more than a tiny amount of distortion (none if the DAC buffer has no output "safety" resistor).
Also note the test resistors were 1W or 0.5W so for 22K we are talking not even a milliwatt at line levels. Do people even read what is posted (I know in this case it's hard)?
Your wording and the wording you quoted do not imply exactly the same meaning. 'Adding distortion' and 'changing the signal' are not necessarily identical consequences.
Cracking off the encapsulation of most resistors will destroy most of them. Without a laser trimmer under a microscope the resolution would be pretty bad. This is just a non-starter.
Last edited:
Infinitesimal is okay, just so long as people don't jump to conclusions that a very small effect is inaudible. If someone says they swap resistors and hear a difference then maybe they do and maybe they don't.
If there is an audible difference then there has to be a plausible physical explanation. If one takes Martin Mallinson not to be a liar, then ESS found out that humans can hear 'exquisitely' low levels of signal modulated noise. How many resistors are measured for that performance parameter? Also, what pathological circuit conditions might be sensitive to different resistor construction details?
My point is that there are plenty of things to think about before jumping to conclusions.
The ESS IMD "hump" is hardly 'exquisitely' low level. For the conditions here most resistors are -180dB or lower, if you can't measure them you can only make up stories.
Wasn't assuming the ESS hump is a signal correlated modulation of noise, although maybe it is. Would be interesting to see a time-domain notched residual. Some say the old residual distortion seen with ES9018 that you posted before is fixed in the later generation ESS chips. Don't know if that's true.
- Status
- Not open for further replies.