Reality is hard for some to face...
If it cant be measured then it aint gonna be heard, that is a fact.
Attachments
Bruce Hofer had a nice presentation at the latest AES convention, making this precise point.
His solution: if you have a feedback network of say 22k to 1k, build the 22k from 22 ea 1k from the same batch as the one 1k. Lots of cancellation - V coeff, thermal, etc.
I tend to listen when Bruce talks.
Jan
Hmm, just to review there is a common often true saying "The chain is only as strong as its weakest link"
The application to audio seems obvious - the fidelity of an audio chain is then logically most profoundly affected by the poorest sounding component.
From a perceptual standpoint the audible deficiencies in the poorest component in a chain can be shown to be capable of masking lesser audible problems elsewhere.
For example as soon as CD players and CDs became readily available back in the 80s our audio club experienced an improvement of the ability of our listening tests to reliably discern differences among certain other components that had been apparently masked by the vinyl and analog tape program material that we formerly had to rely on.
There was a presentation AD used to do to university students (and possibly others) in the 80s that made that point. They went further and said that, if you need close matching between 2 resistors pick consecutive ones off the reel. Mind you that was talking about precision analogue, not audio
I listen when traditional analysis says use bigger parts.
Now that we are frequently building more powerful amplifiers, the dissipation in the loop feedback network has become a potentially serious technical issue.
Where one could once throw in a 1/2 watt resistor without a great deal of thought, a simple traditional analysis will guide the designer to specify parts capable of handling far larger amounts of power.
I can even think of an bottom priced amplifier sold by a vendor who is famous for making highly cost effective gear that uses paralleled higher wattage resistors in the main feedback loop.
Yes, I once had to change an anode resistor from 0.6W to 1W. This is because the 0.6W resistor had insufficient voltage rating so after a few months operation it became crackly and eventually failed. The 1W replacement is still working fine.nicoch58 said:
Who is Erno?
That's good to hear. Some people on here never seem to learn anything.
I hope "and so on" includes people like Stuart Yaniger and Jan Didden - to just pick two normally reliable people at random. They don't make much money from audio, and their names rarely feature in the popular audio magazines.
The response variations and distortion generated by loudspeakers are orders of magnitude larger then what we commonly see with even mid-priced electronic components.
That audiophiles may fret over cartridge load resistors and give a pass to all the foul things that even the finest loudspeakers and room acoustics impose on real world systems is instructive.
It is true that some high end CD players from that era had some pretty severe technical problems. One source of these problems were naive attempts to put the DAC in a separate box.
This precise stacking of equal resistors was suggested by a Member here to try to measure the distortion of a properly rated single resistor.
I wish I knew who.
We often see multiple resistors in the feedback network to increase the dissipation. But some place them in parallel instead of in series and lose the benefit of voltage coefficient cancellation.
An example: 5off 2k7 in parallel for an effective 540r seeing almost 40Vpk.
5off 110r in series would have been a better solution.
Even that does not completely solve the problem, as each resistor is still exposed to ~280mW. Rather a lot for a 500mW, or 600mW resistor.
Sy's suggestion of a decade factor for dissipation requires at least 20 resistors.
Last edited:
- Status
- Not open for further replies.