John Curl's Blowtorch preamplifier part II

[simon7000]

Originally Posted by bear

Simon7000,

Are you testing the thermal stressing of resistors WRT signal?

If so, can we not merely design with higher wattage resistors and avoid the problem entirely by not designing "on the edge"??

_-_-bear

A reasonable question. I expect a reasonable design would use a resistor at 1/2 of it's power rating or less. So at first I tried to do the tests at 7.74597 volts rms into a 1K ohm, (about 1/4 power rating) as this is not an untypical use. As the signals are around -165db I suspect the level may need to be higher.

[bear]

So, again your focus is only on the possible effect of thermal stress in a resistor WRT possible "distortion"?

_-_-

[john curl]

Resistors might be a great candidate for a differential subtraction test, with FFT analysis, after the difference subtracted by the best instrumentation op amp available. Then, more than just sine waves could be tested, such as single pulses, etc.

[bear]

Presumably then a plan might be to have a "carrier" to bring the thermal level up to a certain point, and then "bang it" with a pulse?

Now would we want the "carrier" to be DC or an AC signal, and if AC, what freq?

_-_-bear

[john curl]

Depends, maybe nothing is best as a carrier. Just a change from low to high.

[simon7000]

Originally Posted by john curl

Resistors might be a great candidate for a differential subtraction test, with FFT analysis, after the difference subtracted by the best instrumentation op amp available. Then, more than just sine waves could be tested, such as single pulses, etc.

A bit more complicated, but basically that. The thermal noise level is around -134 and the distortion -165 from the excitation. A dc or slow ac would separate thermal from intrinsic, level changes vs % would show discontinuities vs non-linearities.

[bear]

And after these tests you expect that it will illustrate quantitatively the audible difference(s) between Mr. Curl's old Resista, maybe Holco and all others??

Or?

_-_-bear

[john curl]

Bear, that may be more than he can do, but at least he is trying.

[john curl]

I suspect that a current summation test, with the 1K resistor being driven with 50 ohms or less, drive impedance into a really good op amp that is set up for summing input, would be very effective with many tests, especially with signal averaging, where useful.

[bear]

John C, I sent you a PM, btw... (like me, you may not have your PM notify turned on in your prefs here...)

I am all for Simon7000 doing tests. I'm only trying to focus in on what he plans to do and what will actually be tested!

(fwiw, I've seen other sorts of 'tests' where lots of effort was made to set up and control everything, but the thing that was actually being tested and the conclusions drawn did not have a direct correspondence(!) - not that I am an expert of any sort...)

As i said, a separate thread on resistor testing might bring forth some expertise, prior work and new ideas - why duplicate earlier work, if it was done right?