Thermal Capacitance of Resistors

[Samuel Groner]

Hi

For some research regarding thermally introduced distortion from resistors I'm looking for thermal capacitance data of standard metal film (0.25-0.5 W) resistors. Any ideas? I've found data for precision wirewound parts where time constants have been some 100 seconds, but I presume for standard parts things will be faster. Also available is data on thin film arrays and bulk metal resistors, again not particularly helpful.

Thanks,
Samuel

 

[JohnRoberts]

I am aware of a voltage coefficient distortion mechanism.

I presume you are talking about a temperature coefficient for resistance, and looking for thermal mass? I would expect thermal effects to be modest at audio frequency time constants and conservatively rated resistors. I am aware of (very low) audio frequency distortion in fuses near their marginal trip current, so it is plausible resistors could exhibit that mechanism iof banged pretty hard.

I wonder if IC designers have researched this due to their smaller Rs, but their Rs are heat sunk to substrate so not trivial to parse out. I guess thin film specs would be similar to bulk resistance in ICs, not IC pinch Rs which is a different can of worms.

Interesting question.

JR

 

[Samuel Groner]

I presume you are talking about a temperature coefficient for resistance, and looking for thermal mass?

I have data for thermal mass; Welwyn says 110 °C/W for the RC55 series. Together with the 15 ppm tempco this gives a rather drastic distortion level of -80 dB for a 500 ohm resistor at +14 dBu. This is assuming instantaneous heating, i.e. zero thermal capacitance (or frequencies below the thermal time constant). Presumably the thermal time constant is somewhere around 1 s, so these effects are fortunately attenuated in the audio frequency range.

Samuel

 

[abbey road d enfer]

Resistors have a dual thermal time-constant, one is determined by the thermal capacity of the resistive element and thermal exchange (radiation and convection), the other by the exchange between the resistive element and the substrate.
The Joule energy first increases the temperature of the resistive element in accordance with its thermal capacity, and radiates energy and heats the air, in addition it heats the substrate (through the thermal resistance of the bond), increasing its temperature according to its thermal capacity again. You can figure it as charging a capacitor through a resistor. I wouldn't be surprised if this time-constant was several tens of seconds.
In fact it's more complex than that, because it's an iterative RC network from the outside to the inside of the core.
In most film resistors, the thermal capacity of the resistive element is much smaller than that of the substrate, so the latter cannot be neglected.

 

[Samuel Groner]

Of course a first-order model has its limitations, but if we don't even have data for such... Unfortunately direct measurement of resulting distortion is difficult as at low frequencies most oscillators show quite high (and unsteady) residuals; additionally I'd like a worst-case estimate, and typical samples will show lower than guaranteed tempco. Perhaps I'll go ahead and measure thermal step response, but that costs some serious time again .

Samuel

 

[Samuel Groner]

I've done some measurements on a carbon composite part which looked like it could have a response similar to modern metal film parts. Time constant was about 40 s.

Samuel

 

[tv]

slightly OT (and mundane): http://www.diyaudio.com/forums/chip-amps/20605-reason-exotic-components-9.html

 

[pstamler]

Unfortunately direct measurement of resulting distortion is difficult as at low frequencies most oscillators show quite high (and unsteady) residuals;

What about using computer-generated signals, and jumpering the coupling caps on the soundcard? Or is the 1/f noise of the soundcard's opamps big enough to swamp what you're measuring?

Peace,
Paul