From reaike–(at)–x.netcom.com Mon Dec 4 10:41:56 CST 1995
Article: 3924 of rec.audio.tubes
From: reaike–(at)–x.netcom.com(Randall Aiken )
Subject: Resistor noise
Date: 4 Dec 1995 05:19:09 GMT
References: <49g3al$ji--(at)--lassic.iinet.com.au> <49js0m$d3--(at)--ewsbf02.news.aol.com> <49lk11$3v--(at)--anix2.panix.com>
X-NETCOM-Date: Sun Dec 03 9:19:09 PM PST 1995
In <49lk11$3v--(at)--anix2.panix.com> mgarvi–(at)–anix.com (Mark Garvin)
>In <49js0m$d3--(at)--ewsbf02.news.aol.com> jsuh–(at)–ol.com (JSuhr) writes:
>>There is a common theory that Carbon Comp resistors sound better than
>Comps are supposed to have more musical noise spectrum. Verdict’s not
>in on that yet, as far as I am concerned.
>Carbon comps DO handle transient overloads better than films, so
>certain applications may benefit.
I’ve accidentally fried both types of resistors, and I can vouch for
the fact that a carbon comp will hang in a bit longer…however, the
value usually changes drastically. Metal films, on the other hand can
make a very pleasing firey display when overloaded.
Of the three basic types of resistors, wirewound, carbon comp, and
metal film, wirewound are the quietest, carbon composition are the
There are several different types of noise; thermal noise, shot noise,
and contact noise and others. For resistors, thermal noise and contact
noise are the predominant players, shot noise occurs in vacuum tubes
and semiconductors and is due to random electron emission or random
diffusion, generation, and recombination.
All resistors have thermal noise, which comes from the thermal
agitation of electrons in the resistance. This thermal noise is
proportional to the square root of the bandwidth and the square root of
resistance, as well as the square root of temperature.
This means that in order to reduce the contribution of thermal noise in
a circuit, you can lower either the bandwidth, the temperature, or the
resistance itself. This noise is independant of type; for example, a
1K carbon comp resistor has the same amount of thermal noise as a 1K
In most vacuum tube circuits, high resistances must be used because of
the operating impedances of the tubes themselves. This makes it
difficult to reduce the contribution of thermal noise by lowering the
resistance. Bandwidth is generally fixed to the audio spectrum and the
temperature is usually hot. Because of this you are pretty much stuck
with the thermal noise problem. The frequency distribution of this
noise is uniform; it appears as white noise.
Contact noise (also called flicker noise, excess noise, low frequency
noise, or 1/f noise) is the noise that results from fluctuating
conductivity due to imperfect contact between two materials. It is
particulary bad in carbon composition resistor because the resistor is
made up of many tiny particles molded together. Contact noise is
directly proportional to the current flowing through the resistor; if
you increase the current five times, the noise is five times greater.
The noise power density varies with the reciprocal of the frequency
(1/f), rounding off at the point where it equals thermal noise. Because
of this, the noise can become very large at low frequencies,
particularly when large currents are flowing.
Wirewound resistors have the lowest noise; their noise is no greater
than the thermal noise. If no current is flowing in a carbon comp
resistor, the amount of noise approaches the thermal noise. Wirewound
resistors however, do have inductance that may affect the circuit
One final factor in resistor noise is the power rating. For the same
resistance value and current, a higher power resistor will generate
less noise. This is due to a factor “K” in the contact noise equation
related to the geometry of the material itself. There is as much as a
factor of 3 between the RMS noise voltage of a 1/2 watt carbon comp
resistor and a 2 watt carbon comp resistor.
In summary, the best way to reduce resistor noise is to use a wirewound
resistor if the inductance doesn’t cause a problem; secondly, use the
largest wattage resistor practical; third, use the lowest current
practical for the circuit; and fourth, use the lowest resistance value
practical for the circuit.
In terms of a “musical” noise spectrum, the carbon comp resistor has
more noise, period. The wirewound resistor has only thermal noise which
is constant density with respect to frequency. The primary difference
between carbon comp and metal film is in the amount of contact noise.
The 1/f response of the contact noise in the carbon comp or metal film
may impart a pleasing tonality to the noise, since it is rolled off at
a rate of 6dB per octave, but I would prefer having no noise to having