Checking for noise in resistors

[Calambres]

What's the best way to check for noise in resistors?

I'd like to test old carbon resistors used in high voltage circuits, i.e.: old vacuum tube amps and the like.

Is there a testing jig and/or testing circuit I can build to check for that?

[Xena E]

Hello Calambres...
It can be done but why bother? If the resistors are old carbon composition types they will have likely drifted high in value and be very noisy, (the two problems go hand in hand).

Carbon film types are usually more stable and lower noise.

Just swap them out for metal oxide types, they're cheap enough.

Regards Xena.

[TimFox]

I agree about metal oxide to replace old carbon composition resistors.
In this context, you are looking for “excess” (aka 1/f or pink) noise, which is added to the unavoidable thermal noise of a resistor when current flows through it (i.e. with voltage applied across it), due to fluctuations in the path taken by the current.
Any fixture to measure it requires a quiet voltage source, and the resulting noise is at low frequency (1/f dependence);  you need to compare low-frequency noise without and with bias voltage, and to avoid confusion from noise in the bias source.
Compare against a premium resistor known to have low excess noise (ideally wirewound, but good metal film is usable).
You could carefully measure the noise voltage at the junction of your tested part and a comparable value premium resistor, with a polypropylene capacitor coupling the noise voltage to a high impedance noise meter.

[Calambres]

Well, while I agree it should be easier to change all old carbon resistors for newer and better ones, I still would like to have (build) some type of testing device. I feel courious...

[Vovk_Z]

But what do you want to find? As I understand, the noisier the tube amp is - the better. Or not? If you don't want noise at all - use metalfilm resistors.

[CatalinaWOW]

In today's world the conceptually simplest way to test is to sample the voltage often enough to meet Nyquist for the frequencies of interest to you, and sample long enough to get solid data on 1/f noise and then data whip the resulting file.  You still have to do tests along the line that Tim mentioned to isolate the source(s) of the noise you find, and you have to assure that your sampling system is not the dominant noise contributor.  The latter requires either a lot of thought or a lot of money, or some combination thereof.

[TimFox]

But what do you want to find? As I understand, the noisier the tube amp is - the better. Or not? If you don't want noise at all - use metalfilm resistors.

Who said that noise is a good thing in tube amps?
You are probably confused by advocates of single-ended amplifiers who like the even-order distortion from the circuit.
Metal film resistors work well with tubes, but sometimes the high resistance required may need metal oxide for practical applications.

[Teledog]

Embarrassed to say it now, ..but as a "beginner", I went to an electronics surplus shop and asked if the big wirewound resistors were inductive, the shop owner said "oh, certainty not!".
Live & learn..
I wish the place was still around actually, was on Main St., Vancouver & had a creepy basement (no, not Lee's or Main Electronics)
 

[Vovk_Z]

Who said that noise is a good thing in tube amps?
You are probably confused by advocates of single-ended amplifiers who like the even-order distortion from the circuit.

Noise is like a natural disering to digital signal. With added some grain of noise and some grain of even harmonics makes tube sound what it is.

[coppercone2]

1) plug in good resistor to current source
2) measure ACV
3) compare with bad resistor of similar value
4) amplifier may be required


you measure noise with the AC volts function of a multimeter

[Calambres]

I think most of you are getting me wrong. I know about replacing the resistors for better ones, I already said that as well as I said that I'm just courious about the possibility to have a noise testing circuit. I'm sure there should be some designs out there. I'm aware of an old Heatkit tube signal tracer that can perform noise testing in components but such gear is almost impossible to get here in Europe, at least at a reasonable price.

Anyone know of some noise testing circuits out there? I'm sure there must be some around...

[CatalinaWOW]

I think most of you are getting me wrong. I know about replacing the resistors for better ones, I already said that as well as I said that I'm just courious about the possibility to have a noise testing circuit. I'm sure there should be some designs out there. I'm aware of an old Heatkit tube signal tracer that can perform noise testing in components but such gear is almost impossible to get here in Europe, at least at a reasonable price.

Anyone know of some noise testing circuits out there? I'm sure there must be some around...

I haven't thought it all through, but I strongly believe you can accomplish what you want using the sound input of your computer.  This should offer adequate signal to noise for the measurement but is uncalibrated.  You can use FOSS software like Audacity to capture the data, or there are other open source programs more directly engineering oriented.  Calibration could be achieved by measuring a good film resistor and using the the calculated Boltzman noise to scale the measurement.  Read the Wikipedia article on Johnson-Nyquist noise for a quick run through of the calculation.

[TimFox]

To understand resistor noise, I recommend reading Chapter 12 of https://pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Motchenbacher_Connelly/Low-noise_Electronic_Design.pdf , which is an update of the classic monograph by Motchenbacher and Fitchen: "Low-Noise Electronic Design", Wiley 1973.
I'm not aware of any equipment from that era specifically designed for measuring excess noise in resistors, but the books discuss test equipment designed to measure noise of amplifiers and semiconductors.
The fundamental difficulty is that the excess noise only rises above the unavoidable thermal noise in a resistor at relatively low frequencies, where measurements take a long time.
In some semiconductor devices, the "corner" frequency where excess noise exceeds thermal noise is much higher.

A recent technical article on measuring resistor noise:  https://www.mdpi.com/1424-8220/23/3/1107
It includes a good review of measurement techniques since 1960.
The reasonable ones for experimentation use Wheatstone bridges (presumably with three known good resistors) and a spectrum analysis, which could be done possibly with a good sound card after a preamplifier.
Remember that excess noise only appears with substantial voltage applied to the resistor, while thermal noise is always present at a finite temperature.

Back in graduate school, I measured excess noise in thick-film high-voltage resistors by very carefully applying 300 V from a dry battery (no longer available) and coupling the output through a good capacitor to a low-noise current preamplifier (from Princeton Applied Research, no longer available) into a strip-chart recorder.  One must be very careful not to smoke the input of such a preamplifier with spikes from applying the voltage.

[Calambres]

Very interesting those last two postings. Thanks!

[TimFox]

One last thing:
In all of that literature, note that “carbon composition” resistors are always the worst for excess noise, while wirewound and metal foil are the best (although limited in maximum resistance), followed closely by metal film and thin-film.
I used to joke with the younger guys at work that 1960s pop music was better than later stuff because of the ubiquitous Allen-Bradley carbon resistors.

[coppercone2]

you can make a 300V battery using coin cells. You need alot, it needs a good 3d printer, and you need plastic tweezers.

[floobydust]

From the era of tube gear, many signal tracers have a "NOISE" mode switch built-in. Example Heathkit IT-12.
That applies high voltage bias to the probe about 115VDC/1.7mA and with the audio amplifier you can listen to the result.
A noisy resistor or capacitor with bad dielectric can be flushed out. It was a common problem in the day.

Mr. Carlson's Super Probe has a "noise" switch but I think he's got it wrong and it's the actually RF probe mode.

[Calambres]

From the era of tube gear, many signal tracers have a "NOISE" mode switch built-in. Example Heathkit IT-12.
That applies high voltage bias to the probe about 115VDC/1.7mA and with the audio amplifier you can listen to the result.
A noisy resistor or capacitor with bad dielectric can be flushed out. It was a common problem in the day.

Yeah, that and the more capable T-3 are the Heathkit units I talked about in a previous post in this thread. It is unobtainium here in Europe at least at a reasonable price.
That was more or less the circuit I was trying to make, albeit somewhat modernized, i.e.: no tubes 

[strawberry]

humidity spoils carbon resistors therefore drift

[floobydust]

From the era of tube gear, many signal tracers have a "NOISE" mode switch built-in. Example Heathkit IT-12.
That applies high voltage bias to the probe about 115VDC/1.7mA and with the audio amplifier you can listen to the result.
A noisy resistor or capacitor with bad dielectric can be flushed out. It was a common problem in the day.

Yeah, that and the more capable T-3 are the Heathkit units I talked about in a previous post in this thread. It is unobtainium here in Europe at least at a reasonable price.
That was more or less the circuit I was trying to make, albeit somewhat modernized, i.e.: no tubes 

It's not hard- take an LM386 amplifier board $1.50 and run that with added blocking cap+protection diodes, and low noise HV power supply.

[TimFox]

The T3 in noise mode is described in its manual, available online.
It applies something like +125 V through a current limiting resistor and amplifies the resulting audio through a high gain tube amp to allow listening for snap-crackle-pop noise.
The LM386 suggestion is a reasonable alternative for the 12SH7, but probably less bulletproof.