Audibility of Johnson Noise?

[pozz]

I was reading the Radiotron Designer's Handbook posted by @Wombat in another thread. There's this passage about a "rushing noise" in high gain tube amplifiers.

Can anyone shed some light? @SIY?

I thought Johnson noise described the minimum noise floor of any circuit. Was that so high in the 40s and 50s that you could hear its modulations?

 

[Blumlein 88]

Apologies for using wikipedia on you:

https://en.wikipedia.org/wiki/Johnson–Nyquist_noise

Handy calculator. You specify bandwidth and resistance and it gives you the noise that results.
http://www.sengpielaudio.com/calculator-noise.htm

 

[Blumlein 88]

So basically just resistance creates a low level noise. Attached to no power supply the resistance from thermal agitation of the material causes a current of electrons to flow. The attachment you had says beyond just the electron flow in the tubes you have resistances in the grid circuit of the valve as being responsible for it. Temperature also has a small effect. Higher creating more noise. At absolute zero we'd have no such noise.

An example from microphone preamps which deal with very small signal levels and employ very high levels of gain. A typical condenser microphone has 150 ohms of output impedance. This will set the lower noise level limits. The calculator would show noise will be -133 db V just from the resistance. And yes if a microphone has a higher impedance it will have a higher noise from that.

It isn't uncommon for someone not experienced to get a microphone preamp and complain it is noisy with the gain turned up. The noise added by the preamp might be only 4 or 5 db above that -133 db noise floor. But you add 60 db of gain to that and noise floor with nothing going on is now -70 db or greater and you could hear it as a slight rushing sound. Tubes often having much higher impedances in the circuit are going to have higher Johnson-Nyquist thermal noise levels and coupled with high gain you could hear it depending upon the particulars.

For an extreme example suppose you have a tubed preamp with a high output impedance feeding a small tube amp with a high gain, and connected to horn speakers. You'll hear that Johnson noise quite clearly.

If you looked at the thread I had on can we get a recording and playback with 120 db of dynamic range, these noise issues in the microphones and the gain we have to use makes it very hard to reach that level of dynamic range. Best recordings I've found so far have an intrinsic SNR of something in the mid 70 db area (though most of that is noise in the venue rather than amplified Johnson noise). Of course you have masking when music is playing. If we listen at an average level in the 70-80 db area it isn't too evident on most music though at louder levels it can be.

 

[SIY]

I was reading the Radiotron Designer's Handbook posted by @Wombat in another thread. There's this passage about a "rushing noise" in high gain tube amplifiers.

Can anyone shed some light? @SIY?

I thought Johnson noise described the minimum noise floor of any circuit. Was that so high in the 40s and 50s that you could hear its modulations?

Triode tubes have a combination of Johnson noise and 1/f noise. Actually, so do solid state devices, but they tend to have a lower and more consistent noise corner than tubes do. So the 1/f noise can be far more noticeable in tube circuits. Pentodes have an additional noise source called "partition noise," due to the randomness of division of charge between plate and screen grid.

Probably the most definitive treatment of the noise sources in tubes (with measurements) was Merlin Blencowe, "Noise in Triodes with Particular Reference to Phono preamplifiers," JAES 61(11), 911 (2013). The most interesting part was his demonstration that higher transconductance does not always equate to lower noise, and that most tubes have an idle current where noise is at a minimum.

 

[pozz]

Thanks. I'm just surprised about the quoted passage because I'm used to thinking of Johnson/thermal noise as being very, very low in modern designs, way below the threshold of audibility.

 

[SIY]

Thanks. I'm just surprised about the quoted passage because I'm used to thinking of Johnson/thermal noise as being very, very low in modern designs, way below the threshold of audibility.

If the circuit is well designed, the Johnson noise of resistors will not contribute much to the overall noise of the circuit. But active devices also have Johnson noise (as well as 1/f), so it's one of those inescapable facts of nature.

 

[pozz]

If you looked at the thread I had on can we get a recording and playback with 120 db of dynamic range, these noise issues in the microphones and the gain we have to use makes it very hard to reach that level of dynamic range. Best recordings I've found so far have an intrinsic SNR of something in the mid 70 db area (though most of that is noise in the venue rather than amplified Johnson noise). Of course you have masking when music is playing. If we listen at an average level in the 70-80 db area it isn't too evident on most music though at louder levels it can be.

Obviously I didn't take the entire the chain into account. The preamp example is very sensible.

What's interesting is that thermal noise can rise and fall that severely, instead of being a constant, static background. It makes sense given the increasing resistance as current and temperature rise, especially in tube designs. Although I wonder hear how much of the audibility is to do with flicker noise, which is signal-dependant and more sensitive to fluctuation.

 

[scott wurcer]

Obviously I didn't take the entire the chain into account. The preamp example is very sensible.

What's interesting is that thermal noise can rise and fall that severely, instead of being a constant, static background. It makes sense given the increasing resistance as current and temperature rise, especially in tube designs. Although I wonder hear how much of the audibility is to do with flicker noise, which is signal-dependant and more sensitive to fluctuation.

There is another source of fluctuating noise in older tube circuits, excess resistor noise especially in carbon composition resistors. This noise is dependent on the current in the resistor so hence on bias level.

BTW noise variation with temperature is a very weak effect in circuits operating at normal room temperatures. Noise vs temperature is a function of the square root of relative T on a degrees K scale (room temp ~300K) . So increasing temperature by 100C only increases noise by root(4/3).

 

[DonH56]

There's also shot (current) noise...

A 1 k-ohm resistor produces about 4 nV/rtHz ("every" engineer remembers this) or about 0.6 uV in a 20 kHz bandwidth. That does not seem like a lot but is about -125 dB relative to a 1 Vrms signal so resistor (Johnson) noise can be enough to help set the noise floor of some of the components Amir has measured. Note a MM phono cartridge input is typically 47 k-ohms so that can present a fair amount of noise. Records are noisy enough on their own so it is not usually a big deal, but using the wrong op-amp can exacerbate the noise. Low-noise op-amps using bipolar transistors have higher current noise due to their input (bias) current requirements and are not usually suitable for high-impedance inputs like phono stages; JFET-input op-amps are typically used. MOSFETs often exhibit high 1/f and flicker noise due to charge traps and such in the device so again are often less suitable for low-noise inputs.

As Scott implies, IME noise rise and fall with temperature, especially at audio frequencies, is often due to other noise sources than basic resistive thermal (Johnson) noise.

Years ago a fellow engineer used some wideband (microwave) GaAs devices and then later SiGe HBTs from an RF LNA (low noise amplifier) design of mine to (attempt to) create a low-noise audio circuit. Alas, I had to point out after he built it that the devices he chose had 1/f corners in the hundreds of MHz and thus "low noise" only applied at microwave frequencies... Need the right tool for the job.

Some of my undergrad classes included tube circuits, but I've not designed a tube circuit in many years, so will pass on digging into them again.

 

[SIY]

Note a MM phono cartridge input is typically 47 k-ohms so that can present a fair amount of noise. Records are noisy enough on their own so it is not usually a big deal, but using the wrong op-amp can exacerbate the noise. Low-noise op-amps using bipolar transistors have higher current noise due to their input (bias) current requirements and are not usually suitable for high-impedance inputs like phono stages; JFET-input op-amps are typically used. MOSFETs often exhibit high 1/f and flicker noise due to charge traps and such in the device so again are often less suitable for low-noise inputs.

This is why I have been pushing Amir to use an actual cartridge across MM stage inputs for noise measurement instead of a short, both the current noise aspect and the fact that the high inductance of an MM causes less shunting of the input resistance with increasing frequency. I have a calculator for Johnson noise of phono cartridges interacting with preamp input impedances- this sets a hard limit to SINAD in phono systems even with a "perfect" noise-free preamp.

 

[pozz]

This is why I have been pushing Amir to use an actual cartridge across MM stage inputs for noise measurement instead of a short.

What would be a good representative cartridge? And wouldn't this be equivalent to testing speakers to measure other components?

What about measuring a single cartridge with a particular phono stage and then using the results as a SINAD reference?

 

[SIY]

What would be a good representative cartridge? And wouldn't this be equivalent to testing speakers to measure other components?

All the cartridge is used for here is the source impedance- you're not actually playing it, just measuring the noise of the preamp with the cartridge plugged in. See the noise spectra data at the end of this post. The difference is significant, and for preamps having bipolar transistor inputs or using opamps like the AD797 (and hence higher current noise), the differences are yet more significant.

Choice of cartridge is fairly non-critical- something with a DCR of 500-1000 ohms and an inductance of about 1H will be pretty representative.

 

[syn08]

In despite of a very common High End Audio fetish, I've recently confirmed that reading a MC cartridge into a virtual ground (that is, very low impedance, much lower than the MC cartridge DCR) is perfectly fine. I was unable to identify any frequency response aberrations using a test record. This is not unexpected, since it was pretty obvious that a cartridge is not a reciprocal motor, so shorting does not amortize the needle movements (as many audiofools assume).

 

[SIY]

Yeah, reciprocity is very, very low. But the loading fetish continues, likely because "MM loading can be critical, so MC must be as well." My buddy Anatoly likes to use the expression "rule of dumb," and it's appropriate.

 

[syn08]

Yeah, reciprocity is very, very low. But the loading fetish continues, likely because "MM loading can be critical, so MC must be as well." My buddy Anatoly likes to use the expression "rule of dumb," and it's appropriate.

Don't you love the "100ohm optimal loading" specification you can find in almost all MC cartridges data sheet? I'm pretty sure that most manufacturers don't even bother to determine an "optimum", they go with the legend while thinking how many expensive MC head amps are implementing it. It would be a business problem to specify another value, although given the 1000% and up margins that the MC cartridge industry enjoys, they could probably afford the hit.

 

[JJB70]

Sadly Johnson noise is all too audible over here in Britain

 

[syn08]

Sadly Johnson noise is all too audible over here in Britain

In general, filtering is the only way to lower noise. In your GB case, a kick in his shadowy place could work.