Just as the title says. If you reduce the current through a triode would this reduce the shot/flicker noise?
Triode or pentode?
I read that a guy called Robert several decades ago discovered that noise was reduced reducing the heater/cathode temperature and then the illumination from it. Noise in tubes came from various sources, mechanical, electrical and perhaps, chemical.
I read that a guy called Robert several decades ago discovered that noise was reduced reducing the heater/cathode temperature and then the illumination from it. Noise in tubes came from various sources, mechanical, electrical and perhaps, chemical.
Noise in tubes has entire books on the topic. It is a very complex theme. If you have time to read, I can suggest the Radiotron, from Langford Smith and is easily downloadable for free in the web. Mathematics of it is well outside of the forum, and from my brain 🙂 .
Cascodes are by its own nature, high gain and low noise, in DC, audio and UHF.
Cascodes are by its own nature, high gain and low noise, in DC, audio and UHF.
Last edited:
shot noise increases only with sqrt (current). So if signal scales with current, SNR increase with sqrt(current) as far as shotnoise is concerned.
Triode noise is dominated by the Equivalent Input VoltageNoise, with scales (?) with 1/transductance (IIRC) .. ans since transductance increases with current, triode noise should be lower with higher current.
Triode noise is dominated by the Equivalent Input VoltageNoise, with scales (?) with 1/transductance (IIRC) .. ans since transductance increases with current, triode noise should be lower with higher current.
tschrama is correct about a triode's white noise: higher current means higher noise current at the anode, but lower equivalent noise voltage at the input and higher signal to noise ratio.
However, most valves have substantial 1/f noise at audio frequencies. The equivalent input noise voltage due to 1/f noise increases with increasing anode current.
So all in all, there is an optimum anode current that depends on the triode type and frequency range and frequency weighting you are interested in.
Merlin Blencowe, Merlinb on this forum, wrote an interesting AES paper about it. There is a thread about it:
Flicker Noise dominates triode noise in audio (AES)
However, most valves have substantial 1/f noise at audio frequencies. The equivalent input noise voltage due to 1/f noise increases with increasing anode current.
So all in all, there is an optimum anode current that depends on the triode type and frequency range and frequency weighting you are interested in.
Merlin Blencowe, Merlinb on this forum, wrote an interesting AES paper about it. There is a thread about it:
Flicker Noise dominates triode noise in audio (AES)
Last edited:
Thanks. Heavy reading but I’ll fight through it.
The other thing that confuses me is that noise reduces as you reduce heater voltage. This document suggests otherwise, although it is aimed more at higher frequencies going up into RF. My concern is with audio frequencies.
http://www.keith-snook.info/Stuff-y...ise in Vacuum Tubes and Attached Circuits.pdf
The other thing that confuses me is that noise reduces as you reduce heater voltage. This document suggests otherwise, although it is aimed more at higher frequencies going up into RF. My concern is with audio frequencies.
http://www.keith-snook.info/Stuff-y...ise in Vacuum Tubes and Attached Circuits.pdf
The quick answer:
For each valve type there is an optimum anode current at which SNR will be maximised. This optimum becomes broader for high gm types, so the operating current is less critical for these. Below the optimum anode current, gm falls and EIN will rise. Above the optimum, gm does not increase fast enough to overcome the increasing flicker noise. From my paper I suggest these optimums (optima?):
ECC81 = 1 to 2mA
ECC82 = 1 to 2mA
ECC83 = 1 to 2mA
ECC88 = 3 to 4mA
6J52P = 4 to 6mA
(These are for a 20Hz-20k noise bandwidth. There is a lot of sample variation, of course)
If a valve is operating below 1mA, a small reduction of heater voltage may improve the SNR. Above 1mA the opposite is true; flicker noise dominates and increased heater voltage has a beneficial effect on SNR, but at the inevitable expense of reduced heater/cathode lifetime.
This is covered in my book, too. See page 203:
Designing High-Fidelity Valve Preamps - Merlin Blencowe - Google Books
Last edited:
Increasing heater voltage means the cathode is hotter, hence the cathode space charge is hotter, so you get more noise.
Reducing heater voltage means the opposite, except that it also weakens the cathode space charge which reduces space charge smoothing of shot noise. There is an optimum heater voltage for noise, which may depend on cathode current. It is likely that in many cases the optimum heater voltage for low noise is just below the usual heater voltage.
Reducing heater voltage means the opposite, except that it also weakens the cathode space charge which reduces space charge smoothing of shot noise. There is an optimum heater voltage for noise, which may depend on cathode current. It is likely that in many cases the optimum heater voltage for low noise is just below the usual heater voltage.
The noise optimum also depends on what kind of weighting you use. For example, with A-weighting you will find a different (higher) optimum current than with a 20 Hz to 20 kHz unweighted measurement and amplifiers with RIAA correction will have a lower optimum current than amplifiers with a flat response.
6922 gain of 15 into tone stack ( -20db flat) into gain of 10 SS amp is delivering 103db s/n at 1w of output. I can hear hiss with a 95db/w/m tweeter pressed to my ear.
Triode noise about doubles the SS amp noise.
Triode noise about doubles the SS amp noise.
Would I be correct in thinking that I can use the noise sources for Tina here: http://c-c-i.com/sites/default/files/OpAmp_Noise-part_4.pdf and the information in Merlin’s book to simulate the noise of triodes. Nothing too exact, as there is so much variation between samples but maybe just to have an idea of best case, worst case?
I guess something has already been done. If so is anyone aware of where the information is?
I guess something has already been done. If so is anyone aware of where the information is?
Ooh nice find! Yes if those noise sources work then you could do that. But you're going to have to edit the noise macros every time you alter the valve anode current, if you want realistic results (although -depending on the circuit- you probably only need to do this for the input valve so I guess that's not too much of a burden).
Why only the input valve? Is it because that noise amplified by the following valves just completely swamps the noise they generate themselves?
Exactly. Unless the input stage has very little gain, or there's a volume control inbetween etc.
Well spotted! That's exactly why Neumann chose a lower than usual heater voltage for the VF14 pentode in their famous U 47 large diaphragm condenser microphone.
Best regards!
it's why I use the lowest noise transistor I can find for the front end. low noise Transistors are much better then tubes for noise levels.
Wow !
I wasn't expecting an ecc83 to be lower noise than an ecc81 !
Ra is so much higher for the 83 and the transconductance is lower for the 83 too.
Is there a simple explanation, I tought Gm was the key for low noise ?
I wasn't expecting an ecc83 to be lower noise than an ecc81 !
Ra is so much higher for the 83 and the transconductance is lower for the 83 too.
Is there a simple explanation, I tought Gm was the key for low noise ?