Does Valve Noise Reduce if you Reduce Plate Current?

[MarcelvdG]

That's only true for white noise (seen as an equivalent input noise voltage), not for 1/f noise. One of the causes of 1/f noise is randomly varying resistance between the cathode nickel and the oxide layer that actually emits the electrons. Push more current through it and there will be more 1/f voltage drop across that resistance.

 

[Frederico Acardi]

I see, thank you for your explanations.
I read Merlin Blencowe chapters on noise in triodes. I understand better now.

If you had a tube microphone, for the first gainstage what tube would you use ?
Is a non microphonic ecc83 the best in term of noise, availability (I mean cheap) and overall SNR at the first stage ?
Maybe an ecc81 will be better than the ecc83 ? the mic source impedance is quite low.

I don't understand if in figure 5.13 p202, the EIN include Ig (grid current noise or not).

If I want to calculate Ig(shot) (5.14) wheredo I find the value of Ig in nA ?

Too bad Spice does not model all of this...

 

[MarcelvdG]

For a quite low impedance source, grid current shot noise is bound to be negligible as long as you bias the grid sufficiently negatively with respect to the cathode, as is normally always done.

Valve datasheets almost never specify the DC grid current. Philips datasheets often specify how negative the grid voltage has to be to keep the grid current below 300 nA, but that is all there is to find; for example, the 2 February 1955 Philips ECC83 datasheet specifies that VGK < -1.3 V for IG < 300 nA and the 5 May 1960 Philips datasheet relaxes that to VGK < -0.9 V.