That's why I didn't bother to do any more measurements on the Amperex valve that had 1 uA of grid current. All others had no noticeable grid current. Theoretically that can still mean that two currents cancel each other, like an ion and an electron current, but it is rather unlikely that that happens at all anode bias currents.
Note that 2q*1 uA is already some 4 dB below 4kT/20 kohm, so grid shot noise would not even have dominated on the crappiest Amperex, although it would have caused an error of about 1.5 dB. I don't know/remember if there is such a thing as grid 1/f noise though, do you?
Besides, for both Amperex devices at low currents (few hundred uA), I couldn't hear or see any difference in noise between resistor shorted and resistor not shorted. That is, their very large low-frequency noise at low currents must have been equivalent input voltage noise rather than equivalent input current noise.
For the second Trigon I also measured the A-weighted noise of a complete RIAA amplifier with shorted input using a true RMS meter. Divided by the gain and the root of the noise bandwidth, the result was 10.26 nV/sqrt(Hz) at 1.9 mA bias current, so very much in line with the measurements on the valve itself.
For the second Trigon I also measured the A-weighted noise of a complete RIAA amplifier with shorted input using a true RMS meter. Divided by the gain and the root of the noise bandwidth, the result was 10.26 nV/sqrt(Hz) at 1.9 mA bias current, so very much in line with the measurements on the valve itself.
Hey SY,
I am puzzled!
Did you really get those results on the bench? When loading a Gm with 10k like in your example, simmed results are on par with the same pentode-circuit wrt bandwith and distortion. The Gm has 3dB higher gain though.
It is when unloaded the Gm works(extremely high gain), so you must add a buffer and use a CCS load for the D3a as FrankB indicates.
Also can you explain the original Gm circuit that JC shows that have an ECC82 as Gm-element with its anode fed from B+?
I am puzzled!
Did you really get those results on the bench? When loading a Gm with 10k like in your example, simmed results are on par with the same pentode-circuit wrt bandwith and distortion. The Gm has 3dB higher gain though.
It is when unloaded the Gm works(extremely high gain), so you must add a buffer and use a CCS load for the D3a as FrankB indicates.
Also can you explain the original Gm circuit that JC shows that have an ECC82 as Gm-element with its anode fed from B+?
Oh, I thought you used 47k||1M. Not that the error is huge of course.
No, haven't heard of one.
If by "those results," you mean results similar to Frank's when he had the plate connected and high current running through it, yes. The gain was lower than my earlier prediction which I didn't fully understand, but thanks to the corrections of a few of the posters here, it's because I overestimated the impedance of the emitter.
I'm not familiar with the JC circuit you mentioned.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.