Implications for pentodes
it seems reasonable for the results of Merlin's paper to also apply to pentodes. We could add the screen grid's noise sources to the model, though that may not be necessary, as Frank has shown that we can minimise those by adopting the BestPentode approach with adequate bypassing to ground.
It's plausible that, for pentodes as with triodes, there will be a local minimum of the sum of flicker and shot noise, a low-noise a sweet spot. Merlin's paper indicates that, for high gm tubes, the sweet spot is broad. Hence, it's likely that it's unnecessary to use very high currents to achieve the best noise results, and it's likely that we won't have to be too finicky about finding the exact minimum for high gm tubes, especially considering the variability that already exists between tubes' noise characteristics. This all seems to me to be pretty good news.
it seems reasonable for the results of Merlin's paper to also apply to pentodes. We could add the screen grid's noise sources to the model, though that may not be necessary, as Frank has shown that we can minimise those by adopting the BestPentode approach with adequate bypassing to ground.
It's plausible that, for pentodes as with triodes, there will be a local minimum of the sum of flicker and shot noise, a low-noise a sweet spot. Merlin's paper indicates that, for high gm tubes, the sweet spot is broad. Hence, it's likely that it's unnecessary to use very high currents to achieve the best noise results, and it's likely that we won't have to be too finicky about finding the exact minimum for high gm tubes, especially considering the variability that already exists between tubes' noise characteristics. This all seems to me to be pretty good news.
This is indeed the case. The noise spectral density for a pentodes behaves exactly like that of a triode; the white noise is simply at a higher level, due to the extra component of partition noise (so just think of it like a triode with rather less gm). But if you're truly concerned with absolute minimum noise then pentodes are probably best avoided. I'm not yet conviced of the efficacy of Frank's trick, since you're substituting transistor noise for screen noise.
True, transistor noise should be included in the BestPentode model. From Frank's article, I estimated that base current was low, which would tend to reduce noise from the MPSA44As employed. The collector current was also likely to be fairly low, in the region of a few mA - though maybe not in a low noise "sweet-spot" ?!? Nonetheless, I thought that the transistor's noise contribution was likely to be somewhat less than the estimated screen grid shot-noise included in the classic pentode noise equation.
This may not be a good assumption. For instance, Frank's results showed that BestPentode operation had a less-than-expected noise advantage over ordinary pentode operation. This might suggest that that screen grid noise isn't as large as thought, which would also go some way toward explaining the similarity between the input-referred noise figures for the same tube operating as a pentode and a triode.
On the other hand, the BestPentode/pentode results might indicate that transistor noise is only a little less than screen grid noise. Some transistor measurements might help clear this up.
This may not be a good assumption. For instance, Frank's results showed that BestPentode operation had a less-than-expected noise advantage over ordinary pentode operation. This might suggest that that screen grid noise isn't as large as thought, which would also go some way toward explaining the similarity between the input-referred noise figures for the same tube operating as a pentode and a triode.
On the other hand, the BestPentode/pentode results might indicate that transistor noise is only a little less than screen grid noise. Some transistor measurements might help clear this up.
Old thread, but so are the vacuum tubes.
1959's 109 page M.S. Physics thesis by R.A. Finston on measuring xray and nuclear chamber particle energy with vacuum tubes. There is extensive discussion of determining lowest noise operating point properties starting about page 34 thru about page 54. An example three-pentode low-noise preamplifier optimized for their unique usage appears in Figure 7. It's described as a modified cascode with pentode current source. Very low voltages (30, 81, 84 VDC) are used. Pages 34-54 were most interesting for optimizing low noise. There are several tubes listed as having the desirable characteristics, but they were selecting roughly 10 out of 50 tubes as meeting their requirements. They are not audio circuits and they are using extremely small signals, but the principles are interesting. You'll either find something useful in it, or discover what is not worth trying because the range of voltage and current may exclude certain techniques for audio signals. Except maybe tube mics might be closest to this. Very high input impedance, small signals, operation at low voltages. I could not find section on filament voltage this time. The PDF is not searchable and I didn't want to read the whole thing again.
https://www.osti.gov/servlets/purl/4225786
1959's 109 page M.S. Physics thesis by R.A. Finston on measuring xray and nuclear chamber particle energy with vacuum tubes. There is extensive discussion of determining lowest noise operating point properties starting about page 34 thru about page 54. An example three-pentode low-noise preamplifier optimized for their unique usage appears in Figure 7. It's described as a modified cascode with pentode current source. Very low voltages (30, 81, 84 VDC) are used. Pages 34-54 were most interesting for optimizing low noise. There are several tubes listed as having the desirable characteristics, but they were selecting roughly 10 out of 50 tubes as meeting their requirements. They are not audio circuits and they are using extremely small signals, but the principles are interesting. You'll either find something useful in it, or discover what is not worth trying because the range of voltage and current may exclude certain techniques for audio signals. Except maybe tube mics might be closest to this. Very high input impedance, small signals, operation at low voltages. I could not find section on filament voltage this time. The PDF is not searchable and I didn't want to read the whole thing again.
https://www.osti.gov/servlets/purl/4225786
Measurement of Electron Flux in Irradiated Media by AC Methods R.A. Finston
{from 2013...}
$65-$300 old pulp, but archive.org has a scan (32MB PDF download):
https://archive.org/details/in.ernet.dli.2015.462206/page/n3/mode/2up