![[-]](https://theultimatetone.com/images/collapse.png "[-]")
01-01-2021, 08:48 PM
Hi Noel
Note that the input stage is not a cascode; rather, a standard common-source gain stage with a current-source load. The tube contributes nothing to the gain per se and does not limit it either - the gain is all from the LND150. Output impedance is essentially the same as for a cathode-follower as 1/gm - pretty low even with a 12AX7. This means you can change how the signal is attenuated into the lower channel and reduce some potential for noise.
The alternate way to pad the gain and drive controls - any level control - is to add a shunt resistance across the pot to work against the series R feeding the pot. The lower part of the divider is then the prallel of R-shunt and the pot, where R-shunt can be used to swamp the ot's value out of the divider calculation in situations where you may be constrained to using high-value pots. For example, a 1M pot parallel with 100k is 90k net. If R-series is 90k then the divider ratio is 0.5 - signal-max cut in half. If R-series is 900k, then the ratio is 0.09 (1 / 11). You could make R-shunt much lower, say 33k, then R-net is 32k. Now R-series can be reduced in value to 320k for the high attenuation of 1/11 - a bit lower noise.
Was the input stage stock, or did you modify it to be as it is? I suspect a straight 12AX7 stage would be fine here, both gain-wise and noise-wise, and it might improve control sweep utility.
In both channels the Drive control does not go to zero, with signal feed-through allowed by R75 (upper) and R82 (lower). The lack of grid-stops for most stages is a bit worrisome, but as you say, one channel is copied from Mesa practice.
The mosfet source follower depicted uses a voltage divider to set the gate voltage and thus the source voltage for the output. This makes deciding on the source current independent of the mosfet parameters. The divider sets Vg and Vs at B+/2, about 165V. You set R15 to 47k so current is 3.5mA and Pd is 577mW - maybe a bit hot for the LND150.
Input impedance of the divider-bias follower is simply the parallel combination of the divider elements. Since both are 1M, then the net is 500k, making the effective load on V2B 175k. The mosfet gate impedance approaches infinity, same as a triode, so it can be ignored. Output impedance is device-dependent and will be 1/gm.
Personally, I do not like bright caps on level controls as it makes the tone versus sweep irregular. The two channels have lots of gain and either can easily be an overdrive channel. They are out of phase though, which some people care about and some do not.
It's usually sunny here even in winter, so winter is quite bright
Have fun
Note that the input stage is not a cascode; rather, a standard common-source gain stage with a current-source load. The tube contributes nothing to the gain per se and does not limit it either - the gain is all from the LND150. Output impedance is essentially the same as for a cathode-follower as 1/gm - pretty low even with a 12AX7. This means you can change how the signal is attenuated into the lower channel and reduce some potential for noise.
The alternate way to pad the gain and drive controls - any level control - is to add a shunt resistance across the pot to work against the series R feeding the pot. The lower part of the divider is then the prallel of R-shunt and the pot, where R-shunt can be used to swamp the ot's value out of the divider calculation in situations where you may be constrained to using high-value pots. For example, a 1M pot parallel with 100k is 90k net. If R-series is 90k then the divider ratio is 0.5 - signal-max cut in half. If R-series is 900k, then the ratio is 0.09 (1 / 11). You could make R-shunt much lower, say 33k, then R-net is 32k. Now R-series can be reduced in value to 320k for the high attenuation of 1/11 - a bit lower noise.
Was the input stage stock, or did you modify it to be as it is? I suspect a straight 12AX7 stage would be fine here, both gain-wise and noise-wise, and it might improve control sweep utility.
In both channels the Drive control does not go to zero, with signal feed-through allowed by R75 (upper) and R82 (lower). The lack of grid-stops for most stages is a bit worrisome, but as you say, one channel is copied from Mesa practice.
The mosfet source follower depicted uses a voltage divider to set the gate voltage and thus the source voltage for the output. This makes deciding on the source current independent of the mosfet parameters. The divider sets Vg and Vs at B+/2, about 165V. You set R15 to 47k so current is 3.5mA and Pd is 577mW - maybe a bit hot for the LND150.
Input impedance of the divider-bias follower is simply the parallel combination of the divider elements. Since both are 1M, then the net is 500k, making the effective load on V2B 175k. The mosfet gate impedance approaches infinity, same as a triode, so it can be ignored. Output impedance is device-dependent and will be 1/gm.
Personally, I do not like bright caps on level controls as it makes the tone versus sweep irregular. The two channels have lots of gain and either can easily be an overdrive channel. They are out of phase though, which some people care about and some do not.
It's usually sunny here even in winter, so winter is quite bright
Have fun
