ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

previous: Stephen Conner Pentode cathode follower? What dat? -- 2/3/2000 12:40 PM

pentode cathode follower explained

2/3/2000 3:44 PM
kg	pentode cathode follower explained hi steve, in order to operate as a pentode, the potential between the screen grid and the cathode must remain fixed. this is not normally a problem in plate loaded stages, since the cathode sits at pretty much a fixed potential anyway. thus bypassing the screen to ground accomplishes this exact thing--but, in a cathode follower, the cathode is swinging all over the place. this method will not work. with a pentode connected cathode follower, you've got to cap couple the screen to the cathode. obviously this also requires that the screen be fed through some series impedance--otherwise the cap would be just shunting ac to the screen supply rail. when connected in this way, the performance of the CF goes way up. this is due to the limitiations of the triode connected CF, in that the Vak is constantly changing with signal, and therefore the tube "runs out of steam" on the positive peaks where the effective plate voltage is lowest. with a pentode, the screen will always remain a fixed potential above the cathode, which allows it to swing much higher on the positive peaks. the cathode's voltage can be pulled much closer to the plate's, and the peak current capability goes way up. obviously neither of these connections will ever net you voltage GAIN from a CF, but the key is that the gain _does_ go up... closer to unity. the mu of a pentode is much higher than that of a triode, and that factors into the equation for ultimate gain/loss through the CF. so essentially there's a smaller Vgk (bias voltage, you might call it) on the positive excursions when in pentode mode. there's also less distortion, since the effective "plate voltage" (meaning Vg2-k, in the pentode case) does not change. in terms of sizing the cap and resistor, your best bet is to use the highest tension in the amp, with the largest series resistance you can get away with to feed the screens. rememeber that the screen current is typically 1/10th that of the plate. that series resistance will be one factor into the total "R" of the F=1/2piRC equation for the necessary capacitance from screen to cathode. the screen itself also has a dynamic impedance that can be computed from mu=rpgm. if you can find the triode characteristics for the tube in question, you can determine the mu and the gm, since triode mode has the screen acting as the plate. if you've got the plate curves, read the plate impedance right from their slope. knowing the effective dynamic "plate" impedance allows you to parallel this with the value of series screen resistance, which then allows you to size your cap for a given LF breakpoint. in my case, i need about a 100k series screen resistance, fed from +300VDC, and a 1 to 2 microfarad cap to break at <20Hz. what makes it easier for my calculations are the current sources on each el84's cathode, pumping out a steady 15mA per tube. therefore i know that each screen will be pulling about 1.5mA each, which enables me to size the 100k stopper very easily to drop 150VDC. the plates are fed straight from a 150VDC supply. i could have fed the screens from the same supply, but i had this higher rail available, and it will allow a larger series resistance and therefore a smaller cap value for a given TC. the cap will see essentially an unchanging DC voltage across it, from cathode to screen, and in this case is about 150VDC - (-85VDC) or 235VDC. i'll use a 300VDC rated cap for safety. all of the above will net me less voltage loss through the cathode follower, and better performance from the cathode follower as a whole in terms of steadiness and consistency of output Z. NOTE: connecting a CF as a pentode can also be thought of as cascoding the "top" half of the tube, almost like putting a current source on the plate. along with the CCS in the cathode circuit, 2/3rds of the distortion producing mechanisms in the CF are eliminated (variation in gm due to delta Ik, and variations in Vgk due to delta Vp). not that those were design goals for ME, but merely byproducts of performance enhancements. ALSO NOTE: using a MOSFET as a source follower makes this kind of thing moot, since it has such a high dynamic drain impedance to begin with... the Vds does not matter nearly as much as it does in a low plate impedance triode. the MOSFET SF can* be helped by the CCS in the source circuit however. whew. i hope this has been clear enough to follow. ken

2/3/2000 3:44 PM

pentode cathode follower explained

hi steve,

in order to operate as a pentode, the potential between the screen grid and the cathode must remain fixed. this is not normally a problem in plate loaded stages, since the cathode sits at pretty much a fixed potential anyway. thus bypassing the screen to ground accomplishes this exact thing--but, in a cathode follower, the cathode is swinging all over the place. this method will not work.

with a pentode connected cathode follower, you've got to cap couple the screen to the cathode. obviously this also requires that the screen be fed through some series impedance--otherwise the cap would be just shunting ac to the screen supply rail.

when connected in this way, the performance of the CF goes way up. this is due to the limitiations of the triode connected CF, in that the Vak is constantly changing with signal, and therefore the tube "runs out of steam" on the positive peaks where the effective plate voltage is lowest. with a pentode, the screen will always remain a fixed potential above the cathode, which allows it to swing much higher on the positive peaks. the cathode's voltage can be pulled much closer to the plate's, and the peak current capability goes way up.

obviously neither of these connections will ever net you voltage GAIN from a CF, but the key is that the gain _does_ go up... closer to unity. the mu of a pentode is much higher than that of a triode, and that factors into the equation for ultimate gain/loss through the CF. so essentially there's a smaller Vgk (bias voltage, you might call it) on the positive excursions when in pentode mode. there's also less distortion, since the effective "plate voltage" (meaning Vg2-k, in the pentode case) does not change.

in terms of sizing the cap and resistor, your best bet is to use the highest tension in the amp, with the largest series resistance you can get away with to feed the screens. rememeber that the screen current is typically 1/10th that of the plate. that series resistance will be one factor into the total "R" of the F=1/2piRC equation for the necessary capacitance from screen to cathode.

the screen itself also has a dynamic impedance that can be computed from mu=rp*gm. if you can find the triode characteristics for the tube in question, you can determine the mu and the gm, since triode mode has the screen acting as the plate. if you've got the plate curves, read the plate impedance right from their slope. knowing the effective dynamic "plate" impedance allows you to parallel this with the value of series screen resistance, which then allows you to size your cap for a given LF breakpoint.

in my case, i need about a 100k series screen resistance, fed from +300VDC, and a 1 to 2 microfarad cap to break at <20Hz. what makes it easier for my calculations are the current sources on each el84's cathode, pumping out a steady 15mA per tube. therefore i know that each screen will be pulling about 1.5mA each, which enables me to size the 100k stopper very easily to drop 150VDC. the plates are fed straight from a 150VDC supply. i could have fed the screens from the same supply, but i had this higher rail available, and it will allow a larger series resistance and therefore a smaller cap value for a given TC. the cap will see essentially an unchanging DC voltage across it, from cathode to screen, and in this case is about 150VDC - (-85VDC) or 235VDC. i'll use a 300VDC rated cap for safety.

all of the above will net me less voltage loss through the cathode follower, and better performance from the cathode follower as a whole in terms of steadiness and consistency of output Z.

NOTE: connecting a CF as a pentode can also be thought of as cascoding the "top" half of the tube, almost like putting a current source on the plate. along with the CCS in the cathode circuit, 2/3rds of the distortion producing mechanisms in the CF are eliminated (variation in gm due to delta Ik, and variations in Vgk due to delta Vp). not that those were design goals for ME, but merely byproducts of performance enhancements.

ALSO NOTE: using a MOSFET as a source follower makes this kind of thing moot, since it has such a high dynamic drain impedance to begin with... the Vds does not matter nearly as much as it does in a low plate impedance triode. the MOSFET SF *can* be helped by the CCS in the source circuit however.

whew. i hope this has been clear enough to follow.

ken

Replies:

R.G. q{the MOSFET SF *can* be helped by ... -- 2/3/2000 5:13 PM