Output impedance of various MOSFET followers at low (<10mA) currents

[bigwill]

MOSFETs are great for followers, and supposedly can drive heavy loads, but what are values of output impedance one can expect for the average MOSFET such as an IRF820? The Gms of such devices are huge in the whole amp rages, but the mA ranges are just too small on the scale of the drain curves in the datasheets!

[cerrem]

Simply the output impedance is: 1/gm in parallel with Source resistor....
Doesn't make sense to use this device in such an application in the low mA ranges...
The voltage swing will push this into cut-off in typicall audio signals, when biasing it this low, I would have to see the circuit....
At 500mA your gm is about 1.1 S
At 300mA your gm is about .4 S
At 100mA your gm is about .1 S
Very non-linear in this region.... The only way to linearize it somewhat is with a matching resistor in the Drain as you have in the Source....This will maintain constant voltage across the device durring operation and keep the gm more stable....
Any lower than this your gm is in the gutter....
I suggest to pick another FET....

Chris

[Miles Prower]

Originally Posted by bigwill

MOSFETs are great for followers, and supposedly can drive heavy loads, but what are values of output impedance one can expect for the average MOSFET such as an IRF820? The Gms of such devices are huge in the whole amp rages, but the mA ranges are just too small on the scale of the drain curves in the datasheets!

Don't worry about it. For my current project (845 SET) I intend to use the FPQ2N60C N-Channel MOSFET as a source follower grid driver. According to the spec sheet, this one has: Gm= 5000mA/V @ Id= 1000mA. I figure that a static Id= 34.5mA with a 10K tail resistor. Guesstimate the Gm:

Gm(34.5)= 34.5(5000/1000)= 172.5ma/V

Even if this is off, it still beats the living hell out of just about any VT. Figure the Zo:

Zo= 0.1725 || 10E3= 5.8R

Even if "off", that still comes quite close to the ideal Zo= 0, and much closer than any VT.

As for linearity, the 100% NFB inherent to follower design takes care of that.

[bigwill]

Originally Posted by Miles Prower

Don't worry about it. For my current project (845 SET) I intend to use the FPQ2N60C N-Channel MOSFET as a source follower grid driver. According to the spec sheet, this one has: Gm= 5000mA/V @ Id= 1000mA. I figure that a static Id= 34.5mA with a 10K tail resistor. Guesstimate the Gm:

Gm(34.5)= 34.5(5000/1000)= 172.5ma/V

Even if this is off, it still beats the living hell out of just about any VT. Figure the Zo:

Zo= 0.1725 || 10E3= 5.8R

Even if "off", that still comes quite close to the ideal Zo= 0, and much closer than any VT.

As for linearity, the 100% NFB inherent to follower design takes care of that.

Cool. I'm sure my IRF820s will be fine, although I will be running them at only around 6mA - I'm making a Pimm style CCS/mu-follower. I may have to rethink the design, because even taking into account what you said, I think 6mA may just be too low for this MOFSET. If I was driving output grids then obviously I'd have no problem with running at higher currents!

Perhaps I could use a signal device as the lower mosfet (in a way similar to Gary Pimm's battery biased CCS)

[bigwill]

Originally Posted by bigwill

Cool. I'm sure my IRF820s will be fine, although I will be running them at only around 6mA - I'm making a Pimm style CCS/mu-follower. I may have to rethink the design, because even taking into account what you said, I think 6mA may just be too low for this MOFSET. If I was driving output grids then obviously I'd have no problem with running at higher currents!

Perhaps I could use a signal device as the lower mosfet (in a circuit similar to Gary Pimm's battery biased CCS but that uses IRF820s for both the upper and lower devices of the cascade)

[Miles Prower]

Originally Posted by bigwill

Cool. I'm sure my IRF820s will be fine, although I will be running them at only around 6mA - I'm making a Pimm style CCS/mu-follower. I may have to rethink the design, because even taking into account what you said, I think 6mA may just be too low for this MOFSET. If I was driving output grids then obviously I'd have no problem with running at higher currents!

Perhaps I could use a signal device as the lower mosfet (in a way similar to Gary Pimm's battery biased CCS)

If that's what you're trying to do, then how about a Kimmel Mu Stage design instead? That way, you can run the MOSFET at a higher current, and still give the VT the 6mA it wants.

[gingertube]

If driving a pair of EL84 or other easy to drive tubes then you don't need huge current in the MOSFET follower. 3mA is ample for EL84 drive. The problem with MOSFET Source Followers is that the input capacitance can vary with signal as Vds changes. If you look at any MOSFET data sheet you will see that this capacitance varies a lot for Vds between 0 and about 20 volts but is fairly constant for Vds above 20V. So design such that on peak positive signal swing at the source you still have 25 to 30 V across the MOSFET. Also choose a MOSFET just large enough for the power dissipation expected. Larger MOSFETs have large capacitances. I like little ZVN0545 for this reason.
The other major improvement you can make is to current source load the source follower. A simple "ring of two" BJT current source is adequate.
My 2 cents worth!
Cheers,
Ian

[Bandersnatch]

hey-Hey!!!,
One method to get high gm( at low current) with high voltage and high heat dissipation limits is to cascode; small device on the bottom, and large one on top. It will also do away with the vairable capacitance due to d-s voltage change.
cheers,
Douglas

[bigwill]

Originally Posted by Bandersnatch

hey-Hey!!!,
One method to get high gm( at low current) with high voltage and high heat dissipation limits is to cascode; small device on the bottom, and large one on top. It will also do away with the vairable capacitance due to d-s voltage change.
cheers,
Douglas

I have considered this but I don't know any parts that have high Gm at low currents, unfortunately

Care to recommend any easy to find devices?

[Bandersnatch]

Originally Posted by bigwill

I have considered this but I don't know any parts that have high Gm at low currents, unfortunately

Care to recommend any easy to find devices?

Check out Allen Wright's stuff; he uses a small FET on the bottom of his cascode circuits, a DN2540N5 on top should take care of the rest.
cheers,
Douglas