Source follower replacement for 6C4 w/300V supply

[SoulFetish]

So looking for some advice on implementing this. Not overly complicated, but not used to designing with mosfets as amplifiers (love them as switches and regulators).
Here’s the application: want to replace the 6C4 driver stage in vintage SVTs. 6C4s are miserably noisy, IME, and the 7 pin sockets are often in terrible shape
i have some I some IRL820s to use. But looking for some advice on biasing and choosing a source load.
here is a link to the SVT preamp section to see the operating parameters:
(Open to other fet suggestions if theres a better choice)

:
https://ampeg.com/data/6/0a000509221...plication/pdf/

[SoulFetish]

Just a thought: I wonder if a high voltage n-channel depletion mode mosfet might be the way to go.
the LND150* has a Vds of 500V and is readily available as TO-92 package and cheap.
This might be easier to self bias using a LED, in that the gate is capacitively coupled to the previous stages, being used as a mixer of channels one and two in the SVT. This ultimately supplies the driver/output amp.
it does have an Rds on resistance of 1k, but that shouldn’t be a concern, should it? ( mean we’re talking about milliamps here)

[Helmholtz]

The 6C4 runs at a current of 3.5mA.
Source auto-bias can only be used with negative gate-source bias.
With an LND 150 this would only work up to a source current around 1mA.
Also a dissipation of around 460mW is somewhat high for a 700mW device at elevated ambient temperature.

The IRF 820 needs a positive gate-source bias of around 2V.
So a voltage divider bias supply for around 170V gate voltage is required.

[SoulFetish]

The 6C4 runs at a current of 3.5mA.
Source auto-bias can only be used with negative gate-source bias.
With an LND 150 this would only work up to a source current around 1mA.
Also a dissipation of around 460mW is somewhat high for a 700mW device at elevated ambient temperature.

The IRF 820 needs a positive gate-source bias of around 2V.
So a voltage divider bias supply for around 170V gate voltage is required.

So a divider consisting of R1/470k and R2/600k* with a 50k trimpot connecting them should allow me around +/-5V of 170V to dial it in.
is that overkill, just get close?

Edit: * I happen to have an IRC 597.6k resistor for some reason. Figured we’d call it 600k if y’all will forgive the non-standard value

[Helmholtz]

Seems the forum doesn't update correctly. I saw an email that you gave me a like, but it doesn't show.

I don't think a +/- 10% deviation from the 170V will make a difference.
Consider that the 2 divider resistors in parallel load the preceding circuit and coupling cap.

[SoulFetish]

Good point!

[R.G.]

Generate +170V with resistors. Bypass this to ground with a cap to keep it clean. Place a 100R resistor in series with the gate of the MOSFET to suppress VHF and/or UHF self-oscillation. Move the 1M to run between the 170V bias source to the 100R series resistor. The source should sit at 170V minus a couple to a few volts. Precision is not needed.

The MOSFET tab will be live to +300V. The MOSFET will dissipate 300V-170V = 130V times the drain current of about 3.5ma, or about 0.46W. If you used a TO-220 MOSFET, it can dissipate up to 1.5 to 2W in air without a heat sink.

The MOSFET Follies article at geofex has more info on MOSFET followers in tube amps, btw.

Edit: forgot - IRF820 should be fine.
Edit: dang, I need to keep my mental threads straight. Using a resistor divider bias and a series 1M to the gate, where the coupling cap will come in also removes any loading from the biasing resistors. It will be remarkably like the original as regards loading the previous stage.

[SoulFetish]

R.G. I appreciate you chiming in. The ‘follies were the source of the inspiration, but I wanted to better understand how to design a voltage follower using an enhancement mode device if i had to. Not used to that kind of operation in amplifier stages.


edit: this article looks like it has some helpful info, even though the circuits are voltage amplifiers -
http://www.geofex.com/article_folder...t/mosboost.htm

is this what you mean as far as looking at the gate input circuit?

[Helmholtz]

It will be remarkably like the original as regards loading the previous stage.

1M input impedance surely is good enough. But due to bootstrapping the input impedance of the original tube circuit is much higher.
Shouldn't make a noticeable difference, though.

[R.G.]

is this what you mean as far as looking at the gate input circuit?

Yes. The one exception is that the schematic should show a separate 100 ohm gate stopper resistor in addition, between the Cin and the gate. For a follower, Rdrain = 0. The gate goes wherever the bias voltage and signal tells it to. The transconductance of power MOSFETs is in the amperes per volt range, so the bias voltage won't change much for any sub-1A followers.

[R.G.]

1M input impedance surely is good enough. But due to bootstrapping the input impedance of the original tube circuit is much higher.
Shouldn't make a noticeable difference, though.

Yes, right on both counts. It won't matter, and the bootstrapping will be even higher impedance.

It's possible to bootstrap the MOSFET circuit as well: a cap from the source voltage to the bias string resistors puts the output voltage on the other side of the 1M biasing resistor. I suspect that this might get the input impedance well up to and possibly above the impedance of the tube circuit.

[SoulFetish]

There’s 2 ways to look at it I figure: on 1 hand, this stage functions as a mixer input, and buffer. Bootstrapping decreases the channel isolation your looking for at the input. Also, SVTs of this vintage can suffer all kinds of noise issues, and sticking an input in the 10’s of Mohms right in the middle of the amplifier could be more of problem than solution.
on the other hand, I don’t know if I’ve ever seen anybody play through both channels unless they’re jumping them, so Little channel interaction might not be a huge problem in this case.

[Helmholtz]

Also, SVTs of this vintage can suffer all kinds of noise issues, and sticking an input in the 10’s of Mohms right in the middle of the amplifier could be more of problem than solution.

Actual circuit impedance is determined by the source impedance in parallel with the follower's input impedance (Thevenin's rules), so more like 100k.

[Helmholtz]

It's possible to bootstrap the MOSFET circuit as well: a cap from the source voltage to the bias string resistors puts the output voltage on the other side of the 1M biasing resistor. I suspect that this might get the input impedance well up to and possibly above the impedance of the tube circuit.

I agree.
But then you couldn't use the recommended bias filter cap across the lower voltage divider resistor - or would need to add some "isolation" resistor.

[SoulFetish]

Actual circuit impedance is determined by the source impedance in parallel with the follower's input impedance (Thevenin's rules), so more like 100k.

I was referring to RG’s suggestion to feed signal back from the source and boot strapping it.