LND150 MOSFET as first gain stage

[xtian]

I'm preparing to build a half-wattage (two EL84s instead of four) Liverpool, using Blues Junior iron. My OCD hates the idea of the unused triode, and I'd like to delete one preamp tube, so I'm experimenting with replacing the first gain stage with a LND150. Because it's nearly impossible to overdrive the first stage with humbuckers, a linear MOSFET seems like an acceptable choice for the first gain stage.

I ran a successful test this morning, using my Black Butte Desktop #4 amp as a test device. It went perfectly. I'm using 100K drain, 1K source, and 33K gate.

IMG_2892.jpeg

In this example schematic, what purpose do the Zener pair serve? I've seen other drawings with 12v Zeners there. And is there any benefit to bypassing the 1K source resistor with a cap?

LND150 2Q Preamp.gif

[martin manning]

The Zeners protect against Vg-s getting anything larger than ~8V. Bypassing Rs with a cap has pretty much the same effect as in a triode stage.

[Roe]

vox had a budget ac30 - the ac30s1 - that used FET for the first gainstage. it seems to have worked fine

https://www.thegearpage.net/board/index ... d.2009054/

[xtian]

The Zeners protect against Vg-s getting anything larger than ~8V. Bypassing Rs with a cap has pretty much the same effect as in a triode stage.

And is the only possible source of this voltage whatever the user puts into the input?

[martin manning]

In that schematic, yes.

[Ten Over]

The LND150 has a maximum Gate-to-Source voltage of +/- 20V. Unlike tube maximum ratings which we routinely exceed, they are serious about the LND150 ratings and the device will fail if you exceed 20Vgs. Heat will also destroy the LND150 in short order.

The Reverse Transfer Capacitance is the equivalent of the Grid-to-Plate capacitance in tubes. Despite what the data sheet says, this capacitance is nearly identical between the LND150 and a 12AX7.

This is a handy graph:

[R.G.]

And is the only possible source of this voltage whatever the user puts into the input?

Yes - including scuffling across a carpet and then touching the end of the guitar cable to plug it in. Several kV is pretty common for things like that. Then there's the static electrical gradient, about 120V/meter on a dry day. The LND150 has a specified maximum of +/- 20V on the gate to the channel, so raising the end of a cable by 20/120 = 1/6 of a meter could theoretically puncture the gate isolation. Then there's mistakenly plugging in the cord to a phantom-powered socket on a mixing board.

Putting the zeners in is a far superior way to do the circuit.

Interesting side-trip. Static electricity zaps happen on all input cords. It's common for old bipolar transistor gear to get noisy. It turns out that zapping a bipolar base is non-fatal, as the junction recovers, but every zap/breakover of the base-emitter junction of a bipolar permanently degrades the trasistor's noise figure. Fixing old noisy bipolar inputs can be as simple as replacing the input transistors with a new low-noise device. If you ever get into this, add a 1N4148 or similar in parallel with the base-emitter junction, normally reverse biased. This completely stops static zaps from breaking the base-emitter and degrading noise performance.

[Ten Over]

And is there any benefit to bypassing the 1K source resistor with a cap?

The 1k Source resistor is there to set the gain for that stage. You wouldn't want to bypass that resistor because it would dramatically increase the gain.

[bepone]

1k is there for bias........

[martin manning]

You get a bit more boost bypassing the source resistor than you do with a 12AX7, but you can do it, and use the bypass cap to shape the response. With 1k and no bypass you get 33.5 dB, fully bypassed you get 40.8 dB. If you want less gain, increase the source resistor.

[Ten Over]

1k is there for bias........

No. It is biased way off center. It is there to set the gain.

[xtian]

1k is there for bias........

No. It is biased way off center. It is there to set the gain.

How would we center bias this input stage?

[martin manning]

Try a 470 ohm source resistor, and measure voltage on each end of the drain resistor.

[Ten Over]

How would we center bias this input stage?

For a 300V supply:

EPSON639.JPG

The voltage swing is from 30V to 300V, so the center voltage is 165V. The Vgs at the intersection of 165V and the load line is -0.12V. The current at the intersection is 1.37mA. The required resistor is 0.12V/.00137A=88 Ohms. The AC load lines were not included in order to avoid a confusing cluster of frequency dependent lines. The value of the source resistor would have been about the same with mid to high frequency AC load lines.

The grid lines get bunched-up at the bottom of the graph, so it might be desirable to reduce the value of the source resistor in order to get a more symmetrical signal. The second stage is biased at 0.0V for this reason.

[bepone]

1k is there for bias........

No. It is biased way off center. It is there to set the gain.

affecting the gain but is for the bias, simple, depletion mode fet