> have NO idea where I got the idea to put a coupling cap between the plate of the previous triode and the grid of the mosfet?
Cap and _no_ gate resistor. That's the dubious/bad idea.
You didn't get it from R.G.'s page: all those gates connect to the previous plate or to a DC voltage through a resistor.
In a TUBE circuit, _ALL_ grids have a resistor to some DC voltage. Usually zero/ground. Sometimes a bias voltage. Sometimes to some part of a larger circuit which will assume a DC voltage. But any grid which has NO resistor (infinite resistance to anything) is gonna give trouble.
It is in fact 90% the same as a tube. Look at 5F6a. The cathode follower is "direct connected" to the plate before it.
OK, do it your way. What is the DC voltage at the gate; also at the source ("cathode") resistor? Let's see: coupling capacitor at left blocks all DC. Gate is a thin layer of glass, blocks all DC. The gate voltage could be zero to infinity (plus or minus infinity!). The source could be zero DC voltage (zero current, won't work) to +300V (resistor may cook and MOSFET clips half the signal). The stage's bias is UNDETERMINED.
That's as professional as showing-up to a gig with "undetermined" pay. You usually want some general idea: free beer? $5/4 hours? $7,000/40 minutes? You must pay to play? Likewise the amplifier stage "wants" to be 100V-200V or so, not "undetermined".
> had a 12v zener there and the amp would play for a while and then start to cut out and lose volume. IF I banged a power chord, the volume would return.
> I traced the problem down to the zener and removed it. Played the amp maybe an hr and 15 minutes without the zener and it worked great and sounded wonderful.
That's about what I'd feared. Yes, with good luck and no bad moon, the "undetermined" bias MAY work. With a JFET (you can't find hi-volt JFETs) the device leakage is kinda-sure to find an off- medium bias, and a few hard slams may keep it centered. With an unprotected MOSFET, however, the leakage is too tiny to know what will happen, and a few hard slams will probably pop the gate. I suspect you have a protected MOSFET with some gate leakage.
With one external Zener, the leakage will tend to drift the bias one way, eventually to a non-working condition. A few hard slams may re-center it into operation, only to drift again.
A brutal analogy: if you disconnect a car steering wheel, the car still goes down the road a while, but "undetermined". It is pretty sure to drift. If there's walls along the road, the slams will tend to center the car in the lane, unless it drifts gently and ends up scrubbing the wall.
"Steer" your circuits so the output is somewhere the center of the power supply. With plate-output triodes, you grid-leak to ground, use 100K and 1K resistors with 12AX7, and it will self-center near-enough. With cathode/source followers, you set the grid/gate near the center, and the cathode/source, uh, follows.
> Replace the cap with a piece of wire.
> Normally, you might use 2 resistors at the gate to define the bias
Either way can work. In this case, you know the previous plate is at a reasonable DC voltage. Do like 5F6a and just direct-couple it. In other cases the previous stage is not at suitable DC voltage (after a volume/tone control or input jack) and you use a cap and a couple 1Meg resistors to get the gate up near half-supply. (There's another way for vacuum triodes and JFETs but won't work for common MOSFETs.)
Yes, put the Zener back. When the gate is properly guided to a reasonable DC voltage, the Zener does no harm and WILL save grief. The Zener was not the problem; it was the lack of any idea where the gate should sit.
Topic: Tweed Overdrive w/ 2 CF (Read 12411 times)