backwards mosfets

[duck_arse]

I just tested my 2sj49/2sk134 power mosfets as a single supply complementary pair with a fet buffer in front, a bit like a Dis'Creep. they worked fine, just like I was expecting, just like they was a bs250/bs170 pair. nice rounded valvesound output on the cro, with gain. only, when I look at the eti477 amplifier, for instance, the N channel is to the positve supply and the P channel is to the negative.

is there a reason "old smokey" was connected what looks like backwards? and which side is most "up"?

[R.G.]

All MOSFETs have a built-in reverse diode from source to drain. N-channels have to have the drain more positive than the source if you want them to act like MOSFETs and not diodes. P-channels have to have the drain more negative than the source. There are odd exceptions for very low voltage situations like synchronous rectifers and such.

That being said, you are free to use N and P MOSFETs as an output stage as either common drain (source follower) or common source configuration.

The follower setup will have the N-channel from the positive supply to the output on the source, and the P-channel with drain to negative supply, source to output.

The amplifier setup will have the P-type with source to V+, drain to output; the N-type will be drain to output and source to V-minus.

The difference is that the follower setup is a follower, the amplifier has voltage gain.

[duck_arse]

thanks rg, I promise I will read this and compare to some circuits untill it makes sense.

[R.G.]

Then I didn't do it right.   

Think about an N-channel MOSFET used as a source follower. It's drain is connected to V+, the output is on its source. If you want voltage gain, you hook an N-channel up so its source is to the most-negative voltage and the output is on its drain. The common source connection has gain, the follower does not.

In each case, you can hook up a complementary P-channel equal-but-inverted to the opposite power supply and have a complementary circuit. If the N-channel is drain to V+ and the P-channel is drain to V-, and both sources are tied together and used as an output, it's a follower, with unity gain. If the N-channel has source to V-, P-channel has source to V+, joined drains with output from the drains in the middle, it's has gain.

[duck_arse]

thanks again, but I'm STILL going to try and understand it .....

I'm taking the advice you gave me earlier, and I'm going to flog sell my 4 unused mosfets for the most I can gouge for what I can get for them on the eebay, and then pull some from a junked amp to make the originally planned booster. I say I'm going to, but we'll have to wait and see.

[PRR]

> I'm STILL going to try and understand it .....

R.G. told it straight. (The bit about the diodes was a semi-digression.)

Let me hammer a different way.

Imagine _one_ transistor. And say it is a BJT (just easier for me to think).

Do you know the difference between a BJT Emitter Follower (common collector) and a BJT Voltage Amplifier (common emitter)?

Emitter Follower has Unity voltage gain and low distortion.

Voltage amplifier has significant voltage gain and significant distortion.

When you have both N and P type devices, you can put two together flip-flop push-pull. Still the emitter follower has unity voltage gain, and the voltage amplifier has voltage gain.

In *hi-fi* amplifiers, Emitter Follower (or Source Follower) is the usual way to work the output stage. The low distortion is key. The (nearly!) unity voltage gain is a slight problem, because the stage before the output must make a large swing with low distortion and preferably without extra power supplies; but the problem can be solved satisfactorily.

When low-low distortion is not a prime goal (or when voltage gain is vital), you wire as Voltage amplifiers. CMOS logic is a classic because logic needs HIGH gain in minimum stages (and huge distortion is not a problem).

The main problem with voltage-amplifier connection in complementary push-pull is getting a good bias. In CMOS the center bias is allowed to be high and poorly controlled because in logic use it will never linger near center. In audio we linger near center most of the time. With unity-gain connections we can simply stick a voltage source between the bases/gates. With amplifier connection we need two voltage sources one for each base/gate, because the "input" is between base/gate and supply, not base/base.

[duck_arse]

I do know the difference between follower and amp, yes. I **STILL** can't remember a decent way to remember how to "polarise the arrows", even after 35 years of looking at them. pnp/npn/n-ch/p-ch ..... I need diagrams to get them right way round. I re-read what rg wrote, and drew a diagram, and it seems my first post may have been inverted, because it didn't add up straight when I looked.

I'll try again, w/ data sheets and diagrams.