Single mosfet amp

Yes this will work, but remember to AC couple it to the load since you'll have an offset at the MOSFET's output. Also an IRF610 isn't going to last long at that dissipation you are talking about. Try an IRF540 if it must be in a TO220 package.

If possible, you might include the mosfet inside a NFB loop to help compensate for the non-linearities from the mosfet. Hexfets(like some of the IRF types) are quite non-linear at lower currents, and the capacitances inside the mosfet(Cgd, Cgs) change with operating point but certainly not in a linear fashion. A driver for the Fet is another idea. I'm not so familiar with tubes, but driving the Fet at higher frequencies will be more like driving a capacitor, so if the tube isn't up to that challenge, a driver would effectivly reduce this capacitance that the tube stage sees. Keep in mind that you have to drive the gate voltage up, and then pull it down to turn off the device....like a totem pole configuration or something similar.

Generalisations...

A couple of things that are similar about a FET and a tube are:
1. The output current (Id) is related to the gate-source voltage.
2. There is virtually no dc current into the gate

Not so similar:
3. The output resistance (dVds/dId) id is much higher for a FET
4. The maximum useful output current and power dissipation is much higher
5. The parasitic capacitances, Cgs, Cdg, Cds are MUCH, MUCH higher in a FET (2 or three orders of magnitude) and are VERY non-linear.
6. FETs operate at much lower voltages, but characteristics start to go weird when Vds is low, meaning 5V or less.
7. The transconductance is probably higher for a FET at the Id you are likely to use but I'm not familiar with tubes.

If you are used to using tubes I would look into these characterisc differences very thoroughly. You can find this info on most datasheets.

I suspect the linearity and the capacitance differences will be the biggest shock. There is a tendency for people to think a FET is like some half-way device between a bipolar and a tube, but IMO this is far from the case and so just as much care is needed as if you were using a bipolar.

Simpleton said:

The tube voltage amplifier runs at aproximately 3mA, probably just enough to run the mosfet. Don't the capacitances change the low frequency response, not the high? Maybe a opamp buffer stage before the mosfet will do the trick is necessary.

Click to expand...

I was not refering to the frequency responce as I was the drive required. The capacitance between the gate and source-drain channel is what the driver must drive. It is easier to drive a fet at lower frequencies because that impeadence is higher. At higher freqs, the impeadence drops and it takes more current to drive the fet, so I guess it would effect the frequency responce as well. With all of the different non-linear variables, sometimes placing the device inside the NF loop will allow the error correction of the circuit to help with this...but can also cause instability. This is why hexfets are popular as switches where they are either on or off like in SMPS type circuits instead of linear applications. That is not to say they can't be used, but it requires more complexity. There does exist lateral Fets that are more linear but they typically cost more. If you research the construction of a hexfet, you will see why they have non-linear properties.

Simple circuits are the building blocks of the more complex ones.
I've split the thread, please continue with the interesting part.

/Hugo

In this circuit will be mosfet destroyed after switching of mains... And linearity of tube by 25 V ? Funny question....

Some tubes run pretty fine at low voltages, no problem there. Seems nobody gets the fact that simpleton has said he only wants a few watts, mW actually.
This setup will easily swing enuff for at least 2watts into 8ohms.
But: You must nor AC couple the MOSFET, it needs a DC voltage at the gate. This sets the bias of the MOSFET together with the CCS or resistance in its source. So lets say the tubes anode sits at approx 15volts, so will the gate of the MOSFET since you direct couple it. Then you will have about 4volt drop to the source pin, and about 11volts to play with. With a resistor, it will be a very simple, but likely decent sounding for a watt or so. Into an 8ohm load it will get rather unlinear as the voltage swings more and more, but mostly second harmonic, which will sound pleasing so no bigdeal. Using a LM317 might make it more linear, but give less available voltage swing. I d go for a simplest possible solution, a resistor is easily replaceble later.
Hey, this is a simple and cheap way to have fun, go for it. Btw you mention in one post the tube runs at 3mA, but then that you'll have a 20kohm anode reistor. Not possible with a b+ of only 25volts ya know.

You cannot use ideal load lines with a resistor load and such low supply voltages. The idle of 3mA gives you a anode resistor of 7V/3mA = 2260, so use 2200ohms. The cathode will have a resistor of 0.5V/3mA = 160ohms. (Rounded off to standard values)
This gives a gain of about -4 to -5. (About +12dB gain).
Bypassis Rk only gives a slight increase in gain, but limits the input range to about max 1voltpp.
Using a 2200ohm resistor gives a relatively low impedance to drive the MOSFET so this should be good. You dont need much gain in a small power amp like this, so I think you'll be pleased. With a resistor at the MOSFET source of 8ohms the IRF530 (or equivalent) will have an idle of a little over 1.6 amps. Perfect. Just remember the resistor disipates more than 25watts and must be seriously heatsinked, same for the MOSFET. Id rather chose a IRF150 because teh larger package can more easily transfer the heat to the heatsink.

Another point of using a resistor to bias the MOSFET rather than a CCS is that the tube will have a nonsymetrical voltageswing at the output that is more negative than positive.(second order harmonics). This because it turns itself on more easily than off and the output is inverted. The exact opposite will happen for teh MOSFET since the passive CCS the resistor is, will make the MOSFET go more on than off, but this time noninverted, so the two devices distortion will cancel a little bit.

trying to post schematic of your amp. FYI, the peak dissipation will be almost 50watts for both the MOSFET and resistor!

man Im talkin to myself again! But realized you only need to bias the MOSFET at 1amp or so if all you need is 2watts. That means you can get away with a 15ohm resistor which then only dissipates 13watts quiescent and some 20 peak. Much more pleasant.

You know what? I feel like building something really simple tonight, and its May1st tomorrow, so a workers amp is long over due. Im gonna make this now and post results in an hour or so...I think I have an old Phillips ECC86 laying in some dusty box...