Mosfets as a simulated pentode?

[PenPen]

It's been some time since I've been here, I've been out of the hobby for a bit and trying to jump back in.

Project I'm cooking up at this point is a small headphone type amp, but I want it to be interesting. Long ago I'd built a "Fetzer Ruby" with a Bassman tone stack, that's long gone now.

However, I was thinking this time I want to just do away with the lm386, and try if possible to replicate my Univox amp using silicon. Years back there was the jFets to replace 12ax7 circuits, those were fun. Now, however, I have my eye on finding some way to replace a 6L6. I read on Mosfet Follies that standard Mosfets are very pentode-like, so I'm looking in that direction. So, how do we make the class AB in the style of the Bassman output stage using Mosfets? Without an output transformer, can this be done with a pair of BS170s? How exactly would these be hooked up, since in Fender output stages the cathodes were hooked up to either side of a transformer, would I have to come up with some kind of substitution scheme to drive the output? What kind of power output could we get with this driving a pair of headphones?

Basically, I've spent all day searching here and Google looking for info on this, just looking for some more pointers and direction on the matter.

Thanks.

[SonicVI]

6L6's are tetrodes, not pentodes.    Not that that helps you in any way 

[Brymus]

That article really intrigued me too.
I would like to know what to substitute(mosfet) for EL84 and 6V6.
Sorry I cant help either.
I bet RG knows...

[Cliff Schecht]

6L6's are tetrodes, not pentodes.    Not that that helps you in any way 

Well. The 6L6 is actually a beam power tube, which was designed as a way around using the conventional pentode. They have a lot of advantages over standard pentodes and have very similar curves. It's not wrong to call a beam tetrode a pentode, but it's wrong to call any pentode a beam power tube!

To answer the question at hand, yes FETs all  (usually) have very pentode like curves. Speaking of which, has anyone ever tried a tube as a voltage controlled resistor? It's voltage in and current out with a small portion of linear operation (like FET's, near the cutoff) and I assume it's been done before, I just don't know where or how. Anybody?

[Renegadrian]

There's a ROG project that is just what you said - a MOSFET for a pentode.
Look for the MATCHBOX.

[PenPen]

Okay, yes the 6L6 is a beam tetrode as a way around the conventional pentode patent. They behave mostly the same way.

Thanks for the reference to the Matchbox, I'm familiar with that design but that isn't what I'm looking for really. When you compare the Matchbox to the DC/30 it's not set up the same way, I'm looking for any reference to using MOSFets as power amp stages, such as in my Univox amp (which is very similar to a Bassman, so for the sake of discussion that's what I am referring to) power stage of a pair of 6L6 tubes in a class AB arrangement.

Does this not work as I expect it should?

[R.G.]

Power MOSFETs are indeed very much like power pentodes or power beam tubes when you look at the drain and plate curves side by side.

However (you knew that was coming, right?  ) there are differences that take some dinking with, and the results are less than perfect.

The big differences between a power MOSFET and a pentode/beam tube (which are, frankly, identical to the outside user in almost all cases) are
- big gate capacitance compared to smallish grid capacitance
- much higher transconductance for the MOSFET
- different behaviors at the 'edge' conditions.

Big gate capacitances mean that the driver circuits need changed to drive the gates, and that leads you off into a chain of tinkering trying to get back to the way tube amps operated. Lots of places to go astray.

Higher transconductance is easier to cope with - you simply change the drive voltage smaller (which is the easy direction) to feed the things a corresponding voltage input.

The different behaviours at the edges is the real toughie. In listening to a tube amp, we pick out as different from solid state amps not so much the ordinary operation where everything is within bounds, but the way it sounds when it's overdriven. For a tube power amp, that's when the power tubes are running the output transformer right up at the edge of what they all can do. The transformer bandwidth decreases, it clips (that is, saturates) on the lowest notes, the tubes run out of current to drive it, the leakage inductances in the transformer are highly active, the tube grids are driven near or into forward conduction, and other minutiae. It gets difficult to replicate all the jots and tittles.

One place a power MOSFET does work nicely in tube power amps is to provide a constant current drive with no signal on it for "converting" a push-pull output transformer to single ended. The reason single ended is so hard on OTs is that there is a net DC current. You can get the single ended operation of one output device in a non-specialized OT intended for push-pull by feeding a MOSFET (or indeed, a tube) a constant bias current on one side, but no signal. The other side is fed a signal and biased so that it sid doing all the signal work. You get the sound of the single output device, although not the one-sided operation of the OT over the ful B-H range.

[PenPen]

It brings a smile to my face to get an R.G. post in my thread. Thanks for the analysis, basically I was looking for "yeah, what you are wanting to do has been done before, see xyz", or "no, this isn't going to work the way you expected it to, there's no easy way to do this".

I have a project to work on now. If I get something decent I'll post the results. I just didn't want to start beating my head against a wall then find out later that it had already been done and I could have saved myself a whole lot of time.

Thanks RG!

[Gus]

http://www.bonavolta.ch/hobby/en/audio/2sk135se.htm

[davidallancole]

Here is schematic I found for a push pull MOSFET amp being fed through a transformer.  It uses a differential front end feeding a set of MOSFETs in push pull.

http://www.turneraudio.com.au/solidstateamps4-50w-mono-mosfet.htm

R.G.

Does the 2 dc currents in a push pull transformer cancel each others magnetic field?  Is that why the bias currents don't push the BH curve towards saturation?

[PenPen]

Very interesting. So it looks like I'll have to incorporate a small transformer into this for the headphone amp if I'm to keep it to a literal reproduction. That's one method to try, thanks for the info!

[R.G.]

Does the 2 dc currents in a push pull transformer cancel each others magnetic field?  Is that why the bias currents don't push the BH curve towards saturation?

Yes. The two DC currents to the two output tubes in a push-pull transformer are roughly equal and in opposite magnetic directions, leaving the transformer core nearly at zero magnetically.

This is also one reason that matching the tube currents with bias is important.

[davidallancole]

Thanks for the response R.G.  Gus provided a link showing an amplifier using that principal.

[teemuk]

Trace Elliot's Road series of amps used that idea (transformer coupled MOSFET output) too. They just had an Opamp -based phase inverter instead of BJT. GMT did the same thing in the early 1970's but with BJTs instead of MOSFETs. Not surprisingly, both amps had very "tube like" characteristics (rather soft clipping, high output impedance etc).

[PenPen]

I love reading RG's replies, because I might not fully grasp the entire thing on the first read, but after multiple viewings you eventually follow everything he says.

Just occurred to me, though, does this mean in a headphone amp type implementation there's the potential to burn out the OT if you don't have headphones plugged in? Will I need some kind of permanent dummy load hooked up to this?

[Brymus]

Here is schematic I found for a push pull MOSFET amp being fed through a transformer.  It uses a differential front end feeding a set of MOSFETs in push pull.

http://www.turneraudio.com.au/solidstateamps4-50w-mono-mosfet.htm

OMG
DUDE thank you for posting that,seriously that build report and explaination is a must read for anyone intrigued by RGs mosfet folly article.
I followed the link back to his main page and, he has alot of good info there.

[Brymus]

Look at his pre-amps he has one using transistors as buffers on his cathode follower set up.
He claims it lowers the THD of the 12AU7 that way.
Sorry its off topic just thought you would find it as interesting as I .

[J M Fahey]

Well, I *have* built  many "tube type" MosFet amplifiers, with Push-Pull center-tapped transformer output, and driving them with both sides of an TL072.
I sold quite a lot of them, as 60W RMS 12V battery powered mini PA combos.
They work, but the results are not good, the sound is hard and brittle.
That's due to their being switching Mosfets , very nonlinear, specially referring to their threshold area which requires a very high bias current to minimize crossover distortion.
If you look carefully , you'll see that their curves and those of a Pentode are absolutely different.
In regular Linn type amplifiers (99,99% of all modern SS amps), the very high open loop gain allows you to use up to 40dB negative feedback to correct/iron out a lot of small defects; in "tube type" ones, it's difficult to apply much more than 6dB (10 dB tops) because of the phase shifts caused by the transformer.

[davidallancole]

Hi J M,

Would you be able to tell us what MOSFETs you were using?