MOSFET signal inverter?

[bluefoxicy]

I'm trying to figure out how to do a MOSFET signal inverter, rather than use half a 12AX7. Trying to figure on if depletion or enhanced MOSFET works better for this application. The inverter is for doing Class AB1 push-pull power output; a phase splitter might work, but will cause some trouble losing the part of the signal near 0 (i.e. class B operation, no good).

I'm using this for a vacuum tube amp. I want to source follow a MOSFET into a tone stack as well, rather than cathode follow a triode; however the first gain stage is essential for this as well, and I'm not sure if I can mimic that with a MOSFET. I wouldn't mind replacing both stages of the tube with a MOSFET (depletion or enhanced here?) for the tone stack; see the second 12AY7 (in the center, cathode followed) here:

http://www.schematicheaven.com/fende..._5d8_schem.pdf

My only concern is absolute linearity. I don't want this to sound like a BJT op-amp; but MOSFETs should behave themselves anyway.

[Steve Conner]

I take it you've already read RG Keen's excellent "Mosfet Follies"
http://www.geofex.com/Article_Folder...osfetfolly.htm

To make a MOSFET phase inverter, a good start would be to copy an existing PI circuit and substitute MOSFETs.

The only difficulty is that you won't be able to find depletion-mode MOSFETs. Nobody makes them. Only JFETs, and they are low voltage.

Therefore you can't just drop them into a tube circuit: they'll do nothing. You need to bias them differently. I think the concertina phase splitter circuit would adapt easiest to a MOSFET.

http://www.sheldonamps.com/jargon.htm

[bluefoxicy]

I take it you've already read RG Keen's excellent "Mosfet Follies"
http://www.geofex.com/Article_Folder...osfetfolly.htm

Yep. I'm exploring the possibility of removing tubes from non-voltage-gain circuits. The MOSFET switch seems interesting since I can get any old SPST switch to act as an N pole double throw switch. Source follower would be useful if I can replace the whole tube for the tone stack (possibly doable with a JFET into a MOSFET, but I don't know how that would affect the signal).

To make a MOSFET phase inverter, a good start would be to copy an existing PI circuit and substitute MOSFETs.

The only difficulty is that you won't be able to find depletion-mode MOSFETs. Nobody makes them. Only JFETs, and they are low voltage.

Therefore you can't just drop them into a tube circuit: they'll do nothing. You need to bias them differently. I think the concertina phase splitter circuit would adapt easiest to a MOSFET.

Phase splitters have that middle point issue; I need a tiny bit of cross-over for good fidelity in class AB1, rather than straight Class B. As I understand, there'd be a little loss in the middle (between +/-n volts the signal would vanish, just become 0 volts), whereas inverting the signal would give the whole wave with voltages reversed.

Enhanced-mode MOSFETs would need to be biased to pull the voltage up so it's over a certain level; as I understand, they kick on at some charge, then kick off. Is it possible I could bias the MOSFET so that it gets hit with a signal with a DC offset, thus outputting a signal with a DC offset, and then use a filter on the output (i.e. 0.47uF cap with a 2M ground reference on each end)?

[Steve Conner]

No. All phase splitter circuits in tube amps operate in Class-A, and put out what you called "the whole wave with voltages reversed."

The power tube grids of a push-pull amp are driven by two complete copies of the whole signal: one normal, one inverted. The conversion to Class-B, or Class-AB, or whatever, happens in the power tubes themselves.

So that "little loss in the middle" you mentioned, which by the way is called crossover distortion, will only happen if you've biased your power tubes too cold, such that they both turn off completely at the same time.

To bias a MOSFET in a concertina phase splitter, I'd just hang a voltage divider between B+ and ground, and connect the MOSFET gate to the junction of these two resistors. Say 1M to ground, 2M to B+, since the optimum bias point for the concertina is around one-third of B+. Then just put a 22k resistor from source to ground, and another 22k from drain to B+ and you're done.

I suggest 22k since I think the average power MOSFET might appreciate a bit more current than a 12AX7. The little TO-92 ones RG Keen recommends might need larger resistors than 22k to stop them overheating.

Of course, if your motive is to save money, bear in mind that 12AX7s aren't really that expensive, and will save you a bunch of time since you know they'll work already. The most expensive parts in an amp are the power and output transformers, and I don't know of any way of getting rid of these that doesn't amount to cutting the amp's balls off.

[bluefoxicy]

No. All phase splitter circuits in tube amps operate in Class-A, and put out what you called "the whole wave with voltages reversed."

Someone told me a pair of resistors and diodes makes a cheap phase splitter, because it voltage divides and only lets half the signal go in either direction. Problem being it loses a little in the middle.

The power tube grids of a push-pull amp are driven by two complete copies of the whole signal: one normal, one inverted. The conversion to Class-B, or Class-AB, or whatever, happens in the power tubes themselves.

So that "little loss in the middle" you mentioned, which by the way is called crossover distortion, will only happen if you've biased your power tubes too cold, such that they both turn off completely at the same time.

Ah, yeah. I'm aware of the general concept (BARELY, I've taken forever to collect enough tube theory for this). I think I found out you need a center tapped OT for this like ... 3 days ago <_<;

To bias a MOSFET in a concertina phase splitter, I'd just hang a voltage divider between B+ and ground, and connect the MOSFET gate to the junction of these two resistors. Say 1M to ground, 2M to B+, since the optimum bias point for the concertina is around one-third of B+. Then just put a 22k resistor from source to ground, and another 22k from drain to B+ and you're done.

I suggest 22k since I think the average power MOSFET might appreciate a bit more current than a 12AX7. The little TO-92 ones RG Keen recommends might need larger resistors than 22k to stop them overheating.

Of course, if your motive is to save money, bear in mind that 12AX7s aren't really that expensive, and will save you a bunch of time since you know they'll work already. The most expensive parts in an amp are the power and output transformers, and I don't know of any way of getting rid of these that doesn't amount to cutting the amp's balls off.

The 12AX7 has an impedance/current thing going on, but I suppose the output of a MOSFET will have different Zout and such than just bare signal. My intention is to not have a tube where I'm not coloring the signal, if I can make a FET not do anything to the signal I don't want (i.e. like BJT op-amps like to do).

The reason for this isn't power saving, but more academic. I want a functional amp; if I have a way to get the exact same functionality but use a different technique, then maybe I can learn something from the experience. Once I've done this a few times it's like walking; I can build a Class A single ended amp to whatever specs in my sleep now, it's time to move on to more complex things.

[Steve Conner]

Someone told me a pair of resistors and diodes makes a cheap phase splitter, because it voltage divides and only lets half the signal go in either direction. Problem being it loses a little in the middle.

Well, that might be technically true in some sense, but it's not relevant to tube amplifiers. I can't think of any use for a circuit like this in an amp.

The reason for this isn't power saving, but more academic. I want a functional amp; if I have a way to get the exact same functionality but use a different technique, then maybe I can learn something from the experience. Once I've done this a few times it's like walking

In that case, you should probably start by making a clone of one of the famous vintage amps, like a 5E3 (the "Tweed Deluxe") or whatever. If you're short of cash, get an old PA or organ amp off Ebay and modify it to the Fender circuit. Use it stock for a while, then start sticking MOSFETs into it and listen to see what it does for the tone.

Of course, you never mentioned what tone you're looking for. If you like death metal and own a guitar that looks like a medieval weapon, then a 5E3 is unlikely to do anything for you.

[bluefoxicy]

Well, that might be technically true in some sense, but it's not relevant to tube amplifiers. I can't think of any use for a circuit like this in an amp.

It was what I was recommended. I immediately saw a problem in the design :|

If you're short of cash, get an old PA or organ amp off Ebay and modify it to the Fender circuit.

http://www.wattstubeaudio.com/

25 cent resistors, $1 capacitors, $4 electrolytic power caps, slab of plexiglass and some turrets...

The expensive part is the output tranny and PT; other parts of a circuit ... a whole tube circuit costs like $10, socket and shield and all. Power supply has those big cans that cost like $5 each, so costs like $20 in itself.

Of course, you never mentioned what tone you're looking for. If you like death metal and own a guitar that looks like a medieval weapon, then a 5E3 is unlikely to do anything for you.

Santana-like sound is a good goal, but anything that sounds good is okay too. I've actually got a few simple design goals I'm experimenting with...

• I'm going for slow preamp overdrive-- meaning I want a long preamp chain (3-5 12AX7 tubes) that overdrives a little, attenuates, then hits the next tube and overdrives again, warming the signal and thickening the harmonics each stage.
• Bypass to use a single stage 12AX7 preamp, with more headroom and less overdrive when you do break that stage. (MOSFET switch bank)
• Push-pull 20 watt with 1 6V6 and 1 EL84 pair
• Bank switching: EL84, 6V6, or both at the same time
• 3 knob tone stack

I find that filtering the signal (i.e. with a tone stack) lets me control how the tubes overdrive, and EL84 vs 6V6 is a significant factor for tone. I should be able to get a good variety of blues tones out of this design, as well as rock and heavy metal sounds.

[Steve Conner]

http://www.turretboards.com/images/c...rem_vjpurp.JPG

What the...