MOSFET Driver for Power Pentodes

[bradivarius]

I'm a new member...Thanx to all for the great info sharing.

I ran across this MOSFET power tube driver circuit design and I need a little help to implement it. I want to try this in a guitar amp using 6V6s or 5881s with a 435 V B+ (Fender style output section).

I'm not sure what I need for the MOSFET +Vs and -Vs supply. Is this a seperate supply, or can I devise a circuit to tap off the existing B+ supply. What voltage on the MOSFET am I shooting for ?

I know just enough to be dangerous, but smart enough NOT to be ;-).

Any help will be greatly appreciated !

[Miles Prower]

Originally posted by bradivarius
I'm a new member...Thanx to all for the great info sharing.

I ran across this MOSFET power tube driver circuit design and I need a little help to implement it. I want to try this in a guitar amp using 6V6s or 5881s with a 435 V B+ (Fender style output section).

Way too much voltage for a 6V6. The max voltage for this type is 315Vdc. You're 120V over spec, and very likely to poof the 6V6s. Get that voltage down!

I'm not sure what I need for the MOSFET +Vs and -Vs supply. Is this a seperate supply, or can I devise a circuit to tap off the existing B+ supply. What voltage on the MOSFET am I shooting for ?

I know just enough to be dangerous, but smart enough NOT to be ;-).

Any help will be greatly appreciated !

The MOSFET's bipolar rail needs to be a separate supply. You could possibly derive the positive voltage from the normal DC rail, but then you still need a negative rail. Better to add a regulated bipolar supply. As for voltages, the 6V6 requires a max of 38Vp-p of input (+/- 19Vp). Since MOSFETs work best with higher voltage margins since internal device capacitance rises rapidly with diminishing voltages, figure on a +/- 40Vdc supply at the very minimum. +/- 100Vdc would be even better.

The other prime consideration is getting a MOSFET with as small a reverse transfer capacitance as possible. In source follower operation, this becomes the bulk of your input capacitance. The smaller the better, and the easier it will be for the driver tube to source the current needed for a good slew rate.

[gingertube]

For some ideas, see my 6V6 design here:

6V6 Ultralinear

Its a HiFi design but using MOSFETS to drive the 6V6's.

Note the zener protection diodes from gate to source on the MOSFETS. They are REQUIRED.

Cheers,
Ian

[SY]

Asa practical matter, the IRF820 as source follower won't load down the plates significantly.

[HollowState]

What Miles said. But what he didn't say is you simply don't need it. These low impedance hi current drivers are needed for transmitting triodes that draw grid current. 6V6s and other small pentodes are very easily driven with standard coupling capacitors. IMHO adding such a mosfet driver is overkill and would provide no beneficial return.

Victor

[gingertube]

Victor,
You are correct in saying that the source follower is not required to drive the output tubes and a capacitor coupled drive would suffice.

In my HiFI Amp I added the source followers for two different reasons - 1) To get the load impedance on the input stage as high as possible so as to not compromise the shunt feedback scheme.
2) To eleiminate blocking distortion on overloads.

That 2nd reason would apply in a guitar amp where overdrives are common.

Cheers,
Ian

[aletheian]

I have seen that circuit before... I think the idea was to prevent grid blocking distortion when you really slam the output tubes directly from a 12ax7 LTPI. Really, it is just a band-aid and you can design for a heavily overdriven power amp that will not go into blocking, but the mosfet will work... there are about 100 different ways to rig up a source follower though, many of which do not require a negative rail. Like Miles said, you just need to be sure that the mosfet stays linear throughout the entire range of voltage swing because clipped mosfets sound like pppppppbbbbbbtttttt!

I hope that the IRF820 was just for the sake of the diagram, bcause I wouldn't put on of those directly in an otherwise all tube signal path... actually, I wonldn't put them anywhere. If you really want to do this, pick out a more modern mosfet... maybe with built in gate and source to drain protection and an isolated tab... or just say screw it and add a vacuum tube follower between your LTPI and the output grids, or even better, a low Zout stage that will actually add a few dB of gain to drive the output grids harder. I did that once when I really wanted to see how a pair of KT88's would sound if they were clipping like mad... I think it was a 6SL7 LTPI into a pair of 6SN7 drivers with low plate loads and just a few X gain.... but that was a while ago.

Since it is just sitting there, you could experiment with it non-destructively by building up a small, bipolar external power supply and just slip it into the circuit to tinker with. You shouldn't need too much current for it.

[Miles Prower]

Originally posted by HollowState
What Miles said. But what he didn't say is you simply don't need it.

I didn't say because he didn't ask. However, since you brought it up, yeah, a source follower is a bit of overkill for a 6V6. My preference for driving 6V6s would be a cathode follower made from 6SN7s or the 6FQ7 DC coupled to the 6V6s. That's about all you need since 6V6s aren't a difficult load, but not so easy that you can connect 'em to something like an LTP splitter made from 6SL7s or 12AX7s. Those high gain, high r(p) triodes just don't cut it when driving power finals, even if you keep them out of clipping.

I also don't recommend capacitor coupling either. During an overdrive transient that turns on the grid/cathode parasitic diode, the resulting current negatively charges that coupling capacitor (this is how RF amps derive most, if not all (bad design) of their operating grid bias) and puts the finals closer to cutoff, and less linear operation until that excess charge leaks off through the grid DC return resistor. Even if you don't hear obvious distortion, this can still degrade sonic performance. Add gNFB and the situation becomes even worse since gNFB makes clipping harder (more solid state like).

[aletheian]

Originally posted by Miles Prower



I also don't recommend capacitor coupling either. During an overdrive transient that turns on the grid/cathode parasitic diode, the resulting current negatively charges that coupling capacitor (this is how RF amps derive most, if not all (bad design) of their operating grid bias) and puts the finals closer to cutoff, and less linear operation until that excess charge leaks off through the grid DC return resistor. Even if you don't hear obvious distortion, this can still degrade sonic performance. Add gNFB and the situation becomes even worse since gNFB makes clipping harder (more solid state like).

Agreed.... and grid clamp is ugly. Simple as it is, direct coupling between driver and output grid is fairly uncharted territory where typically employed AC-coupled guitar amp techniques include dropping the value of the grid to ground resistors, upping the grid stoppers and limiting the low freq bandwidth coming off the phase inverter.

And +1 also on the gNFB evaluation. At worst, the dynamics are gone and output tube clipping feels like a switch... you lose the compress-ey transitional area between slightly driven and flat out clipping. The phase inverter clipping gets nasty and blatty as well... "blatty being a technical term of course

[Miles Prower]

Originally posted by aletheian
And +1 also on the gNFB evaluation. At worst, the dynamics are gone and output tube clipping feels like a switch... you lose the compress-ey transitional area between slightly driven and flat out clipping.

I do hiFi designs, and I don't want "compress-ey". gNFB is quite valuable there, and as for the clipping behaviour, I design so that the clip occurs at the finals, and not some intermediate stage. Add final clip to front end clip and things get real ugly real fast. Add a good final grid driver, and the "clip" is limited to OPT saturation, when the finals are somewhat into Class AB2.

Back in "the day" there weren't any silicon diodes or integrated FWBs, so providing a negative rail wasn't so easy.

The phase inverter clipping gets nasty and blatty as well... "blatty being a technical term of course

Williamson topology helps prevent that.