Ultra linear and phase inverter drive

[Tage]

Ok so I have now finalized my design and am in the process of making up my PC boards and prepping the chassis. I decided to go with the Mosfet drivers, just because the more I've learned the more benefits I've seen from them. really there seems to be no drawbacks. I'm running them after a pretty standard 12ax7 split load PI, with a big grid stopper, running ~350V on it's plate with .1 coupling caps. The signals then go to a potential divider of 100k/470k. the 470k going to ground and the bias voltage being fed in @ their junction. there shouldn't be much if any voltage dropped across the 100k, were it then enters the gate

I just have a couple questions still... My phase inverter will be driving with an AC load line of 100K//570K? I suppose this is the advantage of using the mosfet drivers, the PI is not heavily loaded and the Power tube can have nice low grid leaks and high grid stoppers, I'm using 33K grid leaks/source resistors and 10k grid stoppers. Are the grid stoppers still as necessary after the Mosfet buffers?

I'd love to upload my schematic but I don't have any good software for drawing out the schematic on my Macbook...

[kg]

Are the grid stoppers still as necessary after the Mosfet buffers?

yes. always use grid stoppers, pretty much 100% of the time. the values can be tweaked of course, and the lower the output tube stoppers in this case the more the drivers can pull the tube into grid current, but you absolutely should use SOMETHING.

[R.G.]

kg's right. Grid stoppers 100% of the time.

The reason is that grid stoppers are there to prevent local oscillation of the tube they're "stopping". The wires leading to the tube grid have inductance and capacitance, and also capacitance to the output/plate. This can and does make the tube go into RF oscillation if the stars line up just right/wrong.

It's the same reason to always use gate stoppers on MOSFETs used as source followers. These things easily oscillate at frequencies you can't even see on oscilloscopes with only 20MHz bandwidths.

[Tage]

I've got 10k stoppers drawn up, more than sufficient, but I figure I can go bigger because of the follower set up being able to sink the grid current.

[R.G.]

Generally stoppers don't usually need to be big. The idea is not to choke off current flow, it's to damp the resonances that set the thing up to oscillate. Beyond that, you're making a lowpass filter from the capacitance from grid to cathode and the Miller-amplified capacitance to the plate. For instance, usually 100R to 1K is OK for MOSFET followers.

[jpfamps]

We built a 50W head with DC coupled grid using a 12AU7 cathode follower and initially running the output stage UL.

This amp sounded good clean and had plenty of headroom, but for guitar sounded better in pentode mode when overdriven, although it did sound nice "clean" in UL.

Initially we were using KT77s which had just been reissued, and are designed to be run UL, but then ended up using EL34s which aren't (although several hi-fi amps have done this).

We were trying to build an amp that would not go into blocking distortion, so the idea was to use a CF to DC couple the pre-amp to the fixed bias phase splitter (we've used this before with good results), DC coupled CFs to the output valve grids, and UL output stage without global negative feedback.

The amp ended up being quite complex, and ultimately we didn't pursue the design further, so I can't really comment on the suitability of UL, and at some stage I would like to revisit this configuration.

As discussed above the advantages of the DC coupling where elimination of blocking distortion, reduced grid resistance, possible AB2 operation, and reduced loading on the output of the phase splitter, so no "nipple".

It also allowed us to easily add a post phase splitter master volume.

One issue we were aware of was the heater cathode insulation of the CF driver stage. MOSFETs seem a very good engineering solution all round.

We were also quite wary of oscillation. The relatively high output impedance of standard long-tail pair forms a low pass filter with the Miller capacitance of the power valves. You now would have low impedance driving the power valves so more potential for instability as you have raised the HF roll off.

[jpfamps]

That's probably true. Blackstar use a lot of solid state components in their amps, and have copped a lot of flack from prospective buyers, because their not 'all tube'. Though for the more technically inclined home builder, I think on some level we just like to bring our haphazard creations on stage and gloat. Sometimes I tell people my amp is solid state just for fun (it's not far from the truth either!)

Most of the flak Blackstar received was not so much for using solid state components, but more for their marketing which implied that certain products were all"valve", when they weren't by most people's definition of all "valve".