[Voodoochile]

Specifically, I've been looking into depletion-mode mosfets as a CCS for a tube amp. The SOHA uses one, for example. In researching this further, I thought I'd check out Gary Pimm's CCS projects, which is really interesting as well. Of course using a mosfet cascode improves the performance further. That wasn't too surprising.

Upon reading his "Active loads and signal current control" paper, it looks as though perhaps using an enhancement mode mosfet cascode may be more interesting, as you can then 'take the output from the source of the lower mosfet, rather than directly from the plate of the triode'. I like the idea of that.

But there is still a little more baggage regarding the isolation of AC and DC paths. So Gary suggests adding a shunt regulator between the B+ and the CCS feeding the triode. This shunt reg can be as simple as another CCS feeding a glow discharge tube, setting the upper CCS for the burrent to bias both the vr tube and the plate load CSS.

Steve Bench also has a lot of great info on shunt regs and vt regs, which could also be implemented.

I'm curious if anyone has implemented this scheme, and if so, how's it working out for you? Seems like it would be a nice improvement.

 

[Rescue Toaster]

Quote:

Originally Posted by Voodoochile I'm curious if anyone has implemented this scheme, and if so, how's it working out for you? Seems like it would be a nice improvement.

I did this, using Gary's battery bias in the CCS fed-shunt, and two of his self bias for the ECC99 loads. I do not use the mu-follower, as too low an output impedance makes the parafeed capacitor enormous.

Performance-wise, the CCS-fed gas regulator shunt is unbeatable. The trick is though, either the loads must be truly constant, or the power supply signal current (for instance from a cathode-follower circuit) must not cause the regulator tube's current to go out of spec. mu-follower outputs exhibit the same problem as cathode followers in this regard, the output impedance is very low, but the output current is limited by the range the shunt can handle.

http://s58.photobucket.com/albums/g241/gridstop/

 

[Voodoochile]

Thanks man, that's very encouraging! Thanks also for showing the schematic. Your vr glow is cool

to boot. Do they run quietly for you? Very nice PT in your amp, also.

Also, what are the specifics of the battery bias for the shunt reg... enhancement mode mosfets, or another arrangement? Also, how is the battery life?

Thanks a bunch for sharing your project. I was hoping some one had tried this, and more importantly, that they liked the outcome.

 

[Rescue Toaster]

Quote:

Originally Posted by Voodoochile Thanks man, that's very encouraging! Thanks also for showing the schematic. Your vr glow is cool to boot. Do they run quietly for you? Very nice PT in your amp, also. Also, what are the specifics of the battery bias for the shunt reg... enhancement mode mosfets, or another arrangement? Also, how is the battery life? Thanks a bunch for sharing your project. I was hoping some one had tried this, and more importantly, that they liked the outcome.

The glow is misleading, due to the camera's response. The front VR tube does not look that bright, and of course the ECC99 to the left & diodes in back are not purple, they're orange. By running quietly do you mean electrical noise or mechanical? You actually can hear them as they turn on, they sound like a tiny tiny version of an arc lamp, but once on they're silent. They have about 1 mv of electrical noise (wideband) but the cap across them helps, and of course I use CCS's as loads which totally eliminates it.

The battery bias CCS is http://home.pacifier.com/~gpimm/Battery_biased_ccs.htm Which uses the same enhancement mode mosfets. Battery life is equal to shelf life if you use capacitors that don't leak. The trim pot lets you adjust for fading battery voltage over time as well.

The sound is fantastic, and the amp is totally noise-free. I haven't tried the amp with a traditional power supply, but I see no reason to. This isn't very expensive and the advantages are obvious.

 

[Voodoochile]

Yeah, I meant noise in the physical sense. I guess is some of the more standard applications those can be a bit on the noisy side.

Thanks for the details. I was already pretty interested... now even more.