Anyone ever played with GU-50s?

[thorpej]

Well, I'm cooped up in the house with a cold, so I fiddled around with this on paper a little yesterday. After finding the triode curves, I'm thinking running them triode-strapped, which should be just fine even with the low screen voltage rating (as @BinaryMike pointed out with Bonseki, the plate still sucks up all the current because the plate never drops below the screen).

I drew up a load line based on 450V B+ and 6.6K Rpp:



My thinking here is that if I run them Class AB2 and can drive them to around +20 (using a MOSFET source follower), that should give me around 40W (42W on paper in a frictionless pulley universe) of output power into an 8 ohm load from a push-pull pair. I'm ball-parking the +20 -- from looking at the curves, +20 might actually yield more half-cycle swing than 375V, so the output power might actually be a bit higher. And this operating point keeps it well below it's max dissipation rating, which bodes well for tube life.

Over coffee this morning, I threw together one for 5K Rpp just to see:



Less swing, but higher current would get me to just shy of 50W, but I'm not sure it's worth the trade-offs (beefier power supply, having to run the tubes hotter, etc.)

Now, the issue with this is I need a ton of driving voltage -- biased at -80V and driving up to +20, that's 200Vpp. And if I want to have any headroom available for NFB, I'm going to need a lot of gain. So, I suppose I'll fiddle around with that over a second cup of coffee this morning.

 

[PakProtector]

If you tweak the delivered output impedance to match that triode set of curves, the gain will be higher. The exact plate-to grid values will depend on the voltage you choose. IMO, tweaking the OP for 5 more watts is not going to be anything anybody but you and your calculator will know about. At these power levels, even 7-8 Watts less is going to take a lot to notice.

How comfortable are these beasties with g1 current? I'd go trouble wavebourne if I needed an answer I could trust...
cheers,
Douglas

 

[thorpej]

A cursory search turned up a couple of discussions about successfully running them in AB2, but pinging Wavebourne is a good idea.

 

[thorpej]

Thread bump!

Wavebourn said g1 current was doable, but pushing to +20 is probably ill-advised. So I think I'm back to pentode mode. After eyeballing the pentode curves, a ~500V plate supply and a 5K Rpp looks like ~60W into an 8 ohm load if operating AB2 and driving to +5, with significantly lower drive signal swing requirements than triode mode. So I'm now leaning towards that with a 5-20 style front-end with a cascode comprised of 6N1Ps in the first stage (which would yield plenty of headroom for feedback).

 

[PakProtector]

Thread bump!
So I'm now leaning towards that with a 5-20 style front-end with a cascode comprised of 6N1Ps in the first stage (which would yield plenty of headroom for feedback).

Either get a real pentode, or a different triode to build one with. That particular position is running with a minimum of voltage headroom, and thus is not the place to expect built in gain headroom. You can get some of this voltage back by using a MOSFET on top, instead of a triode, and don't forget to put a solid Zener clamp on it to protect the gate-source voltage limits that are typically around 20V.
cheers,
Douglas

 

[prelius]

GU-50 were designed to be used with the special "can holders", which provided several functions: 1. additional mechanical anti-vibration protection; 2. Proper placement of the tubes; additional heat dissipation.. These cans have a notch in the top horizontal ring that match the notch on the glass envelope (See attached pic). Therefore, the tubes can be inserted only one way. The sockets cost as much (or more) than the tube itself, but I think they are worth the money. On top of everything, they look cool!
Cheers, Paul