Re: 12au7 as SE output



flipper wrote:

snip,

> >A couple of mosfets would really drive it well.
>
> A couple? The model I did was simply a FET follower just like the
> tube.

I meant one mosfet for a single outpyt tube.

>
>
> >Maybe too well, because there is a limit for how much +ve voltage one would want to apply
> >to an output grid, and if its too much the grid overheats and begins emitting
> >electrons and the tube thermals away.
>
> I understand. But you saw the grid currents. Are they too high?

Obviously if you force the grid to rise to a higg +ve value the
current becomes excessive since the grid has a low power dissipation limit.


>
> >You never see class AB2 guitar amps; the tubes may not last long.
> >
> >
> >> I also seriously considered,and spiced, a FET follower since that
> >> eliminates the heater issue as well as drive power since I have a
> >> couple of hefty hi voltage FETs on hand.
> >
> >Well yes, but see above.
> >
> >>
> >>
> >> >The CF must be capable of producing the grid current drive without itself being forced
> >> >into grid current conditions so I would prefer to use a 6SN7 as a CF driver rather than
> >> >a 12SL7, which is a rather gutless tube like a 12AX7.
> >>
> >> I hear that. I would too but this was a 'make do' kind of thing after
> >> I figured 'why not?' and added the triode switch. Then had the
> >> positive grid drive issue.
> >>
> >> I hadn't looked at the 12SL7 grid but it starts going positive at
> >> about 2.5W. Interesting. It's GOT to be affecting it but it sure as
> >> heck isn't as dramatic as the 6V6 +ve was. I can't visually tell with
> >> it inside the NFB loop.
> >
> >With no loop FB, the amp should be substantially linear.
>
> Do you mean 'should' as in a goal or as in what I've got 'should' be
> linear the way it is?

if tou have a CF driving a tube where the Rout of the CF is about
1/10 of the minimum value or Rin to an output tube then the
operation will be substantially linear.
The AB2 op will introduce additional harmonics to the brew which can then be reduced with NFB.


> >> By comparison, though, the 6V6 is just entering +ve drive at 1.3W so
> >> even staying under the 12SL7 +ve drive at 2.5W I'm getting close to a
> >> power doubling.
> >>
> >> Yes, the 12SL7 is a lot like the 12AX7 and, in fact, I'm using the
> >> 12AX7 in the spice model.
> >
> >12AT7 has a lot more current ability, and about the same gain as 12SL7.
>
> Yep. I could use a 10EM7 too but I don't have 'em.
>
> I'd like to stay octal and the 10em7 is cheap but it seems a shame to
> waste a triode with almost as much plate as the 6V6 in a 'driver' for
> it.
>
> >> >A 6V6 can dissipate about 12 watts, and if triode efficiency is 20% you
> >> >should get 2.4 watts, class A1, and in A2 probably E = 35%, so maybe you get 4.2 watts.
> >>
> >> Yeah. 4.2W is bleeding edge pushing it and I don't consider that an
> >> 'operating' number. Curve flattening is glaringly obvious. It's just
> >> the max I've set the gain for so that actual max is below 100% on the
> >> volume pot. I was shooting for 3 watts actual.
> >>
> >> Even with some +ve in triode mode at 3W spice says 2'nd and 3'rd are
> >> 60 and 50dB down so maybe I'll just live with it.
> >
> >I doubt you will ever see 2h+3h at < 0.1%
> >Most SET near full power with say 15db of global FB measure about 1%
> >at a dB below clipping.
>
> Yeah. I don't expect it'll come out that low either and I imagine the
> disparity is at least partly due to the minimal, 'too good',
> transformer model.

Building and measuring is always going to be different to a simulated
outcome.
I roughtly calculate then build and measure.
I neve use spice programs; I don't need to.
I have a brain.

>
>
> >
> >> Maybe say it's a 2W
> >> amp with some hefty headroom ;) That's closer to the truth anyway as
> >> it's obvious the NFB is 'working at it' to keep it in line at 3W.
> >
> >If there is 4W at the tube, expect 3W at the OPT outlet because many little
> >OPT have maybe 25% winding losses.
>
> Yeah. I took note of your earlier mention of that. I've got the OPT
> winding resistance in the model but not capacitance leakage as I had
> no convenient means to measure that. At any rate, I get about 20% loss
> that way.

The Lp, LL nor Cshunt will affect what happens at 1 kHz.

>
>
> >> One thing that puzzles me a bit. I have an Rk bypass cap, of course,
> >> on the 6V6 but the bias voltage shift when over driven is completely
> >> different between triode and pentode modes.
> >
> >For A2, you should have fixed bias.
> >The cathode cap will charge up under the effect of the non linear
> >tube currents and grid current.
>
> Ah. Well, that explains it then because, as I re look at it, the 'odd'
> behavior is when positive grid drive takes place. But the grid drive
> tracks it so even though it's 'different' it doesn't seem to be any
> worse at the power levels I'm driving it at.

?

>
>
> >At low level cathode bias is fine though.
>
> I had thought about fixed bias, anyway, using the DC heater supply
> but that would conflict with the 12SL7 heater needing to be elevated
> for the CF.
>
> I'll rethink the options and whether I want to go 'impure' with a FET
> or maybe just limit the triode mode power output. That's certainly
> easy enough to do as I have to switch gain resistors to keep the power
> roughly equal between the two. Just leave it the same cuts max power
> down to about 3 watts (at full 100% pot that is a dirty 4.2W in
> pentode mode).

3 watts of triode power will sound heaps better than 4 watts of
beam tetrode power.

Patrick Turner.

>
>
> >
> >Patrick Turner.

.