When can screen voltage be higher than plate???

[rsumperl]

Hello everyone,
Looking at the 12BY7A, how is it that the screen can be higher than the plate? My understanding has always been the plate is higher.

Thanks,
Ray

[Sch3mat1c]

Typically, receiving tubes are safe with plate voltage higher than 30-100V, or 1/4 to 1/2 of Vg2. Transmitting tubes are typically higher, of course.

As shown, the tube can never draw excessive screen current due to the large series resistance. However, this affects distortion and makes the input impedance particularly bizarre, because as the tube goes into saturation, screen current shoots up, screen voltage falls and miller effect comes into play on Cg1g2. Effectively, the amplifier slows down as it goes into saturation, which is undesirable.

I wonder why the designer chose 12BY7 over a more suited pentode like 6AU6, or a triode such as 6AV6.

Tim

[Eli Duttman]

I wonder why the designer chose 12BY7 over a more suited pentode like 6AU6, or a triode such as 6AV6.

Considering the designer was Stu Hegeman, you can be certain he had good reason. An obvious fact is the high gm of the 12BY7. That high gm provides resistance against slew limiting. The Cit. 5 employs Mullard style circuitry and a good deal of GNFB is present. Slew limiting is always a concern under those conditions. A possible tweak to the Cit. 5 is employing an ECC99, instead of a 6CG7/6FQ7, as the LTP. Stu didn't have access to the '99, with its high gm, back in the 1960s.

IIRC, the graphic is not error free. Check Jim McShane's site out for info. on tubed H/K Cit. units. Jim's expertise in this area is unrivaled.

[HollowState]

I believe your schematic in incorrect. Here is a partial scan of an original HK document. Notice the voltage chart in the lower left. It calls for 145v on the plate and 140v on the screen. And the screen bypass capacitor is in the wrong place on yours as well.

[Miles Prower]

Originally Posted by Sch3mat1c

I wonder why the designer chose 12BY7 over a more suited pentode like 6AU6, or a triode such as 6AV6.

The 6AV6 is a real pipsqueak with a Pd= 500mW. Not much drive capability there, and the 6CG7 LTP probably isn't gonna present a 6AV6 with a very friendly load.

The 12BY7A is a real beast, having ratings more like a power pentode, as it was intended as a color TV video amp. As such, it likes high voltages (no shortage of that in a color TV) and the plate characteristics in the spec sheet are plotted against a screen voltage of 180Vdc, though it has a nice looking plate characteristic down around the -6.0V region (where this seems to be operating: Ip= 6.0mA). Dropping screen voltage and bias would keep the plate current the same, and get the screen voltage under the plate voltage where you want it.

6AU6's seem to be at their best with Ip= ~3.0mA, so may have been lacking a bit in the drive department, and perhaps y(f) too, despite working the 12BY7A at such a low plate current?

[Bandersnatch]

Originally Posted by Eli Duttman

Stu didn't have access to the '99, with its high gm, back in the 1960s.

Sure he did, it was type number 5687....

FWIW, the Citation II also ran the 12BY7 with its g2 higher than its plates.
cheers,
Douglas

[DF96]

Originally Posted by Eli Duttman

That high gm provides resistance against slew limiting.

Could you briefly explain why?

[Miles Prower]

Originally Posted by DF96

Could you briefly explain why?

Almost every text on hollow state tells you that tubes not driven into grid current require just voltage, not current. That isn't just an over simplification: it's just plain wrong. Even if Vgk never goes positive, the grid still had capacitance associated with it: Cgk + Cmiller + Cstray. The grid won't see the true signal level until that capacitance is fully charged, and it takes current to charge a capacitor. The faser you want that capacitor to charge, the more current you need to pump into it.

Take a look at this loadline. The driven stage will be demanding the most current when the 12BY7A is least able to supply it. When the plate voltage comes up, the potential difference across the plate resistor goes down. The ability to supply current to the load is:

I= (Vpp - Vplate) / Rp

That's why it's a good idea to make sure the Q-Point current is at least five times the peak current at the highest frequency of interest, and why the 12BY7A was designed in the first place: as a color TV video amp. Even though you try to make CRTs in such a manner to minimize internal capacitance, you can't completely eliminate it. Since you need to drive the cathodes of the color CRT at a max rate of 4MHz, it takes a pretty stiff current to charge that capacitance, even if it's held to picofarads.

If you use a tube with a high y(f) -- like the 12BY7A -- you can get the gain and the linearity with a much smaller plate load. The smaller plate load, the more current reserve.

This is something that gets overlooked a lot in audio designs. No way can something like a 12AX7 drive the grids of an audio power final, and keep up with the current demand. If there isn't enough current, the control grid(s) of the final(s) won't track the signal. That's the slew rate problem.

[Cassiel]

No way can something like a 12AX7 drive the grids of an audio power final,

How about driving the grids of two KT88's in parallel? Yes, I wouldn't do it but Jadis did.

[DF96]

So Mullard were completely wrong to drive a LTP phase splitter with a pentode with such low gm and drive capability as an EF86, and then drive a pair of EL34s from this ECC83 splitter?

I accept that if you think you need a low anode load but still want high stage gain then a high gm valve is necessary. A low anode load may be necessary at this point if the LTP is biased into the grid current region - an easy way to do this is to run the LTP at too high a current (perhaps in order to 'drive' the output stage) or too low quiescent anode voltage. The design being considered here avoids that problem, fortunately. However, it has a large capacitor at the pentode anode (430pF) which creates the problem I thought we were trying to avoid. A 6CG7 LTP should not be particularly hard to drive so I am puzzled why such a low impedance input stage was chosen.