• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Output (beam) pentodes / tetrodes suitable for screen grid drive

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Something that came up in another thread:

Apart for the oft mentioned sweep tubes (PL509/519), which other tubes would be suitable for screen grid drive? Any experiences, schematics, curves, or just musings on the subject are highly welcome!

(SY, you seem to have experience with this, I would greatly appreciate your input)
It depends on what you can find these days- of new production, the 509/519 is pretty much it. Conventional output tuibes are quite poor in this configuration- high perveance horizontal output tubes are most suitable.

Older tubes of choice would include 6LF6 (the King of Screen Drive) and 6JN6. For a lot of sweep tubes, the data sheets include zero g1 bias curves at different g2 voltages. Signs of suitability would certainly include having those curves be reasonably flat, more-or-less evenly spaced, and very high current for relatively low (say, 100V) g2.
For power output tubes, I would think that low screen voltage rated horizontal output tubes would be the place to look.

For low power tubes, the 6LE8 is rated for higher than usual screen dissipation since it was intended for dual control applications, but no reason it can't be used for screen grid drive. (put about +12 Volts on the g3 grids to square up the plate curves, to get back to normal beamer like curves) The dense g3 grids also provide good shielding against Miller capacitance and a high plate output impedance.

Another type of small tube with interesting possibilities were the "shadow grid" tubes, 6FG5, 6FS5, 6GU5. These have an extra aligned, grounded, repeller grid in front of the screen grids that greatly reduces screen grid current. (they were intended for low partition noise in TV tuners) These tubes had remote cutoff characteristics as far as grid 1 was concerned, but this is irrelevent to G2 operation. (actually, the screen grid in these tubes is G3, and the suppressor is G4, repeller or shadow grid is G2) I bought some a while back to try out for screen grid operation, but they are on the very back burner. If the repeller grid loses its effectiveness at high screen grid voltages, then the screen currents could potentially be rather non-linear, don't know. Totally unexplored territory as far as I know.

i`ve recently got excited bout screen drive too.. i can`t get 519 sockets locally so that was out of the question. i had a look at the graphs of 6l6gcs and they looked okay. not fantastic, but they were much better that those of the el34. so, i see some potential.

can anything be done to make the screen more linear?
6L6 will require hideous amounts of drive voltage. Let's look at a simple comparison. At zero volts on the control grid, it will take about 175 volts on g2 to swing up to 100mA. By comparison, a 6JN6 will require about 60V to swing the same current.

Screens as loads are very, very nonlinear. The design burden is transferred to the driver stage.
Everyone, thanks for the great input!

I've recently looked at some datasheets, and, surprisingly, at least for the european tubes, I tend to find a Vg2 vs Ip @ Vg1 = -1V graphs, for vertical sweep TV tubes more often than the horizontal ones.
Out of the smaller tubes, PL805 and PL81 seem to be more linear with G2 drive, quite surprisingly so. The ECL/PCL 82 also has such a spec as it was used both as field scan and audio output (no doubt this will give people some ideas ;) ).. The rare PL508 also has this specified, but the curves, though very evenly spaced, show a definite kink at lower voltages.

I've ben wondering about the PL36 myself - as well as the PL508. Hard to believe a miniscule EL92 will do 12W anode dissipation, supposedly the same as a PL36 or PL508 - considering the size of the latter bottles and structural elements. The sheer dissipative area is probably one order of magnitude larger for the PL tubes... Anyone care to share some insight?


Joined 2003
Paid Member
> suitable for screen grid drive?

Big heater, low G1-G2 amplification factor (Mu).

Even though it says it is a pentode, triode-data tells you a lot about the basic tube performance.

For a simple triode power stage, grid swing can be estimated as 0.6*Vp/Mu.

The Mu tells you the ratio of G1 and Plate (or G2) swings needed to get the same effect on the electron flow. If Mu is 10, you can swing 1V on the grid or 10V on the plate, and get the same current change. In a Pentode, the G2 Mu has the same significance as Plate-Mu in a triode. (Pentodes also have a plate-Mu but it is so very-very high that it can usually be ignored.)

Taking Jason's plots for PL36: at plate voltage of 150V, swinging current from 100mA to 400mA, we can either swing G1 from -5V to -20V (15V swing) or we can swing G2 from 60V to 140V (80V swing). Dividing 80V/15V, we get G2 Mu of 5.33. Looking at the second page of Frank's PL36 sheet, it says Mu(g2g1) is 5.6, substantially the same (the factory data is taken at Vp=Vg2=100V, Ip=100ma, not Vp=150V and Ip=150-400mA as I did).

Now if a triode power stage needs ~0.6*Vp/Mu of drive, then a G2-drive with static Vg2 equal to plate voltage needs about 0.6*Vp of drive. This leads to driver supply voltage much higher than output supply voltage (or audio transformers or other complications). It appears that Mu has dropped-out of the picture; not quite.

So what we want is a tube that can pass huge current at low Vg2. How do we pick that?

The static (not dynamic) plate resistance is the cathode resistance times Mu. Lower Mu will pass more current at the same voltage. This is obvious in triodes; the same applies to pentodes only with Mu(g2) instead of Mu(plate). Cathode resistance 1/Gm is pretty nearly fixed once you pick a cathode size (and use most of it), and cathode size sets heater power, though there was some improvement between 1936 and 1960.

So for a given cathode (heater) size, you want a low Mu for maximum output current. With a Pentode, you can set the G2 voltage fairly low and keep the Plate voltage high to get the most power out of your current (in a triode you don't have this split).

> Conventional output tubes are quite poor in this configuration

Correct: audio output designers like easy-to-drive tubes for low distortion and easy ripple filtering. Mu(g2) of many audio tubes is over 10. 6L6 is not the worst of these: Mu is only 8. (BTW: 0.6*Vg2/Mu = 0.6*320V/8 = 24V peak G1 swing, which is indeed about-right for 6L6 with 320V on G2.)

Speech/music has needs other than maximum power per bottle. But TV-set design was strictly about power per bottle. For maximum current swing, especially at the lower B+ voltages in line-operated (no power transformer) home TV sets, they went to lower Mu. With R-C coupled driver working on a B+ similar to the output tube, wihtout cathode-bias, Mu of less than about 5 is as low as you can go without the driver clipping before the output. (What is the Mu of 300B or 2A3? More like 4 because the idea was to use cathode bias "whenever possible"; otherwise they would tend to Mu=5.)

So: big heater and low Mu(g2) for easy Screen-drive. Driver has to be fed much more B+ than the output G2.

It would seem that another way to look at it is: extra high plate voltage rating, so you can extract big power with modest G2 and Ip swings. However the very high voltage tubes generally have lower Gm (because they can't use the hyperactive cathodes with hi-volt electrons bouncing around), and therefore higher static plate resistances.

TV tubes are always a good bet for power.

> for vertical sweep TV tubes more often than the horizontal ones.

H-sweep tubes have additional problems: super-high pulses, multiple highly-stressed loads, possible extended overload. They burned out a LOT: not when working as designed, but in real life. Getting them to hang together with low-bid sweep coils was at least as important as raw power. Some of them are fairly poor performers in more normal applications. H-sweep design was always "try and see"; you can't calculate everything that is happening, and the economics don't allow conservative H-sweep design (half the TV's total B+ power flows through the H-sweep).

The V-sweep system of a TV "is" an audio amplifier. No great stress, added loads, less chance to get into trouble. And V-linearity is important for an undistorted picture. Some V-sweep tubes make very fine audio amplifiers if you drive them well. Problem is that the biggest TVs only needed a few VA of sweep energy, so you don't get a lot of watts per bottle. Like a 6V6, not like a 6550.
I have nothing of real substance to add except some notes I got from somewhere on the Net

"Jim Dowdy’s 6AV5 SRPP integrated amp uses a choke-loaded 7119 directly-coupled to a pair of screen-driven (quasi-triode) 6AV5s; pair of top tubes are standard triode-connected 6AV5s in parafeed with nickel/M3 pinstripeMagnequest OPTs. "

"Check out Sound Practices issue #14 for two amps by JC Morrison that are both PSE non-triode amps. The first is a screen drive 6AV5 amp, and the second is a 6L6 amp. Both are direct coupled. The 6L6 amp is supposedly Blackie Pagano's amp for daily listening, and the 6AV5 amp JC claims rivals some 845 amps he's heard. On the Svetlana site there is a screen driven EL509 circuit by Bob Danielak (creator of the Darling amp) that also is supposed to sound very good. It even uses the 6BM8"


Joined 2003
Paid Member
> There is an extremely strong resemblence between screen-drive and zero-bias power triodes.

It "is" a triode, no doubt.

It isn't scaled for "zero G2 bias", as the triodes made for zero-bias use are.

You do have a lot of flexibility by picking a G1 bias. However negative G1 reduces current, which we are usually short of. Taking G1 positive will increase cathode current at any given G2 voltage, but some part of that cathode current is captured by G1 and does not flow to Plate where we (usually) take output. The G1 split of current varies a lot (rises) when voltage on P and G2 falls low. Anyway most fist-size tubes won't stand much G1 current without melting. Which pretty much fixes G1 at near-zero, sightly-negative because the drop of plate current is small, while barely-positive leads to trouble.

> One big advantage is the linearity at very low currents.

Hmmmm. Most G1 are proportioned for low drive requirements, because any sane designer wants a simple/cheap driver. To get high gain the G1 is wound very close to the cathode, often as close as the grid-lathe can manage without making too many rejects, and with wire that is fairly large compared to pitch and spacing. Some extreme tubes set G1 at the edge of the cathode electron cloud, what sand-heads would call a "graded junction". So it gives a highly non-uniform field at the cathode surface.

G1 control of current is usually not much like Theory. (If you know FET geometry: long-channel FETs follow the theory, short-channel FETs deviate badly from what simple theory says but work better, more gain, in many applications.)

OTOH, G2 is not required to have large gain, may even work better if it has "little" control of the electron flow. So it is further out from the cathode, with dimensions large enough that grid-lathe and wire-size lumps are small. The G2-K field is much more uniform than the G1-K field. (It is a small stretch to compare it to a long-channel FET.)

There is however a big difference between a Triode and a G2-driven Pentode: G3. This makes the G2-drive tube work a little like a True Tetrode with the near-Plate grid at zero volts. Current would be low, except G2 is held much more positive than we would hold G1 in a classic Tetrode. That's why I'm not seeing a clear picture of G2-drive operation.

> I've printed your post in order to read it quietly and more than once.

Let me know if you find anything in it. I started off thinking I had a simple re-mapping from published G1-oriented operation to G2-drive operation. But something is missing, probably G3.

Perhaps what we really have is a Junction Transistor. Output impedance is very-high, unlike conventional VT triodes. Input voltage is positive of cathode/emitter; much more positive because vacuum is thin compared to silicon. So soft that we don't have anything like the 0.6V threshold of a BJT. Input terminal draws current, and I suspect we could derive a Beta just like the BJT: not dead-constant, but a general guide to design.
PRR, thanks for some of the best data I have read on this subject.

But you do raise a question: since most sweep tubes have G3 connected externally, what if it was at closer to plate potential? i have seen a couple of examples where G3 was strapped to the plate to make a pentode into a triode... thoughts? Given the PL5xx g3 shape, one has to wonder about what it can dissipate in that sort of connection...
For those in Oz following this - and possibly others.

The tube you will find a "heap" of, dirt cheap from Oz TV sets is the 6CM5 (EL36 I think??). I don't know what its availability is like in other countries. I've got something like 20 of them and had been wondering if there was anything worthwhile for which I could use them. They don't have a good reputation for use in standard g1 driven push pull amps (high distortion). It seems it was one of those tubes that TV "Tube Jockies" used to pull and replace routinely regardless of if it needed it or not eveyrtime they serviced a set - or maybe they were just smart amd knew that the sweep tubes were subject to stress (as pointed out by PRR) and therefore prone to failure.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.