diyAudio (
-   Tubes / Valves (
-   -   Screen driven sweep tubes - the return of the evil kink? (

Tubes4e4 22nd April 2013 09:05 PM

Screen driven sweep tubes - the return of the evil kink?
3 Attachment(s)

quite some discussions on this board deal with screen driven sweep tubes. I would like to draw your attention to my observation, that when screen-driving certain types which make use of beam forming plates for secondary electron emission suppression one rather obviously loses the benefits of said secondary electron emission suppression - at least this is my interpretation of the plots I made, please correct me if I am wrong.

In datasheets or application notes, plots for screen drive of sweep tubes are not that commonly published. The 1st attachment shows a graph for a screen driven sweep tube from an application note.

Only few Spice models are available, but they give an extremely over-simplified view on reality; I suppose the curves from the 1st attachment were generated by such a model.

When measuring and plotting the same type of tube in real life using the same conditions (like prebiasing @ -33V Eg1) I get what you see in the 2nd attachment.

Now you can argue I didnīt connect the beam forming plates to the cathode with this certain sweep tube type, but I did. Just to make sure, I attached the plot of a sweep tube type which has an internal connection between the beam forming plates and the cathode, see 3rd attachment. Please note there is not only one massive kink, but also a much slighter additional kink (both marked red).

How do you interpret these graphs, what is your theory about the physical effects that (seemingly) cause the return of the evil tetrode kink when screendriving such sweep tube types? (Bonus points for giving an explanation of the much slighter 2nd kink).

What would be the consequences to consider when designing an amp with such screen driven sweep tubes? I mean, obviously it would be rather hard not to enter the instability / kink area at lowish Ea with real life reflected loads like real life speakers...

Regards, Tom

Wavebourn 22nd April 2013 09:36 PM

Hi Tom;

I believe the answer is in your experiment setup. I can't imagine negative resistance in triode.

DF96 22nd April 2013 09:54 PM

It is a triode with a rather positive grid so secondary emission complicates things. Do the kinks disappear if the beam plates/g3 are more negative?

It might also be interesting to see the g2 current - does it go negative at any point?

wrenchone 22nd April 2013 10:28 PM

1 Attachment(s)
Shown below are the results I got measuring a 6CD6GA in screen driven mode. These were static measurements taken a point at a time, rather than swept.
A source follower was used to drive the screen grid.The control grid was tied to cathode. I should probably try the tube in question at the beginning of this thread, as I have a few handging around.

Tubes4e4 23rd April 2013 07:23 PM

Hi Wavebourn,


Originally Posted by Wavebourn (
I believe the answer is in your experiment setup.

For the very slight kink, yes, that may be. For the large kink I donīt think so, since at least two people confirmed with their own measurements. When I showed these plots in a lecture at ETF06, some known experts there didnīt object, one of them showing a screen driven EL509 based design in his own lecture. I am sure your also can check yourself, if still in doubt. Just make sure to use appropiate PSUs when doing the measurements...


I can't imagine negative resistance in triode.
What do you think about Dynatron - Wikipedia, the free encyclopedia as a cause and explanation? As I understand it, it is about the same effect that causes what is called (translated) "Barkhausen oscillation" in German literature. Also, "Dynatron" seems to be known in Russian literature, but unfortunately I cannot read that language.

Regards, Tom

Tubes4e4 23rd April 2013 08:02 PM

Hi DF96,


Originally Posted by DF96 (
It is a triode with a rather positive grid so secondary emission complicates things. Do the kinks disappear if the beam plates/g3 are more negative?

I didnīt check this but I dimly remember to have seen PL509 based TV sweep circuitry that runs the beam plates at negative voltage level referenced to cathode. Maybe -30V or something like that, but I am not sure. If so, that may be a hint that designers of such TV sweep circuits knew about how to suppress such instability area effects caused by secondary emission by lowering Ug3 below cathode level, indeed.


It might also be interesting to see the g2 current - does it go negative at any point?
The pink curves in the EL509 plot show Ig2 at least over a small range where I expected something interesting to happen. In the kink valley area, you see Ig2 rise when Ia falls. Simplified, Ia + Ig2 = Ik still holds true within expectations, supposing Ig3 to be negligible. I didnīt measure Ig3, though.

For the other specimen EL36, there is no chance to change Ug3 from Uk, since the beam plates are connected to the cathode internally.

Regards, Tom

Tubes4e4 23rd April 2013 08:36 PM

Hi wrenchone,


Originally Posted by wrenchone (
Shown below are the results I got measuring a 6CD6GA in screen driven mode.

Extending your measurements with Ua below 100V down to, say, 10 or 20V would be most interesting, since the kink valley area with types EL509 and especially EL36 showed up below Ea=100V, unfortunately not covered by your plot. Also, prebiasing g1 with a constant voltage (as in my graphs) might make a difference. This g1 prebiasing is recommended in real world circuitry (using auto / cathode bias) to keep those fat sweep tubes from current runaway....


These were static measurements taken a point at a time, rather than swept.
Same measuring method used with my plots. I donīt trust swept measurements for power tubes, because average Pa will be much less than when when running power tubes at, say, 80% Pa iddle in a real world amp circuit. Swept measurements just donīt show the real world thermic effects on internal electrode geometry, often causing a steeper gm close to the Pa,max hyperbole. Think "thermal runaway".

Regards, Tom

DF96 23rd April 2013 09:30 PM

For normal tetrode/pentode operation to get more current you reduce the negative voltage on g1, while keeping g2 constant. The extra electron flow will establish (in a beam tetrode) a negative space charge near the anode - a virtual g3, but aided by beam plates.

For g2 drive, to get more current you increase g2 voltage. This means that at low anode voltages g2 is much more positive so is more likely to collect any secondary electrons, as well as more likely to intercept the primary electrons. The very positive g2 may also tend to eliminate the virtual g3 formed by a space charge near the anode. My guess is that you need to make g3 more negative so that anode secondaries are returned to the anode.

I'm not an expert on valve innards, so I could be wrong.

wrenchone 23rd April 2013 10:35 PM

I ran the 6CD6GA curves originally to get enough data to establish the operating point for a screen drive SE amp, where it really doesn't make much sense to slam the positive drive when you can only take the negative excursion down to cutoff. Geroge Tubelab was able to yank the plates pretty much to ground in his P-P setup - in that case, the screens can act as surrogate plates and encounter a lot of abuse as a result. If I recall, he was actually able to melt some screen gids with that sort of treatment.

It would be interesting to take my 6CD6GA screen drive amp out of the living room and look at the plate excursion for varying levels of drive (something I've never done). Right now, the plate supply is at ~400VDC. The control grid is tied to the cathode through a 330 ohm resistor. I'm running fixed bias with the idle current set at around 65 mA. This is over the stated plate dissipation of the 6CD6, but the pair I'm using have held up nicely. I'm actually using a pair of CBS Hytron 6CD6Gs, as these have a beefier plate setup than the corresponding RCA units, which use a bizarre setup with paraleled elements. The Hytrons also have the old "coke bottle" envelopes, which impart a neat retro look to the amp. I've also run GE GA type bottles, as well as a pair of
Sylvania 6CB5As, all of which bias up satisfactorily in my circuit.

I was actually considering doing a higher power version of my 6CD6 amp using the 6P45S. It appeared to require some negative bias to tone down the screen grid sensitivity, but I never understood the necessity for the -33V used by others. I was thinking of starting with -8 to -10V. I suppose running a set of curves would be revealing.

To be honest, though, I think that the sweep tubes are kinda wasted in SE apps, as it's difficult to take advantage of the high peak cathode current capability in that application. The hungry filaments also make selecting appropriate power iron difficult - it's embarassing to be blowing more filament power than you're actually delivering to the speakers...

avincenty 24th April 2013 02:18 AM

1 Attachment(s)
A while back i was playing around with my analog storage oscilloscope and plotted a 12bk5 screen driven. There is a slight kink.

All times are GMT. The time now is 06:03 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio