Grid leakage at power tubes: share measurement and experience

[es345]

We have discussed in the KT120 thread the different scenarios of positive grid current (grid conductance) and negative grid current (grid leakage) where the grid starts to behave like a cathode. http://music-electronics-forum.com/t25271-2/#post224289.

I have measured the grid leakage at my KT120's with the result of 1,65 uA - worst case measurement.

My question:

Does anybody have measurements and/or experience of grid leakage values of power tubes,
expecially over life time ?

Thanks in advance

Hans- Georg

[es345]

Ok,

there seems to be not too much experience regarding that topic,
perhaps not so important in the real world.

[Gingertube]

Hans - Georg,
NO it is important in the real world, its just a mostly overlooked thing and not often measured.
It is also something which 90+% of commecrial guitar amps and almost as many HiFi tube amps get wrong in that they use way too high a grid leak resistor resulting in a high incidence of failure due to output tube run away.
I was working on a 300B PP HiFi Amp last night and measured grid voltage of +3.0 Volts which was being developed across a 220K grid leak resistor, that infers a grid current for the 300B of approximately 1.5mA.

I also note that the latest bias servo board offering from Tentlabs (Menno Van de Vinne and Guido Tent design) is deliberately designed to handle up to 2mA grid current.

Just a couple of observations - hope this helps.

A little theory here:

What we are talking about with power tubes is NOT the grid current which is associated with grid leak bias in small signal tubes.

What we are NOT talking about -part of the electron cloud (space charge) around the cathode makes its way to the grid and causes current flow INTO the grid. This results in the grid becoming more negative which adds to the bias and if the grid leak resistor is high enough can develop the full required bias. This is what is known as grid leak bias.

What we ARE talking about is what happens in power tubes. As the electrons accelerate up from the cathode toward the anode, some collide with residual gas atoms in the tube (Helium and Hydrogen in particular). This collision is energetic enough to strip outer orbit electrons, turning those atoms into positively charged ions. These ions accelerate back down the tube toward the cathode and the grid (the 2 tube elements with the most negative potential). When some of those ions hit the grid then current will flow OUT OF the grid. This current will develop a voltage across the grid leak resistor BUT in this case it is a POSITIVE voltage which subtracts from the bias, resulting in more current => more collisions with gas atoms => more ions =>more current OUT OF the grid => more positive grid voltage and hence run away in the tube.
This can be countered to a degree by cathode bias and that is why you will see (on tube data sheets) a certain maximum value for grid1 resistor for cathode bias and a SMALLER maximum value for fixed bias.

Perhaps someone can chime in here, one of these mechanisms is called NEGATIVE grid current and the other is called POSITIVE grid current but I can never remeber which is which.
I don't want to give anyone a "bum steer" here.
I seem to recall that:
The current OUT of the grid (from space charge electrons hitting the grid) is called POSITIVE grid current and results in a more NEGATIVE grid voltage
AND
The current INTO the grid (from residual gas ions hitting the grid) is called NEGATIVE grid current and results in a more POSITIVE grid voltage.

Cheers,
Ian

[es345]

Ian,

I know I was a little bit provocative

Thanks for you explanations!
Now I know that I have used a wrong explanation/word, but that we are talking about the same phenomen.

Interesting to read that the grid leakage current of a 300B is about 1,5 mA, much higher than at the KT120.
Would be good if we can collect the values for other tubes. I will do it for the EL34. Regarding design margin for the source follower current an estimate between 5-10 mA seems to be ok as already mentioned by Ken and you.

Some more reasoning for my request:

Coming back to the result of my test setup of 3mA injection (So...is anyone using the KT120's yet? - Page 2): the measured impedance with an injection of 3 mA is less then 40 Ohm.
That means: as long as the Source follower is in normal working conditions the value of the source resistor doesnt count in the chain - as long some mA are flowing through the source.
This current is basically the "design margin" I am looking for - not only for the KT120 but also for other tubes.

Cheers
Hans- Georg

[Enzo]

I have never played with KT120, so any questions about them are moot as far as my input.

However, at 1.65ua, I would be curious as to what that would suggest one do differently with regard to circuit elements, compared to any generic circuit. For example, if my math is accurate and appropriate, 1.65ua across typical 220k grid resistors results in about 0.35volt. That would knock my -45v (an arbitrary bias voltage for the example) down to -44.65v. COnsidering the lack of criticality in bias settings, I wonder if that is significant in some way I don't realize.

[es345]

The discussion was coming up in the KT120 thread and I want to get a better overall understanding how big the grid leakage current could be - not only regarding that tube.

In fact I haven't observed any instability at my amplifier with the KT120 even before I have implemented the mentioned improvement proposal in the other thread. But increasing the design margin by exchanging 2 resistors without any functional impact I would always do.

The choice of implementing a source follower in the powerstage is caused by the use case: its a Bass Amp powerstage. I want to avoid any crossover distortion.

[Gingertube]

Another measurement,
From a matched quad of JJ KT88 - worst of the 4 had 12uA grid current.

Of-course grid current goes up as you run the tubes harder. The above was at 32 Watt anode dissipation (42 W max for KT88).

If you stick to the max Rg1 values on the data sheets then you will not have any problems but hardly anyone does - 6V6 (max 100K in fixed bias) and 6L6GC (max 100K in fixed bias) have to be the most abused tubes in history. You rarely see a guitar amp design where these resistors are not at least 220K. While tubes are brand new you will get away with the X2 of recommended maximum - as tubes age or if you want to use NOS tubes then you will run into problems.

Depends on whether you design for 30+ years life time or a lifetime just longer than the warranty period (oops - I think I just pressed one of my own buttons)

Cheers,
Ian

[trobbins]

How do you guys measure down at 1uA? Do you account for the meter in the measurement?

Just interested ;-)

[es345]

I have measured the voltage over the grid stopper (10K in my case) with a high impedance high resolution digital voltmeter.
At the KT120 the result was 16,5 mV which leads to 1,65uA.

[TransLucid]

Ian - 3V across 220k ohms is 13.6 uA.

Bob

[Gingertube]

Bob,
Quit right - should have used the calculator on my desk rather than the one in my head. Must have been a Wednesday (my bad day), although most mistakes apparently happen on Tuesdays when even the calendar for the next 3 days says WTF.
Cheers,
Ian

[Gingertube]

I usually measure voltage across the grid leak resistor and calaculate from that although I can also read it as a direct current measurement (on gas test position) when using my AVO MK3 Tube Tester.
Cheers,
Ian