6922 acting strange........or is it ?

[ashok]

Hi ,
I rigged up a 6922 (Sovtek ) tube as a simple cathode follower with +/- supplies and I get -2.4 volts ( measured) at the grid . Something must be wrong. I tried a 6922 EH tube also , with similar results.
I later checked the simulated circuit which also shows the expected zero volts at the grid ( with respect to ground).
What could be wrong. In one tube one section died after a few hours use. That grid now shows the negative supply voltage at the grid !
I can't think straight now. Anyone have any ideas what could be wrong? I checked the board several times to see if there was any connection errors. I can't find any.
Thanks.

[Sch3mat1c]

Try a different one... any grid voltage other than what it should be means grid current...

Tim

[ashok]

Thanks Tim,
I will check it out with other tubes.
I might add that I later added a 1N4007 diode from grid to cathode directly. Cathode of diode to grid and anode of diode to cathode of tube. Hoping to restrict any + voltages to the grid during turn on.
Ashok.

[SY]

If I understand your drawing correctly, you've got the grid grounded (through the grid leak resistor) and a 10K resistor in the cathode circuit going to -40V? If so, and if you've got voltage on the grid, you've either got a wiring error or a bad component.

[Sch3mat1c]

Originally posted by ashok
Thanks Tim,
I will check it out with other tubes.
I might add that I later added a 1N4007 diode from grid to cathode directly. Cathode of diode to grid and anode of diode to cathode of tube. Hoping to restrict any + voltages to the grid during turn on.
Ashok.

Absolutely no need, CFs are inherently stable and will find their own bias point. When you have *many* stages direct coupled and not stabilized the best, you might have to worry about such.

Besides, as soon as the tube starts conducting, it draws grid current just as the diode does if grid is positive. AND besides, the grid leak is hopelessly larger than the cathode resistor - even if a G-K short occured, or the diode there burnt out shorted, the resistor only has so much pull on the much stiffer Rk.

Tim

[ashok]

Hi Tim and Sy,
Thanks for all the suggestions. I found the Sovtek 6922 seemed OK but he Electro Harmonix 6922 was not with one channel significantly different from the other .
It never occured to me that their actual operating characteristics could be different. From the curves on the Web for the 6922 , I expected certain operating voltages. I found them to be totally different !
I later used a current source of about 1mA ( original plate current was about 4mA) and supply of +/- 40 volts.
I got -0.74/-0.83 volts grid cathode voltage for the Sovtek 6922
and -1.2/-1.3 volts for the EH 6922 ! They certainly are very different tubes !

What looked like a faulty tube earlier ( EH6922) now settles down quite well at 1mA plate current . I heard that many like to run the 6922 at higher current as they sounded better. Well it can't be done with the tubes I have and low plate voltages. I have some NOS ECC88's , but I guess they can't really be compared.

What are the other dual triodes that work well at low voltage ?
Preferably newly manufactured. The NOS tubes can get really pricey and I've burned my fingers more than once by getting NOS tubes that were bad.
Thanks everybody.
Ashok.

[Sch3mat1c]

No one said tubes had to be exactly on in terms of curves. Particularly if you have that unlikely bad egg. I would try a few to see if any match up to the expected data... then I'd check the circuit if none work. (My collection of 6U8s and 6GH8s aren't the most predictable as I've seen from trying them in Frankenhouse. Ain't used tubes great...)

Tim

[ashok]

Hi Tim,
You are right about tubes not behaving exactly as the curves suggest. Might be worthwhile getting a curve tracer. Unfortunately they are too expensive. I wonder if we can build one that can display on the computer screen ( with an external exciter box ).
I saw some curves from a Russian curve tracer ( using dos software). It would do. I couldn't find any info on it.
Cheers.
Ashok.

[SY]

I'd love someone to do a decent and buildable diy curve tracer for tubes. I've seen a few commercial efforts, but they're pretty expensive.

The funny thing is that we tube guys get terribly surprised when the curves for some tube are 20 percent different than the published curves. Yet the FET guys do just fine with parts that routinely vary in their critical parameters (like gm and idss) by five times. Kinda puts things in perspective.

[EC8010]

Yes, the phrase "sample variation" was invented for FETs.

It's worth remembering that the grid of a cathode follower is a stunningly high impedance point and if you had accidentally fitted a 6M8 grid leak (rather than a 680k), then connecting the paltry 10M input resistance of your DVM would be quite sufficient to cause the problems you are seeing.