Questions on vacuum triode behavior

[rubidium]

I'm exploring various alternative vacuum tubes for use in an audio amplifier. These are basically low-mu triodes having equivalent, or similar, characteristics to the 6SN7 family. One of these of interest to me is a WW-II vintage single triode - the 2C22 (AKA 7193 in military designation). To characterize it, I have assemblesd a bunch of GPIB-controlled power supplies and multimeters and have put together some MATLAB code to essentially map out families of characteristic curves for these tubes.

My questions concern some strange behavior that I'm seeing with the one 2C22 (7193) that I have. With the filament activated for several hours, I make an initial run to map out the characteristic curves. But when I immediately follow that run with subsequent runs, the subsequent runs all group together with a small shift in performance from the initial run. This can be seen in the first image. To explore this further, I simply pick a single operating point (V_anode=250V, V_grid=-8V) and plot the anode current vs time. The current starts out high, and then settles to a stable value. Removing the anode voltage for about 10 minutes and then reapplying it shows the same behavior. This is seen in the second image.

So what is happening? Is the dissipation of power causing thermal expansion of the rectangular anode, with a corresponding increase in cathode-to-anode distance causing the anode current to drop? Is is gas in the tube? The other strange behavior is a slight photosensitivity I can observe. Compared to operating the tube in total darkness, the anode current increases by about 0.5% when I shine a flashlight on the tube from the top along the cathode axis. This is repeatable - over and over. I don't have any other tubes that behave this way.

Any thoughts?
Jim

[Roehrenonkel]

Hi rubidium, hi Jim,

To characterize it, I have assemblesd a bunch of GPIB-controlled power supplies and multimeters and have put together some MATLAB code to essentially map out families of characteristic curves for these tubes.

My tube-test-system is a lot like yours.
The multimeters (seven by now) are in a VXI-mainframe and all is controlled by HP-VEE.
The 2C22/E1148 is a very  strange tube with Grid and Anode on top-caps.....

Are all tubes behaving like this?
If yes, then i'd blame it on thermal expansion.
If not, then you got a dull tube.

How is the current messured?
Maybe swap the meters?
To what potential is the filament connected?
Can you meassure grid-current?

Photosensitivity:
I know this just from "sand". ;-)
....and stabilizer-tubes.
Try another flashlight, maybe emf from pulsed LED?
Add a 1k-resistor and/or ferrites at the grid-cap.

Good luck, best regards

[TimFox]

One possibility:  when measuring grid current in small-signal tubes, I noticed that running cathode current for several minutes caused noticeable changes in the grid current, probably due to outgassing the grid wires.  This might contribute either directly or by making small changes in the actual grid-cathode voltage to your measurements.
Another possibility:  when biasing an amplifier with low-mu, high g_m power triodes, I ran the audio power up to the maximum anode dissipation of the tubes to measure distortion, then removed the audio drive.  This resulted in a large shift in the quiescent current, possibly due to a change in the grid-cathode spacing due to thermal effects.

[rubidium]

Roehrenonkel, I have a few more 2C22 tubes arriving in a couple of days. I will have to wait until then to see if my observations with just a single tube can be observed with the others. The tube that I have is 80 years old, so I wouldn't be surprised by anything at this point.

To clarify my setup, I am controlling a Sorensen LHP 300-3.5 (300V @ 3.5A) power supply via GPIB as my source of plate voltage, with positive connected to the 2C22 anode, and negative connected to one side of an Agilent 34401A set up as a current meter. The other side of the 34401A goes to the 2C22 cathode. In addition, an Agilent E3631A power supply serves as my grid supply, with positive connected to the 2C22 cathode, and negative to the grid. This arrangement has worked flawlessly with other tubes, such as the 6SN7 - giving me very agreeable sets of curves to published curves and, the results are stable and repeatable. A separate and isolated power supply is driving the filament with 6.3V, and 300mA  is drawn - which is the correct level. Again, this supply is completely isolated from everything else, e.g. the filament is not connected to the cathode.

I can and should introduce an additional current meter to check grid current. (At one time I looked at grid current for some 6SN7's, and it was as much as 10-20uA, but only when the grid was very close to 0V. I think that's to be expected.) But I don't think it's gas. The time constant for the anode current to stabilize after a cool-down period appears to be *minutes*, and also quite repeatable. It also drops, rather than rises, as it stabilizes. All of that strikes me as being due to thermal effects. But I'll see if the effect is present in other tubes of the same type once I get them. The one tube I've been experimenting with is a National Union military edition (i.e. JAM-CNU-7193). I'm getting a few more of the same, along with an RCA commercial edition 2C22. I'm not sure if there's much difference between the military and commercial construction.

The slight photosensitivity is baffling. I can cause it with a DC incandescent bulb in an old-school flashlight, so I shouldn't be generating any EMI. At a plate current of about 8mA, I get an additional ~40uA with the illumination. It is very repeatable. I'll see if the same happens with other tubes.

Jim

[jonpaul]

Rb:

1. See high end audio forums for lots of vacuum tube amps and alternate devices.

2. Use a curve tracer, like TEK 575/576/577.

3. Possible to DIY a vacuum tube curve tracer, or adapter for tubes on the above TEK curve tracres,

4. See RCA/Varian/Eimac industroial/transmitting tube catalogs, manuals for well characterized tube specs and typical curves.

Jon

[Roehrenonkel]

Hi Jim,

.......
A separate and isolated power supply is driving the filament with 6.3V, and 300mA  is drawn - which is the correct level. Again, this supply is completely isolated from everything else, e.g. the filament is not connected to the cathode.

I can and should introduce an additional current meter to check grid current. (At one time I looked at grid current for some 6SN7's, and it was as much as 10-20uA, but only when the grid was very close to 0V. I think that's to be expected.) But I don't think it's gas. The time constant for the anode current to stabilize after a cool-down period appears to be *minutes*, and also quite repeatable. It also drops, rather than rises, as it stabilizes. All of that strikes me as being due to thermal effects. But I'll see if the effect is present in other tubes of the same type once I get them. The one tube I've been experimenting with is a National Union military edition (i.e. JAM-CNU-7193). I'm getting a few more of the same, along with an RCA commercial edition 2C22. I'm not sure if there's much difference between the military and commercial construction.

The slight photosensitivity is baffling. I can cause it with a DC incandescent bulb in an old-school flashlight, so I shouldn't be generating any EMI. At a plate current of about 8mA, I get an additional ~40uA with the illumination. It is very repeatable. I'll see if the same happens with other tubes.

Jim

Tubes can get one puzzled, indeed. See for example the two E88CC/6922 i tested would be ideal tubes for an epay-sale when checked only on an emission-tester. I wonder how the heck it got through the Siemens-QC.......
Only the first tube has a defective system (low control-function, too much current), the second is really good.
 
Given that the 2C22 is a UHF-oscillator-tube it might be "hand-sensitive", have you checked for oscillation
(shorted scope-probe with the ground-wire as antena)?
Filament-supply should be connected to some potential (cathode or a bit positive), Rfc-max is stated in some data-sheets.

Also i meassure currents not directly but over a 1 Ohm (Caddoc MP900) resistor in the supplys minus-lead to ground.
So one can better isolate tube from instrument (and wiring!).

The rise in grid-current near 0V is normal, the grid is slowly becomeing an anode.

Photosensitifity: It's just 0.5%, can you modulate it with a squarewave and see what the maximum frequency is?
Maybe it's a hidden photo-multiplier..... ;-))
Maybe i'll ask around at https://www.gfgf.org/Forum/ on my Tube-radio-forum where the wise geezers are.

The 15 pieces of E1148 (2C22, VT232, CV6, CV1135) from N.U., GenRad, GE and Marconi are in storrage at the other end of town, my curve-trace-system is in a upgrade-stage (Keithley 237 SMU for grid-supply) and i was building myself a CNC-mill/-Laser. So much to do, but sometimes i'll see if i can replicate your findings.

Best regards

[rubidium]

I should receive a few more tubes of this type today to look at. The single tube I've tested thus far could always be corrupted in some way, despite it being "NOS" - since one must emphasize the "O".  Also, time to bring on the oscilloscope to check for any oscillations somewhere (in what should be an otherwise DC setup.)
Jim

[rubidium]

Well, I had the chance to look at a few other tubes of this type that I just received - an RCA 2C22 dated 1939, a Ken Rad military 7193 dated 1942, and another National Union military 7193 dated 1943 - and they are all well-behaved and exhibit none of the strange behavior that I noted with the original NU 7193. So it's quite clear that initial tube that I tested was just some quirky outlier. Who can say what conditions the tube experienced during 80 years of traveling the earth. 

At least I'm relieved that this particular type of tube isn't unstable, and even more relieved that my test setup didn't have some issues. I'll still play around with the original tube a bit more when I get some spare time away from more pressing things, as it's become sort of a science project now.

Thanks to everyone for the comments and suggestions ...
Jim