Measuring Tube Gm

[kevinkr]

I guess one of the things I am wondering is under what conditions is the Mercury actually testing your tubes?

Your test jig is a whole lot more rigorous and I suspect that the mercury is using dropping resistors to drive the screen and that the screen current in some devices may be a bit higher or lower than the mercury expects and hence the different readings.

You might actually want to measure screen current on your jig and observe whether the ones that correlate best with the Mercury also happen to all have very similar screen currents whilst the ones that don't differ significantly. Just a thought..

[TheGimp]

Yes, I didn't consider the screen other than setting the voltage to 100.0V.

I've come to the conclusion that the chassis I chose for my test set is too small considering the changes I've made. I'm going to start over with a larger chassis, and include a set of SIMPSON 135-0-0-13-0 LCD meters to display DC Plate Voltage, DC Plate current, AC Plate Current(with a 1vrms input =Gm), DC Screen Voltage, DC Screen Current, and DC Grid Voltage.

I'm also going to rework the Plate and Screen supplies to allow me greater adjustment latitude and current capacity in testing so the test set will be more generic. I'll ditch the 0B3s in favor of fuylly regulated supplies (FET pass transistors probably).

Not exactly what I started out to do, but I think it will pay off in the end.

[disco]

Reading your interesting journey I made two observations.

- tubes in their circuits will oscillate, though conditions under which this manifests may vary. Did you take precautions like a grid stopper etc?
- seems you did not make use of the stabilisation of your G2 supply. When a tube under test oscillates, it may develop G2 current which in turn will drop the G2 supply voltage over the 5K pot you used. As you probably know the penthode is very susceptible to changes of the G2 voltage, more than small changes to the actual plate voltage (!).

Readings in the 1940/50s were made with 10Kohm equipment, from what I usually see in old Philips publications. Modern SS equipment has higher impedance so test outcome may vary somewhat.

[Michael Koster]

Originally Posted by TheGimp

Yes, I didn't consider the screen other than setting the voltage to 100.0V.

I've come to the conclusion that the chassis I chose for my test set is too small considering the changes I've made. I'm going to start over with a larger chassis, and include a set of SIMPSON 135-0-0-13-0 LCD meters to display DC Plate Voltage, DC Plate current, AC Plate Current(with a 1vrms input =Gm), DC Screen Voltage, DC Screen Current, and DC Grid Voltage.

I'm also going to rework the Plate and Screen supplies to allow me greater adjustment latitude and current capacity in testing so the test set will be more generic. I'll ditch the 0B3s in favor of fuylly regulated supplies (FET pass transistors probably).

Not exactly what I started out to do, but I think it will pay off in the end.

While you're rethinking...

I attached the 6BA6 gm curve for reference. I think the measurement needs to be done in the area of the chart I highlighted in the enargement. To measure gm between 4000 and 5000 the grid swing needs to be <= 1.5V peak-peak (~0.5 VRMS) or less and biased up so that Vg1=0 on the positive peak. I think for small signal pentodes the 0.1V RMS might be a better choice if you can get the S/N ratio of the test set good enough. Your current 1VRMS produces too much current swing at the anode.

Michael

[TheGimp]

I had suspected that and tried lower drive levels.

Thanks for the attachments.

Right now I'm working on better regulation of the plate and screen supplies. I've got an FET based current source working up to 180V (with an appropriate heat sink it should be good to 400V, or better with a different FET). This with the 100V minimum output (drive for the Plate and screen) should allow me to control the supplies from 100 to 350V with a 400V supply input. I plan on using this with a variable resistor to ground to set the output voltage. an Op Amp for a sense amp and another FET for the pass transistor should do it. Big Heat Sinks will be required. Say 75mA with a drop of 250V just to play it safe means 18.75W dissipation. 0.4 W/K with a temp rise of 50C looks like it will be close for the IRF7222/723 which should handle 20W. I may have to find another FET if I cook any of the IRF722/3s.

Did I mention I need a bigger chassis?

Man, I wish I hadn't passed up on the sloped panel test sets that were scraped at work last year. The guts were useless, but those cabinets would have been great. Oh well.

Steven

[magooamps]

[mogliaa]

Hi everyone,
I'm trying to adapt my curve tracer to also measure transconductance using an externally 100mV 1KHz feed. As I'm also using a digital panel meter to switch between anode, screen and grid voltage measurement and avoid impacting the grid bias due to the meter input impedance, I changed a simple bias circuit to a simulated inductance (gyrator) to provide a stable DC bias point. That's works (simulates fine).

Now, when I'm adding a gyrator load and placing a 10 ohm sensing resistor in the anode to measure GM using an AC mV panel, when I simulate it in LTSPICE I get very tinny signal close to 15uV

Am I making a basic mistake here?

Thanks
Ale

[euro21]

Hi Ale,

15uV?

Static anode current 1.52mA (15.2mV on 10R).
Anode current swing 309 uApp (0.1V input) on 10R resistor, thus 3mVpp between measuring points.

[mogliaa]

Hi,
Thanks for your sim! Need to check my simulation again (not in front of my PC now)
So this should work then?
Thanks!
Ale

[euro21]

Yes, it's works, but...

1.) If your negative bias supply (gyrator) not ideal, it makes as divider.
Divider with 1Meg resistor (DC), and divider with signal generator, then shunt input signal.

2.) I think anode current (variation) is to small to measuring in 10R, without amplifying.

p.s:
add this spice directive to sim:

.MEAS TRAN Output RMS (V(Meter2)-V(Meter1)) TRIG time=20*period TARG time=40*period