having trouble calculating transconductance

[yourownfree]

I have a bunch of 6AQ5 tubes I am testing. I am trying to test them by hand. I do have a basic Sencore tube tester but I want more precision. I cant see how how a good tube tester can give the correct transconductance if Gm changes with Ip. I see people selling tubes and stating their Gm measured on their tester, but my question is.. at what voltage and current? I have made several measurements with a digital meter inline with the Ip and another meter I use to check the plate voltage which I have a 0-300 volt dc regulated power supply. The bias is supplied from a couple of batteries and a couple of resistors (not regulated yet)to have choice of -12.5 volts or 13 volts. If they stray say to -12.53 volts, I take note as well, and use that in my calculations. I understand that if I keep the plate voltage at +250 and check my Ip with the grid bias at -12.5 volts and again at -13volts I use that as my calculation which is change in Ip/ change in Grid voltage answer is Gm. Well I definitely get different answers for the same tube. That is if I change the plate voltage to say +220 or +200 and then do the process all over again. The only thing I see that is more constant is the plate resistance. I had one 6AQ5 make me fall out the chair and I had to recheck it as I could not believe what I saw. This one had an Rp at +250 volts of 1714 ohms. It drew 66.8 ma with -12.5 volts on the grid. Most of the others are from 46ma to 52ma. Anyhow, is there a standard as to how the tube manual checks their Gm. The RCA tube book states that an RCA 6AQ5 has a Gm of 4800. Are they going more negative with the bias or more positive, and how much of a change in bias do they use to get their answer? I read in the tube manual an example was .5 volt change on the grid. I cant seem to get a constant on my calculations. You would think that doing this by hand I would get good results compared to a tester as I am trying to be precise. How does a tube tester make a change in the grid voltage and calculate the Gm? If I get this thing down I would like to make a tube tester that uses the computer to calculate the values and control the grid voltage and plate voltage. What the heck am I doing wrong? I am not going to run out and get a tube tester, I would rather build one if I have to. I saw one somewhere a guy built that was suppose to good called a rabbit or something like that. The bottom line is here, I want to be able to measure Gm accurately as well as all the other measurements. What are methods that would work for me?

[nicoverduin]

The transconductance you measured is at the BIAS point with the given voltages. Tube manuals usually state this info by giving VA, Vg2, Vg1. Is you wish to find comparable info, you need to do the same. The transconductance is the tangent line along the curve at that specific point. In other words, with different voltages you will get different measurements. The book usually gives the typical operating parameters.
Also keep into account that the book is based on averages.
As for the differences in plate current, that is normal. I bought a batch of 600 Sylvania 6V6 GTY 's once. Plate current varied per tube from 16mA (Va=250V, Vg2=250V, Vg1=-12.5V) to 57 mA. S (transconductance) varied from 2400-4800.
This was a little confusing in the beginning. Now this can be appreciated as this gives me a great flexibility to use tubes with a high plate current (enabling more headroom) compared to tubes with a lower plate current.
Hope this helps a bit.
As for the tube tester. I have been dreaming/thinking/designing on building my own digital tube tester including a curve tracer.
The voltage problem seems to be resolved as I am currently having a variable voltage SMPS developed for 0-450V, 250mA max. two of thos wil be used for Va and Vg2. If they work well, I will have 2 more developed for Filament and Vg1.
They will be PIC controlled (manual is also possible). The biggest challenge seems to be the dirgital switching circuitry for the different tube sockets. Ikt looks so simple, but I don't want to install some 50-80 relais. I prefer Mosfets, but I still have some difficulty in getting the right design. Maybe we shoudl start a separate thread on this issue.

[TheGimp]

You might want to look at this thread:

Tube testers which is the best?

near the end there is as discussion of the RAT tube tester, along with other info.

Also this thread of my learning experience on how to measure GM.

Measuring Tube Gm

[nicoverduin]

Thanks for the info. But the real challenge is the following:

for every electrode (A, G1, G2, G3, C, F1, F2) there is a rail. Each rail has to connect to a specific pin on the socket. I want this digitally switched. Somehow I haven't found the correct solution with mosfets yet. Possibly a mindset error? The voltages won't be a problem, the digital control circuitry won't be a problem, the measuring of currents and voltages neither... just the switching.

[TheGimp]

Multiple sockets for different pin outs. Minimize the number of switches necessary to take care of the odd cases. Hard-wire your filaments. Select a small set of sockets to handle the common filament configurations. Have maybe two or three 9 pin sockets with the filaments hard wired for those configurations. That leaves fewer pins to switch. Do the same for 7 pin sockets.

Alternately you could have only one 9 pin socket, and have adapters for the filaments and still only switch the anode, cathode, grid, screen 1, screen 2, etc.

Also, if you use relays you will need to turn the HV off before switching, then turn the HV back on or you will arc the contacts. Either that or use HV Vacuum relays.

[nicoverduin]

Originally Posted by TheGimp

Multiple sockets for different pin outs. Minimize the number of switches necessary to take care of the odd cases. Hard-wire your filaments. Select a small set of sockets to handle the common filament configurations. Have maybe two or three 9 pin sockets with the filaments hard wired for those configurations. That leaves fewer pins to switch. Do the same for 7 pin sockets.

Alternately you could have only one 9 pin socket, and have adapters for the filaments and still only switch the anode, cathode, grid, screen 1, screen 2, etc.

Also, if you use relays you will need to turn the HV off before switching, then turn the HV back on or you will arc the contacts. Either that or use HV Vacuum relays.

I was planning on switching the right connections first. Then apply The HV at the end to avoid that problem. Filaments will probably be switched with relays. HV with mosfets and the neg Vg1 with an opto-fet.
But like I said I am not out of that problem yet...

[yourownfree]

Oh yeah It was a RAT not a rabbit. Looks like I was on that thread too. I need to go back and review that diagram. Ok at least I know I wasn't crazy, and that Gm does vary. I tend to forget things easier now, places I have been and things I have read. It is a constant battle to stay on top of things anymore. I am still trying to make new dendrites, or at least make room for more info. I just recently started to understand tubes more, the reason I am pounding my head trying to soak all the info. I built some transmitters etc from scratch but never was interested in audio til now. I seem to have a bit of crossover distortion trying to decide between the two. I will start a new thread as per suggestion on the pc tube tester. Seems that has taken over. Thanks for the replies as I will keep viewing and learning.

[TheGimp]

Nico, I think your idea of using a micro to run the process is a good idea. It can eliminate, or at least minimize the probability of executing the process in the wrong sequence. It is very easy to forget to turn the voltage down before flipping a switch.

I'm still working on my Vreg. I think I might have something with this design:

[TheGimp]

Believe it or not, that started out from this:

http://www.national.com/ms/LB/LB-47.pdf