Tubes?

[Earl]

Hi Earl,

Yes, I did miss the pull up resistor on the setup. Just to make sure I have this straight, the pull-up resistor should be tied between the zener diode and the +300V plate supply? Should I use a huge value like, say, 1MOhm for this?

Use a resistor that doesn't dissipate too much heat.  Not critical, maybe 100k to 1M ??
Can also maybe use a zener of 24V and a separate power supply of 30V with a much lower value resistor.  What you want to achieve is a relatively stiff voltage supply that is simple.  A zener power supply with pull-up resistor is about as simple as it goes.

I realize that I probably had a far from ideal setup for this test, but I didn't want to invest time in making a PCB or finding a way to solder the chips directly to my vacuum tube before doing at least some preliminary testing on it. In the final version, I would hope to make a small PCB that would fit right under the tube itself and just 'plug' the tube right into the PCB.

For experiments, I cut a piece of PCB where one or both sides has thin copper platting.  The ICs are soldered to this with their legs in the air, held to the copper sheet by soldering ground to the copper (usually on one corner) and Vcc (usually the opposite corner) to ground plane via a chip ceramic C.  The rest of the wiring is done "in the air".  Thin copper or brass sheet would also suffice.

I also meant to ask you earlier about the capacitor. What size range cap is appropriate for this arrangement? I just wily nilly picked the size that I had (0.1uF),

0.1 disc ceramic should be OK, better is 2 or 3 in parallel, each with a different value, but for first experiments one 0.1 disc ceramic is good enough.

but I honestly don't understand how this circuit actually functions. Could you explain it in detail to me? I would probably help me not make stupid choices in picking out the appropriate parts and things.

See attached drawings.  When cathode to ground (grid) is +30V, then grid to cathode is -30V.  -30V should drop the plate current to zero, I would think.  When the zener is shorted by the switch (FET) grid to cathode is zero Volts and the tube should conduct very well, probably enough current to turn the plate red hot.  When conducting, current flows in the green loop.

A 6AS7 triode needs much too much negative voltage to turn off; maybe -200 Volts on the grid.  An EEC88 needs only -4 to -5 to turn it off.  A 6N1P audio triode needs about -10 V on the grid to turn it off.

Thank you again for all your help!

God Bless,
Jason O

[Jdo300]

Hi Earl and others,

I did some more tests on the tube circuit today and have scope shots galore below. The first modification I made was to add the 100 kOhm resistor between the zener and the +300V supply. (By the way, I just discovered from the last experiment that the supply is actually putting out 468VDC and not 300 so I am overdriving my tube quite a bit).

There are two problems that I am still faced with after fixing the zener bias. Though the bias did help to sharpen up the rise and fall times on the FET significantly, I am having trouble getting the vacuum tube to turn all the way on. I'm also having problems getting it to switch on fast.

In the first scope shot attached below, you can see how much the tube is switching on. (I replaced the coil with a 20K resistor since I realized that it was the low inductance of the coils that was not allowing me to see the full pulse when it switched the tube on).

I've been having problems getting the tube to conduct even when just applying a 0V bias to the grid to allow it to be fully on. My first guess is that I may simply have too much voltage applied to it so I will have to find a way to drop the input voltage down to 300. According to the scope shot, there is still 150V between the bottom of the square wave and the 0 line. Is it simply the fact that the tube is overdriven causing this or something else??

The second issue is the slew rate. In the second scope shot, you can see a comparison of the MOSFET switching on (bottom trace) to the tube switching on (top trace). The third attached image shows the waveforms zoomed in a bit so you can see it better.

The MOSFET is switching on very fast (~20 ns) whereas the tube is coming on in about 380 ns. I'm guessing that most of that problem can be fixed by using shorter wire lengths and I will tackle that issue last after fixing the turn-on problem.

@Earl,

After fixing the turn-on issue, then I'll try your suggestion with the PCB setup. But I'm not sure what is stopping the tube from fully conducting. According to my calculations, if I look at the tube and the 20k resistor as a voltage divider, the tube seems to have an equivalent resistance of 9433 Ohms. Assuming a 468V B+ supply, the tube would be dissipating about 2.4W which, according to the datasheet, is about the limit that it can handle. Would simply getting the DC input voltage down fix the problem?

God Bless,
Jason O

[Earl]

Hi Jason,

One of the first things to do is look at the plate characteristic curves.
How much negative grid bias is necessary to turn off the tube?

I would like to see one trace with drain or plate, and the other trace the zener voltage, during a on/off switch cycle.  The last two photos only show the falling transition, not the rising edge.

If the grid (zener) voltage is sufficient to turn off the tube, and zero Volts sufficient to turn it well on, then the plate voltage should drop to zero if the impedance of the plate coil is high enough.  However the plate resistance is nowhere near comparable to the R_on of a power FET.  Plate resistance of a tube can be, say 10k to 100k.  Take a look at:
http://www.angelfire.com/electronic/funwithtubes/Basics_04_Triodes.html

This is a good page about using a 6SN7 in pulse circuits:
http://www.6sn7.com/
Also has info about the 12SX7GT that can work with 28V on the plate.

It appears that you need a tube with low plate resistance and turns off at convenient grid voltage.

What are you presently using for a plate load coil?

Earl

[Jdo300]

Hi Earl,

I can easily turn the tube off with the voltage presented by the Zener diode. The problem I was having was getting the tube to conduct fully. Unfortunately, I didn't have a chance to take some more scope shots of the circuit before I dismantled it. I changed the setup to the simplest possible arrangement; just the tube with the 20k resistor in series with the B+ and a DC power supply to set the grid bias. I wanted to find out what exactly was stopping the tube from conducting fully.

From running the test, I could see that tube would shut off easily with less than 30V negative bias on the grid (I forget the exact number off hand). But when I set it to 0, I still measured 150V across the tube. For this experiment, I also changed the B+ supply to 300V just to test my earlier notion that the tube simply could not  dissapate the power fast enough.

It turns out though that you were correct about the plate resistance. I was actually able to get the voltage drop across the tube down but that was after I added an extra 100k resistor in series with the 20k. I also tried adding positive voltage to the grid, which also seemed to push the tube all the way on, though I heard that you're not supposed to do that right?

Anyway, I'm going to see if I can get my hands on some of those 6SN7 tubes that you mentioned. They sound like the perfect fit for what I need, plus the fact that there are two triodes in the same tube gives me ideas to use it in a push-pull setup. I've done push-pull setups using MOSFETs and it works great to keep the current input into the coil down.

God Bless,
Jason O 

P.S. I was originally using a ~30 mH inductor for the load coil, it was just what I had laying around. But for the TPU, the coil inductance will likely be down in the low uH's. But the ultimate idea is to put in very sharp short pulses to smack the coil before a significant current will start to flow.

[Earl]

Jason,

the first thing to find out with one 6SN7 section is if the plate resistance is low enough to bring one end of a several uH choke down to ground during a short pulse.  If not, try both sections in parallel.  Driving a grid positive turns it into a plate; it starts drawing current, so this offers no advantage.

Whether in the end you stick with tubes or return to semiconductors, you will will still learn a lot.

It is written that

At a local radio shop Tesla bought 12 vacuum tubes, some wires and assorted resistors, and assembled them in a circuit box 24 inches long, 12 inches wide and 6 inches high, with a pair of 3-inch rods sticking out. Getting into the car with the circuit box in the front seat beside him, he pushed the rods in, announced, "We now have power," and proceeded to test drive the car for a week, often at speeds of up to 90 mph.

It is also said that

The box is said to have been 24 inches long, 12 inches wide and 6 inches high. Out of it protruded a 1.8 meter long antenna and two ? inch metal rods. Inside the box was reputed to be some dozen vacuum tubes -- 70-L-7 type -- and other electrical parts. Two wire leads ran from the box to a newly-installed 40 inch long, 30 inch diameter AC motor that replaced the gasoline engine.

The 70L7 has two sections, a Rectifier and a Beam PowerTetrode.

The electric motor was 80 kW, so how twelve 70L7s could power it is a big mystery.

I myself, am a FETman or perhaps an AVALANCHEman.

Regards, Earl