Measure vacuum tubes (triodes, pentodes) as they did in the 1950's with a remake of the Tektronix 570 "Characteristic Curve Tracer".

I have a heap of vacuum tubes lying around for years that I took from tv's and radios in the previous century. They just sit there gathering dust and I don't know if they are any good anyway. Curious to see if they're worth anything, I looked up a few on Ebay. I hit upon a seller from France that accompanies each vacuum tube with a photo of its "characteristic curve",  which he makes with an old Tektronix 570 Characteristic Curve Tracer. Search for "Tektronix 570" on Ebay.

This tube tester successively applies several different grid voltages (0V, -1V, -2V, etc) and then varies the plate (anode) voltage from zero to a couple hundred volt. Plate current is drawn against plate voltage. In this way a handful of graphs is automatically displayed on an oscilloscope screen.

I want to have that too!
The Tek570 was a huge machine weighing 75 pounds, 42 cm high, 33cm wide and 62cm deep. With modern electronics, I'm sure I can do this at a fraction of the weight. I'll use an existing oscilloscope. Maybe later, I can even measure things like transconductivity etc. automatically and display all that on an LCD display or something. But let's start with the Neptronix 570!

My design
I am using an ATMEGA32 because it has lots of I/O ports, is available in DIP, is cheap and I have several of them lying around. The only drawback is that it doesn't have a digital to analog converter (DAC), so I'm using a two-output AD7302. Disadvantage is that it is no longer available in DIP (at least for a reasonable price), but in 0.05" grid SMD package, so I have to solder it on a small converter board to be able to put it on a breadboard. The 8 bit wide parallel interface is very convenient from a firmware standpoint.

The microcontroller and the DAC co-operate to put a voltage on output B of the DAC that goes up stepwise from 0V in 20 steps of 10/256*5V up to 3.9V. As the grid voltage needs to be negative, I convert the 0-3.9V output from output B of the DAC to minus20-0V with opamp IC3A. I added R9 and R10 later to do fine adjustment of the delta between each step (1 volt) and the offset. Opamp IC3B was also added later and is explained below.

The plate voltage has to go up to +300V. First, opamp IC2A amplifies output A of the DAC 3 times and adds 2.4 volts. Then T2 does an amplification of -16. T3 is an emitter follower to increase the current. T2 and T3 are MPSA42, which can withstand up to 300V. Whereas the grid voltage has to be exact in steps from 0V, -1V, -2V, etc., the plate voltage doesn't need to be exact, it sweeps from about 50 to about 300V and is being plotted on the oscilloscope screen. There is a switchable resistor in the plate circuit, which limits current and power dissipated in the tube and is copied as-is from the Tek 570.

In order to be able to measure the kathode current, R26 is placed between kathode and ground. To compensate for the voltage drop across this resistor, IC3B adds the kathode voltage to the output voltage of IC3A. The result is given to the grid.

The following supply voltages are needed:
  • +5V for the ATMega and the DAC
  • +20V for IC2
  • +320V for the MPSA42's to generate the plate voltage
  • -20V for IC3A to generate the grid voltage
  • Heater voltage, depending on the tube used.
Note that the maximum supply voltage of the LM358 is 32V, so don't use +20V and -20V on one LM358!
I generate the +320V with an MC34063, a mosfet and a power inductor. For the -20V I used a buck converter with mosfets and capacitors, connected to some free ports of the microcontroller, but later I removed that again.

The graph that accompanies this article was produced with this circuit. So it works! As an experiment, I wanted to indicate visually in the graphs which one is the -1V, which one the  -2V, etc. If you look closely, you will see one or more bright spots in the graphs which indicate just that! This was done by connecting the brightness (z) input of the oscilloscope to PD3.

Future plans
I suspect the the LM358's are a bit too slow to generate a clean step function. Maybe I'll replace them with something faster. Any suggestions?

I also want to be able to draw a graph of kathode current against grid voltage, with stepping plate voltage. For this to be of any use, the plate voltage needs to be controlled exactly. The current circuit is not able to generate a reliable step voltage on the plate output, there are lots of ringing and other unreliable behaviour when I try to do that. I'm open for suggestions how to improve the plate circuit.

Also the grid circuit is not perfect. I would want to get rid of those nasty potmeters. Any suggestions?

I want to be able to test pentodes. They need yet another supply voltage for grid 2. Or one can tie g2 to the anode.

It would be nice to use the Analog inputs of the ATmega to measure the voltages and show the results an a graphic LCD display. Then we could measure really slowly, eliminating any AC behaviour. We could do without an oscilloscope. Or measure and playback on a scope anyway!