Marko Cebokli S57UUU

Presented at the EME conference Prague 2002

ABSTRACT: When one wants to get useful RF power from TWT's bought randomly at fleamarkets and junkyards, the testing of these tubes usually presents a big problem. A suitable power supply is seldom available, especially for the more powerful tubes. Often these tubes are near (or beyond :-) the end of their useful life, the focusing is degraded etc. Sometimes they can still give useful power, provided one can experiment with them a little. This paper describes a relatively simple device for testing and safe (for the tube) debugging of TWT's with videly varying power requrements.


Here are some of the problems the TWT experimenter encounters in his life:

- you've got a nice big tube but no power supply. Before starting to build or otherwise invest in the big and complicated supply, it would be nice to know if the tube works at all!

- you've got the tube and supply but no heatsink. A few hundred watts of heat will fry the tube in seconds without a heatsink. How to test it?

- you've got everything, but the helix protection trips instantly when you apply the power. Maybe only a slight adjustment of the voltages or a small magnet would help, but how to find the solution without exposing your tube and supply to the repeated stress of aborted switching?

- you've got an 'UFO' tube, no data, not even who the manufacturer is. Using some tube sleuthing you can, with some confidence, find out which terminal is which electrode, but how to find out what voltages to apply without burning the tube out?

- you've got everything but that big RF dummy load. Now, you could hook up that horn, but your neighbours brain has been washed with those fairy tales about extremely dangerous microwaves...


One solution to the above problems is to switch the tube on just for a few (tens of) microseconds. For such a short time, it's possible to supply a lot of power from a capacitor which is charged by a simple small low current high voltage supply. Another advantage of using short pulses is that because of the thermal mass of the tube elements, the maximum ratings of the tube for continous operation can be exceeded by a big factor without endangering the tube's health. Fig 1 shows the basic idea.

a simplified schematic of a pulse tester
Fig 1. Principle of pulsed TWT testing

In fact, only the helix supply has to be switched since the TWT will then switch its own collector current:
a schematic diagram
Fig 2. Separate supplies for helix and collector

The switch is connected with its negative side to ground. This way a tube (or maybe a string of transistors) can be used as the switch. This causes some problems with the capacitor charging supply, which must float. The switch is controlled by a monostable multivibrator which determines the pulse width. If a suitable storage oscilloscope is available, it can be triggered by a debounced pushbutton in a single shot way. Otherwise, a low frequency (tens of Hz) oscillator can trigger it periodically so that any oscilloscope can be used to view the pulses. The duty cycle shold be kept below 0.1% to keep the average power requirement and dissipation low.


Because of the huge dissipation margin, the stages of a multi collector tube can be connected together and operated at the first collector's voltage. With continous operation this is not allowed, because the last collector would take most of the current and overheat. When pulsed, the collector(s) can even be operated at helix voltage (fig 1), however a separate energy storage capacitor is still desirable to reduce the voltage drop on the helix during the pulse. Both can be fed from the same supply, with a resistor between them for decoupling. In the repetitive pulse regime, this resistor can be dimensioned so that the voltage drop across it will give the correct collector voltage. In single-shot mode a full voltage divider (add R2 across the collector capacitor - fig 3) is required for that purpose. Observe the power and voltage ratings of the resistors! Collectors are not very sensitive about their voltage as long as it is above the minimum required.
yet another schematic diagram
Fig 3. Common supply for helix and collector

The collector current is not accesible at ground potential, but since it is pulsed, a small (well insulated) current transformer can be used to measureit (fig 3).


Their capacitance is determined by the voltage drop that can be tolerated during the pulse. The basic equation is

dU/dt = I/C

and therefore

C = I*dt/dU.

For example, a 50V drop during a 100us pulse at 100mA load current gives

C = 0.1*0.0001/50 = 2E-7 or 200nF (0.2uF).

Since the helix requires a more constant voltage but draws less current than the collector, a similar value of capacitor can be used for both.


I have used a PL519 pentode that once served for TV line deflection. It is specified for up to 8kV of pulsed anode voltage. Since here the voltage is present across the tube most of the time (it's the current that is pulsed), I'm afraid it will flash over one day. On the other side, the current and dissipation in this circuit are far below the tube's maximums. This is a "P" series tube, built for 300mA series connected heaters. It's a bigger tube, so it needs 40V filament voltage at 300mA. I used a 20+20V transformer and derived all the required volotages (40V filament, +12 for the solid state circuits, +200 for second grid and -100V for first grid) from it:

schematic of the low voltage power supply
Fig 4. Modulator power supply

A simple circuit with two 555's provides a pulse train to drive the PL519:

schematic of the pulse generator
Fig 5. Pulse modulator

The bottom one is connected in astable mode and can be set with the "freq." potentiometer between cca 2 and 20Hz, to enable viewing of the pulse waveforms on a non-storage oscilloscope. It triggers the upper one, which is wired as a one-shot and determines the width of the pulse, which can be set with the "width" potentiometer between cca 5 and 100 microseconds. The three transistors amplify the pulses and shift the levels to values appropriate for driving g1 and g2 of the PL519.

A photo of the hardware Fig 6. A photo of the prototype modulator

I soon found out that at g1=-100V and g2=+200V, the PL519 starts to 'leak' current at about 2kV of anode voltage. On the other side, with g1=0V and g2=0V it won't open fully. That's why I added the circuit for g2 modulation. So the tube is operated with g1=-100V, g2=0V for cutoff and g1=0V, g2=200V for conduction. Maybe with -200V bias for g1 it could be operated with g2 fixed at +200V.


The only non-trivial demand on the HV supply for charging the energy storage caps is that it must be floating (well insulated transformer). It is nice if it's regulated, but not mandatory. If one monitors the voltage with a suitable HV voltmeter, even a variac/neon transformer + rectifier combo can be used.


Usually, the TWTs have some extra electrodes between cathode and helix. The required voltages vary widely, but luckily the current consumption is low. That means one can use simple resistive dividers. Manufacturers mostly specify that at switch on, these voltages must rise in proportion with the helix voltage, or else must come up last. If the divider is connected as shown in fig 7, the voltages will rise proportionally. The current through the resistors will flow only during the pulse.

schematic of the TWT grid circuit
Fig 7. Grid voltage divider

To prevent the stray capacitances of the wires and electrodes to influence the shape of the voltages, the resistance of the divider should be relatively low. Since current flows only during the short pulse, the power rating of the resistors is not the problem, provided they can stand the voltage (flashover). The lower limit on their resistance is set by the droop they would cause on the helix capacitor voltage during the pulse.
Alternatively, one could add a small capacitor between grid and cathode to delay the voltage rise on the grid.
In some tubes, the beam can be turned on and off by a single grid. In that case, one can eliminate the big modulator for the helix voltage and make only a small one for the modulating grid, with the helix voltage permanently applied to the tube. The problem is that the grid modulator has to work close to cathode potential. Perhaps a pulse transformer could be used.


Switch on the TWT heater. This is usually 6.3V, up to a few amps for bigger tubes. Use a well insulated transformer, since it must withstand the full helix voltage between primary and secondary! Also try to reduce stray capacitances here (short wires etc). Wait for the prescribed warm up time (a few minutes typically). Connect an oscilloscope to monitor at least the helix current and desirably the collector current on a second scope channel. Initially, I intended the 10 ohm resistor in the cathode of PL519 for helix current monitoring, but sometimes the G2 current of PL519 will distort the picture, so it's necessary to put a separate resistor in the helix (ground) lead of the TWT to monitor the helix current. Apply the high voltage and switch on the modulator. If you have a (digital) storage oscilloscope, turn on the modulator at lowest frequency for just a few pulses (the 'STOP' switch, Fig 5). If the tube is OK, the helix current should be something like fig 8 below

A drawing of the helix current pulse shape
Fig 8. Helix current shape

and you should have full collector current. The 'horns' on the helix current pulse are caused partly by stray capacitances and partly because of poor focusing during the rise time of the voltages. These horns can be several times higher than the maximum allowed helix current, but that's normal - don't worry about them. The steady state helix current, which is important in CW operation of the tube equals the midpoint 'valley' value of the waveform. If the valley isn't flat in the middle, try lengthening the pulse. If you have a suitable HV oscilloscope probe, also check the rise time of the voltages to see if they are ok (less than 1/4 of the pulse width) and when the current overshoot should end. When tuning up the tube DC-wise, your goal is to get most of the electrons through the tube to the collector - that means the lowest possible midpoint helix current. Most of the cathode current should end as the collector current, which should be close to the specified value for the tube. If the currents aren't what they should be, try varying the voltages and applying magnets to the (cathode end of the) tube. Changing the heater voltage could also help. If you manage to get good values for the currents, then you can apply RF. Use a suitable coupler at the output which will give you cca 10..100mW of peak power and a diode detector connected to an oscilloscope to monitor power. The detector can be calibrated usin low power CW RF and DC coupling on the scope.


A reatively simple circuit, using mostly off the shelf components enables one to do many measurements on TWT's, including ones that are not possible or safe for the tube with normal power supply in continous regime. This includes reviving and refocusing of old and/or damaged tubes. By using short pulses, one can also avoid thermal and RF loading problems and test high power tubes with a small low-current HV supply.


[1] Jim Vogler WA7CJO: Optimizing TWT Power Output for Narrow Band CW-SSB Operation, Proceedings of the 9th int. EME conference, Ro de Janeiro 2000.

[2] Marko Cebokli S57UUU: A TWT Power Supply, Proceedings of the 7th int. EME conference, Washington DC 1996

Up to S57UUU EME Page

Copyright info