ron soyland
The circuit is a standard half-bridge type amplifier. The FETs are alternately switched between cutoff and saturation to cause the output to essentially be a signal that has a P-P voltage equal to the power supply voltage.
Power for the amplifier is derived directly from the mains input through the full wave bridge rectifier. This provides a full wave rectified pulse train at the input line rate. There is no filter capacitor to produce a DC power supply. The 4.7 mfd bypass capacitor has no effect to filter the mains frequency but is used to provide a very low impedance path for the upper FET to switch power to the load.
The FET gates are protected by two zener diodes that are directly soldered to the FET terminals for minimum inductive voltage drop. Power is delivered to the output tuned circuit via the two coupling capacitors. The required capacitance is broken down into two parts for safety reasons. The signal ground of the circuit is pulsing at the full mains voltage at all times the unit is plugged in. Thus it is desired to completely isolate the output coil from this voltage to reduce the shock hazard. By putting half the capacitance in the common lead and half in the output, the tuned circuit is isolated from the power line by the impedance of the capacitors, which is very high at the mains frequency.
The output power is determined by the size of the coupling capacitors. If the values are zero, there would be no output power, and if the values were very large the output would be virtually limited only by the current drive of the FETs. Thus it can be seen that by selecting the size of the capacitors any drive level desired can be implimented. The size of the capacitors for 220 volt mains is about half that for 120 volt mains for equivalent output power. The values shown above are for the 120 volt mains.
The output tuned circuit consists of the work coil, which is made of such size as to be proper to flash the tube getters. A set of four capacitors are tightly connected to the coil. The layout of this tuned circuit is very critical and even slight variations in layout can result in poor results. The use of 4 capacitors in parallel to obtain the .12mfd capacitor is to distribute the extreme current as much as practical. The capacitors are rated at 6 amps current capability, each, at that frequency. The operating current of the flasher at 500 watts output is in excess of 100 amps. Thus the resonating capacitors are being overdriven by a factor of at least 4. This is the major time limiting factor in the on-time of the circuit. The capacitors will heat internally and short out if the power is left on for more than about 20 seconds continuously. This is far more time than is necessary to flash a getter so it is not a problem. For longer duty cycles the much more expensive RF mica capacitors can be used, but the cost of each capacitor is nearly as much as the total cost of the entire rest of the unit!
Tuning of the unit is done with a small lamp that is across the tuned circuit. The total voltage across the tuned circuit when at resonance is approximately 600 volts P-P so a limiting capacitor is used in series with the lamp to prevent it from burning out. The tuning is somewhat sharp but easy to do while operating the unit. Once set it doesn't need to be adjusted again unless you change the coil type.


The circuit is quite critical in the design and layout of the circuit. Follow the plans closely as possible or you may end up with serious problems: low output power, lots of blown FETs, etc.


The flasher is built in a commonly available aluminum utility box available from Mouser or other electronics suppliers. The main consideration is that the outer case IS COMPLETELY ISOLATED FROM ALL CIRCUITRY! The only connection to the outer case is to the center wire (ground) of the power cord. NO OTHER PARTS OF THE CIRCUIT CAN TOUCH THE OUTER CASE! This is important in the design because the entire electronics circuit is at line voltage when the unit is plugged in!
Thus, the outer case must be carefully grounded for safety.
All parts are mounted on a piece of 1/4 inch thick delrin or other plastic sheet. The screws are positioned and countersunk such that no connection is possible betweent the circuit and the case.


The circuit is built on a circuit board that is enclosed in a shield box. This shield is necessary to prevent feedback of the powerful RF field generated by the output circuit from causing erratic operation of the 555 oscillator. The input power to the shield box is particularly sensitive to the interference. The 1 millihenry choke coil is used to block the RF. This choke coil has a DC current of over a half amp while the circuit is activated, so the DC resistance of the choke must be low, less than 5 ohms is best.
A ground plane is necessary on the circuit board. The layout is easily done using only one side of the board so the top side of the board can be used as a ground plane. If you make homeade circuit boards with no thru hole plating, use small strands of wire in the grounded holes to connect the related pin to the ground plane.


The driver electronics is powered by a 12 volt unregulated power supply. It is not necessary to have a regulated supply, and indeed the 3 terminal regulators have serious trouble with the strong RF feedback flowing around the circuit and will go out of regulation. Pick the transformer secondary voltage such that when it is loaded to a half amp the voltage is between 10 and 14 volts on the filter capacitor. A secondary RMS voltage of 10 volts works well. These transformers can be found in small wall power units for portable electronics. Junk stores have them for a few bucks each.
The main power for the output amp is unfiltered full wave rectified AC mains voltage. This gives a peak voltage of about 300 volts (on a 220 volt mains) and 140 volts on a 120 volt mains. The bridge rectifier must be capable of handling 10 amps in the 120 volt unit so it will need to be on the heatsink with the power FETs. The current in the 220 volt unit will be less than 3 amps so the bridge does not need to be placed on a heat sink.


The power FETs dissipate approximately 50 watts each during the time the unit is activated. Thus, a very good conductivity of the FET to the heatsink is necessary. Use thin mica insulators for the insulation. DO NOT use the rubber insulator pads! These are not satisfactory for high powered use and will result in blown FETs! Be sure to use heat sink compound on both sides of the insulator. Firmly tighten the mounting screws.
The heatsink actually does not require a fan, due to the short operating time of the unit. But, it is prudent to use a fan. A small CPU cooler from an old PC works excellently and is small enough to fit in the case. Find one that has enough flat surface area to mount the FETs and if on a 120 volt unit, the bridge rectifier. It is highly recommended to mount this entire heatsink on the plastic sheet that isolates all the electronics from the outer case just in chance that there is a fail in a mica insulator.


The unit is made in one box that is about 50mm by 75mm and as long as is necessary for your components. The Hammond box specified is 12 inches long (30cm) so it can be cut down to the length that you need.
All of the electronics for the driver is mounted on a flat sheet of 6mm perspex or delrin sheet. Of course wood board can be used if the plastic is not available. The screws for all parts are positioned and recessed such that there will be no possibility of any of the electronics to touch the outer case.
Begin at one end of the box, mounting the power connector (robbed from an old PC power supply) and the mains fuse. Then the small power transformer. The filter capacitor can be held to the plastic board with heat glue. Then the shield box. Be sure the frequency adjust pot sticks out the side of the box.
The heatsink is mounted to the plastic with small screws and large holes drilled in the correct places to allow air flow though the fins.
Wire the connections to the output FETs with short as practical wiring. The common connection should be of at least #12 solid buss wire (3mm dia) and should connect to the fins of the heatsink as well as the FET.


The end of the box is cut out and a flat piece of perspex or delrin is cut to exactly fit. Small screws are used to hold it in place. Two slots are cut in the plastic and copper sheet wrapped tightly through. (see photos) These sheets form the low impedance output terminals for the tuned circuit. The capacitors for the tuned circuit should be soldered directly to the copper using more strips of copper sheet. This copper sheet does not have to be thick. Due to the skin effect at this frequency, all of the current is flowing in the outer 1/10 mm of the metal. So making the sheet thicker than 1/5mm doesn't add to the function at all. Surface area is what improves operation, not thickness. Place the copper strips right up to the wires exiting the capacitor body. Do not use the wires of the capacitor to run to the terminals since they are very thin and have poor conductivity at that frequency. Solder the copper strip right at the wire where it comes out of the capacitor body. If you use the radial lead type capacitor do the same and solder the copper strip to the terminal along its entire length.


The work coil has certain diminsions that give optimum performance. Too small of diameter and the necessary distance to the getter will be reduced to an impractical value. Too larg of a diameter results in a small coupling coefficient, so the power transferred to the getter is reduced. For most tube getter flashing uses, a 30mm I.D. coil works best.
The number of turns is also fairly critical. Too many turns will increase the wire resistance and thus the Q will drop. Low Q is equal to low circulating currents and low output. Too few turns and the number of amp turns is reduced. Coupling equals amp turns times coupling coefficient so lower amp turns equals less energy transferred. A broad median in the factors (from many experiments) shows that a 4 turn coil made up of two layers of two turns each gives optimum Q and amp turns. The wire size should be at least #12 and #10 (awg) (equal to 3mm and 4mm diameter) solid wire is necessary. Heavy formvar is the best wire since the insulation is good to 500 F. If you only have bare wire, obtain kapton tape (ebay) and carefully wrap the wire with it to insulate it. This tape is good to 500 F as well.
Wrap the wire around a form, wood or metal, and do two turns side by side. Then carefully work the third turn up over the previous turn. This is somewhat tricky to do but is possible with some fooling around. Then put the last turn right next to it over the first turn. This should give you a more or less sqaure cross section coil of 4 turns. Bend the connections to give about a 25mm extension from the box. Solder copper screw lugs to the ends to connect it to the flasher terminals. If you are not going to change out coils, you could solder the coil directly to the terminals.


Do not connect the output of the bridge rectifier + terminal to the FET drain yet.
Do not connect the output tuned circuit to the amp FETs yet.
Power up the flasher and use a scope to observe the waveforms on the gates of each output FET. The waveform should be like that in the photo. Note that in that photo there is noticeable dead time on one transistion and marginal dead time on the other. This is OK but not optimum. Adjusting the dead time on the other edge would give a greater safety margin but as long as the FET gates are not both driven above the 3 volt turn on threshold at the same time there will be no problem.
If the two gates are driven positive at the same time, reverse the connections to the coupling transformer secondary.
Connect an external power supply to the drain of the power FET. Set the voltage to around 20 volts and use an ammeter in series to measure the current. (if the power supply has meters that is fine) Trigger the circuit and note the current. It must be well less than an amp.
Observe the output waveform of the amp. It should be a good solid square wave at the operating frequency and an amplitude equal to whatever you have the power supply set to. If this waveform is not correct, find out why before continuing!
Now, disconnect the power supply and connect the + terminal of the bridge through an ammeter to the drain of the output FET. This will allow reading the current while the circuit is delivering power.
Trigger the flasher and observe the current. Note that this is no load. The current should be well less than an amp. The waveform on the output should now be a square wave of close to 300 volts peak to peak. Note that the peak voltage of the wave will be following the envelope of the mains voltage.
If this is all working fine, connect the coupling capacitors to the output circuit.
Leave the amp meter in the circuit and trigger the flasher. Adjust the tuning potentiometer for brightest light from the tuning indicator lamp. The current should be between 2 and 3 amps at the point of resonance. Do not hold the button on for more than a few seconds at a time. The coil will start to smoke from the tremendous heat dissipated in it. The output power at this time is between 400 and 600 watts, which is all dumped into the resonating capacitors and the work coil.
If the current draw is over 3 amps, reduce the value of the coupling capacitor until the current is below 3 amps. The unit is now ready to close up and use! Connect the bridge output directly to the FET drain and close up the case, making absolutely sure that no part of the electronics can touch the outer case.
Construction notes and photos are on the next page.