Protection of the screen in tetrode RF power amplifiers.

All too often, I am looking up something on the internet and come across something else that ends up being far more interesting than the original search. This was how I came to discover the IXYS IXCP10M90S current regulator. Functionally, it is similar to a depletion mode FET. Current flows at a maximum with zero gate-source voltage, and is reduced with increasingly negative gate source voltage. What was intriguing was that the maximum voltage is 900V. With the addition of one resistor, a current limiter capable of delivering up to well over 50 mA could be configured.

Immediately I thought of my NCL-2000 and the need to keep the screen current below 40 mA or risk losing the expensive 8122’s. In typical linear SSB/AM operation the screens are at 400V and the amplifier is tuned up to about 15mA of screen current. The screen current has always been touted as a measure of the tetrode’s operation as it peaks at plate resonance and is reduced by output loading. It was obvious that the IXCP10M90S might be a solution to the problem.

A study of the application notes revealed that the simple current limiter circuit would have little impact on the screen supply voltage until the current limit level was reached, at which point the limiting takes place. As the threshold is passed, the voltage drop across the device rises sharply, thus dropping the voltage on the screens of the tetrodes, besides limiting the current. This seemed to be so much simpler than other screen protection circuits that employed opto coupled current detection with reset functions.

IXYS prefers to call what would normally be called the Drain and Source terminals, the Anode and Cathode. IXYS has a spice model and with LTSpice, I was able to experiment with the basic 400V screen supply of the NCL-2000 and also determine power dissipation in the part during current limiting action. From the datasheet, it looked like a 51 Ohm resistor would result in about 36mA as the current limit. Spice simulation agreed with the current limit. If the entire 400V were dropped across the device at 36mA there would be about 14.5W to dissipate.

The part is available in a TO-220 type package with the Anode lead connected to the mounting tab. Conventional circuit topology would place the limiter between the positive side of the screen supply and the tubes. This would require the use of a mica insulator as well as drilling a mounting hole. But, by placing the limiter in the negative lead of the supply would have the mounting tab at ground potential, with the possible re-use of a mounting screw already in service in the amplifier, and no mica insulator.

Ten devices were ordered from Digikey for $30.00. The current limiting was dry tested by soldering a 51 Ohm ¼ W resistor between the gate and cathode leads. A 9 Volt battery was used for quick testing and the current limited at 30mA. A second part was wired up with a 39 Ohm resistor and it limited at 38mA. The next step was to hook up a 600V supply, turned down to 400V with a variac and with a 5K 20W safety resistor, placed in series with the limiter circuit. With well over 200V on the part the circuit limited at 38.5mA.

Now to incorporate the limiter circuit into the NCL-2000: A 20.5 Ohm screen current meter shunt resistor is on the negative side of the screen supply with one side connected to ground. The meter is a multimeter used to measure plate voltage, screen current, grid current, or relative RF output. With a little bit of close quarter surgery, I was able to install the limiter circuit and use an existing ground mounting screw for one of the doorknob caps to affix the TO-220 package to the chassis. I took this opportunity to check the accuracy of the built in multimeter by using a 9V DC battery to drive the limiter into service an comparing the meter to a DVM. It read close enough.

A 1N4007 was reverse connected across the limiter to protect the circuit in the event of negative screen current, a phenomena I had heard of but never experienced. A 1000pF cap was also close wired across the circuit to prevent any stray RF from causing mischief.

I fired up the amplifier into a dummy load and followed all regular tune up procedures with no ill effects. The output was clean in both CW and SSB modes as viewed with a monitor scope. I then tried the limiter by deliberately reducing the loading of the amp. As I slowly reduced the load, the current rose up to and limited at about 38 or so mA on the meter.

D1 mounted on meter switch

Pictures of the rework are available. For amplifiers requiring more screen current, place duplicate circuits in parallel.