load power 4w

Hi,
I designed a 4W power amplifier 300Mhz-350MHz with an LDMOS MRF1513.
The transistor blows off when I disconnect the load and give an input signal.
My application requires the amplifier to withstand all posible faults at the output (short, disconnected load etc...)

Can someone suggest ways of protecting the amplifier ferom those situations?

My application is a mobile radio so space hungry solutions are not good.

ps. Does anyone have a spice model or s-parameters for MRF1513 or MRF1518 transistors?

The only thing that comes to mind is a real estate hog so it might not be
a good solution.

Use a 5W-10W Isolator / Circulator.

Hi,
thanks for the reply, but as I said my application is a mobile radio and this protection cannot be higher than 5mm and can take max 1inch x 1inch space on an FR4 PCB.
I have 2 weeks to do it.

You can protect it in a couple of ways:
1) output isolator and 4 Watt load (probably too big at 300 MHz)
2) balanced amp with two 2W stages in parallel, and 4 watt load (the hybrid combiner is probably too big or lossy, but worth a look in case someone makes a lumped element one)
3) a DC power supply crowbar circuit

I recommend the later approach. I would set up a minicircuits ADC-26-52+ directional coupler between the antenna and the 4W output device. I would set it up so that it measures power reflected from the antenna. On the coupled port, I would have some sort of shottky diode detector to turn the reflected power sample into a DC voltage. I would send that detected voltage into an op-amp, and compare the detected voltage to a trimable "threshold" voltage. If the voltage exceeds the threshold, the op-amp (or comparator) would throttle back the Drain current (or maybe just VDS) to a level that CAN survive a long term output open circuit, protecting the device. You probably have to design in some hysteresis in the op amp circuit to keep it from oscillating.

Rich

The reason that the transistor is burning up so quickly is because you are pushing to get 4W from MRF1513 which is specified for 3W output power.
Try to use MRF1518 for a 4W application (using just a little bit lower Vcc than 12.5V, and also lower input power) and you will see that the transistor will survive.

Output matching network and the PCB design are very important to get good ruggedness performances.

A fourth way would be to just use something like a 10 watt device set up for 4 W output. That way, when the reflected power hits, it can still handle it.

Hi,
I did the test with 1W output power and the transistor still burns up.
MRF1513 in theory can handle a lot more power.
The transistor seems to be very sensitive to mismatch as it burns so quickly that the overcurrent protection in the power supply circuit does not even react when the transistor burns up.
My test equipment is very limited but the amplifier seem to work very well when a proper load is connected.

If I just disconnect the load at no input signal (50ohm at the input ) the power supply current goes up to almost the value at full output power but the transistor survives.

Could the problem be related to the bias network?

We forgot to ask you the all important question...when you open circuit the output, does the transistor start to oscillate? If it is oscillating, it will put out maximum power, irregardless of what output power you thought you were measuring into a power meter attached to the output. You could tell by a big jump in DC power just a moment before it self destructs.

Keeping it from oscillating is easier! All you have to do is figure out what output impedance it can not oscillate with, and load it with that. For instance maybe by loading the output with 300 +j50 ohms makes it unconditionally stable so you put a shunt resistor the right distance from the output, and it never oscillates again.

Hi,
Without the input signal and with disconnected load I see an increase in the supply current but it is less then during normal operation and the transistor survives a long time. I cannot be sure whether it oscillates but it is not the problem until I give the input signal. As soon as I turn the signal on , the transistor burns very quickly.
I was wonder what is causing the transistor to fail?
It is not the power, because it burns in microseconds and all transistors have the gate shorted to the source.

Added after 19 minutes:

would it be possible to add protective diodes between the gate and drain/source?
how to do that? what diodes?
would it help?

Use MRF1518 supplied at 10V and you will see that the transistor will survive whatever load you place.
Bias is important for ruggedness. Linear power classes behave better under poor load conditions.