Input Power Over-Drive Switch
(Factory assembled PCB module)
- 50 Ohms in/out SMA ports
- Series and Shunt PIN- Diode Switch
- 1 MHZ - 2.5 GHZ
- -20.....+50 dBm RF power input - 100 Watts max
- 30 dB isolation if the input power exceeds the threshold
- 100 W / 50 Ohm load resistor to dissipate the input power (not included)
I have received some emails from friends asking if we have anything available for protecting expensive RF Power LDMOS transistors because they can easiy be destroyed with excessive input power.
Here is a wideband solution (LF to >2.5 GHZ)** which uses an inexpensive PIN-diode RF Switch. Actually, there are two PIN diodes, one shunt-connected across the output (which connects to the input of a Power Amplifier) and one connected in series with an external 50 Ohm load resistor. A wideband log-amp, the Analog Devices ADL5513, (super fast like a diode) monitors the input power continuously. Its output voltage, if below a threshold, allows for the input signal to be applied to the output connector directly without any loss except this caused by the PCB trace and the series capacitor. Both PIN diodes are not biased and there is no loss of power going to the output connector. But if above a threshold, a voltage comparator drives the MOSFET driver IC (U3) so to apply bias to the shunt PIN diode in order to place a short at the input of the expensive amplifier. At the same time, another PIN diode connects a terminating 50 Ohm resistor to the input signal. This lasts as long as the input signal is above the threshold. This arrangement provides a wideband isolation of approx. 30 dB between the input and output ports. When the input power gets below the threshold it is driven to the output port again.
The RF Power Overdrive Protection Switch module needs +5 VDC for its operation and a higher voltage (+12V or +24V, +35V max) for biasing the PIN diodes. Since we don't care for any distortion introduced by the PIN diodes to the signal because it will not pass to the power amplifier, if only 15-20 dB of attenuation is enough for your application, then +5 VDC can also applied to the high voltage input too. Maximum attenuation between the input and output ports is achieved with higher voltage. You can start measuring normal RF power of the signal driving the power amplifier to be protected. Then measure the dc output from the log amp as applied to pin 3 of the voltage comparator. This voltage should be 0.21...2.84 V depending of its power level and the setting of potentiometer P1. Then adjust P2, the threshold voltage to a higher value which you need the protection to be activated so to stop feeding the power amplifier. The module can also be used as a level controlled limiter too. Its action, i.e., disconnecting the input from the output, lasts only as long the input power level is above the threshold.
VNA Photo 1: The input to output insertion loss from 1 MHZ to 2.5 GHZ
** A Note about the "wideband solution" mentioned above:
My effort was to provide a wide-band solution from overshoot and spikes at the output of tranceivers or driver amplifiers before these reach the expensive power amplifier. This solution is not perfect. Specifically, when the input power is over the setpoint and the 2 diodes are biased, sure the power amplifier which follows the switch is protected. But the driver amplifier may "see" the short of the shunt diode and the 50 Ohm loading resistor and have a mismatch. This may not cause any problem since we are dealing with low power levels, but if I would like to improve the system, I would add a quarter wavelength long (lambda/4 x velocity factor) coaxial line between the first and the second diode. This will allow the power amplifier to get protected by the short and to deliver its signal to the load resistor and not "see" the short at the power amplifier's input. But with this solution, the design stops being "wideband". Please take this under your consideration. It is very easy to cut the PCB transmission line at a point and insert the lambda/4 line at the frequency you are working on. For low frequencies a pi-network (C-L-C where XC = XL = 50) can also be used. The quarter wavelength long tranmission line has a very useful feature. The one end being shorted by the shunt diode, causes an "open" to be seen by the driver amplifier's end.
Also, the driver amplifier will get DC when the switch operates, so please ensure the driver amp is AC coupled with a capacitor, or add a capacitor in series with J1 to block the DC from going into the driver's amplifier output.
Any comments are welcome! Experiment removing D1 or D2...change the topology or whatever.
If you think of a better idea please tell me first!