Click here to learn about designing for high peak power (new for September 2011!)
Click here to go to our main page on nonlinear devices
Click here to go to our page on PIN diodes
Click here to go to or page on LNAs
Click here to go to our page on receivers
Click here to find limiters on everything RF
New for September 2008! Limiters are used to protect weak components such as low noise amplifiers (LNAs) from stray signals. The power transfer characteristic (shown below) behaves somewhat like an amplifier with just a little loss rather than gain. Above some critical input power, the output power can behave quite strangely, often described as having a "kink" to it. From the plot you can read off the small signal loss (insertion loss), one-dB compression point, and the flat leakage.
The limiter is used most often to protect a low-noise amplifier in a receiver chain. LNAs are carefully designed for low noise figure, and use very small devices to accomplish this; small devices can't handle high input power. Two mechanisms can kill an LNA, heat from the RF input signal, or overstress due to the RF voltage that appears on the input transistor. Often, an LNA's performance can be changed slightly from high input power, without causing it to fail. Having a damaged LNA might be worse that having a destroyed LNA, you never know when it might fail.
In a radar the stray signal that most likely will damage an LNA comes from the transmitter, so it is a pulsed signal. Damage threshold of an LNA might change somewhat with duty factor of a pulsed signal, but we recommend that you perform LNA survival testing with CW signals because it is easier, and it will give you a worst case result.
One method of evaluating an LNA damage threshold is a stepped stress test, which we'll explain later. There doesn't seem to be any accepted standards for survival testing across the industry, but at least we'll all soon have one method to refer to.
Flat leakage
Flat leakage of a limiter refers to the CW output signal that bleeds through it under high input power conditions (see figure above). Guess what? The term "flat" is a misnomer, in real limiters the leakage will have a slope to it at any power level.
Spike leakage
Spike leakage refers to the very short part of a high-power pulse that blasts through a limiter, before it clamps down on the signal (there is some delay in turning on PIN diodes). The spike leakage is often referred to in units of energy, not power. For example, if he limiter allows a spike of 1 watt for 10 nanoseconds, the spike leakage would be 10 nano-Joules.
We said spike leakage, not Spike Leekage
Terminating or reflective limiters
Terminating limiters will attempt to provide an impedance match at any power level. This is a trickier design to pull off than a reflective design, and it won't have as good a response (more insertion loss, or higher flat leakage for example) Plus, you might have to think about sizing the load termination for all the power you might have to dissipate, which could be an issue if you are designing a limiter on a MMIC.
Besides "terminating", another correct adjective for a limiter that absorbs power is "absorptive". If you say "absorbtive" or "absorbative" your smarter coworkers will snicker and say mean things behind your back...
Limiter technology
Solid state limiters are most often comprised of PIN diodes, but Schottky diodes, FETs and other devices have been used. A shunt PIN diode acts like a small lumped capacitance to small signals, and matching networks or pairs of diodes separated by a quarter-wavelength can bring the network back to fifty ohms.
PIN diode limiters can be realize monolithically (as a MMIC), but the best performance comes from a chip-and wire limiter. In this case, silicon PIN diodes can be used (they perform better than GaAs believe it or not). One problem with assembling limiters is getting a wirebond connection to a diode where the mesa is list 1 mil in diameter... if your wirebond smooshes out beyond the mesa, you've just increased the capacitance of the diode.
More to come.