Epic Win offers his profound advice on power combiners:
Push the limit, every time! Much inspirational. No pain, no pain!
New for May 2011. We broke this topic off from our power combining page, to underscore its importance. Note that on this page, we take the term "solid state power amplifier" , or SSPA, to mean a network that combines two or more solid state power amplifiers, typically MMICs. Yes, that is a stretch of logic, but it seems to be the industry standard.
There seems to be a big misconception in the field about combiners and the "magic" that they provide. For example, it is true that when you combine two identical amplifiers with ugly reflection coefficients using quadrature couplers, the reflection coefficient of the combined amplifier (SSPA) can be ideally zero (perfect match).
So what happens when you get reflections back into the SSPA? This is often the case in an steerable antenna array (phased array) when the beam is scanned off from boresight... your SSPA can be looking into a 3:1 VSWR at 45 degree scan. Yikes!
Here is where the general misconception is. If you ask working engineers, many will tell you "the reflected power goes back into the isolation loads".
Epic fail here, comrades. Perhaps this is because we have moved to a "smart buyer" approach to microwaves, where the systems guy might think a Lange coupler is something he encountered once in Home Depot's plumbing section when he was looking to repair a leaky pipe. We're not just talking about the Sarah Palins, the Sean Hannitys and other cretins, we've seen Engineering Fellows make this mistake. Why would an antenna guy ever need to know anything about load pull? Indeed...
Couplers are passive, reciprocal networks. If you don't understand that statement, go here and study basic network theory. This means that transmission coefficient from terminal i to terminal j is exactly the same as from terminal j to terminal i. If your coupler has 0.5 dB loss from amplifier to output terminal, it has 0.5 dB loss going the other way. Thus any reflections at the antenna find their way back to the amplifiers, not the isolation loads.
The "loss" due to this reflection is not "mismatch loss" and many engineers have mistaken. The problem is load pull. Your amplifier(s) were designed to see fifty ohms, if they see 100 ohms, power will drop far more dramatically than a simple 2:1 mismatch would suggest (go to our VSWR calculator so verify that 2:1 mismatch equates to 0.51 dB). You might lose 2 dB or more looking in to a slight mismatch.
There are perhaps two things you can do about this. It turns out, quadrature combined amplifiers can be less sensitive to load pull. But this could be a mere Band-Aid on a gaping wound, it doesn't solve the problem.
The real answer is to always configure a ferrite isolator on the output of an SSPA, to isolate your amplifiers from antenna mismatches. Why do you think they call it an isolator? Chances are you will lose 0.2 dB in the fifty ohm condition. But you will also lose 0.2 dB in 3:1 VSWR condition, instead of 2 dB or more.
Epic win!