"RST Engineering" wrote in message
...
Several reasons come to mind. The aforementioned numb unused input.
They
found some interesting motorboating (low frequency) proclivities and
started throwing capacitors at it until they found the right value. The
applications engineer was two weeks out of school on his first real RF
circuit.

Something I've found myself -- and several considerably experienced RF
designers have verified -- is that RF coupling and bypass capacitors are
often nowhere near as critical as you might think they are. On a real
circuit board, by the time you start getting close to the SRF of even a
100nF capacitor, there's usually enough parasitic coupling that what the
capacitor is doing just doesn't matter that much. I've used 100nF coupling
caps in paths all the way up to 3GHz, and while you *can* see a difference
in S21 compared to using, e.g., 10pF, it's really, really small... like,
under 1/4dB, and thus the overall response is usually swamped by active
devices, loads, etc.

Thus I'm not at all surprised the OP saw absolutely no different when he
reduced the 10nF caps to 200pF.

That being said, I wouldn't throw a larger-than-necessary cap into an RF
path without good reason. My main point is that the problem of providing
wideband coupling is often nowhere near as difficult as it first appears.
On the other hand, this implies the well-known corollary that isolation is
often a lot *harder* to achieve than we'd like.

---Joel