I've made a microwave freq counter using ( among others ) four HMC492 chips. The counter seems to work properly, each of the 492 chips seems to divide the input frequency by exactly two, as expected. Scope observations of the chip output signals also clearly shows a waveform, while a signal is applied to the counter input.
The problem I see is this : If the signal is removed and the complementary outputs of each chip are examined, no significant voltage difference is observed, being less than 5 mV between the two outputs. ( one should be high, the other low ) The outputs are capacitively coupled to the inputs of the following chips, but even if I add external pulldown resistors of 4.7K to each output, the voltage difference between the two outputs is still insignificant.
I had intended to read these outputs and use the results as data bits in the overall frequency measurement of the signal, but it seems ( for reasons I cannot explain ) that these chips behave like divide-by-2 frequency counters, but they do NOT behave like regular bistable flip-flops that hold their last triggered state ( high or low ) when the input clock is removed.
This isn't a design / assembly problem, the circuit is identical to the one in the Hittite EVAL pcb, and the counters work properly ( dividing the frequencies by exactly two ) all the way from 1 to 13 GHz, which is the upper limit that I tested.
These chips behave almost like there is an internal capacitance, in series with each output pin of the chip. That is almost impossible to believe, but that's how they behave... anyone have any idea what is happening ?