I've got the HMC631 eval board and I've noticed that it's performance is significantly different than what the datasheet specifies.
The issue is with respect to the I and Q voltages - per the datasheet and ref these should be ranging around 0.5 -> 2.5V and the null point should be around i=1.5,q=1.5. My test setup is the dev board, powered with a 8.3V psu, and I/Q voltages controlled by a 0-5V buffered 2 channel DAC. The rf input signal is a 1MHz BW QPSK waveform @ 2 GHz
If I do a computer controlled sweep of the i and q voltages, each over the range 0.5 to 2.5V, I find that the minimum output power point occurs at I=2.48V, Q=2.48V @ -46 dBm The maximum output power point occurs at i=1.03V, Q=1.14V at -35 dBm. Clearly there is an issue here as the observed null point is far off from the expected i=1.5,q=1.5 point. Additionally, the gain between min and max power is only 11 dB while the device specifies gain variation being around 40 dB.
Next I do the same sweep of both voltages but this time I sweep them over the range 1.5 to 3.5 V. Now I find that the min output power is at I=2.62V and Q=2.92V @ -74.5 dBm. The max output power point occurs at I=1.52V and Q=1.62V @ -35.58 dBm. This looks more reasonable as the gain between min and max is ~ 40 dB as the data sheet specified - however the null point is really far off from the ideal point of (1.5,1.5).
One final test was to sweep both i and q voltages over the entire 0 to 5V ranges to find the minimum gain point - again it found I=2.62 and Q=2.92 V @ -74.5 dBm. This further confirms to me that this is my devices null point.
TLDR: My measured null point is I=2.62 and Q=2.92V - the device works as expected using this as my null point but It seems very strange to me that the null point is so far off from the (1.5,1.5) ideal and that I'm forced to use i/q voltage ranges that are far out of the (0.5,2.5) nominal range spec'd in the datasheet.
add qpsk center freq
[edited by: mike64b at 8:07 PM (GMT -4) on 1 Oct 2018]