I have a question regarding the AD8339. I'm not sure this is the right place, because this part works at radio frequencies but is intended at ultrasound imaging applications.
The 8 mixers are made of Gilbert cells and the LO signal is internally generated (from the 4LO inputs) by ÷4 frequency dividers. So I assume that every mixer merely makes a ×(+1) and a ×(-1) alternatively. This causes the mixers not to reject the input signals whose frequencies are close to the odd harmonics of the LO frequencies.
For example, let's say 4LO = 40 MHz and LO = 10 MHz ; a 11 MHz input signal would yield a 1 MHz output, but a 31 MHz signal at the input would also yield a 1 MHz output signal. Also, the unavoidable third harmonic of my 11 MHz, hence 33 MHz, would result into a 33 - 30 = 3 MHz output.
I am right, and how can I quantify this ? Shall I suppose the mixing gain at 3×LO is 1/3 of the "main" (LO fundamental) mixing gain (and 5×LO is 1/5, and so on) ?
Needless to say, I would filter those unwanted components out of the input signal if I could. But due to bandwidth requirements, the third harmonic proves hard to filter out.
Thanks for your hints :-)