I'm using the AD8333 to perform mixing of ultrasound transducer drive signals down to baseband in order to perform high accuracy (~12-14 bit) phase and amplitude measurements. The drive signals are output from a switch-mode amplifier with some conditioning applied, but they still have a fair bit (~10%) of power in the harmonics. I'm using the AD8333 to make a measurement that can determine how much fundamental power is in my signal, ignoring the power in the harmonics.
As a first step towards this I tried presenting different RF waveforms at the RF inputs on a custom board with the AD8333 on it. When I present a sine wave at the RF inputs slightly detuned from one quarter of the the LO oscillator frequency I get nice I and Q sine waves at the output that oscillate at the detuning frequency, as below:
However when I present a square wave at the RF inputs, I get I and Q triangle waves at the output as seen below:
Based on the block diagram in the datasheet I had expected that the AD8333 would just take the product of the square wave input with the square wave LO and give me a square wave output. Is there an undocumented integration step happening somewhere in the chip?
Ultimately what I'd like to be able to do is read the baseband signal and be able to infer the harmonic content of my original RF signal and so determine how much power is in my fundamental. It seems promising that the baseband waveform changes a lot in response to changes to the RF waveform but I need to understand what the internal signal processing steps are to properly interpret it. Any help you can offer is greatly appreciated.