Options for receiver architecture
时间：04-05
整理：3721RD
点击：
I'm having to build what is essentially the functionality of a several synchronized VNAs operating in parallel (so synthesizing and sampling many RF channels at once), and I'm considering the practical tradeoffs of implementing direct downconversion vs low-IF downconversion for the receiver parts. The RF will be in the range of 50-200MHz, but with only around 100kHz of measurement/modulation bandwidth at most. I'm thinking of using one or more AD9959 (quad channel DDS) for direct sysnthesis of my RF outputs, and the LOs for the receivers. As I understand it, the basic tradeoffs are as follows:
Direct downconversion:
Need quadrature LOs
Need two mixers per channel
Baseband IF amplifiers, which bring in offset errors
Low IF downconversion:
One LO, but at offset frequency
One mixer per channel
Bandpass IF amplifiers, slightly more complicated but no offset errors
But one issue which I can't wrap my head around is how absolute phase measurement works in the Low-IF case. For example in a direct downconversion chain my LO and RF are always phase locked, and thus measuring the absolute phase of the baseband signal relative to the RF is trivial. But in the case of a Low-IF case, there is no permanent phase lock between the LO and RF, which are at different frequencies. So dependending on when exactly I start a measurement, I will measure different phase on my IF signal, leading to uncertainty in the phase of my received RF. The only way I can see to get the absolute phase of the received RF is to also sample the transmitted RF (via directional coupler), put it through the same downconversion chain, and measure the phase of that resulting IF. But then I need double the amount of effective receiver channels, which sort of kills the hardware advantages that Low-IF seemed to have.
So if there a practical way to avoid the baseband offset errors of direct conversion, while still getting absolute phase measurements without tons of additional hardware?
edit: also, if I were to go with low-IF downconversion, how should I select the sampling rate of my ADC? For example if I choose an FIF of 150kHz with a BW of 100kHz (so from 100-200kHz) then in theory don't I only need a sampling rate of 200Ksps instead of 400ksps, assuming my IF filters have good rolloff?
Direct downconversion:
Need quadrature LOs
Need two mixers per channel
Baseband IF amplifiers, which bring in offset errors
Low IF downconversion:
One LO, but at offset frequency
One mixer per channel
Bandpass IF amplifiers, slightly more complicated but no offset errors
But one issue which I can't wrap my head around is how absolute phase measurement works in the Low-IF case. For example in a direct downconversion chain my LO and RF are always phase locked, and thus measuring the absolute phase of the baseband signal relative to the RF is trivial. But in the case of a Low-IF case, there is no permanent phase lock between the LO and RF, which are at different frequencies. So dependending on when exactly I start a measurement, I will measure different phase on my IF signal, leading to uncertainty in the phase of my received RF. The only way I can see to get the absolute phase of the received RF is to also sample the transmitted RF (via directional coupler), put it through the same downconversion chain, and measure the phase of that resulting IF. But then I need double the amount of effective receiver channels, which sort of kills the hardware advantages that Low-IF seemed to have.
So if there a practical way to avoid the baseband offset errors of direct conversion, while still getting absolute phase measurements without tons of additional hardware?
edit: also, if I were to go with low-IF downconversion, how should I select the sampling rate of my ADC? For example if I choose an FIF of 150kHz with a BW of 100kHz (so from 100-200kHz) then in theory don't I only need a sampling rate of 200Ksps instead of 400ksps, assuming my IF filters have good rolloff?