Hello,

In a full duplex RF transceiver, what are the potential problems of a receive IF strip at a frequency that's equal to the separation between the transmit and receive frequency? 

For instance, for an 800MHz narrowband radio transceiver, assume the receive channel is 806 MHz, the receive IF is 45MHz, the receive LO is 761Mhz, and the transmit signal is directly generated at 851MHz (no transmit IF). The receiver IF is sampled by an ADC. Is there a thumbnail for "general project risk" in terms of the transmit signal degrading receiver performance with a 45MHz IF as compared to a 70MHz or 90MHz IF?

It sounds risky, but I'd have to do some math.  The general way that you do this is to make a chart of the frequency of every mixing product of every signal that you know is an issue -- so, every RF signal that may come in (including the transmit RF) and all of their harmonics, then all the harmonics of the LO.

Anywhere that there's a combination that falls within, or even close to, your IF, that's a potential problem.

Like this document, Table 1.

I will say, if you can show yourself that it isn't a problem, any RF engineer in the room is going to be instinctively flinching at the apparent harmonic relationship.  Especially if there's an oscillator anywhere in the system at 45MHz -- that may combine with the transmit signal in the front-end, and be "seen" as a valid receive signal.

As Tim said, check where all the intermods land wrt harmonics, etc., but the architecture described might not be too bad.  Since the Tx is direct conversion, as long as it is sufficiently filtered that there are no out-of-band images that violated transmit mask, the only worry would be in-band images.   If the Tx and Rx are sufficiently electrically isolated from each other, then in-band stuff shouldn't be too much of a worry, either.  If the splitter/combiner/diplexer is reasonably linear over the region of interest that should help prevent issues there.

So the bigger concern is just whether the Rx is susceptible to issues from the Tx.  The 45 MHz Rx IF LO might mix the Tx signal in-band, which would be the big problem, if the splitter/combiner/diplexer can't sufficiently isolate the two, which might be a big ask.

If for some reason the application doesn't need a lot of sensitivity or SINR, it might be manageable, but otherwise that'd be the biggest concern, I'd think.  

On the splitter/combiner/diplexer.  My basic rule of thumb is that 20dB is no problem, 40dB is about as much as you want to expect from a commercial circuit, 60dB is doable for military/scientific stuff, and if someone's demanding 80dB or more, then you should start singing "Lab Queen, she only works when she's clean, it was a working phenom and then the furnace kicked on" under your breath.