Phase shifting?

I've been considering building my own ground penetrating radar unit rather than pay outrageous rental fees and mind boggling full prices and I keep coming back to the same problem: time of flight is haaaard. My most recent thoughts on how to go about collecting that data would be to use 2 clocks/crystals that are offset from each other by fractions of a percent(one would be straight from a rubidium standard, and the other from a frequency multiplier). One would be in charge of timing the emitter and the other for timing the receiver.
This is the part where I make tons of assumptions about things I know nothing about, so correct me if I'm wrong.
     The emitter is running at f1 and it emits a short pulse.
     The receiver is running at f2 and collects data as fast as it can until the reflections dissipate.
     The emitter pulses again
     The receiver, due to the f1:f2 phase shift, is offset by a tiny amount and again collects data as fast as it can.
This cycle repeats until the collective phase offsets brings it back around, building the image . My thinking is that this shift, given the time to cycle around, could provide data that is functionally in the range hundreds of GHz, provided the target doesn't move(which at milliseconds per dataset, is functionally true).

Could this work?

There is a reason they cost a lot of money. The equipment needed for the job is very precise and not at all cheap. Particularly the timing circuits needed to time the audio signal's bounce back since it happens in fractions of a second. The kinds of clocks that con only loose fractions of a second over the course of a year and have to be re calibrated often. Which also costs a good bit of money because of the equipment involved to do the calibration accurately. Your basic clock crystal isn't going to have that kind of accuracy.
The Ideanator (author)  mpilchfamily3 years ago
     I figured there was a reason because DIY setups are exceedingly rare and don't show documentation. As for clocking, I'd use a rubidiium standard (a cheap atomic clock in your pocket) or a crystal oven, neither of which require any particular tuning.
     My intention is to bypass the need for expnsive and specialized parts thaat can't be easily had by one such as myself by cheating. The system I'm trying to describe would take a series of readings, each one shifted in time by an offset of fractions of a percent, then stack those data together to form a much higher resolution image in the timescale at the expense of a longer measurement time.
     I did some time-of flight calculations for a system running at 1 GHz., it turns out that the best distance resolution you could hope for was with an error of about a foot, nowhere near what I need. Expanding on those, What I aim for is something in the neighborhood of 100GHz or more (necessitating a difference of 0.01% from the 10MHz rubidium standard and requiring 10000 run cycles with 1-10ms for a full loop).