FMCW measuring system, do you mean the radar measuring the volocity of car?
That radar need PLL circuit. What is your VCO drift of frequency? Or frequency accuracy?
First of all, thanks for your attention. Let me explain: some peoples design things at work. And the other ones do some DIY just for fun. When we try to use working-dude-design proces then DIY becomes a nightmare. For example, when i tried PICs and Cortex DSPs, i did not used working-dude-process: "how many cycles we need to perform signal processing, how many milliwats our DSP consumes, how big is on-chip RAM and ROM". Instead i just bought randomly available DSP, some capacitors, put all this things on PCB, installed software, read some tutorials and it worked! I was happy, learned many things about why we need capacitors on VCC, why program does not compile, how to debug things. So actually there is no any "design parameters", buy what is cheaper and can get some real result.
Now about measuring FMCW system. It is FMCW level meter gauge, can measure distance inside a barrel like this:
Most barrels are smaller than 2.5 meters, so i guessed/invented empiric formula for the barrel: distance = max_barrel_height x 2 = 5 meters.
No known drift frequency, no accuracy, because I am not designing things. I make dummy DIY device, so accuracy can't be known before it ICs bought and PCB is implemented. Accuracy is fully depends on what component can be bought on ebay or extracted from old broken FMCW device / LNA head.
So here is second formula: Drift = Ebay *k1. Accuracy = Ebay*k2
Let's see what real design steps still here: we only know that 1) range is small, maybe smaller than 1 meter = we need small RF power
2) we have frequency modulation = we need small RF power, because modulated signal is easy to extract
3) things in a barrel does not move = we need small RF power, because need small SNR ratio, because we can accumulate signal for a long time. Actually signal can be below noise.
The third empiric formula is 0.1 x 0.2 x 0.3 / k3 = 0.006 * k3 - so we need ever smaller RF power. k3 - is coefficient, am I happy with result. Maybe RF power is too small, and there is no any signal at all. Then k3 becomes small and needed RF power increases. If I am happy with result then k3 becomes big, and RF power decrases.