nRF51 - time of flight / distance

[travisf]

I'm interested in creating a time of flight algorithm and am trying to calculate what to expect for accuracy.

Is it possible to get an IRQ from when the sync arrives at the RX modem? If so, how deterministic is the timing of the RX modem? Can this IRQ trigger a timer? Can this timer save information using DMA?

Thanks!

[Janek Mann]

As far as I can see from the reference manual the earliest EVENT you could set up would be for an address match (in BLE address format). And afaik you should be able to capture the timing of that with the PPI and the TIMER peripherals. However, consider that even if you were able to resolve down to a single on-air symbol (unlikely) at 2MHz (500ns), your distance resolution at the speed of light would be ~150m, best case!

(Note that radio TOF distance measurement for small distances is only really done with "spread spectrum" transceivers which use individual pulses which can be timed to sub-symbol resolution, with considerable in-silicon effort)

[travisf]

1-3m would be nice. I ran a field test where I measured the time between transmitting a signal to a radio and then RX a packet back using TIMER0 at 16MHz. I found that even if I average 20,000 times, I wasn't getting very good results. I could reasonably distinguish between two points 24 feet apart, but at that averaging amount it's not practical. I suspect the distribution of times is not uniform, so I don't think this approach would work. Theoretically a 16MHz clock should get 10-20m of accuracy, and I was hoping noise (dithering) would drop that to 3-5m with accuracy, but I suspect the weak link is in the radio itself.