Forgive my ignorance, but how do you achieve 10CM accuracy using a 38.4GHZ clock on the DW1000? If light travels 29,979,245,800 cm/s, and one clock cycle at 38.4GHZ is 26nS, wouldn’t that be about 800CM? In order to get 10CM of accuracy, you would need a clock of about 3GHZ. Or is my math wrong? Please enlighten me. Thank you.
Hello Neil,
as stated in the DW3000 manual in chapter 2.4.3:
A 63.8976 GHz sampling clock is associated with ranging for the IEEE802.15.4 standard [1], where a 15.65
picosecond time period is referred to, it is an approximation to the period of this clock.
This is presumably done by multiplying the system clock frequency of 499.2MHz (see chapter 8.2.2.16 and the description of “RX_STAMP”).
Kind regards
Fhilb
Hi @NeilGaede ,
In addition to @Fhilb 's comment, there is a Fourier relationship between frequency and time. So as you increase the bandwidth of your signal, you can get narrow pulses in time and hence more accurate timestamp information. Therefore, when you translate the time stamps to distance measurement, we can achieve 10 cm accuracy.
Kind regards,
Emre
A large bandwidth isn’t required for accurate signal timing.
GPS uses a relatively low bandwidth signal (L1 C/A is around 9 MHz wide, the widest signals are around 30 MHz) but can achieve distance accuracies of a couple of cm.
The timing accuracy is based on the high sample rate and the leading edge detection algorithm. It’s detecting the start of the pulse. How long the pulse lasts for isn’t that important, it’s only the start of it that is used for timing measurement.
While the high bandwidth means you can have a very short duration pulse of radio energy this doesn’t translate directly to timing accuracy. What short radio pulses gives you is very good reflection immunity, something highly beneficial for indoor ranging. If you have a short duration pulse then you must also have a large bandwidth, that’s just how the maths works out.
So a wide bandwidth doesn’t help position accuracy. A short pulse helps indoor positioning and requires a large bandwidth. - Edit - as @Emre_Ozbas_Qorvo indicates you want to signal to start rapidly, going from 0 to maximum amplitude as fast as possible makes it easier to find the exact start of the signal. That feature is implied by the fact that the pulse is short but is important enough to explicitly state. It is this rapid change in signal amplitude, turning on or off quickly, that requires a large bandwidth.
If you set things up correctly you can actually get noise levels of around a 3 cm standard deviation and a 2 cm error on the mean range.
1 Like
Thanks @AndyA ,
The fast rise time of the pulse enables accurate positioning and as well as resilience to multipath signals and noise.
@NeilGaede
For more information, you can also check this e-book.