[time-nuts] timestamps on downconverted data streams

Bob Camp kb8tq at n1k.org
Sun Dec 20 16:34:03 EST 2015


HI

One simple way to look at it is:

The conversion process is a filter. It has an impulse response and a group delay. If it’s a digital downconversion 
process, those things are very well known (no analog drift …). You can, if needed, put in correction filters to
take care of any messy phase artifacts in the conversion process. All-pass FIR’s are often used to do this after
CIC conversion. The net result is a well defined (flat) time delay.

Once you have all that done, put an impulse into the input and look at the peak in the output. Net result is your
time offset due to processing. 

Bob

> On Dec 20, 2015, at 9:40 AM, Jim Lux <jimlux at earthlink.net> wrote:
> 
> Here's an interesting problem.
> 
> You have a fast sampler that is collecting samples off-the-air (e.g. the end of LORAN) with a fairly wide bandwidth: say 10 Megasamples per second.
> 
> Those samples get post processed in a digital downconverter (not necessarily in real time) to a narrower band representation at a lower sample rate.
> 
> You know when the input samples were acquired: e.g. you've got a good oscillator, and a reliable sync pulse.  For instance, your handy GPSDO (or the ensemble of H2 masers in your garage) might give you a 1pps tick good to, say, 20 ns, so you know when your 10 MHz samples were taken (to 20ns)
> 
> Is there a consistent (and standardized) way to calculate and report the time of the output samples.
> 
> Each output sample is composed of information from multiple input samples.
> 
> One could test the system by digitizing a signal with known timing (e.g. a 1 MHz sine wave, where the zero crossing is "on the second") and then look for the zero crossing in the downconverted output.  Depending on the filtering in the downconverter, there's some time vs frequency characteristic that could be used to back out any deltas for other frequencies.
> 
> So you could report the time of the low rate output samples in terms of the time of the input sample, at least for the 'center frequency' of the downconverter.
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.




More information about the time-nuts mailing list