[time-nuts] Low SNR GPS reception and cheap LNAs

Magnus Danielson magnus at rubidium.dyndns.org
Sat Apr 26 15:36:14 UTC 2014



On 04/26/2014 02:59 PM, Attila Kinali wrote:
> On Sat, 26 Apr 2014 08:20:47 -0400
> Bob Camp <lists at rtty.us> wrote:
>
>> The ability of these receivers to handle noisy signals depends on a lot of
>> things. The good stuff seems to have a massive number of correlators. Going
>> from a 1.3 to a 0.3 db nf amp likely only helps you by 1 db. The low
>> correlateor GPS’s are / were 10 to 20 db less sensitive than the newer stuff
>
>
> Yes, the sub-optimal correlation system in GNSS-SDR seems to be one of
> the problems. But still, the LNA will help with two things: It will
> enable me to put the antenna in a better spot while compensating for
> the cable loss (i need about 20m of additional cable, which is about 6dB
> of loss) and a stronger signal at the bladeRF such that less gain is
> needed in the radio chip there.
>
> My current goal is to get a feeling how GNSS-SDR works. After i get
> an understanding for the code, i will try to improve on the correlators
> in order to improve reception.

You probably want to move from the Tong-search (traditional channel 
search) to FFT based cross-correlation search. Once the FFT has 
determined coarse phase and doppler frequency, you can setup the normal 
correlation channel with that and lock into the signal. You can also do 
a secondary FFT in order to get higher dopper frequency resolution once 
you have the phase. With both phase and doppler found, channel lockin 
becomes much quicker as you leave it very near the actual balance-points 
and that way you can allow the PLL bandwidth to be much smaller and thus 
suppress noise better.

I've toyed with FFT phase correlation and hand-over to the channel, and 
for my case it worked really well without too much of code.

Cheers,
Magnus



More information about the Time-nuts_lists.febo.com mailing list