[time-nuts] CGSIC: FW: New NANU 2014090

Magnus Danielson magnus at rubidium.dyndns.org
Tue Dec 16 20:01:59 EST 2014


On 12/17/2014 01:46 AM, Jim Lux wrote:
> On 12/16/14, 4:29 PM, Magnus Danielson wrote:
>> Jim, Bob,
>> There is a fair amount of work along the full path.
>> LNA with some L2 and L5 filters is pretty easy.
>> I think you still want to have a correlator baseband processing in say
>> an FPGA.
> well, yes.. but I don't know if there's any handy open source free cores
> out there for that.
>   I do know <grin> of an implementation that does the acq and track in a
> pair of Xilinx 2-3000 parts and does the nav solution in a SPARC V8, but
> it's not open source and it's definitely export controlled.

For that application it needs to break both the limits at the same time, 
for sure.

> Seems that what's out there is mostly "record bits" and "postprocess in
> C++ or Matlab" Several textbooks even include it.

It's good for many purposes as you get yourself up to speed.

>> There is naturally stuff to be done on the L2C and L5 modulated signals,
>> but it goes in a relatively slow paze so that even modest processors can
>> keep up with it.
> Indeed.. we do 24 channels (where channel is one PRN at one frequency)
> with a 3 frequency solution without making a 66 MHz LEON2 based SPARC
> sweat too much.
> That's why it would be intriguing if someone had the FPGA stuff out there.

Indeed. I did a GPS correlator core once, but it had issues to fit the 
FPGA I had at hand at the time. The software receiver I did was not at 
all doing real-time, but it did do many of the crucial points and was a 
nice exercise.

> It would still be an expensive project, I suspect.  Either you'd have a
> few $50-100 boards that would need interfacing and a lot of time, or a
> $1000 board with less time.
> One hopes that in a few years, multifrequency stuff will become available.

Indeed. Maybe a complete implementation just needs to hit the web..


More information about the time-nuts mailing list