[time-nuts] GPS module recommendation for Pi timing

[jimlux]

On 3/14/20 7:54 AM, Brian Lloyd wrote:
> On Fri, Mar 13, 2020 at 8:49 PM Bob kb8tq <kb8tq at n1k.org> wrote:
> 
>> Hi
>>
>> If you always operate in fixed locations, your chances of seeing a slip
>> are not that high. The original post question and reference here were to
>> timing in a *mobile* situation.
>>
> 
> Thank you all for your inputs. Here is a little more data, prompted by
> comments and questions. (Some of this I wasn't thinking about at first.)
> 
> I realized that, while I expect the majority of units to be stationary with
> the ability to properly position an antenna, there is a chance that one or
> more of the units may be mobile. Regardless of mobility, the purpose will
> be time and frequency reference, not position. Does that change my choice
> of device?
> 
> Some of the units may have access to the Internet and I could run NTP on
> them but the majority will be using GPS as the single stand-alone source of
> UTC. The goal is for all the units to be sync'd to UTC as they need to
> perform repetitive functions concurrently.

NTP provides advantages, as mentioned by Tim Shoppa, in that it also 
does a "smoothing" of the time from your GPS.



> 
> The starting point for me was going to be the Ublox LEA-M8T because it is
> current generation and Ublox offers a common footprint so that, if
> production runs into the future, there will be newer modules that can be
> dropped onto the same board. I want the second output to use as a 10MHz
> reference. I have need for a 3e-7 accuracy 10MHz frequency reference in
> some cases and that looked like it would do what is needed without going
> full-on GPSDOCXO.

Do you actually need 10MHz "control" or do you need "knowledge"?
If you count your clock with the 1pps, you get knowledge to 1E-7 (for 
10MHz), but not necessarily control.
If you don't need good close in phase noise, or smooth ADEV statistics 
you could probably use the count to drive a simple loop to steer a VCXO 
every 1 second or every 10 seconds or something.

Conceiveably, one could also do something like continuously reprogram a 
SiLabs part


> 
> So is my first choice of the Ublox LEA-M8T a good one or should I be
> looking at other units. While I don't want to spend $50(us) on a receiver,
> if it does the best job, it will fit the budget and that extra timing
> output I can use to generate 10MHz is quite attractive.

Of course, cheap 10MHz VCXO (not VCOCXO) might get you there also.
Or use the Abracon
https://www.mouser.com/datasheet/2/3/ASG-C-9844.pdf
which has both a voltage and a programming input

$8 in single quantity from Mouser.

>