[time-nuts] FE5680 GPS Disciplined Controller

EWKehren at aol.com EWKehren at aol.com
Fri Jun 20 09:52:39 EDT 2014


I have stayed quiet over the last six month on your effort but allow me to  
make a few comments and recommendations.
First getting 1 E-13 per second accuracy is not possible with a FE5680  
because of tuning steps of 6.8 E-13. Even 1 E-12 may not be possible.  We  are 
working on braking  1 E-12 but it is not easy since parameters like  
barometric pressure start making a difference and have to be compensated  for.
My question is : is your project an time nut project or an Aduino project.  
For time nuts a low cost good performing project that can be done with a 
vector  board and better than 1 E-9 is for instance the PLL version that J.R 
Miller has  done. As GPS receivers are getting better that circuit will give 
better  performance.. See attached.
Now if you want to do a digital filter you have to first recognize the GPS  
limitations and the source you want to control. Let me start with the easy 
part.  Exceeding 1 E-10 you will have to monitor GPS for a longer period of 
time and if  you choose analog control you will not be able to control your 
source wether Rb  or OCXO without a DAC or other circuits outside your 
Aduino. Some Rb's like the  FE 5680A have DDS control via RS232 and it will most 
likely be a solution.  Please take a hard look at that alternative.
On the input side of the filter you want to capture small changes and we  
have tried a few and went back to some thing we know that works reliable and  
without any adjustments and for good measure we also have a GPS input 
filter  that is user adjustable something we found very handy. Something you may 
want to  consider.
With all your effort and help from other time nuts it would be a shame if  
the outcome is just a low cost solution not making performance a key  
requirement. Looking at the last six month your approach may get better results  
because so many true experts have an opportunity to contribute and in the 
final  analysis create a product that expands the availability of better 
performance at  an affordable price to time nuts. That is the underlying goal of 
our  projects.
Think about it, I prefer not to get in to long discussions.
Bert Kehren
 
 
 
In a message dated 6/19/2014 2:13:11 P.M. Eastern Daylight Time,  
albertson.chris at gmail.com writes:

I've  been working in the same thing BUT I don't want anyone who builds it
to  need a PCB.  And I want the firmware to load over USB so there is  no
need to ship programmed chips or deal with external  programmers.   I think
I can get the cost below $20.   That said I doubt I'll get 1E-13
performance out of my  Rb.

My little Arduino based controller has been running now for a  couple months
and keeping a crystal in lock.  The board has a pins  left over for a serial
port that I'll hook up to the Rb.

The trick  to getting the cost down is NOT to do a custom PCB.  Take
advantage of  one of the uP development boards and then for under $5 you get
the USB  interface, D/A and A/D, serial ports, timers and quite a bit of
logic  all  1/3rd the size of a credit  card.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: schem.gif
Type: image/gif
Size: 54604 bytes
Desc: not available
URL: <http://www.febo.com/pipermail/time-nuts/attachments/20140620/5231dc72/attachment-0001.gif>


More information about the time-nuts mailing list