[time-nuts] Re: Small NTP appliance

[Bob kb8tq]

Hi

If you are doing this on a micro, handling the network stack is *probably* not
something you will do from scratch. Often bringing in a network stack brings 
in an RTOS with it. You may well get some fancy time code “for free” with 
the RTOS. 

Once upon a time this RTOS / network stack stuff cost you money. That’s 
becoming less and less true as time marches on.  Free makes it a *lot* easier
to drop into a project :) 

Bob

> On Apr 25, 2021, at 3:20 PM, Hal Murray <hmurray at megapathdsl.net> wrote:
> 
> 
> folkert at vanheusden.com said:
>>> Most modern kernels have a side door used by ntpd to adjust the clock 
>>> frequency.  Typical values are few 10s of PPM and it's easy to measure down 
>>> 0.001 PPM or better.  The NTP world calls that drift.  If you have a PC or 
>>> Raspberry Pi running Linux or *BSD and ntpd you can find the drift.
> 
>> Hmmm, I believe Eamonn is going to use a microcontroller. They usually don't
>> run linux or something like that. 
> 
> You can still use the same trick.  You may have to rewrite the clock code.
> 
> If the clock code does something like
>  time = time +1
> where time is in ms and gets bumped from a 1000Hz interrupt, then you have to 
> change that "1" to be the carry out of a slot with enough precision.  If you 
> want to correct to 1 PPM, that's 1E6.  So the code becomes:
>  calibrate = 1000000;
>  partial + partial + calibrate;
>  while (partial > 1000000) {
>    time += 1;
>    partial -= 1000000;
>  }
> 
> Now fudge calibrate to make your clock run slower or faster.
> 
> The trick is that you have to scale things such that your equivalent of 
> partial has enough working bits to match the resolution that you want in your 
> correction.
> 
> A million is 20 bits so you are probably using 32 bit arithmetic.  You might 
> as well use a billion.
> 
> If you want time in ms and you are getting 1024 interrupts per second, the 
> code may already do something like the above.
> 
> -- 
> These are my opinions.  I hate spam.
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.