[time-nuts] Brooks Shera
Bruce Griffiths
bruce.griffiths at xtra.co.nz
Mon Mar 25 19:49:23 UTC 2013
There are essentially 4 clock domains in the circuit
1) PPS
2) Divided down 10MHz (~300kHz)
3) 24 MHz
4) The microprocessor internal clock ( the micro probably has internal
synchronisers for at least some external inputs).
Depending on internal delays and jitter this may be regarded as
synchronous with the 24MHz signal.
The required fix is relatively simple:
1) use a 2 stage synchroniser (clocked at 24MHz) plus an additional FF
to generate a 1 clock period duration pulse on each leading edge of the
PPS input
2) use a 2 stage synchroniser (clocked at 24MHz) plus an additional FF
to generate a 1 clock period duration pulse on each leading edge of the
~300KHz input
The synchronised PPS and 300KHz signals are now in the 24MHz clock
domain and can be used to generate a synchronised gate (or enable)
signal with a duration equal to the time interval between the
synchronised PPS and the next synchronised 300KHz pulses for a ripple
(or synchronous) count chain.
Bruce
EWKehren at aol.com wrote:
> Bruce, would you mind being more specific and offer a solution.
> Thanks Bert Kehren
>
>
> In a message dated 3/25/2013 7:09:41 A.M. Eastern Daylight Time,
> bruce.griffiths at xtra.co.nz writes:
>
> The lack of synchronisers when crossing clock domains is a design flaw
> that should be corrected.
>
> Bruce
>
>
More information about the Time-nuts_lists.febo.com
mailing list