[time-nuts] Soekris NTP server
Poul-Henning Kamp
phk at phk.freebsd.dk
Sun Dec 31 02:56:15 EST 2006
In message <4596F3A7.3020904 at febo.com>, John Ackermann N8UR writes:
>Brendan Minish said the following on 12/30/2006 06:04 PM:
>> On the soekris board wiring the following instruction is given on the
>> febo site
>> http://www.febo.com/time-freq/ntp/soekris/index.html
>>
>> Run a wire from the junction of R61/R62 (near the backup
>> battery) to one of the GPIO pins (I used PIO0, which is on JP3,
>> pin 3) and from there to a TTL-compatible PPS signal such as the
>> output of the FatPPS.
>>
>> what is the purpose of this wire from r61/r62? on my board, a 4521 I
>> can't find this junction with any degree of certanty, we we pulling the
>> GPIO pin low here, or is something else going on?
>
>R61/R62 is the access point for the high-resolution timer in the CPU
>chip. Poul-Henning's extremely elegant hack has the PPS signal start
>the timer and at the same time bring the GPIO pin low. Since the GPIO
>is interrupt-driven, it will always have jitter based on the Hz value.
>So, when the interrupt hits the GPIO pin and it triggers, the code looks
>at the timer register to find out how long ago the line actually went
>low. That's how he gets nanosecond resolution with a millisecond
>interrupt rate.
>
>I have no idea where the equivalent point would be on the 4521 board, or
>if it even has the high-res timer available.
It does, and it is pretty much the exact same place on the PCB.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
More information about the time-nuts
mailing list