PRU clock accuracy

148 views
Skip to first unread message

Jonathan Ross

unread,
Jul 29, 2015, 6:08:05 PM7/29/15
to BeagleBoard
I'm having trouble figuring out the accuracy of the PRU 200mHz clock (in ppm). Datasheet seems unhelpful, am I missing something?
Thanks,
JR

Graham

unread,
Jul 29, 2015, 6:51:27 PM7/29/15
to BeagleBoard, jon...@nephology.org
If it is derived from the main CPU clock, then the accuracy will equal that of the main clock.

So you need to look at the specs on the main (24 MHz) crystal, not the CPU datasheet.

If you want, on any of the Jessie releases, systemd-timesyncd will measure it for you.

Boot your BBB, with access to the Internet via Ethernet.  Let it run for a while, so the time system converges

cat /var/log/syslog

the ppm number at the end of any line starting with "systemd-timesyncd" is the ppm correction the time system had to apply to the CPU clock to keep time with the reference NTP server.  So, if it says -42 ppm, then your CPU clock is 42 ppm high, as will be all the derived clocks.

You will find that the error is typically 30 to 40 ppm, which is greater than the 20 ppm spec on the crystal.

You can reduce the error to low single digit ppm by changing capacitors C25 and C26 to 24 pF.

If you need better than that, you will need to drive the main CPU clock from an external source.

--- Graham

==


On Wednesday, July 29, 2015 at 5:08:05 PM UTC-5, Jonathan Ross wrote:for
Reply all
Reply to author
Forward
0 new messages