Beaglebone Black Hardware bug report - Crystal Y2 load capacitors
Gerald:
I follow a “time-nuts” email reflector where there is a lot of discussion
of precision time related things, and there is a group using the BBB as ntp
time servers.
There were some comments that a small portion of the BBB units were more than
500 ppm off frequency and ntp would not attempt to correct them.
In reviewing the schematic for the BBB, I note that the 24 MHz CPU clock
crystal (Y2) on the bill of material is specified for 18 pF load capacitance.
I also note that the value of C25 and C26 is 18 pF to ground.
If you examine the circuit, with respect to the load on the crystal, that is,
the end to end capacitance across the crystal, the two load capacitors are
effectively in series. So to get 18 pF load across the crystal, each of the
capacitors should be 36 pF, ignoring any circuit parasitics. If you assume
that the parasitic capacitance of each of the I.C. leads is about 7 pF each,
then moving down to the next smaller standard capacitance value of 33 pF for
C25 and C26 would be just about right. Or, you could use the same value caps
(30 pF) you are using on the Ethernet 25 MHz crystal (Y3, C141, C142) if you
do not want to introduce a new value and part number into the manufacturing
system.
You may or may not want to consider making a running change on the capacitor
values the next time you have a hardware update on the BBB. I would ask that
you review the circuits in the X-15, so that you are getting the accuracy
you are paying for in the crystals.
Thanks,
— Graham