[time-nuts] GPSDO for my rubber duckie
msokolov at ivan.Harhan.ORG
Tue Aug 2 16:12:34 UTC 2011
Thank you all for your recommendations. It looks like I will go with a
ThunderBolt: I have found Trimble's manual online, read all of it, and
it looks like the unit will do exactly what I am after.
I will feed all 3 outputs from the TBolt (10 MHz, 1 PPS, EIA-232) to a
custom timekeeper motherboard I'm going to build. My board will feature
a daughterboard seat for the HECPQ CPU module (a type of PowerPC, the
same kind I've decided to use for my otherwise totally unrelated
non-Ethernet WAN router work) and an FPGA for the timing functions.
The 10 MHz and 1 PPS inputs will go to the FPGA which will generate
millisecond interrupts to the PowerPC module per the logic which I have
described in my previous posts. The way the TBolt generates its 10 MHz
and 1 PPS signals (according to Trimble's manual at least) is perfect
for my logic.
Why millisecond interrupts and not something else? Well, my firm desire
and intent is to run my own HECBSD on this thing, a stripped-down
embedded version of 4.3BSD-Quasijarus. The latter is my very own
operating system and I obviously know it very well inside out. That
includes the timekeeping code, and I know that feeding it timer
interrupts that are spaced 1 ms (SI) apart and which have a known
relation to MCAT PPS will make it easy for me to do what I want in terms
of HECBSD in-kernel timekeeping and UTR timescale generation.
I'm also going to write a formal spec for my UTR timescale; I will post
it both here and on leapsecs.
More information about the time-nuts