[time-nuts] 4046 experiment for gpsdo

[Jim Harman]

On Sat, Sep 26, 2015 at 9:54 AM, Magnus Danielson <
magnus at rubidium.dyndns.org> wrote:

> If you have an oven, it is wise to discard beat-note frequencies outside
> of some suitable range, so that it first needs to go within that range
> before any attempt to lock it is done, so that worst part of the heat-up
> frequency deviation and drift has ended.


Yes I have found this to be very helpful. Building on Lars' implementation,
my code currently does the following to speed up lockup from a cold start:

1) It sets the initial DAC value to the most recent 3-hour average value
from previous runs. This is stored in EEPROM so it is remembered through
power-down. On the very first run it uses half scale as a starting point.

2) It holds this DAC value until the frequency error is <= 3 counts at 5
MHz (600 ns/sec or 6x10^-7) and the TIC difference from the previous second
is <50 counts (50 ns/sec or 5x10^-8). The first check is what ensures it
does not try to lock on to the wrong cycle, which would be off by 5 counts
or more.

I usually leave the oscillator (a surplus C-MAC STP2322) powered on, so it
meets these criteria right away after startup. If the oscillator has been
powered off for 15 minutes, it meets these criteria in less than 10 minutes
after power-up. The spec for the oscillator is that



-- 

--Jim Harman