[time-nuts] Some ES100 WWVB BPSK success

[Graham / KE9H]

I also received mine.
I have it hooked up to a Beaglebone Black (BBB) running Python for control
of the ES100.
Since the BBB is a 3.3V device, no issues with the Voltage of the
interfaces.
On the second floor of a house in Austin, TX, so far, it provides the time
every time I ask it (about a 2 minute process for the ES100.)

There is a deep drop in WWVB signal here every day at sunrise and sunset,
when the AM receivers will temporarily stop receiving.
I want to explore what happens to the ES100 during these signal nulls.

Learnings so far:
The chip does not accept a concatenated I2C read transaction.
(That is, write the number of the desired register, then read the register,
in a single transaction.)
You must do the register write, then a separate read transaction to get the
data.
This the first time I have encountered an I2C chip that behaves this way.
Upon re-reading the data sheet, this is documented.
No big deal, but if you are having problems reading the I2C registers,
split the transaction into two.

The 6 pin JST connector will not reliably accept a standard 0.1 inch plug.
I had to remove the JST connector and replace it with a 0.1 inch header, to
get a reliable connection with a 0.1 inch 6 pin socket.
The pin lengths in a 0.1 inch JST connector are shorter than in a standard
0.1 inch header.

I am having some kind of timing read race condition, such that if I trigger
the I2C reads from the fall of the IRQ line,
I get all zeros, which was the previous contents.  Apparently the BBB,
which is a 1 GHz processor, can get to
the I2C registers faster than the ES100 updates them after the ES100 drops
the IRQ line.
So, a small read delay is needed.

Next step is to time the IRQ against a known time source.
The data sheet says the IRQ time accuracy is +/- 100 ms.
So intended for setting the display of a human readable clock face, not
"time-nuttery" class performance.

The ES100 reliably reports time, on request, in a location and at a time of
day where an AM receiver
performance is spotty, so the benefits of the phase modulation plus error
detection-correction are
real.

--- Graham

==