Page 1 of 1

Rain error last day of year (9am rollover)

Posted: Sat 02 Jan 2016 4:09 pm
by Adrian Hudson
Hi Steve,

I can't see this mentioned in other posts...

Just before the yearly rollover - at 08:20 was 4.0 mm and then at 08:30 went to zero. This meant the day total was zero and this was reflected in the dayfile.

Nothing was untoward with the machine I was enjoying a lay-in so it wasn't me fiddling with it.

My rollover is 9am and I have a davis VP2.

Re: Rain error last day of year (9am rollover)

Posted: Sat 02 Jan 2016 5:13 pm
by steve
If you zip up the MXdiags folder and attach it, I'll take a look. Are the clocks on the console and on the MX platform correct?

Re: Rain error last day of year (9am rollover)

Posted: Sun 03 Jan 2016 5:56 pm
by Adrian Hudson
My initial statement was only partially true :-)

Whereas I was indeed, asleep at 08:20 - and therefore not fiddling with the PC, a script woke up at 08:25 and ended the program (nicely) for a couple of minutes so that WeatherLink could download from the VP2. It does this every 5 days.

The problem appears to be something to do with the fact the monthly rollover had happened on the VP2 console. In the logs, during the subsequent startup it talks about "finding raintoday from the logfile data\Jan16log.txt" - which wouldn't have existed yet - then about the raincounter being -4 (no idea why) and hence setting it to zero.

So, I guess this sort of thing could happen on the 1st day of any month - assuming it had rained. I will check back through the logs when I get a moment. It could explain why occasionally the VP2 monthly rain does not agree with my manual check gauge - I have always put it down to the VP2 "having a funny 5 minutes".

I've attached a few of the logs as requested.

Re: Rain error last day of year (9am rollover)

Posted: Sun 03 Jan 2016 6:28 pm
by steve
Adrian Hudson wrote:then about the raincounter being -4 (no idea why) and hence setting it to zero.
The Davis resets its annual rain figure to zero at midnight on the 31st December, so for 0900 starts Cumulus caters for this by setting the start of day counter to the negative value of today's total so far (so that subtracting that from zero gives the correct amount). This is fine unless it gets restarted between midnight and 0900 on the 1st Jan, because it uses a negative value to determine that it doesn't have a valid start of day rain counter for whatever reason, so uses zero.

The same thing happened in Cumulus 1; I'm not sure if I ever worked out a solution, if I did I'll do the same thing in MX. If not, I'll see if I can think of something. Implementing an 0900 start of day is quite tricky when the hardware fights against it...

Re: Rain error last day of year (9am rollover)

Posted: Mon 04 Jan 2016 10:50 am
by Adrian Hudson
I can imagine 9am end of day processing is tricky enough as it is so it might be best left well alone! I'll set my weatherlink update to be after end of day.