Cumulus Support

Can't find an explanination for this, but during heavy rain on the 3rd & 4th, (which are still in my graphs),
Both my GW1000 sites were extremely under reading in both the gauges & charts.

Initially thought it to e a combination of Strong wind & the waver in the 6m pole the Ecowitt is mounted on.

At one point after rollover, EW Vs Davis was 0.8mm Vs 1.2mm, and I got watched the Davis climb to 4.6mm in a short time while the EW didn't budge.

So started blaming hardware, the tipper, the pole, network etc....

Later came to realise though that data fed to Ecowitt's site was correct.
Both uploads from the HP2551 & the GW1000 identical.....

I know at one point I compared all data & figured the EW had barely registered 20% of the Davis's Rain reading.

But now it all seems magically fixed.
And I'm at a total loss.

If you check the figures @ now though, Month to date,
Davis:- 40.8mm
Ecowitt:- 36.6mm

Close enough to be considered correct considering the EW's pole mount & much smaller rain collector.
If you look at the charts, they pretty much agree too, but neither did earlier in the piece.

Can't be exactly sure where I was with updates at that point, would need to go back on logs to check that.

Currently the 3 sites for compare are:-

w2.inverellit.com GW1000 on rPi
s7.inverellit.com Same GW1000 being read from a Win10 install
weather.inverellit.com VP2 on a Win10 Laptop.

Only other thing in the back of my mind is something related to "Last Tip" in the GW1000 world.

It was taking two rain tips in an update interval before the last rain date value was being set.
Afaik it wasn't affecting the rainfall totals, but I'd have to check out the old code to be sure.

I was at the point of dismissing it as poor station performance at first (with disappoinment).

Started, stopped & check both MX's with no difference.
Noted that Temp & Wind gauges both locally & Web based were fine.

But seeing the EW uploads then dismissed that.

No way could I force a correction, it just happened at a point in time.

A quick browse of the diags over that period shows that RainStart & RainCounter look completely correct.