Note missing days, yet the values are there in the data.
An oddity is that in the CSV data of the cumulative kWh log there are several NULL values (corresponding to missing vals?), but in the meter_pulse_WhAcc data there are none. There is only a simple scaling value between them so where have the NULLs come from?
I suspect I am misunderstanding the feed processing somehow as the docs are scant at best, so it’s only what I have been able to find by searching the community and trying to work it out for myself.
If I set the graph type to fixed interval and 86400 (seconds in a day) I get the correct plot. No idea why that works and setting the type to “daily”doesn’t work. There is no manual I could find for the graphing module, so its just trial-and-error until the plots look plausible.
The Emonth is running the standard 58s update cycle so the feeds are set to 60 sec logging as recommended.
Any idea where those data gaps are coming from since they are not there in the Wh accumulator output? Or is it because the kWh log is effectively sampling again on a discontinuous sample stream and sometimes falling in the gap between?
What % of data is missing? you should be able to find this towards the bottom of the graph module, click on ‘show statistics’ and there’s a quality count there.
The usual reasons would be a mismatch between the feed interval and the interval of received data, the other would be missed radio packets…
Ah, I have deleted the feed and replaced with the (join) version so I can’t check the %.
The Wh accumulator log is 100% data, so no loss. I’d expect that though since it patches up gaps automatically. The raw pulse log data shows 88%.
I didn’t know about that statistics function. I had a look at some of my temperature feeds and some of them have quite a high proportion of missing data, so I need to look at EmonTh locations and maybe crank the RF power up in the worst ones.
