I've been noticing something weird lately... I'm keeping an eye on my IT-related power usage (read: put a clamp on its circuit), and I notice that my base load (night) graph sometimes gets it wrong.
Last night between midnight and 7 in the morning, my hourly readings ripple between 98 and 100 watts, yet my night reading for that circuit says 116 watts.
Surely that reading should never be above 100 watts? Calculation issue or some other hidden cause?
Bad powerfactor?
All readings come from the Flukso, from the same clamp. The only thing that happens (IIUC) is that the Flukso server takes some average of the readings during the night.
I could understand powerfactor issues when I would be comparing with other powermeters. But I'm not.
A bad power-factor comes by the usage of capacitors and (or) inductors in the power-supply of many electrical equipments.
The usage of a watt-meter with pulse output eliminate those bad readings provoking by bad power factors, especially when running idle and measured by a current clamp.
A current clamp is only accurate with a power-factor of 1 and/or resistive loads like electrical heating and incandescent light bulbs.
Now in my case, i put last week a watt-meter with pulse output in my fuse box and then the load during the night is dropping by almost half !
Edit: The total consumption is not half during night, but only the readings on the Flukso website of course.
AGAIN: This has NOTHING to do with powerfactor.
I'm comparing the FLUKSO hourly graph during the complete night with the night graph which is supposed to give an idea of base load.
The hourly graph is an almost flat line between 98 and 100 watt, the nightly graph says the average was 116 watt. All from one and the same measuring source.
Because I was curious I just had a look at the sources (help yourself ;-)): The night consumption is calculated as an average consumption between 2AM local time plus 3 hrs (=5AM) by the RRDTool based on quarter hour data (900 sec average); refer to \fls\drupal\modules\logger\logger.rrd.inc -> function logger_rrd_night_create($meter) and \fls\drupal\modules\logger\xmlrps.inc -> function _logger_measurement_add($logs)
See https://github.com/flukso/fls/tree/develop/drupal/modules/logger
Yes I saw.... Is there a way to download my data so I could for example run the flukso web code on my own server to debug this? Not that I don't already have tons of things to do ;)
I'm going to check my graphs for all nights I can see, maybe I can spot a pattern. For now, all nights I checked are wrong.
Currently there seems no method to download data from the database running at flukso.net (I bet this is owed to the immense traffic such will cause); you may write to an own database (e.g. by https://github.com/gebhardm/energyhacks/tree/master/RaspberryPi/FLMtoDB - see the how_to that should also work on any other Linux) and look for patterns (second base should be convenient); via REST you may also get some historic data, but as far as I understood Bart, this is aggregated to also keep the volume in an operable boundary...
You may, of course, operate the complete server side of flukso also on your premise; but that will involve some reconfiguration; check the forum (search term site:flukso.net own server) for options...