Weather Notes:

Weather Notes:

Notes:

• The outdoor temperature sensor is located on a patio on the East side of an apartment building. It's shielded from direct sunlight but we still tend to see a temperature drop on bright sunny days when the sun goes behind the overhang of the roof. I suspect this suggests we might get better ambient air temperature readings on sunny days if the sensor were better shielded.
• The brown trace on the second graph on the weather page labeled ET-DP is the difference of the dew point from the external temperature. During steep changes in ET and DP it became difficult to tell graphically just how close together these actually were, so this trace was started. We tend to just start seeing fog or frost when this trace gets around 2 C or lower. Around 1.5 C we tend to observe much more pronounced fog and/or frost. I'm not sure that these effects being observed a couple degrees above a difference of 0 is normal, or an artifact of these instruments and/or their setup.
• Our wind speed setup provides some very noisy raw data. Worse, its location a couple of feet off the roof of an apt complex on a bit of a hillside doesn't help it reflect actual winds in this area of Seattle much at all. The noisy data problem was largely dealt with by filtering out many common bogus values I suspect are caused by 60Hz power inducing noise on the instrument's long, unshielded, Cat 3 cable. However the data still looks pretty noisy even after this filtering as the wind speed is rarely ever consistent from one sample to the next. So we graph a 16 point running average of the wind speed instead of the actual raw data, and this makes things appear much more consistent and clean.
• Wind direction also tends to reflect the somewhat shielded nature of this instrument's location. We do however tend to see directional trends that seem to indicate the instrument is doing more than just whirling around in local eddies.
• Pressure seems to correlate very well to weather.com's readings, and generally seems to be a well behaved system. Spikes we used to see are now being filtered out quite successfully.

Notes:

• To construct the sensor "units" I have just been crimping the one wire temperature sensors directly into RJ11 connectors. Given the nature of the 1-wire bus design, this allows me to connect many of them together in a random haphazard fashion with standard telephone cables, splitters and couplers while still expecting the whole thing to just work.
• This system has proved surprisingly robust, and produces far fewer data anomalies than the weather setup.
• I am somewhat unhappy with the software available for working with these things. For what I am doing I'm pretty sure I could use a much simpler program utilizing orders of magnitude less code. Work on this is underway, I just need to work out a few kinks.

Notes:

• Activity seems to correlate to rain more than anything else at the moment (and only weakly).

Notes:

• In order to not drive the analog inputs on my uC negative (or suffer voltage drop from a diode), i'm floating the current transformers at a little over a volt. I then set the AREF to about twice that. With the oven set to broil and the hot water heater going this provides enough range that i don't hit the aref, or drive my inputs negative.
• I'm assuming the outputs from the current transformers are linearly proportional to current in our power feeds. I'm also assuming that i can then just sample the AC output of the CTs over a couple cycles, look for the max, and call that 1/2 the PP voltage. I then assume our house supply voltage is fixed. I then further assume (hope) that given all the previous assumptions, the CT output is linearly related to our power usage, and i can just convert from one to the other. aaaaannnnnnnnddddd i'm ignoring power factor for the time being. obviously there are some places where this can fall down. if only i got that bsee...
• Once i'm sure everything is as linear as i hope, i'll compare the output of the current transformers with some known load changes and try to get the graphs in watts, and then make sure everything agrees with our power meter.

Logs: