Hi,
jcw posted this ca. 10 months ago:
http://news.jeelabs.org/2008/12/12/better-mousetrap/
Does anyone have success stories or more information about how well such a setup works?
-Marius
This forum has moved to forum.jeelabs.net - this read-only archive is for reference only!
Hi,
jcw posted this ca. 10 months ago:
http://news.jeelabs.org/2008/12/12/better-mousetrap/
Does anyone have success stories or more information about how well such a setup works?
-Marius
Hello,
I can only comment on the sensors part.
I have a similar setup, with an LDR detecting the electrical consumption (digital meter with a blinking led) and a CNY70 detecting the gas meter, just like JCW, by finding the reflective digit '6'.
The sensors are connected to an arduino, which is wired to a server, and sends every pulse via the serial port. I then log to a database (MySQL) every watthour and every 0.01m3 of gas used as a timestamp.
This kind of sensor setup works very well and reliably! mine is working for more than one year with very little issues.
Greetings,
Alex
FWIW, the setup as described has been running without a hitch since the end of last year. There were 2 or 3 power failures, I think.
I'm currently building a second similar setup for someone else (putting it all into a nice enclosure is the hard part!).
Am also still looking for ways to refine this, btw. The electricity meter has one black marker on its rotating disk, but also lots of tiny notches. I wonder whether the light sensor could be made sensitive enough to detect and count their small variations in reflectance. Then I could have a more responsive readout, especially at low usage levels where I'd like to further correlate the energy consumption changes to specific events around the house.
what about a laser diode reflecting off the rotating disk? in theory it should pick up the black notches?
We've got red notches over here. That might make it slightly more complicated..
I've ordered the same reflective sensors that jcw and amvv use, so let's see how it goes..
I like the idea of using a laser - if the beam is focused enough and positioned to hit exactly the notches, then detection might be quite easy with just a simple photodiode. Marius: green laser, perhaps?
Here a commercial solution for a energie consumption meter for rotating disk or
blinking led ,accurate = 10 W.
http://www.wattcher.nl/english
Anyone tried to read the sounds of their meters? My old-style electricity meter does a faint double click. I'm trying to read that with a piezo disc but can't isolate the sound from the background noise - pre-amp maybe and a filter needed. Haven't tried it with my gas meter, water pipes yet but might be another avenue.
Hey amvv,
I have no practical experience on using SQL. Could you describe a bit how you do the logging to a database? And how do you graph the data?
Cheers, Zooer
Hello Zooer!
I can surely describe how I do it, both the logging and the graphing, but I think it does not really fit in the subject of this thread. I will put it in the wiki...maybe that is a better place...if not then I will put it in my blog!
Greetings, and a Happy New Year!!
Hi again,
I keep getting back to this every now and then and I've gotten some SY310's for testing.
However, the black notch which passes by each revolution of the power meter is not black here, it's red, actually a slightly transparent red. I haven't had any success with the SY310, which might be related to the red notch reflecting close to the same amount of IR as the metallic disk.
I'm contemplating building my own sensor using a green LED and a phototransistor/diode/resistor, but before starting on that I just wanted to check back here for any ideas..
Cheers
-Marius
Hm, that's unfortunate. Sounds logical w.r.t. red/green.
I'd like to find a solution which works for as many people as possible. Reading this info is after all what got the whole JeeNode thing started.
One idea would be to write some code which self-configures, i.e. use analog-in to read the sensor values, and then take lots of readings to detect the most reliable threshold. Anything other than IR would be sensitive to daylight, though. That in itself is no big deal, but a gas meter can be off for quite a long time - so an on-the-fly self-adjusting algorithm can't just take the average of readings over say 5 minutes.
How about a semi-manual approach: load a special config sketch, which reports the histogram of values read over the period of a few minutes (by wireless), and which then lets you decide on the threshold and store that in EEPROM. Then you'd re-flash with the real code which uses these presets as thresholds for the analog readings.
We could also add some logic to further filter out false readings: min/max rotation times, for example. Furthermore, there is a fixed relation between on and off times (when the disc rotates at a constant speed).
What else could we do? Use an alternating red/green LED for illumination?
(and then there's the idea of a focused low-power laser...)
Success!
After some trial and error, I got the SYS310 working after all. As jcw pointed out a while ago, alignment is crucial. Well, I can subscribe to that as well now - it helped moving the oscilloscope closer to the energy meter ;)
I solved the alignment issue by lasercutting a cover with a slot for the sensor. This also blocks most ambient light which might confuse the sensor.
For more info and pictures, see this flickr set: http://www.flickr.com/photos/kintel/sets/72157623032210323/
Etching a real circuit board for the sensor and 3D printing an enclosure instead of my CAPA hack, and I think we might be closing in on a good solution for the hardware end of this.
Next up: Gas meter sensor.
Congrats. Heh - grey goo - http://www.flickr.com/photos/kintel/4253080146/in/set-72157623032210323/ :)
Some thoughts on "productizing" this - could the cover be made as pcb as well, perhaps some sort of spacers between the two boards for distance adjustment? Or is distance less critical, just exact positioning? I.e. 2 or 3 pcb pieces which you'd tape/glue together in such a way that it has the circuitry as well as lets you just press the whole thing against the transparent surface?
I hope to soon get both CNC and 3D-printing options going, but I'm not sure it will make a difference here. The Q is how to come up with at least part of a solution which will work for as many people as possible. I'm inclined to focus on the sensor + position, and leave attachment to people themselves because it probably varies a lot.
The sensor should be close enough to the energy meter window to eliminate any IR from reflecting off the window itself.
I also try to avoid designing for tape or glue, so I would prefer smth. screwable or clampable, but this is hard to make universally applicable. The device from wattcher.nl looks similar. I guess they just use a light adhesive (post-it style) to achieve this. Have you gotten a hand on one of these yet?
Having a standard PCB design for this would make all dimensions constant at least, so that's a start (btw., having an LED indicator on the sensor board is a really nice feature for verifying correct installation and quickly checking if things are ok).
Publishing some alternative enclosure designs both for lasercutting and 3D printing should get some people started building them. When you finish your MakerBot, you could "mass" produce enclosures, or have some lasercut and resell in the shop.
I could try out some case designs in OpenSCAD next time I'm in the mood..
I have had my hands on a wattcher, since we had a presentation at work about it, and it is indeed some sort of sticker.
The sensor device also has some line in the box, to aid the user into placing it in the correct position, the line being aligned with the sensor. That is the easiest solution to align it, but the sensor has to be somewhat robust. It also has a LED in the sensor casing as suggested by Kintel, to verify the correct placing.
Furthermore, and to have a standard (open source) solution, the sensor should also be able to detect the blinking leds of the moderm electronic electricity meters. These are very easy (I have one which blinks a red led), but I don't know if a IR sensor would detect the red pulse - I have also seen them in green.
If an IR sensor can't be used on red or green LEDs, an LDR could be used.
How about hooking up some Energy meters that support the s0 (DIN 43864 or IEC 62053-31) pulse output? Could a JeeNode be used to measure and record those pulses? and what if you want to monitor 14 of them (one for each group in the house)?
I am looking at a way to monitor power usage per group, and where needed do socket level measurements if I need more preciese details about a group.
Should be easy (pulse is a switch - or TTL - I assume). With two Expander Plugs you get 16 inputs on a single port.
I envy the level of detail you'll get with that, I only have one Ferraris wheel to watch with an IR reflectance sensor...