Light Sensing Question

[julie_m]

I need a way of sensing the LEDs on my electricity meters, and turning on an NPN transistor each time one of them flashes.

My key meter has an LED that is normally green, but pulses orange for a few milliseconds, 800 times per kWh consumed (or once for every 1.25 Wh; either way, up to a maximum of 11200 pulses per hour or about one pulse per 320 ms., in the worst case with absolutely everything switched on). Using a piece of red plastic as a crude filter, the green light gets absorbed by the red plastic, leaving only a pulse of red light shining through it.

My solar generation meter has a red LED that pulses 1000 times per kWh generated (or once for every watt-hour; up to 2250 pulses per hour) or stays on steadily during the hours of darkness when nothing is being generated. This looks as though it is part of a communications link, allowing a meter reader to take a reading directly with a special device, so the pulses may turn out to be modulated with data.

I want to use a Raspberry Pi to sense and log the times of each of these pulses, to build up a graph of my generation and consumption in the time domain.

I don't really need strictly to monitor the generation meter, since I can get at the telemetry from the inverter to determine how much is being generated. But the best way I can see to get a handle on my consumption is via the flashing LED -- it neatly provides electrical separation, and avoids the need to insert a sense resistor into the high-current cables, with all the potential ways that could go wrong.

So, what is the best device currently available at a sensible price, that I could use to read the flashing meter lights? Back in the Day, I probably would have started with the ORP12 Cadmium Sulphide light-dependent resistor; but there probably is something more modern, and with less noxious chemicals, available now ..... The sensor needs to be capable of being stuck over the LED with Blu-Tack.

Any ideas?

[MrElectronicman]

How about a LDR out of an old photoelectric night light. I have several of them and they are about 5 x 4 mm in size and have dark resistance of several K and bright light resistance of 100R or so.

Would be easy to adapt to a Raspberry Pi with a couple of fixed resistors and they are quite small

[G6Tanuki]

Rather than spending time/effort doing this yourself why not contact your utility-company and ask them what they can provide?

Whether it's a "smart meter" or one of the less-sophisticated clamp-on power consumption meters, for the last couple of decades there have been ~free~ options to log your consumption and send it via WiFi or 433.92MHz RF or similar to a remote display.

Example: http://www.theowl.com/

While I can appreciate the pure intellectual-exercise in "rolling-your-own", I must admit it does seem a bit like reinventing the wheel from first-principles when you could get a solution free/off the shelf and spend more time down the pub rather than hunched over a workbench.

[julie_m]

Even a clamp-on current sensor is still going to need a wired-in voltage sensing connection, if only to get the power factor. And the data still has to be got into the R.Pi somehow for processing; so a ready-made proprietary solution is next to useless.

I don't have a night-light (or anything else with a light sensor, for that matter) to strip down for parts. Are such things still available?

[Alistair D]

http://www.finntestelectronics.com/product/finn/

In a past life I used to use these sensors to detect the displayed colour and intensity of bi-colour LEDs in our automated test equipment. Basically they were photo diodes with a colour filter built in. The output was a voltage in proportion to the LED intensity.

Alas I can no longer remember how much they cost but they were 100% effective

Al

[broadgage]

Most clamp on current sensors for monitoring domestic energy use are NOT used in conjunction with any wired connection to sense the voltage.
They are therefore very inaccurate on loads with poor power factor, and also somewhat inaccurate under varying supply voltages. They are in effect AC ampere meters and AC ampere hour meters and not true wattmeters or kilowatt hour meters.
The difference between 10 amps at 220 volts and 10 amps at 250 volts is significant, but both are within statutory tolerances.

[Diabolical Artificer]

Found these at 50p - http://cpc.farnell.com/webapp/wcs/st...&storeId=10180 and these - http://cpc.farnell.com/optical-slot-...ectric-sensors

I have a couple of photo diode's you can have to play with Julie. I havn't experimented much with them apart from connecting them up to a diode checker and shining a light at them, they're very sensitive.

The other thought is using one half of an opto-coupler.

Andy.

[trsomian]

LEDs work as photodiodes, so a red encapsulated LED will be most sensitive to red light, green encapsulated ones most sensitive to green light. Whether or not the encapsulation would be an effective enough filter for your application I do not know, but it might be worth an experiment