Best detector for flashing red LED?

[Circlotron]

The electricity meter on my house has a red LED that flashes once every watt/hour with about a 50/50 duty cycle. I want to make something that I can hold against the LED and get a readout in watts by measuring the time from one flash to the next. 1kW would be 3.6 seconds.

What would make a good sensor for the red LED?

[Zero999]

An phototransitor would work but you need one without and infrared filter.

[CatalinaWOW]

I see two problems-

1.  You will have widely varying levels of ambient light.  This will make setting up your detector to remain unsaturated at all times a challenge.  Setting it up with a simple telescope focussed on the LED will help as it will reduce the field of view of the detector and allow masking light from other directions.  This can be a simple single lense telescope using a long focal length lens.

2.  Your power company will want an unobstructed view of the meter.  Even if it is normally read electronically.  The telescope approach can help with this problem also.

[StillTrying]

I've often thought of doing the same, I have one of those wireless clip on current transformer things but as they only measure the current they're not very accurate.
I won't do it now because I think smart meters are a waste of time.

1kW would be 3.6 seconds

Check if that's full flash or just a change of state of the LED. 100W would be 36 seconds so it might be useful to detect both edges of the flash to double the update rate.

Could you make a simple sensor to blue-tack over the LED?

I'm thinking something simple like a 20mm diameter, 15mm high piece of wood!
Drill a 6mm hole through the middle of it, bend the phototransistor's legs 90 deg. use a blob of clear glue on the back of the phototransistor to secure it in the hole.

At a wild guessimate, the phototransistor should pass about 3uA/15uA through its collector as the LED flashes off/on.
You'd need a short twisted pair to the phototransistor and a LPF on the detector.

[BrianHG]

Choose a focused 820nm photodiode.  The 820nm photodiode will still respond to red leds & they are much more sensitive than phototransistors.

Apply 3.3v, or, 5v to the photodiode through a 1000uh through 10000uh inductor.  When the red led goes on, you will see a little spike between the photodiode and inductor, then another when it goes off.  This is how you illiminate the low frequency DC bias effect of stray ambient light.  Through a series cap, amplify this signal through an opamp or comparitor & feed your MCU counter.

(This wont work if you red led smoothly increases in intensity instead of going snap on-and-off)

Something like this should do the trick:
http://www.digikey.com/product-detail/en/fairchild-on-semiconductor/QSD2030F/QSD2030F-ND/1049944
This one is more sensitive to visible light and may be flooded unless you darken it's surroundings:
http://www.digikey.com/product-detail/en/tt-electronics-optek-technology/OP999/365-1086-ND/498712
http://www.digikey.com/product-detail/en/fairchild-on-semiconductor/QSD2030/QSD2030-ND/1050960

[Someone]

The electricity meter on my house has a red LED that flashes once every watt/hour with about a 50/50 duty cycle. I want to make something that I can hold against the LED and get a readout in watts by measuring the time from one flash to the next. 1kW would be 3.6 seconds.

What would make a good sensor for the red LED?

Try not to reinvent the wheel:
http://reductionrevolution.com.au/products/watts-clever-smart-digital-energy-monitor-ew4500
Mass produced products are cheaper than your time. But even then there are endless "projects" on the internet to do the same:
https://openenergymonitor.org/emon/buildingblocks/introduction-to-pulse-counting

[mikeselectricstuff]

If it's up against the LED, and reasonably well shielded from ambient light a phototransistor will work just fine. A ~10K pullup and you have a signal that will go straight into a (preferably schmitt trigger) digital input

[BrianHG]

I didn't realize how close you were able to get to the LED.  The photo-diodes I listed will give you a digital square wave with a 10k-100k pull-up or down.  The solution I mentioned with the inductor and op-amp was if you had around 2-4 inches distance between you and the red led without any ambient light shielding.

[cdev]

You may be able to do it with RF using a cheap RTL2832U SDR dongle. ($6-12 on ebay) Some of them act as squitters and send their wattage figures at regular intervals to a local  receiver. The strongest signal is likely your own house/apartment's. They can also receive tons of other interesting stuff. Note: I have never done this because we don't have smart electrical meters where I live yet, as far as I know.

[mark03]

First, the correct unit is watt-hour (watt times hour), not watt/hour, which would be a measure of energy acceleration over time (joules per time squared).  Sorry, pet peeve of mine.

Second, if you measure in an analog rather than digital fashion, i.e. photodiode rather than phototransistor, you could do quite a lot with some clever signal processing:  high-pass filter to look for fast changes, while filtering out slower ambient variations.  It really depends how tight the optical coupling and shielding can be, whether this makes sense.

[BrianHG]

When I made my first comment, I thought the meter's LED was a few inches inside an open clear glass case like some of the ones here in Canada.  After seeing the photos from the earlier post, I noticed how the led is right up against a cheap plastic bezel where you would get an excellent square wave on/off no matter what you did.  As for using photo-diodes, I have created my own IR receiver for remote controls & during development experimented with photo-transistors and photo-diodes.  The photo-diodes outperformed the photo-transistors with regard to sensitivity over distance by a landslide.

[cdev]

Regular LEDs work as photo-diodes, too.