Measuring power in an AC cable with iPhone sensors. How does this work?
August 9, 2014 5:11 PM   Subscribe

My electric utility released an iPhone app called "DTE Insight" which (among other things) measures the amount of power flowing through an appliance's power cord by asking you to hold the cord against a specific part of the iPhone at a specific angle. I have tested it against my Kill-a-Watt and it seems remarkably accurate for what it is. It presumably uses the iPhone's magnetic field sensor to accomplish this, but how on earth does it actually work?

[If you want to try this yourself, you have to select "Tools" and then "Take a Power Scan" from the main menu of the app.]

I understand that current flowing through a conductor creates a magnetic field around the conductor, and I imagine that this app is using the phone's magnetic field sensor to try to measure the current flowing through the cable, then multiplying this by an assumed voltage of 115 or 120 or whatever in order to estimate power.

What confuses me is that household appliance cables contain two current-carrying conductors. Wouldn't the opposing magnetic fields from the two conductors interfere with and almost completely cancel out their respective fields out in the area around the cable? The app does ask whether the cord being tested is a two- or three-pronged cord, and whether the cord is round or flat. Can anybody explain to me how this might work?


I did some tests using the app and my Kill-a-Watt to measure power draw of the same loads. I did five trials of each of the four loads using the app. The loads were:

1. Assorted computing equipment sharing a power strip
2. An electric fan on low
3. An electric fan on medium
4. An electric fan on high

The results are as follows:

Compute Load:
Kill-a-Watt: 632 W, 757 VA
App: 678 W, 667 W, 623 W, 662 W, 644 W

Fan on Low:
Kill-a-Watt: 56 W, 57 VA
App: 75 W, 75 W, 75 W, 75 W, 77 W

Fan on Medium:
Kill-a-Watt: 75 W, 75 VA
App: 103 W, 101 W, 98 W, 100 W, 97 W

Fan on High:
Kill-a-Watt: 97 W, 102 VA
App: 132 W, 129 W, 131 W, 128 W, 131 W
posted by Juffo-Wup to Technology (10 answers total) 11 users marked this as a favorite
I'm not an EE, but would like to point out that a clamp meter or a no-contact ammeter works with a similar process.

Also, in AC, only one wire is 'hot' and oscillates between +120V and -120V; the other wire stays at neutral and is the return path for the current. Regardless, the magnetic field for two wires next to each other with current moving in opposite directions can be seen here. It looks like AC ammeters deal with rectifying the signal; I suspect that the iPhone does this in software by getting an amplitude of the range of the fluctuations of its magnetometer and matching it against a pre-calibrated/pre-tested lookup table, since the exact magnetometer results may vary from iPhone model to model.
posted by suedehead at 5:55 PM on August 9, 2014

I am a EE and clamp meters won't work with normal appliance cords. You have to split out the conductors and clamp just one of them.

I'm really surprised the iPhone app works as you've described it.
posted by ryanrs at 6:21 PM on August 9, 2014 [4 favorites]

I was sitting here biting my tongue not wanting to threadsit, but wanting to mention ryanrs' observation that for a current transformer clamp, you have to split the conductors and clamp [only] one of them.

That is one of the main reasons I am puzzled by the (relative) accuracy of this app.
posted by Juffo-Wup at 6:24 PM on August 9, 2014

...only one wire is 'hot' and oscillates between +120V and -120V...

Well, just to be pedantic, it oscillates between +170V and -170V. 120V is RMS, you know...
posted by Chocolate Pickle at 6:53 PM on August 9, 2014 [3 favorites]

This article talks about using multiple Hall Effect sensors, which is pretty much what powers your iPhone compass. The sensors are 180 degrees out of phase to cancel out noise, like a humbucker pickup on an electric guitar.

I suppose the trick must be that the sensors are closer to one conductor than the other, and you get some usable signal you can extrapolate to the current. Or maybe they do some kind of machine-learning on all eight sensors (like how Fitbit calculates steps). Dunno. Kinda cool though.
posted by RobotVoodooPower at 9:51 PM on August 9, 2014 [1 favorite]

I stand well-corrected, thanks.

RobotVoodooPower - looks like there used to be a clamp meter - the MMC850 that worked on multiple wires - that worked exactly because there were multiple sensors.
posted by suedehead at 10:40 PM on August 9, 2014

It's true that there are two conductors running opposite polarities, but they are parallel and separated by a small distance. They should set up an interference pattern rather than just outright cancelling each other. (They'd have to be coax in order to cancel.)

If you hold an AM radio next to an active AC power line, you can pick up 60 HZ, and I think that's the reason why. I suspect that this program is using those Hall-effect sensors to sample the EMI. The more current, the more EMI.
posted by Chocolate Pickle at 11:04 PM on August 9, 2014

I went to see if I could gank the android version of the app, for my own not-living-in-michigan electricity interests. it says you also need to have the company's billing-meter be a smart-meter (actually I think it said, "Advanced Meter".) which, to me, is going to imply 12b/jeenode RF or probably just ad-hoc wifi since I don't think any phones have 12b in them yet

like, the phone's magnetometer is getting an assist from the in-line meter itself?
posted by dorian at 2:38 AM on August 10, 2014 [1 favorite]

It looks like RobotVoodooPower and suedehead have got it figured out. Very interesting, and pretty cool.

dorian: The smart-meter thing is for a different part of the app; it will query the power company over the Internet and show you your hourly consumption. Unfortunately, it seems that the app won't run at all unless you give it the credentials for a DTE account :(.
posted by Juffo-Wup at 9:43 AM on August 10, 2014

Someone ought to write their own app that just does the hall effect sensing and is not linked to a specific electric company...
posted by Juffo-Wup at 9:45 AM on August 10, 2014 [1 favorite]

« Older timber koi pond: do I need to use any adhesives?   |   Anti-inflammatory diet: do's and don'ts? Newer »
This thread is closed to new comments.