Computing stuff tied to the physical world

SMD current transformer

In Hardware on Sep 16, 2011 at 00:01

Here’s another attempt to detect whether an appliance is on and consuming power.

This time I’m going to use a tiny SMD transformer (DigiKey part# 811-1193-1-ND). It has a 1:100 turn ratio, with 0.007 Ω resistance in the primary “coil” (I think it’s just a single loop), and rated up to 10 A.

The problem with this part, which rules it out for serious use, is that it’s only meant for frequencies in the range 50 kHz to 500 kHz. Whoops, that’ll be off by three orders of magnitude when used with 50 Hz AC mains…

Oh well, since I had one lying around, least I can do is try it – right? Here’s the setup:


The hookup is a bit odd, but will become clear in future posts. The SMD transformer is only a few millimeters square, between the green jumper block and the black mini-breadboard.

I’m using the current transformer (CT) as follows:

JC's Doodles, page 12.png

… except that this schematic is for another experiment, still pending. In this case, there is no 0.1 Ω resistor on the primary side (it wouldn’t make a difference, next to the primary’s 7 mΩ), but there’s a 100 Ω “burden resistor” on the secondary side. This matches the data sheet for getting a 1 V/A sensitivity. For info on CT’s and burden resistors, see this OpenEnergyMonitor page.

Here are some first results, using the same sketch as in yesterday’s post:

    OK 17 164 19 0 0
    OK 17 17 16 0 0
    OK 17 66 16 0 0
    OK 17 42 19 0 0
    OK 17 241 163 0 0   <- 100 W light bulb
    OK 17 107 95 0 0
    OK 17 49 93 0 0
    OK 17 77 94 0 0
    OK 17 134 94 0 0
    OK 17 244 93 0 0
    OK 17 7 94 0 0
    OK 17 151 93 0 0
    OK 17 27 93 0 0
    OK 17 124 90 0 0
    OK 17 119 24 0 0    <- off
    OK 17 125 24 0 0
    OK 17 36 16 0 0
    OK 17 246 15 0 0
    OK 17 109 16 0 0
    OK 17 33 18 0 0
    OK 17 50 16 0 0

These are raw bytes, i.e. a 32-bit long int, sent once a second and received via wireless. I’m powering the JeeNode with an AA Power Board and standing back during testing, the 220 VAC is now a bit too close for (my) comfort…

The average measurement value with no current is now about 16 x 256 = 4 mV.

The high value when the light has been turned on is about 94 x 256 = 24 mV, a 20 mV increase.

Not bad… but it’s less sensitive than the ACS714 Hall-effect sensor. I didn’t even try it with a 1 W load. This is a far cry from the claimed 1 V/A sensitivity, no doubt because the 50 Hz signal is so far off from the design specs of this SMD current transformer. Saturation effects, or something…

I’ll continue these experiments in the next few days with another current transformer which is designed to work at 50 Hz, as well as with a reversed 220 : 6.3 V PCB transformer. Stay tuned!

  1. I got to think about a handmade coil winded around the AC chord and then have it connected to a optodecoupler and a schmitt trigger or similar. Probably some R:s anc C:s aswell, I am no electrical engineer. The idea is that the length of the pulses passing the schmitt trigger would be proportional to the current that passes.

    I remember I watched on tv that they did a logger at the SkyLab space station based on fuzzy logic. It detected current in only one point I think and learnt what different electrical stuff in the space station did in terms of current spikes which translated into textual messages into the log. Pretty neat I think but probably DC based measurements.

    Anyway, interesting experiments, nice work! :-)

  2. Nice going, but what’s a reversed 220:6.3V PCB transformer??

  3. You should use very low ohm (inverted) op to convert I to V in the secondary side.

    If you use 2 ferrit core in serie with 32 turn in each you can get 1:1024 ratio.

    1:(32*32) = 1:1024

  4. Your 50kHz transformer could be used at 50Hz but you would need a lower burden resistor. A AC-coupled transimpedance amplifier (current input, voltage output) works quite well in such a case.

Comments are closed.