New Product: Current Sensor

There are a number of ways to sense electric current — and all of them involve some fuss, at least in comparison to sensing voltages. The chief reason for this is that because current is a flow, rather than a potential (like voltage) we need to measure it in series with the circuit, rather than in parallel.

For beginners, this is the reason that your multimeter behaves so much differently when you switch it into the current ranges. The resistance of the meter is now very low, as it is meant to be inserted in series with a circuit. Consequently if you inadvertently touch a voltage source with the probes, it is like applying a short circuit to the voltage, and the protective fuse in the meter blows out, something that is easy to do by the way, and that I’ve done at least once with all of my meters.

Another popular way of sensing current, is to insert a low value resistor in series with a circuit and measure the voltage drop across the resistor. This is convenient when using small DC circuits, less so when using mains circuits (120 volts or 240 in Europe). By the way, this is the way your meter senses current, and the fuse is protecting the low-value resistor, which would otherwise be liable to a smokey death in inadvertent current-measuring mistakes.

When using mains circuits, often a transformer is employed. Even the electrician’s clamp meters work this way – although the principle may not be that clear. Only one conductor is inserted in the clamp meter, which acts as a one-turn primary and the meter forms the secondary winding of the transformer. Note that if both conductors are inserted into the clamp – say a lamp cord, that no reading is possible, because the two wires have equal but opposite fields, which promptly cancel each other.

Many other small current transformers are available, but for the hobbyist, the problem of tying the circuit to the mains, then finding an enclosure to house the transformer, is also some fuss. Lately hall effect current-measuring chips have become available, for measuring current, but the fuss is still in connecting them to the circuit and housing them.

Wouldn’t it be nice if you could measure current without having to mess with cords or plugs at all? There are small hall effect sensors with linear output that can be used to measure magnetic fields. I thought that it might be possible to use these to measure the current in common electrical wires such as lamp cords. You might be quite right to wonder about the equal but opposite fields from the lamp cord’s other conductor. The trick I have used is to get the sensor physically closer to one conductor than to the other conductor, so the fields are “sensible”.

Two linear hall effect sensors, in surface mount packages are mounted in close proximity  to the conductors.  They are oriented so that each sensor is closer to one of the wires in the cable. The AC voltage from the sensors is then subtracted, so that common mode noise will drop out, and filtered with a low-pass filter to eliminate any signals greater than the power line frequency.

Looking at the schematic of the input of the sensor, we see the two sensors. Actually there are four sensors in the schematic, but only two are populated at any one time. (We wanted to also try the TO-92 format  sensors, to see if they were more efficient for this application.)

The output of the sensors, which is a voltage that is proportional to the current flowing in the power cable, is introduced to the circuit through a 1 uf cap. These caps in combination with R1 and R3 form high pass filters – set at a corner of about 33 hz. So frequencies lower than that roll off quickly and don’t enter into the output. R1 in combination with C1  also forms a low-pass filter with a corner of about 277hz  (f = 1/ 2πR2*C1) so higher frequency noise is also attenuated.

Op amp aficionados will also recognize a differential configuration, so the voltage in one sensor is subtracted from the voltage in the other sensor. One would think that this would decrease the signal, but the trick is that the sensors are actually sensing signals that are 180 degrees out of phase, so the signal is actually increased, and other potential noise sources are diminished.

The output of the first op amp then is an AC waveform – at the same frequency as the power line, but proportional in voltage to the current flowing through the power wires. That small AC signal is then passed through C6 which eliminates any DC voltage offset while passing on our powerline signal. This signal is presented to the non-inverting input of op-amp IC1D, which makes a copy of signal, while also introducing some gain (to the tune of R11 + R13 / R10) so you can see that a range of gain from about 1 to 10 is available. R9, C7 and R8 form a peak detector, and store the height of the AC peaks on capacitor C7. This voltage is buffered by output amplifier IC1C to the output.

The capacitors in the peak detector, slow down its response to eliminate any jitter. The response of the sensor to ambients is thus fairly slow, but I didn’t think that most people measuring current on a powerline would be interested in quickly changing signals, with most users probably sampling their power consumption only about 10 to 30 times a minute or so at maximum.

The hall effect sensors unfortunately draw a bit of power, so the sensor draws about 20 mA when powered up at 5 volts. One strategy for very low- power (read long battery life) users, might be to power the sensor from a microcontroller (e.g. Arduino) pin and sample the sensor only once or twice a minute, and then power down the sensor in between samples. The sensor also needs a bit of time to power up and stabilize (perhaps 3 seconds) , so doing this at a fairly relaxed pace is a good idea.

We have tweaked the sensor a bit to try to get it into the most useful range. The sensor will sense 10 watts at the low end (83 mA @ 120V) or 1500 watts (12.5 A @ 120 volts) on the high end, if the nameplate on my electric kettle is to be believed. There is a fairly significant noise floor of about .25 volt as you might imagine for sensors with so much gain applied. The jitter however is fairly small, since so much filtering has also been applied.

The sensor is ready to go and in the shop.

Paul