Question about Current Sensing

[dan3460]

Working on a project where I need to sense if a machine has been turned on, it has to be as least invasive possible. For a prototype I purchase one similar to this http://www.ebay.com/itm/2PCS-SCT-013-030-Non-invasive-AC-Current-Sensor-Clamp-Sensor-30A-Good-/331746402405?hash=item4d3da06065:g:BVkAAOSwuxFYvTGo The one I purchased the one for 20A (I think), price is not bad but looking to see if I can shave more cost, I can get this units for about US$5 each. I have this sensor feeding on a precision rectifier that tells the MCU that the machine was turned on or off. I don't need to know what amperage is passing through the sensor.
Thanks for the ideas and comments.

[tatus1969]

you can wind your own transformer, because that is what it is. https://en.m.wikipedia.org/wiki/Current_transformer

[Kleinstein]

If accuracy is not important one could use a more normal inductor with with an toroid core as an current transformer.
One might not need precision rectification - if the µC has time available it could directly sample the AC waveform and calculate the amplitude as needed. Accuracy might even be better than with rectification.

If the current is high enough, there is the option to use a kind of Rogowski coil - thus detecting voltage induced by the AC field around the wire without an core. It is more sensitive to external interference, but might be cheaper and even less invasive than the current transformer.

[Natee]

Hall-effect switch plus whatever suitable core you can find? Substitute linear IC if you need some sort of proportional signal.

[dan3460]

Thanks all for the answers.

you can wind your own transformer, because that is what it is. https://en.m.wikipedia.org/wiki/Current_transformer

Don't know if the cost of materials and the rig to wind coils will make it less than $5.

One might not need precision rectification - if the µC has time available it could directly sample the AC waveform and calculate the amplitude as needed. Accuracy might even be better than with rectification

The problem that I see is with the negative side of the voltage coming from the coil, MCUs are not very happy with them. The opamp I'm using for the precision rectifier cost about $0.04 each, plus the ancillary parts cost is about $0.08 for each sensor.

[Buriedcode]

What kind of quantity are we talking here? I don't see the point in winding your own coils unless it is a one-off.  Judging from your replies about the cost of 'rig to wind coils' I'm thinking you need >20 units.

Using offset hall sensors on the mains cable along with a couple of comparators could work, with parts well below $5 US, but then you have to get the PCB's made up.

You mention about the negative voltage from a rogowski coil.  Why does you need to measure the negative voltage at all? Or even why use an ADC?  I thought you were after just knowing when the appliance is on or off, in which case no need for a precision rectifier, just resistors, diode clamps, and a comparator would do.  If you're worried about negative voltage, a simple schottky from ground to signal should prevent the MCU's input clamp diodes from being forward biased.

[Kleinstein]

One can a makeshift current transformer from a ring core inductor. Just add the wire for the load current through the center.  It only needs the external shunt and usually 2 diodes for protection. The negative voltage can be avoided by having the other side of the current transformer at an intermediate DC voltage.

One could get a kind of Rogowski coil from two inductors (e.g. resistor shape) to both sides of the load wire. As the signal can be rather weak amplification might be needed. It might need some kind of shielding, if in an harsh environment.

[dan3460]

We are hoping to be able to sell around 1000 systems, each system has about 3 sensors. I have a prototype working with the current sensor that I posted originally. On the sensor side the two more expensive parts are the current sensor and the actuator, I don't think that I can do better with the actuator, but I though that I could get a simpler (or less expensive) way to sense that a machine was turned on. But it seems to me that all roads goes through some kid a coil or transformer and it looks like the one I'm using is the cheapest solution.

I looked on ring core inductors, but after you add the wire to connect and something to protect them you are over the $5 cost of the one that I working with.

Thanks guys, I really appreciate the help.
 

[TimFox]

This is a typical solid-core sense transformer, where you run a single wire through the hole and measure the voltage across a relatively low-value resistor across the main coil.  You can probably find cheaper ones, I just took a quick look through DigiKey.
https://www.digikey.com/product-detail/en/talema-group-llc/AC1020/1295-1095-ND/3881394
I have used this type of device successfully in measuring 60 Hz current.

[dan3460]

Thanks Tim, I did saw those at Digikey the advantage with the one that I have right now is that it can be easily put on the power wire of the machine and comes with the wire to connect to the sensor box.