Transformer based energy harvesting

[didjcodt]

Hi,
for a personal project, I have a transformer current sensor (the SCT-013-000, it's like a basic current clamp) to measure the intensity of current passing through my home switchboard. This measure is then sent via Bluetooth Low Energy to a base.
The problem is that the sensor circuitry to send the data needs the biggest possible autonomy, this is why I am thinking of harvesting the energy of the current sensor, but I also need to accurately measure the sensor ...

I found some solutions for that :
- Using two current clamps (but it's really hard to put two clamps in my switchboard)
- Switching the output of the clamp to alternatively harvest and sensing (but I have not found a simple circuit for that)
- Sensing the current passing through the harvester to compensate the bias on the sensing (but how can I do that without consuming more energy that my clamp can get ?)

What do you think ? Do you have some ideas ?

ps : sorry for my bad English, I'm French
Thanks,
--
Lichard Torman

[DanielS]

What I would do?

Sense transformer -> bridge -> current measurement resistor -> shunt voltage regulator

Measure the current coming from the bridge and use a lookup table for calibration/compensation, use the voltage across the shunt regulator to power your module. If you have a 1000:1 transformer, that gives you 1mA to work with at 1A load.

[Richard Crowley]

It is not clear why you rejected the conventional method of powering your current monitoring device?
Why not simply use a small "wall wart" power supply? 
"Harvesting" power from another current transformer just seems goofy.

[G7PSK]

You could try winding a few turns of wire around the incoming cable but you might find that you need a battery for when not enough power is flowing to run your gadget. A wall wart sounds like the best idea.

[didjcodt]

Thank you for your quick answers !
I'll try with a shunt voltage regulator (never used them  )

Powering my device is not possible since I need to use an outlet, which is not present in most switchboards, and I can't plug any transformer in an invasive way because of insurances (if there is a fire, I will never be refunded). The current transformer is a clamp that you put on a sheathed cable, so it's not a problem.

I already have a few monthes of autonomy with batteries, but my goal is to get a lot more autonomy. I'll see if it's easy enough to wind a few turns of wire around the main phase (it's very difficult of access).

Thanks

[Marco]

Couldn't you capacitively tap power? This would even work when there is no current flowing.

[didjcodt]

I don't see how it could be possible to capacitively tap power. (by putting two traces near the sensor cable ?)

The shunt regulator seems to be a really good solution
If I understand, this regulator can let me sense the current via a resistor (so i have a direct measure of the current like I actually have with a burden resistor) and then I can use a shunt regulator to take this current when I need to ? If that is the case, it would me exactly what I need !

Thanks !

[DanielS]

The shunt regulator provides a current sink for any excess current from your current transformer to keep its voltage from rising arbitrarily high and also provides a stable voltage for your low-power circuitry as long as the load current is high enough to generate sufficient sense winding current.

As for capacitive tap, I think what Marco meant is using a capacitive voltage dropper. You use a simple RC network to limit current from mains, pass it through a diode bridge, a filter cap, a drop resistor and a zener to shunt excess current. It is a very common arrangement in cheap and often unsafe 120/240V LED lamps and other line-powered electronics.

[Ian.M]

I think there may be some problems with an excessive burden on the CT and also some nonlinearities if you operate it into a fixed voltage e.g a shunt regulator.

Do you actually need to sample every cycle or could you settle for sampling every 5th cycle, which would be ten updates a second?

If that's acceptable, you could have a bridge rectifier on the CT output with a load resistor switched in by a low Rds_on MOSFET.   When the load resistor is switched out, the CT output voltage would rise and charge a big cap (maybe a supercap) through a Schottky diode.  It would also need a beefy Zener across the cap to prevent the voltage rising excessively and exceeding the cap or CT's voltage limits.  The sequence of operation would be:  Turn on MOSFET at zero crossing and wait one whole cycle to eliminate transients.  From the next zero crossing, sample a whole cycle ferequently enough to calculate the RMS current. Then turn off the MOSFET and charge the cap for the next three cycles.

[Marco]

As for capacitive tap, I think what Marco meant is using a capacitive voltage dropper.

No, what I meant is that you could put put 3 say 10 cm strips of copper along the length of the cable, but distributed around it. Then try to rectify the voltage differences between them (you need 3 because two might be exactly at the centre of phase and neutral). Probably won't get anything with normal rectifiers, but with Schottkys they use for energy harvesting you might get some worthwhile energy out.

It's like the capacitive transformer you describe, but the capacitor consists of the wire in the cable and your strip of copper outside it (doesn't work period with shielded cables obviously).

[didjcodt]

For the capacitive solution, I think it may be too complicated for an installation.

I would like to sample every second (don't need more data). The FET solution I interesting, but what would be the consumption of such a system ? (I'm trying to harvest 1-2 mA max., usually it would be something like 300-400 µA, so I need a FET that don't consume a lot of amps)

Can't I just use a battery charger and energy harvester chip like this one (http://www.ti.com/lit/ds/symlink/bq25504.pdf) and put a resistor on the input to monitor the flowing current ? (My goal is to simplify the circuitry)

[oldway]

This is very easy to do 

To power the dc/dc converter when current is low, use a battery in // with the elco and a zener to limit charging voltage on the battery.

[DanielS]

See attachment for what I had in mind.

[didjcodt]

This is more what I understood, in fact.
Just another question: can't we replace the 3 pin zener and the 2 resistors with a basic 2 pin zener ?

Thanks

[Jeroen3]

How many VA is your current transformer? I've seem some of 10VA (x 3 phase), but they have hundreds amps as primary.

[DanielS]

This is more what I understood, in fact.
Just another question: can't we replace the 3 pin zener and the 2 resistors with a basic 2 pin zener ?

The "three pins zener" is a programmable reference / shunt regulator such as a TL431 or one of its countless equivalents. Depending on how much sense current you get out of your transformer, you may need to look into low quiescent current variants.

If you use a regular zener, you are limited to standard zener diode voltages and tolerances, which limits your options. With a 431, you can have anything from 2.5V and up by simply picking suitable resistor values. This gives you the option to experiment with optimizing the balance between ADC supply voltage and measurement non-linearity.

[didjcodt]

As I will put a battery manager that can support 0.5-3V I don't need a precision regulator before, and I can't get more than 2V (/GND) on my ADCs, so if I get 1V for sensing (it's a 12 bit ADC so the precision is large enough to get a good measure) I just have 1V for my shunt regulator, meaning I could even use a standard 0.7V drop, can't I ? The other solution would be to put the ground between the sense resistor and the shunt, but I don't see how I could get a positive voltage (/GND) then...
Could what I put in attachement work ? Another solution would be to use potentiometers on my sense resistor and a TL431 (or another to get any value from 0.5 to 3V). Should I also put some zener diodes on my ADC inputs to protect them ?

The only thing I know about the transformer is that it is 100 A : 50 mA, and the reference I put on the first post. I don't have the VA rating 

[DanielS]

Standard diodes as a shunt regulator might be problematic since their forward voltage varies so much between 0.1mA, the minimum current necessary to operate many 431 variants, and 50mA, the transformer's maximum output current rating. Using a 431 provides a much more predictable value.

If you use a TLVH431, a lower voltage 431 variant, you can go down to 1.24V shunt voltage if you want to.

As far as protection is concerned, in my version of the circuit, that is already taken care of by the filter capacitor if you make it large enough. Depending on exactly what you want to measure and the accuracy you want to achieve, this may or not be adequate.

[didjcodt]

Ok, I changed my current sense resistor to drop 0.55V maximum so I can get a not-so-worse resolution, and I can regulate to 1.3V, meaning the TLVH431 is a very good solution
I'm making the schematic now, I will post the whole version next time !

Thanks !

[didjcodt]

I've made a new post as I'm posting the whole project, link is here : https://www.eevblog.com/forum/projects/sensor-board-using-energy-harvesting/

Thank you