Post Go back to editing

ADP5092 and CT harvester

MPPT for a current transformer based energy harvester may or may not make sense, but either way it definitely needs to be configured differently than you would for a PV.

The ADP5092 datasheet provides an example of how to use a CT with that chip:

The left-most four diodes make up a bridge rectifier.  The right-most pair of diodes establishes what is essentially a constant-voltage burden resistor.  What you end up with is a system with essentially constant voltage (for all but very tiny currents) and current that varies based on the primary.  Like other EH topologies, this charges the main inductor, which switches on and off at some rate and dumps energy into VIN.

However, if you look at how MPPT is set up, it's going to be V(open circuit) * 6.34 / (6.34 + 14.7) = 0.97 volts assuming V(open circuit) is somewhere around 1.4 volts.  The circuit as configured would enable MPPT sampling mode, and MINOP would be 0.22 Volts.

I don't think either of these things make sense.  First of all, as soon as there's any appreciable current  the voltage on VIN will be something close to two diode drops.  In this circuit, I don't think a fixed threshold like that really works.

Second, if you do want MPPT sampling (which I'm not sure you do) then the way to look at that would be that you're looking at the voltage on the 10 uF capacitor.  But that capacitor is going to quickly charge to 1.4V, slowed down only by the real impedances of the components (a CT is often modeled as a current source but it really isn't quite).

My guess is I'm looking at this incorrectly, but where am I going wrong?

Parents
  • Thank you for the carefully thinking about this diagram. There are two conditions for this application.

    1. If the OCV is lower than the maximum operation voltage of ADP5092, the MPPT ratio may be 50% but it would still vary with the material of the CT. The diode or clamp circuit is just for the protection.

    2. If the OCV is higher than the maximum operation of ADP5092, for example 5V, so the clamp circuit would emulate a OCV like 3V. We assume the MPP voltage is 2.5V, the MPPT ratio should be 2.5/3=0.83.   

    I suggest to test the CT I/V curve under your system operation condition then select the clamp circuit and MPPT ratio.  

    MINOP is just for preventing discharging the battery when the VIN is very low and you can adjust the resistance for your CT.

  • I agree, having an I/V curve for the CT would be useful.  I'm not exactly sure how to model what the energy harvester looks like, though.  My thought was this:

    and do two tests: first, make R very large (or remove it entirely) to get the open circuit behavior, though I am pretty sure it's going to be 1.2V and no current.  Second, put an R that represents the ADP5092 input (but what would that be?)  Then for both tests vary Iprimary to build a curve.  Is that how you would approach it?

  • If you want to get I/V curve, you may need remove the clamping diodes. Fix Iprimary and vary R to get the I/V curve cross the R. Then you can get a MPP ratio. You may need vary Iprimary and repeat the I/V curve test. Then you can verify the MPP for different Iprimary. 

  • OK.  The ADP5092 can take up to 3.3V on VIN so can I just replace the two back to back diodes with a 3.2 V  zener diode?  My source can produce this voltage ..... or doesn't it matter to do that, because you are suggesting setting a relatively low MPPT voltage ?

  • You can use zener diode. The MPPT voltage depends on your I/V curve of the CT. So the OCV may be 3V and the MPPT voltage may be 1.5V. Or the OCV is 5V and the MPPT voltage is 2.5V. Then you can determine the MPPT resistor ratio.

  • The OCV of the CT itself is going to be limited only by an 8 volt clamping diode in the CT itself.  Most small CTs have such a diode so that if the secondary of the CT goes open circuit it doesn't overheat.  On larger CTs they can actually cause a fire, so most big CT installations include a shorting block for service.

    We aren't in that domain, though.  In fact, even 8 Volts is too much for the ADP5092.  What we want to do is shoot for 3.3V clamping to stay within VIN's operating range, plus a little margin to account for thermal and tolerance effects.  That's why I was shooting for 3.0V or 3.2V.

    I tried to do what you suggested regarding CT characterization.  Drawing an I-V curve gives you something that's pretty flat, which I think is expected - up to the point where it saturates, the CT is going to behave quite like a current source.  Keeping in mind that we need to stay below 3.3V for VIN, pay attention only from about 3 volts down on this curve:

    It's pretty much flat as I think you'd expect.  But if you take that same data and instead of I-V you express it as P-V you see something a little different:

    So this seems to suggest to me that if you're below the clamping voltage that the maximum energy you can harvest is going to be at the highest voltage allowed - in this case, 3.3 Volts.

    This all suggests to me the following:

    • You should set the clamping voltage to as high a voltage as VIN can tolerate
    • Over the vast majority of the output, you can expect to harvest the maximum amount of energy (power versus time) at the highest voltage, so you should set Vmppt (threshold) to something just below the clamping voltage.

    This leads me to thinking that you really should set the MPPT threshold to somewhere really close to 3.3 Volts ... let's say 90% of the clamping voltage.

    I'm still wondering with a source like this if you're not better off disabling MPPT sampling entirely.

    Does that make sense?

  • It's better to measure the I/V curve at your nominal operation condition which means under the nominal primary current. 

    If the I/V curve is your nominal operation condition, you will waste a lot of energy by limit the voltage lower than 3.3V. But it's still okay for the rest of energy is enough for your system.

    If the energy is not enough, you can try LTC3355 and LTC3225 but these are only for the supercap.

Reply
  • It's better to measure the I/V curve at your nominal operation condition which means under the nominal primary current. 

    If the I/V curve is your nominal operation condition, you will waste a lot of energy by limit the voltage lower than 3.3V. But it's still okay for the rest of energy is enough for your system.

    If the energy is not enough, you can try LTC3355 and LTC3225 but these are only for the supercap.

Children
  • Yeah.  I picked this chip because we're trying to support both CT harvesting and indoor PV with a single design, just by changing a few components (the input side, and MPPT and other voltages).  For the PV setup we want the really low startup voltage, which is one of the best features of the ADP5092.  Depending on how this works out maybe we will decide to do that.  As this MPPT discussion highlights, there are probably features in the 5092 that I don't really need for the CT case.  It would be nice to have something with a buck/boost regulator on the output side as well.

    The nominal operating point is a valuable discussion.  However, I think when the primary is heavily loaded there is plenty enough energy available for harvesting, so if I lose some efficiency up there it's probably ok.  I'm really designing this so that I have enough energy when the primary is at 20% or even 10% of the rated circuit current.  So for example using a 100A CT, can I generate enough power to run my sensor even if the equipment being monitored is only at 10A.  The secondary of the CT at that point will only produce 5mA of current.  At some point I expect that real effects will drag the voltage on the small storage capacitor (the one between VIN and ground, in parallel with the clamp and load) down, and MPPT will regulate exactly how far down we allow it to go.  Your example above seems to suggest 83% which sounds reasonable to me.  The thing I'm not sure of still is whether I'm better off just picking a fixed voltage and turning MPPT sampling off.