I removed the built-in burden resistors from my split core current transformer (CT). This eliminates the power dissipation of that resistor and allows me to harvest as much as possible from the transformer. I was able to charge 4.5 F of supercaps from 2.9 V to 4.9 V in about one hour using a 1300 W space heater (~11 A current in the primary). This means I could charge 3 F in about 40 minutes and get about 5-6 Wi-Fi transmissions out. It looks promising.
There is one big caveat so far. I want to get one report an hour with 0.5 A in the primary (equivalent to one 60 W bulb on the circuit). I did try to charge 3 F with 0.5 A in the primary and it looked like it was going to take about 100 hours. That's not a huge surprise. The energy harvesting circuit is pretty inefficient down at the low secondary current levels. Plus, I had nothing optimized. Finally, the current transformer is not really intended for harvesting, so the turns ratio is a bit off. I do have plans to try to fix these though. A factor of 100 looks bad, but in my experience it isn't necessarily that bad.
First, I'm going to characterize the current transformer I do have. The BQ25504 has a maximum power point (MPP) setting where you set the operating voltage to be a fraction of the open circuit voltage. This is essential to get solar cells operating at their MPP. I will also use it for the current transformer. It is probably operating too close to the short circuit current, and I need to find the MPP. I will generate the IV curve by varying the burden resistance on the CT secondary to find this point. Then I will tune the resistor network on the BQ25504 to hit exactly that maximum power point. I might be able to get a factor of 2 out of this tuning, if I'm lucky. At some point, I will almost certainly get a different and more efficient current transformer that is more optimized to the task at hand.
Another thing I need to do is look at my rectification. Right now, I'm just running the CT secondary through a single diode because the diodes I had on hand dropped the voltage too much when I tried to rectify the CT secondary with a full bridge. I didn't even have a smoothing capacitor There was almost nothing left at the output of the bridge for the charging circuit. If I can get the CT output voltage a bit higher by operating closer to the MPP of the CT, I am hopeful a bridge will work, and that will probably get me a factor of 2. I'll also use Schottky diodes for their lower voltage drop.
Finally, I may have to drop the BQ25504 and look at the Linear Technologies line of harvesters. They are specified to harvest all the way down to 20 mV, whereas TI needs 330 mV to cold start the harvesting. It could just be a difference in how these are specified, but it's something I have to look into. I'm not sure what kind of efficiency improvement I can get out of this. That's the point of the testing.
The other thing I'll need to get clever about is optimizing the Wi-Fi power consumption. I have done no work at all there right now. I'm just running a webserver that autorefreshes periodically. I might be able to get a factor of 10 improvement out of the Wi-Fi by optimizing power use and possibly by selecting a different radio. The ESP8266 is just a good easy one to start with.