0
\$\begingroup\$

I want to build an autonomous outdoor temperature sensor that samples the temperature once per hour and stores the results over WiFi. I would use an ESP32 that will consume about 12 μA in stand-by (most of the time) and 200 mA when awake (15 seconds per hour, with peaks up to 600 mA when using WiFi.)

Because it is outdoors, I thought of using solar panels as an energy source and supercaps to store the energy (as batteries can't stand the temperature fluctuations). Several power harvesting ICs/supercap charging circuits exist, for example this module on Tindie, but they seem like overkill.

I am thinking of directly connecting a 3 V solar panel to 2.5 V supercaps, with a Schottky diode to avoid back flow and a Zener diode to limit the voltage. Then I would need a low quiescent current (maximum 10 μA) but high output current boost converter to convert the 2.5 V from the supercaps to the 3.3 V needed for the ESP32. Ideally, the input voltage of the converter would be as low as possible.

I can't find a converter to convert 1.0 - 2.5 V to 3.3 V with a low (<10 μA) quiescent current and a 600 mA maximum output current. Do they exist?

Ideally, the converter should be through-hole as my soldering skills are limited.

Some calculations for reference. Please correct me if I am wrong.

  • Energy need is (200 mA × 15 s + 12 μA × 3585 s) / hour = about 0.95 mA / hour. This is about 24 mA / day.
  • The sun doesn't shine every day. I would like 4 days autonomy = 100 mA energy storage.
  • Assuming a useful voltage drop from 2.7 V to 1.0 V, this implies a capacitance of about 400 F to store the equivalent of 100 mAh @ 3.3 V. (0.5 × (2.7V2 - 1V2) × 400 F = 1258 J, 1258 J / 3.3 V / 3600 s = 0.106 Ah = 106 mAh)
  • In the calculations, I neglect leakage current and other leakage loss, so I would put two 400 F in parallel.
  • Re-charging the supercapacitors on a sunny day of 8 hours (in winter) would require 12.5 mA / h @ 3.3 V or 16.5 mA / h @ 2.5 V. So I would need at least a 16.5 mA solar panel (but again include a big margin.)
\$\endgroup\$
2
  • \$\begingroup\$ If the 600 mA is short duration, the regulator output capacitor (that feeds the thing that consumes 600 mA) can handle that current with a slight voltage droop. \$\endgroup\$
    – Andy aka
    Commented Jan 31, 2023 at 21:11
  • \$\begingroup\$ That would be great! The rated power usage on wifi transmission is 160 to 260 mA, but some people report spikes. \$\endgroup\$
    – dloeckx
    Commented Feb 1, 2023 at 19:36

1 Answer 1

1
\$\begingroup\$

Use two regulators -- a super-efficient one with low idle current that will get your processor from sleep to waking up. Then turn on the big one to power your WiFi.

You may want to power the WiFi from a separate rail that's just powered from the big regulator.

Note that in your power calculations you'll need to take into account the amount of energy consumed by the "big" regulator just to wake up, and for however long it's turned on. Make sure you have enough energy in your supercaps to fire up that regulator and complete your worst-case WiFi transaction.

\$\endgroup\$
2
  • \$\begingroup\$ Excellent suggestion. Would that work if both regulators are connected in parallel? As the wifi is integrated in the ESP32 itself, it can't be powered from a separate rail. \$\endgroup\$
    – dloeckx
    Commented Feb 1, 2023 at 19:38
  • \$\begingroup\$ It would be much trickier to have them transition without the power rail glitching, and part of the reason I was suggesting a separate regulator was because it'll be more work to have the lightweight converter burdened by the heavyweight converter's output network. I think you'd want to ask how to do this as a separate question. \$\endgroup\$
    – TimWescott
    Commented Feb 2, 2023 at 4:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.