Charging a 3000 mAh LiPo with a cheap 5 W solar panel
post Posted on :: battery experiment lipo power solar

I finally got all the bits in the mail to start testing out solar powered LiPo systems. It's a sunny September Sunday: can I charge the LiPO?

The question

Honestly, I'm a bit scared:

  1. Will these parts charge the LiPo?
  2. Will things catch fire?

The parts

  • Charger: a CN3065 board. It's cheap. I have some nice Adafruit ones on the way too, but what will this one do?
  • Panel: 10W 5V panel with a mount. The only reason I got this one was the mount will help me with packaging later. Another amazon special...
  • Cell: an inexpensive LiPo pillow, 3000 mAh, I think with a little protection pack.

Setup

Rigor is turned all the way to 11 on this one \s. Panel is kinda angled at the sun. DMM to measure cell voltage across its terminals.

Results

Cell voltages today, read by hand. Two phases for the run: the first ended when the cell came unplugged (see Aborted and I fried the charge controller board somehow); the second is on a replacement charger board.

TimeChargingCell voltageNote
10:580:003.30 VStart, board 1
11:120:143.49 V
11:430:453.68 V
12:191:213.86 V
13:342:364.02 V
14:023:044.04 VCell unplugged; board 1 stopped working
15:393:044.03 VResumed on board 2
16:113:364.08 V
16:414:064.10 VStopped; panel partly shaded by a tree

"Charging" is cumulative time on the charger: the hour and a half the cell sat unplugged is left out.

Cell voltage against cumulative charging time: 3.30 V at the start, 4.04 V after 3 h 04 min when the cell came unplugged, 4.03 V when it went back on a replacement board 1 h 37 min later, then 4.08 V and 4.10 V over a further 62 min3.23.43.63.84.04.20:001:002:003:004:00Time charging (h:mm)Cell voltage (V)board swap, cell unplugged 1 h 37 min10:58 — 3.30 V, 0:00 charging (board 1)11:12 — 3.49 V, 0:14 charging (board 1)11:43 — 3.68 V, 0:45 charging (board 1)12:19 — 3.86 V, 1:21 charging (board 1)13:34 — 4.02 V, 2:36 charging (board 1)14:02 — 4.04 V, 3:04 charging (board 1)16:11 — 4.08 V, 3:36 charging (board 2)16:41 — 4.10 V, 4:06 charging (board 2)15:39 — 4.03 V, 3:04 charging (board 2, resumed)3.30 V4.04 → 4.03 V4.10 V
Cell voltage against time on the charger, 21 September 2026. The dashed line marks the board swap: the cell sat unplugged for 1 h 37 min there and drifted from 4.04V to 4.03V.

The first run gains us 0.74V in 3 h 4 min.

After the swap the second board added 0.06V in 62 min, the curve clearly flattening as it approaches 4.2V.

During the hour and a half it sat unplugged the cell slightly relaxed from 4.04V to 4.03V.

I stopped at 16:41 with the cell at 4.10V. The sun had moved far enough that a tree was starting to shade the panel.

Aborted

Just after the 13:34 reading I moved the panel to get a better spot in the sun. In the process the cell came unplugged from the controller. Plugging everything back in in the sun and I heard high-pitched whine — like an old camera flash charging up.

The charge controller was hot to the touch. Not just the ICs, the whole copper layer. The cell wasn't too hot. However, I'm not really sure I trust the cheap components and a small LiPo incendiary device so I paused.

I did a bit of continuity testing and gingerly plugged in a DC supply, but the charge board doesn't seem to be working - no LEDs on. Let's call it fried. I got an identical replacement read and wired another JST-PH to JST-ZH adapter.

I also noticed the solar panel was hot - as you'd expect. When I resumed later the panel had cooled off from being inside. Perhaps this explains the slight bump in charge rate.

What did I learn?

I think it all works. I think I fried the controller by plugging the cell into the charger in fully lit conditions. I have enough confidence in it that I'll start mounting things into a waterproof enclosure and try a discharging load overnight.

And perhaps I should find some fuses.

AI's explanation

Not confirmed by the human.

Two separate things, both explained by the cell falling off:

The heat is probably normal. A linear charger dissipates (Vin − Vbat) × Icharge in the chip. With the panel at a plausible 6.5 V, the cell at 4.0 V, and 500 mA flowing, that is about 1.2 W in a chip the size of a grain of rice, and the copper pour is the only place for it to go. Moving the panel into stronger sun pushed Vin up and made it hotter at exactly that moment. The chip's thermal regulation should throttle before it damages itself, but the board gets uncomfortable to touch well before that.

The whine is most likely the no-battery oscillation, made audible by the ceramic capacitors. With the cell unplugged, the charger's output is just its own small output capacitor. The chip charges it to 4.2 V in microseconds, sees "battery full", terminates; the cap bleeds down through the status LED and divider; it hits the recharge threshold and starts over. That is a relaxation oscillator, and on a board this small its frequency can land in the audio range. Ceramic capacitors are piezoelectric, so a kHz voltage swing across them makes the board physically sing — the same reason a camera flash whines while its capacitor charges. TP4056 boards do the same thing; the familiar "LEDs flicker with no battery" symptom is the same oscillation at a lower rate.

If that is right, the cell was never in the loop, which fits it staying cool. It does not, on its own, explain why the board died: thermal regulation should have held the die below its limit. Either the oscillation did something the datasheet does not describe, or the panel's open-circuit voltage, with no cell to pull it down, went past the CN3065's 6.5 V absolute maximum — a "5 V" panel's open-circuit voltage is typically 6–7 V.1 TODO: measure the panel's Voc in similar sun.

Test (on the dead board, nothing to lose): cell disconnected, panel in sun, listen. If it whines with no cell, that half of the story is confirmed.

TODO: result of the test.

  1. CN3065 datasheet, Consonance Electronics. ↩

Table of Contents