Teardown pics of recalled Sunvilla LED-lit patio umbrella battery/solar charger

[canuckcam]

Hello, I thought I'd post this here before it gets returned. I got a notice that my patio umbrella has a recall on it, specifically the power module being a fire hazard and has caused at least 5 known fires.

The 5.5mm barrel is both output (there's a switch on the side of the umbrella) and also 5vDC charging input. There's two ICs, but the board is has a splash of epoxy so I can't see the markings. I don't have the skill to tell why this is "dangerous" other than temperature-related failure from being baked in the sun, or the circuit potentially not stopping charging? Having the battery terminals so close together on the PCB without protection circuitry on the 18650s is already questionable.

Thanks!



Product recall page: https://sunvilla.com/pages/recall
"... the umbrellas contain solar panels with lithium-ion batteries that may overcharge, posing a fire and burn hazard."
And according to Costco, there's already been 6 known fires.
https://customerservice.costco.ca/app/answers/detail/a_id/10952

[NiHaoMike]

Most likely the combination of high temperature and questionable quality cells is to blame. I wonder if the redesign, in addition to using better quality cells, would put them in the post where they'll stay cool.

[thm_w]

The positive of the battery pack goes directly to the barrel jack. The negative appears to go to U1 which switches it on and off to barrel jack negative.
This could be a "DW01" type chip that protects the cells from being over discharged.

The solar panel goes to a diode, then resistor, then U2, which is likely a li-ion charge IC (TP4056 type). R2 sets the charge current.

The only major issue with the circuit is U1 might not protect from overcharging. Even on cheap $1 USB charge boards generally have done this correctly.
The max the panel would put out is ~5V, so if U2 was corroded/failed/wrong part then the cell could be overcharged.

[tunk]

Maybe this Big Clive video is relevant: https://youtu.be/M88e1r8nvYk

[bdunham7]

I don't have the skill to tell why this is "dangerous" other than temperature-related failure from being baked in the sun,

I'm not sure why anyone thinks being baked in the sun wouldn't be enough unless they live in Iceland or something.  Today here in sunny SoCal it is 101F (~39C) and my microinverters are at about 150F (~66C).  I see no reason that a sunbrella battery would be any cooler and I don't think Li-ion batteries do well at 66C.  In fact if they were charged at a much lower temperature, heating them up that much would likely put them significantly overvoltage.  NiMH batteries work in applications like this but they have short but fire-free lives.

[thm_w]

I'm not sure why anyone thinks being baked in the sun wouldn't be enough unless they live in Iceland or something.  Today here in sunny SoCal it is 101F (~39C) and my microinverters are at about 150F (~66C).  I see no reason that a sunbrella battery would be any cooler and I don't think Li-ion batteries do well at 66C.  In fact if they were charged at a much lower temperature, heating them up that much would likely put them significantly overvoltage.  NiMH batteries work in applications like this but they have short but fire-free lives.

Increasing the cell temperature by even 50C shouldn't have a significant effect on the voltage
https://www.electronicdesign.com/technologies/test-measurement/article/21808722/temperature-impact-on-ocv-of-lithiumion-cells

Increasing temperature will rapidly reduce capacity but will not cause it to explode unless we are talking 130C+
https://www.sciencedirect.com/science/article/pii/S1002007118307536

[NiHaoMike]

I think they intended for the heat exposure to intentionally cause the cells to fail so you'll buy a replacement battery module (the fact it's replaceable is a "feature"), turns out they overdid it and getting not so graceful failure modes.

If I were to redesign it, I would make the solar panel mounted by screws (so it can be replaced but no reason it needs to be easily removable) and put the cells in a module that inserts into the post.

[thm_w]

I can guarantee you no one thought "how can we design this to self destruct so we can get a 1 star rating on costco reviews and no one will buy our product again". They just thought, whats the cheapest possible way to build this product. All dirt cheap solar garden lights are built in this way.

Having the cells in the metal post is better, but more expensive to manufacture.

[jwet]

I have one of these beauties.  My wife really loves the thing.  Its a pretty nice design- poor execution and maybe not feasible having batteries exposed to the Carolina sun 12 hours a day but I still like it.  Each arm on the umbrella frame is lined with small led's and gives off a nice easy light after dark- romantic.   It took my wife a while to find this and she's pretty bummed.  I'm considering redesigning it or fixing it.  There is a plug at the bottom of the umbrella next to the crank for an AC adapter where I can just put some reliable DC source in.  If anyone makes any real progress, please post.  It would make a good big Clive youtube.

[jwet]

In honor of Bigclive, I decided to draw a schematic of this little thing.  See below.  Its pretty simple and nearly the standard circuit for the parts involved.  There are two 6 pin SOT-23's.  One is a dual low Ron (25 m-ohm) N-FET 8205 that is used on the low side of the cell for standard protection for OV, UV and overcurrent.  The other chip is a 312F made by Fortune and others that is the controller that looks at inputs and outputs and drives the dual low side FET to protect the cells.  The solar cell puts out about 6v at 35 mA in good sun.  The battery pack is 2x 2000 mAHr Li Ion cells in parallel.  The output is a standard power coax connector that supplies the charged battery voltage to the switch and LED's in the umbrella stalk.  I won't post the data sheets for the IC's, google the base part numbers and you'll see them easily.  The schematic is show below.  The only squirellyness that I see maybe is the connection of the FET's is a bit unorthodox.  Usually this is show with the two source terminals reversed- it doesn't really matter I guess- this might have simplified layout.  The other slightly strange thing is that the CS resistor is on the source side (solar cell).  I'm going to play with the circuit on the bench some.  If anyone has any brilliant ideas, bring them on.  My plan at this point is to use the solar cell and replace the LiIons with NiMH's- (probably 4) with a simple constant current charger with an undervoltage cutout.  Probably a voltage detector and a transistor- I'll post what I come up with.  NiMH are used in most of these little outdoor solar lights with little more than a resistor as the charger and don't seem to spontaneously combust.  Stay tuned.  Like I said above, my wife has grown attached to this little thing.

[jwet]

SOLVED!


The circuit as designed will overcharge the battery by about 200 mV in the solar mode.  Diode D2 is supplying the Vcc to the 312F controller chip but unfortunately is also the sense terminal for the battery voltage.  The RC is fine and is specified on the data sheet.  Diode D2 drops about 200 mV (schottky) and throws everything off allowing the battery to reach 4.4.  I removed (shorted) this diode and the circuit works as I would expect.  The battery can also be charged from a circuit in the stalk with an AC adapter, I haven't analyzed that.  It would have to repeat the battery protection circuit or have a real charger.  The protection circuit should have been placed downstream of the battery rather than between it and the solar cell.  This way it could protect the battery in both modes with a simple 5v adapter backfeeding charging voltage from the stalk.  I'm going to do a bit more study but will likely implement this change.

[RJSV]

jwet:
   You can research a 4-pin controller IC, used in many of the simpler garden lights:
   Try:. YX8018 4-Pin IC,
    and:. YX8050
   The yard lights can be placed in 'Totem Pole' relation, for having a later switch-on...the top light will hold off, the lower (Umbrella if you want), so it won't turn on until 10 pm, or whenever that first, top light goes dark, from discharge.
   For lower power strings, running on NIMHD type AAA batteries, you could always skip the power wart option and put a few of those singular lights out, nearby.  At least that gets you out of the risks of Lith ion batts.
RJ
P.s. photo shows flip flop assembly, triggered by flashlight.