Magic Smoke Escaping In 2 Cell Lipo Project

[waymond91]

Hello!
I am working on a low-power esp32 camera project that is supposed to fit in a gopro case.
It uses 2 18650 batteries for power. It also has a mosfet and real time clock that can be used to put the device in an ultra low power state until an interrupt occurs from the rtc.

However, in an effort to keep the footprint small I am not using a separate coin cell battery, and rather I am using the power from the first cell of the battery pack to suply power to thep backup pin of the RTC.

You can take a high level look at the schematic below:



And here you can see the rtc power schema:



Here is the sheet for INA219 "PowerMonitor" portion of the circuit:



Here you can see the physical layout of the board for context:




Here is the thing, if I use my power supply or a 2s lipo to power the device WITHOUT hooking up the RTC backup voltage (ie, the leaving the voltage from the lower battery cell unconnected) everything seems more or less ok. 

If I plugin both 18650 cells as intended (with the voltage of the lower battery cell connected to the battery backup line) the both batteries start to smoke and get very hot.

The backup voltage goes through a diode and a small resistor before the rtc backup pin which I thought seemed reasonable.
When I measure for a short between any of the battery positives and ground I dont find anything suspicious.

Any ideas people? Thanks in advance!

[BILLPOD]

Good Morning Waymond91, I only see clips for TWO 18650s on your circuit board.   Where is this 'middle cell' you speak of?

[waymond91]

Sorry for the confusion! It is only a 2 cell pack, wired in series. The "lower'" cell ( Gnd -> 4V ) is being used to provide back up for the RTC.
You can see on the top level schematic the wire going into "VBACK" in the "PowerTimer" hierarchical sheet.

[NiHaoMike]

Those clips have been the source of product recalls before due to the sides having burrs that cut through the cell insulation. Use a different clip/holder design or deburr the clips and add heatshrink over the sides for added protection.
Additionally, since nothing in the design needs more than 3.3V, have you considered having the 2 cells in parallel with a LDO dropping it down to 3.3V? (The voltage over much of the discharge cycle is quite close to 3.3V so not a whole lot to gain using a switcher.)

[tooki]

You can take a high level look at the schematic below - each of the "battery" schematic symbols correspond with a single 18650 clip

Not that I in any way recommend abusing schematic symbols in that way — you should draw a proper footprint so that you can use the schematic symbol properly — no matter what symbols you use, have the good sense to draw the schematic in a way that it’s intuitively obvious what you’re doing. Things that are in series should be drawn in series — i.e. in a line. Especially when you’re already adding complexity by using a symbol in a nonstandard way, you should make every effort to make their context crystal clear. Not just for others, for you, too!

[tooki]

Sorry for the confusion! It is only a 2 cell pack, wired in series. The "lower'" cell ( Gnd -> 4V ) is being used to provide back up for the RTC.
You can see on the top level schematic the wire going into "VBACK" in the "PowerTimer" hierarchical sheet.

Even with this “clarification” the description of the problem makes no sense. Please go back and write a clear, unambiguous description using standard terminology. Not “middle cell” when there is none.

[thm_w]

You left out the PowerMonitor portion of the schematic. We don't know what that does.

As NiHaoMike says, use a different holder and consider switching to just a single cell or two in parallel. There is no need for the higher voltage here that I see.

[waymond91]

Hello all,
I am sorry the post was so incoherent. I updated the original post at the top to show all parts of the corrected schematic.
I also took another crack at describing the problem.
Thank you again for the feedback and please let me know if it is still unclear.

[waymond91]

Thats really useful about the clips!
I have inspected mine and added heatshrink around the cells and I have convinced myself that the clip shorting through the insulation of the 18650 is not the problem.
Unfortunately, I do not have a lot of battery options, as I am in South Africa and larger mounting options quickly exceed my space envelope.

With regards to the making the whole pack 1S2P, I have no qualms with this - I did not consider a linear regulator but now it seems like a good solution.

Still, it would be valuable to get a few test results from this design if possible.

[kevin.gibbs]

Most likely, only one battery is getting hot. You need to look for a short circuit on the board. Also, transistor Q1 may never close. To close it, you need to pull the gate of the transistor to BAT_IN, not half that voltage. Also, the transistor will open when connected to the Hold signal since the ESP32 is powered by 3.3V.

[kevin.gibbs]

You left out the PowerMonitor portion of the schematic. We don't know what that does.

As NiHaoMike says, use a different holder and consider switching to just a single cell or two in parallel. There is no need for the higher voltage here that I see.

Unfortunately, the K7803MT stabilizer will not work from a single supply cell - it needs at least 4.75V.

[kevin.gibbs]

 It is not clear how the power supply is organized. Batteries are not shown in your diagram in an informative way. You can sign the first battery and the second battery. Also, sign where the "+" and "-" terminals are located.

[waymond91]

What is unclear about the battery layout?
Both are labeled as batteries and tagged with "bt1" and "bt2" on the schematic.

[thm_w]

You left out the PowerMonitor portion of the schematic. We don't know what that does.

As NiHaoMike says, use a different holder and consider switching to just a single cell or two in parallel. There is no need for the higher voltage here that I see.

Unfortunately, the K7803MT stabilizer will not work from a single supply cell - it needs at least 4.75V.

But why do you even need it? You are not using the isolation since you are tying the output to gnd.
We are saying to just run everything from the single cell so there is no need to regulate to 3.3V

But if you did have some need for a regulated voltage, you could switch to a good 3.0V LDO, or maybe 3.3V if you don't care about full battery usage.

[waymond91]

Yes, I am onboard with using a 3.3v regulator moving forward.
I believe I found a nice one with low quiscient current, reverse polarity protection, etc
However, I was still hoping to debug and get a test result from this design.

[tooki]

Thats really useful about the clips!
I have inspected mine and added heatshrink around the cells and I have convinced myself that the clip shorting through the insulation of the 18650 is not the problem.
Unfortunately, I do not have a lot of battery options, as I am in South Africa and larger mounting options quickly exceed my space envelope.

With regards to the making the whole pack 1S2P, I have no qualms with this - I did not consider a linear regulator but now it seems like a good solution.

Still, it would be valuable to get a few test results from this design if possible.

For what it’s worth, since your current problem is one of heating, I urge you to put some insulation on the positive clips that is completely resistant to melting or softening — not just heat shrink. A layer of Kapton tape would be ideal, but paper, 100% cotton fabric, or rubber would work.

Having experienced a short through this kind of clips (in my case, a pinhole that was enough to short), I have to emphasize how quickly this can become ferocious. Granted, in my case I had three cells in parallel, tripling the fault current, but as soon as heating began through the pinhole, it melted the insulation on the cell, and the clips’ own force then melted right through the insulation, creating a larger contact area, causing more heating. The thing is that between inserting the battery with the pinhole insulation fault, and having the clip glowing ferocious red hot, only about two or three seconds elapsed. By the time I levered the battery out — which was only a few seconds later since I had a screwdriver within half an arm’s reach — onto the floor, the battery was spewing its electrolyte, which luckily did not catch fire.

I can’t emphasize enough how quickly these batteries can heat up in a short circuit situation, instantly becoming a hazard. So insulate those clips, and keep a bucket of sand at hand.

[waymond91]

Thank you so much for the help everyone! As it turns out, you guys were all right and I was wrong. I didnt put enough clearance between the clips and the clips would make contact when I actually put a battery in them 

At least I can test from here...

But I am updating the design with considerations in this thread and we will repost once I have the schematic updates laid out!

[waymond91]

Ok, I am working on a design update for this using a normal linear regulator.

However, this means I can no longer use diodes to protect the circuit & batteries from reverse current because it is too big a voltage drop for a a 3.3V ldo regulator.

Im sure Im supposed to just a mosfet there now - any recommendations for parts / how to lay this out for multiple lipo cells in parallel?

[thm_w]

http://blog.deconinck.info/post/2017/12/22/18650-Battery-charger-reverse-polarity-protection

Whatever logic level PFET is appropriate for your circuit current.
If you have multiple in parallel then there can still be a short circuit between the cells, so one reverse protection circuit per cell, assuming you want to do that. Or a fuse on each cell.

[tooki]

However, this means I can no longer use diodes to protect the circuit & batteries from reverse current because it is too big a voltage drop for a a 3.3V ldo regulator.

Are you sure? You can get LDOs with wide input ranges. You just need to provide enough cooling. What kind of drop are you envisioning?!?

[thm_w]

However, this means I can no longer use diodes to protect the circuit & batteries from reverse current because it is too big a voltage drop for a a 3.3V ldo regulator.

Are you sure? You can get LDOs with wide input ranges. You just need to provide enough cooling. What kind of drop are you envisioning?!?

We are talking single cell voltage now.
0.2V for a good schottky, 0.2V for a good LDO brings us to 3.7V which is already compromising the cell capacity.

I like to use 3.0V regulator for more headroom but it depends on the design.

[NiHaoMike]

Keep it simple, just use a PTC fuse to limit the current.

[waymond91]

Thank you so much all!
Here is the design Ive come up with:




I suppose its far more important to have the respective battery cells at a comparable voltage now that the diodes have been removed?
Im pretty sure I dont have to connect the feedback pin of the regulator since its a fixed voltage?

[tooki]

Thank you so much all!
Here is the design Ive come up with:

Wait, so now you’re doing just two cells in parallel? Bear in mind that the discharge curve varies by cell model, but that with many lithium ion batteries, a useful part of the capacity can be at 3.0-3.5V. So with your LDO’s 122mV dropout (which is admittedly excellent), your minimum input voltage is 3.422V. So any capacity that requires discharging below that is basically lost. (Or you tolerate the LDO’s output voltage falling below 3.3V, which is perfectly acceptable to a point. As best I can tell, the ESP32-CAM should function down to 3.0V at least, possibly 2.8V.)

Im pretty sure I dont have to connect the feedback pin of the regulator since its a fixed voltage?

I’m confident you can find the answer to that question by yourself. Reading the datasheets to parts you intend to use is a kinda normal thing to do.

[waymond91]

Sorry Im struggling with the quote system a bit.

With regards to the two cells in parallel:

Yeah the switching regulator does seem to add more complexity than it is worth.
Also keeping the cells in parallel make potential shorts between clips less dangerous since they are closer to the same voltage?
Based off the chart on this webiste, Im under the impression that I really wont be losing too much power due to the dropout of the regulator:
https://blog.ampow.com/lipo-voltage-chart/
Unfortunately, since Im in South Africa and Seyhelles most of the time, I often have to "make do" with whatever batteries I can find.
In this case Im using either:
https://www.robotics.org.za/INR18650-35E?search=18650
OR
https://www.clownshobbies.co.za/webshop/1s-lipo/1494-ky101-drone-battery.html

Im assuming this setup is OK as long as the lipo cells are relatively close in charge voltage when I start the device?
With regards to the feedback pin:
Sorry - will check again - just got lazy while posting and thought I should be posing more direct questions