diy solar

diy solar

Monitoring pack pressure for safety

I’m concerned enough about the 2-3 reported fires here
Do we know enough about the cause of those fires? We're any of them caused by swelling? I have had cells swell but never experienced a fire. In both cases the swelling was a result of over discharge at a very slow rate because of human error. The only time I noticed a cell get hot was when I connected it incorrectly and the fast discharge made the cell warm to the touch.
 
Do we know enough about the cause of those fires? We're any of them caused by swelling? I have had cells swell but never experienced a fire. In both cases the swelling was a result of over discharge at a very slow rate because of human error. The only time I noticed a cell get hot was when I connected it incorrectly and the fast discharge made the cell warm to the touch.
I’m aware of three fires, with two caused by the same individual.

In one case there was very likely mechanical damage in a cell (from sitting on a board supported by one high grubscrew).

In the second, (by the same individual), his theory is that a shorted cell heated enough to cause neighboring cells to bloat before something ignited (unclear what caused the initial short).

In the third, the battery appears to have been improperly wired.

So there is no evidence that swelling was the primary cause of any fires but rather some indication a shorted cell will cause neighboring cells to swell before if gets hot enough to combust.

Do I was thinking about monitoring pack pressure on the hope it could provide an easy early indicator of a problem (which I may still do) but another member pointed out that just monitoring individual cell temperatures would be more direct.

So I’m now looking for a budget 16-channel temperature monitor.
 
Some of them will burn internally without oxygen.
They do burn with out external supply of oxygen because the electrolyte creates oxygen when it overheats. That is why Halon and CO2 have very little effect. They operate on the theory of smothering the combustion and depriving it of a source of oxygen. What is needed is some effective way to reduce the temperature so further self combustion does not propagate to the rest of the pack. There are industry examples using etumescent materials.
 
Last edited:
PMS monitoring system idea [pressure monitoring system - not to be confused with that same acronym for something entirely different] for individual cells in a battery pack.

Each cell has a small conductive "probe" mounted just above the vent - something like a 1/8-in gap - and another wire connected to the battery terminal that is not common with the case. Sometimes this is the positive terminal and sometimes it is the negative terminal, depending on manufacturer.

Each wire pair is connected to a micro-controller through an opto-isolator. If the pressure inside the cell rises to a level sufficient to bulge the vent and contact the sensing probe, the micro-processor takes the appropriate action and sounds an alarm. This should occur long before the cell actually vents or the internal pressure rises to a dangerous level.

The probe can be as simple as a looped wire end. The current required to operate the opto-isolator can be very low depending on the device used.

In reality, a micro-controller is not even necessary if all that is desired is an alarm or disconnect, which could be done easily with discrete electronics or even just a simple relay.
 
I’m concerned enough about the 2-3 reported fires here on the forum that I’m considering adding a pressure sensor to my 16-cell LiFePO4 battery pack.

I’ve already got my single-line of cells in a 300kgf clamping fixture which I’m setting to ~12psi @ 50% by counting turns on my 8 calibrated springs so adding a pressure sensor should be straightforward and inexpensive: https://www.amazon.com/dp/B07S9YJH8C

I’ll probably just do this and monitor the pressure range I measure under ‘normal’ operation before deciding what I want to do with this pressure signal and how to use it for an extra level of protection.

Has anyone else done anything like this?

Once I know what the max pressure is at 100% SOC, I’m thinking I’ll run the pressure signal into a threshold circuit controlling an alarm and perhaps eventually a fire extinguisher / preventer.

Any thoughts & advice appreciated…
I had similar concern about the BMS failing to operate properly and wanted some direct, mechanical fail-safe in the system.

I added thermo fuses to each cell in series - so that if any cell temp gets above 77C, the signal to charge bus and discharge bus relays gets cut, regardless of the BMS status.

 
I had similar concern about the BMS failing to operate properly and wanted some direct, mechanical fail-safe in the system.

I added thermo fuses to each cell in series - so that if any cell temp gets above 77C, the signal to charge bus and discharge bus relays gets cut, regardless of the BMS status.

Wow, that is an absolutely BRILLIANT idea!!!!

So just to be certain I understand how to do this:

1/ modify BMS harness, connecting a thermo-fuse in-line with each sense wire.

2/ tape thermo-fuse on top of or on side of each corresponding cell.

If any individual cell heats up to a temperature of more than 77C, thermo-fuse blows and BMS disconnects.

This will (brilliantly) halt any external battery discharge but will not stop a cell that is shorted internally or internal to the battery.

But the same thermo-fuses can be used in series to connect to an early warning alarm.

You may not be able to halt a thermal-runaway event but at least you can be made aware something is going wrong as soon as any single cell hear above 77F…

Where and how did you attach your thermo fuses and have you tested them (in a oven, for example)?
 
I’m concerned enough about the 2-3 reported fires here on the forum that I’m considering adding a pressure sensor to my 16-cell LiFePO4 battery pack.

I’ve already got my single-line of cells in a 300kgf clamping fixture which I’m setting to ~12psi @ 50% by counting turns on my 8 calibrated springs so adding a pressure sensor should be straightforward and inexpensive: https://www.amazon.com/dp/B07S9YJH8C

I’ll probably just do this and monitor the pressure range I measure under ‘normal’ operation before deciding what I want to do with this pressure signal and how to use it for an extra level of protection.

Has anyone else done anything like this?

Once I know what the max pressure is at 100% SOC, I’m thinking I’ll run the pressure signal into a threshold circuit controlling an alarm and perhaps eventually a fire extinguisher / preventer.

Any thoughts & advice appreciated…

BIG YES to pressure monitoring. ?(y)(y)??

Minimum of one pressure sensor is decided for my first 280Ah 4S build. It's a "no-brainer" to me after seeing so much tragedy and loss when things go "puff pastry".

As for how to integrate it, and what sensor to use, I personally am still in evaluation phase.

  • Load cell + amplifier
  • Barometer
  • Force sensitive resistor

These three are the ones I am seriously considering. Force sensitive resistor, as you mentioned, is extremely easy to integrate. For that reason alone I might end up using it as well.

However, the force sensitive resistor is not known to me for its linearity, so I would want to verify its performance myself before installing it, out of an abundance of caution. Nonlinear resistance based measurements are one of those things I keep my eye on in that way.

Barometer.

16476-SparkFun_Qwiic_MicroPressure_Sensor-01.jpg


Like having a heavy duty inflatable whoopee cushion instead of the springs. Rigid outer walls, two heavy duty air bladders, and the pack in between. Fill the bladders to specified PSI at specified SOC and cap them and monitor pressure. Patch holes if leak form, with epoxy or other appropriate sealant. This sensor seen above supports up to 25PSI and reads out 24-bit values which is A LOT OF NUANCE! Alas, haven't found a satisfactory airtight heavy duty whoopee cushion to do this yet. But it would work. I've used that sensor in other remote sensing projects. Good stuff.

Load cell + amplifier.
13331-01.jpg
13879-SparkFun_Load_Cell_Amplifier_-_HX711-01.jpg


load cell https://www.sparkfun.com/products/13331
load cell amplifier 24 bit https://www.sparkfun.com/products/13879

These are what I'm planning on using, because simply do not feel that confident in force sensitive resistor linearity long term for my own needs. However, integrating them is a little bit more annoying than FSR.

It'll be expensive to include 8 load cell sensors and amplifiers, but it's my first pack and I need raw data to get away from all the anecdotes.

1633124265644.png1633124849061.png
Four load cell sensors mounted on an exterior plate which is ridigly fixed. (Black plus knobs) Then a rigid separator (Gray line) and the battery (blue box with nub on top).

Like this, on each side of the pack. Intending to monitor all 8 in realtime and log time averaged values. Very open to other ideas. Feel good about this template design one for my own use.

Hope this helps. Pressure monitoring is a big yes from me. (y)✅ if anyone sees flaws in these assessments, please let me know!
 
Last edited:
Also +1 to thermo-fuses.

To me, pressure monitoring is two utilities in one.
  1. Provide alarm for emergency shutdown
  2. Provide diagnostic data for cell health assessment
Not everyone in this thread will want #2 so I just wanted to mention it because it's extremely valuable to me.
 
Wow, that is an absolutely BRILLIANT idea!!!!

So just to be certain I understand how to do this:

1/ modify BMS harness, connecting a thermo-fuse in-line with each sense wire.

2/ tape thermo-fuse on top of or on side of each corresponding cell.

If any individual cell heats up to a temperature of more than 77C, thermo-fuse blows and BMS disconnects.

This will (brilliantly) halt any external battery discharge but will not stop a cell that is shorted internally or internal to the battery.

But the same thermo-fuses can be used in series to connect to an early warning alarm.

You may not be able to halt a thermal-runaway event but at least you can be made aware something is going wrong as soon as any single cell hear above 77F…

Where and how did you attach your thermo fuses and have you tested them (in a oven, for example)?
My BMS is does not have ahigh powered internal FETs for LVC and HVC, only low current internal optocouplers for signal relays. Like the Chargery BMS or the Orions, it relies on external relays for LVC and HVC. The relays typically require a 'high' signal to stay on. So the thermal fuses are in series with BMS's internal signal relay which in turn feeds the 'high' signal to the external relay.

To my mind, thermal fuse on a sense wire does not really help you. You are still counting on the BMS to function properly to disconnect the battery from charge and discharge. Personally, I'm mostly concerned about over charging causing a fire or a bad connection causing cell heating.

An over heat audible alarm would be a good idea to implement. I might do that.
 
BIG YES to pressure monitoring. ?(y)(y)??

Minimum of one pressure sensor is decided for my first 280Ah 4S build. It's a "no-brainer" to me after seeing so much tragedy and loss when things go "puff pastry".

As for how to integrate it, and what sensor to use, I personally am still in evaluation phase.

  • Load cell + amplifier
  • Barometer
  • Force sensitive resistor

These three are the ones I am seriously considering. Force sensitive resistor, as you mentioned, is extremely easy to integrate. For that reason alone I might end up using it as well.

However, the force sensitive resistor is not known to me for its linearity, so I would want to verify its performance myself before installing it, out of an abundance of caution. Nonlinear resistance based measurements are one of those things I keep my eye on in that way.

Barometer.

16476-SparkFun_Qwiic_MicroPressure_Sensor-01.jpg


Like having a heavy duty inflatable whoopee cushion instead of the springs. Rigid outer walls, two heavy duty air bladders, and the pack in between. Fill the bladders to specified PSI at specified SOC and cap them and monitor pressure. Patch holes if leak form, with epoxy or other appropriate sealant. This sensor seen above supports up to 25PSI and reads out 24-bit values which is A LOT OF NUANCE! Alas, haven't found a satisfactory airtight heavy duty whoopee cushion to do this yet. But it would work. I've used that sensor in other remote sensing projects. Good stuff.

Load cell + amplifier.
13331-01.jpg
13879-SparkFun_Load_Cell_Amplifier_-_HX711-01.jpg


load cell https://www.sparkfun.com/products/13331
load cell amplifier 24 bit https://www.sparkfun.com/products/13879

These are what I'm planning on using, because simply do not feel that confident in force sensitive resistor linearity long term for my own needs. However, integrating them is a little bit more annoying than FSR.

It'll be expensive to include 8 load cell sensors and amplifiers, but it's my first pack and I need raw data to get away from all the anecdotes.

View attachment 67141View attachment 67142
Four load cell sensors mounted on an exterior plate which is ridigly fixed. (Black plus knobs) Then a rigid separator (Gray line) and the battery (blue box with nub on top).

Like this, on each side of the pack. Intending to monitor all 8 in realtime and log time averaged values. Very open to other ideas. Feel good about this template design one for my own use.

Hope this helps. Pressure monitoring is a big yes from me. (y)✅ if anyone sees flaws in these assessments, please let me know!
I’m going to let you take the lead in this (and perhaps also take over the thread?).

As I looked more into that square FSR, I got concerned about the ‘void’ I’d need to create to hide the raised portion at the end into my wooden end piece.

First, it’s be a bit of a hassle to route that out and second, I don’t like the idea of a void in the pressure surface - what if the sidewall of the outermost cell can develop a hernia?

So upon more research, I settled on this as a better FSR design for our needs: https://www.amazon.com/gp/aw/d/B098P2MFYL/ref=sspa_mw_detail_0?ie=UTF8&psc=1&smid=A2YPMGYKJIIDP8

At over 6” long, this could extend right off the end of my wooden endpiece (so no routing / void).

Then I realized tha each of these FSRs are sensitive to only a specific range of pressure (0.1 to 10Kg for this one).

So it’s more of a hairball than I have the time for right now (and the Thermo-resistor idea seems easier and more direct).

Ideally,I would have liked a donut-shaped FSR that would have fit around my threaded rod between two washers between my spring and bolt-end - that would have made this easy-peazy for me.

Best of luck in your quest ;).
 
Last edited:
Thanks for the good quest wish! :D

Interesting, thanks for mentioning the linear ones. Good to know about.

1633128458302.png

Using two flat and rigid plates as a way to prevent the "void" from causing falsely low readings is how I plan to go.

As a random aside, this fundamental design is used in some "Dance Dance Revolution" game pads. The ones I'm talking about use force sensitive resistors as a way of determining when the player has stomped their foot down on one of four arrows to the beat of the music. A flat sheet of metal is on the top, and the foot presses it down onto the serpentine array, completing the circuit, lowering the resistance. This works because air is a poor electrical conductor. Over time, the lateral friction can accelerate the degradation of the sensor by slowly rubbing off the metal, which results in a lower baseline resistance reading. Wiping off the metal dust from the serpentine array and metal plate is a common way of remediating this sensor issue with those game pad pressure sensors.

1633128775491.png1633128845220.png
iu


iu


How I feel :LOL:

hope this helps, not trying to preach, just bumbling through the dark hitting my shins along the way lol
 
1633130882686.png
These two modes are easily distinguishable. Time vs Pressure (assuming continuous charge/discharge cycling)

Hopefully also really easy to distinguish from this mode on a data trace:

1633130974571.png
 
Thanks for the good quest wish! :D

Interesting, thanks for mentioning the linear ones. Good to know about.

View attachment 67148

Using two flat and rigid plates as a way to prevent the "void" from causing falsely low readings is how I plan to go.

As a random aside, this fundamental design is used in some "Dance Dance Revolution" game pads. The ones I'm talking about use force sensitive resistors as a way of determining when the player has stomped their foot down on one of four arrows to the beat of the music. A flat sheet of metal is on the top, and the foot presses it down onto the serpentine array, completing the circuit, lowering the resistance. This works because air is a poor electrical conductor. Over time, the lateral friction can accelerate the degradation of the sensor by slowly rubbing off the metal, which results in a lower baseline resistance reading. Wiping off the metal dust from the serpentine array and metal plate is a common way of remediating this sensor issue with those game pad pressure sensors.

View attachment 67149View attachment 67154
iu


iu


How I feel :LOL:

hope this helps, not trying to preach, just bumbling through the dark hitting my shins along the way lol
What pressure range are you aiming for?

That long strip FSR I linked to supports 0.1 to 10Kg or 0.22 to 22psi.

So in principle, if the ‘plates’ it is pressed between are flat / uniform, it’s actually pretty well-suited to detect the 6 to 18psi range our clamping fixtures are supposed to be designed for (though an ‘over pressure range of only 120% of max design pressure is a bit tight…).
 
need to double check, but 10PSI nominal 12PSI max (i.e. "do not exceed") is what's in my head.

this would be dialed in at low-ish SOC from what i gather, because that's when pressure is lowest, from what i gather. again, my understanding of the exact details are lacking and i hope to fill it in with data.

e.g. if it's 10PSI at 20%SOC and it exceeds 12PSI at "full SOC" (80,90,100%) then I would bleed air until it's below 11PSI at "Full SOC" (this only applies to the airbag method)

as always, i will check the datasheet of the specific cell and do a calculation between kilograms force and the square millimeters area face of the surface being applied to etc..
 
need to double check, but 10PSI nominal 12PSI max (i.e. "do not exceed") is what's in my head.

this would be dialed in at low-ish SOC from what i gather, because that's when pressure is lowest, from what i gather. again, my understanding of the exact details are lacking and i hope to fill it in with data.

e.g. if it's 10PSI at 20%SOC and it exceeds 12PSI at "full SOC" (80,90,100%) then I would bleed air until it's below 11PSI at "Full SOC" (this only applies to the airbag method)

as always, i will check the datasheet of the specific cell and do a calculation between kilograms force and the square millimeters area face of the surface being applied to etc..
At least with EVE, the spec is 12psi nominal (meaning near ~50% SOC or at least in the ‘flats’ well away from either knee) not to decrease below 6psi minimum (near 0% SIC or whatever minimum SOC is being used) and not to exceed 18psi maximum (near 100% SOC or whatever maximum SOC is being used).

With a long-enough spring, it’s possible to maintain close to 12psi across the entire charge/discharge curve, but most of us have mechanical constraints requiring no more than 2” or 3” length for springs beyond the length of our cells/fixtures…

I suspect that designing for a range of 10-12 psi would translate to a pretty long spring.
 
At least with EVE, the spec is 12psi nominal (meaning near ~50% SOC or at least in the ‘flats’ well away from either knee) not to decrease below 6psi minimum (near 0% SIC or whatever minimum SOC is being used) and not to exceed 18psi maximum (near 100% SOC or whatever maximum SOC is being used).

With a long-enough spring, it’s possible to maintain close to 12psi across the entire charge/discharge curve, but most of us have mechanical constraints requiring no more than 2” or 3” length for springs beyond the length of our cells/fixtures…

I suspect that designing for a range of 10-12 psi would translate to a pretty long spring.
thank you for these reference points.

i’ll have to write a python script to graph this out to get a better feel for the specifics, when construction is imminent
 
due to a number of factors, availability and ease of engineering mainly, i have decided to go with an S type strain gauge 200kgf, four units.

it has one M12 attachment on each side. instead of using a threaded rod to compress pack, using four of these with threaded rod on each side.
that way, it's as if there's just four threaded rods holding the pack together between two metal plates (and some insulation..)
but each threaded rod can report the tension on itself.
read by load cell amplifier

for assembly i would observe the four readings and tighten until readings match calibrated target.
like airplane orientation readout.

14282-03.jpg
14282-02.jpg



so there'd be four total. upper left, upper right, lower left, lower right.

each one has 200kgf full scale capacity. due to the geometry splitting the load and 150% load capacity rating i feel pretty decent about it being able to hold.

got four 300Ah cells with threaded studs pre-laser welded. planning on relying on oshcut to get some stainless steel sheets laser cut and maybe bent to create the battery case.

it's gotta happen! for science.

adding this diagnostic feature will increase the BOM by about 300usd for four load cells and four load cell amplifiers. this is for research and data collection. however.. it absolutely will be able to detect a very clear signal if the cells were to swell in a failure mode.

edit: this is the work in progress CAD model
1635100089324.png
 
Last edited:
I like the strategy of putting cells in an outside bunker with inverter. I had a safe room installed in corner of basement when we built house so that is probably good for now. All concrete 6'x16' that I put a 2" vent pipe onto. Maybe phase II we put the batteries in a concrete outhouse with the inverter

In the event of a shorted cell I don't know what can be done to prevent fire. I think that parallel arrays of cells should be fused with a fast acting fuse. A fuse in each string would prevent the whole pack from delivering energy to the one failed cell. This might go a long way towards preventing a fire. In any case, I wouldn't put multiple strings together without fusing each string. I worked at secureplane which burnt to the ground when a battery went critical. Google secureplane fire.

Yes, some of these chemistries have electrolytes that produce oxygen. My boss told me to design a battery pack for a MK48 torpedo. Pack is in a can pressurized for very high pressure as it travels underwater at some secret depth :) He told me to look at the cells the RC airplane guys are using because they pack a lot of energy. I found the cobalt technology produces oxygen while burning. I told boss we couldn't use them. Would be putting a bomb inside a torpedo inside of a submarine. Went with an alternate chemistry.
 
wow.. you are overdoing it. Totally waste of money. It is your money tho. W/e makes you happy go for it
 
Back
Top