Maker Pro
Maker Pro

LiPo over discharge protection

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
So I have destroyed my first LiPo (£20 down the toilet) while desk testing my Quadcopter. So I decided I need a protection module that will power off when the voltage reaches critical levels - this will be used during desk testing not in flight as then the telemetry will warn me of low battery conditions and force return to home and landing at which point the battery will be disconnected. I destroyed the battery as I got called away, for what was supposed to be a short interrupt, but ended up by me not returning till the next morning to a very dead battery. I hope for circuit that will just switch all of when the critical level is reached.
However I cannot find such a module for sale that can handle 40A for a 3S LiPo 4400 maH and my electronics is nowhere near good enough to design my own circuit. I found some circuits that can handle 8A, but that is way too low.
Any help with this greatly appreciated.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Thanks Steve, but that one is only 1S and 4A discharge. That is the problem, I need 3S pack with 40A discharge. I could just use the module to drive a switch and not drive all power through the module, ignoring its switching capability. If I do this, what would be the best way of implementing the switch; a relay will be quite a drain on the battery, so there must be a better way like using FETs? But I would not know how to wire these.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
To do this properly you should really monitor the voltages on each cell.

Admitedly, this is mostly a cell death issue when charging when you only have a few cells such as the three you're contemplating.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
@Steve: This is not for charging - for that I have a proper cell balancing charger for LiPos. This is for discharge protection when I am bench debugging to ensure the lot get switched off if I forget to unplug the battery. I'm getting(is) old and sometimes forgetfulo_O During flight I have an individual cell monitor that feeds back through telemetry to the transmitter and also automatically sets 'Return To Home'.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
This is a series version of one that we use in our company You choose your own FETs, this might work but you would need an adaptor PCB to solder this because it is surface mount.
Hope this helps
Adam
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Hi Adam,
Sorry the link does not work, it says 'file not found'.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
@Steve: This is not for charging - for that I have a proper cell balancing charger for LiPos.

Yeah, sure, but let me illustrate the problem.

Let's say that you decide that you want to cut off the battery when the cells reach 3V (to give you quite some leeway before they get to 2.8V)

You've used your cell balancing charger and all your cells are at 4.2V. However, one of the cells is slightly weaker than the others.

Your plan is to discharge to 9V and stop there because 3V/cell is very safe.

Ad the battery discharges, you find that the cutoff occurs at precisely 9V, but you are horrified to find that two of the cells read 3.5V and the weak cell reads 2V. You've now damaged it more, and the next time the battery gets down to 9V, its voltage may be less than 2V.

The best thing to do is to stop discharging whenever any battery hits your lower limit. To do that requires monitoring each cell.

The difference in useable capacity of your battery will likely be very small, but it will last a lot longer.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
@Steve,
Thanks for that.
I had a look and it seems like it is doable with a single LM139. The three inputs(one per cell) can plug into the balance plug of the battery pack.
This is the one I am looking to get LM139AJ - Can supply directly off 3S LiPo. Then use a zener(3.3V)+resistor(330R) to set the reference voltages at about 3.3V(- input). Connect each cell to a + input. Tie the corresponding 3 outputs together (logic OR) and through resistor to base of transistor to drive the relay. Connect the normally open side of the relay in the positive from the battery, supplying both the quad and this circuit, with a momentary push button that powers the circuit until the relay closes once the battery is connected.

Does that sound about right. Sorry I don't have a program to draw circuits nicely.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
Yeah, sure, but let me illustrate the problem.

Let's say that you decide that you want to cut off the battery when the cells reach 3V (to give you quite some leeway before they get to 2.8V)

You've used your cell balancing charger and all your cells are at 4.2V. However, one of the cells is slightly weaker than the others.

Your plan is to discharge to 9V and stop there because 3V/cell is very safe.

Ad the battery discharges, you find that the cutoff occurs at precisely 9V, but you are horrified to find that two of the cells read 3.5V and the weak cell reads 2V. You've now damaged it more, and the next time the battery gets down to 9V, its voltage may be less than 2V.

The best thing to do is to stop discharging whenever any battery hits your lower limit. To do that requires monitoring each cell.

The difference in useable capacity of your battery will likely be very small, but it will last a lot longer.

That is correct Steve, you will need a 3 cell battery monitor like the one I have linked to in the post below.
Adam
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Adam, I cannot find any stock anywhere of the S-8233 except TSSOP on reels and even those are out of dtock, so I guess I'm back to my homebrew circuit. I know the low voltage is not settable, but it will do the job for what I need, if the idea will work.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
No check out the website for other parts that might work. Search for the manufacturer on RS, Farnell, Mouser, digikey etc. See if they got anything suitable.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Well it seems I can't do it with the LM139 as it has only one ground reference, so I'll keep looking. As this is only for desk testing, I might end up just looking at overall voltage and just regularly check the cells with my plugin cell tester when they are discharged to see if there are any big variance in the discharge state. I will just raise the shutdown voltage a bit as well. Or I might use 3 discrete 741's to do the job.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
I haven't built the 741 solution yet, but gave it some more thought:

It seems a PIC 12F675 should be able to do the job much more elegantly, however I have never used the ADC on one of these and need some help to understand how to use it.

1. As far as I can see the ADC pins accept a maximum of 5V.
2. If I make a voltage divider with 10K/5.6K the middle connection should not exceed 5V max(drop accross 5.6K resistor=4.5V) when the battery is at 12.6V, so I can connect it to the ADC pin directly.
3. If I do the same between the battery negative and each of the 3 cell balance connectors positive, will the ratio of the readout on the ADC be linear. That is if all the cells are at 4.2V , will the first cell(4.2V) render a readout of about 0x155, the next connector at 8.4V=roughly 0x2aa and the last cell 0x3ff as it is a 10 bit ADC. The numbers are assuming perfect charges, I realise that it will be around those values if my assumption is correct.
4. If 3 is correct and the ADC is linear it makes the check quite simple as the readouts can just be subtracted and the result compared against roughly 285(0x11c) to check for 3.5V
5. For a 3S LiPo it will still leave a GP pin free to switch a relay through a tranny.


Any comments on the above will be appreciated.
 
Last edited:

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Attached my proposed circuit with the 12F675

The minimum voltage is set by adjusting RV1 and measuring the voltage at P2 with a multimeter. The program will read the ADC value to determine the reading for the minimum voltage.
The circuit does not include the details of the relay drivers, maybe it would be better to use MOSFET but I have no experience of setting this up. Bear in mind that the circuit might have to provide 80A at 12.6V. GP3 can be used to sound a buzzer.

If I sort out this circuit and program I will post the code here, hopefully with an artwork to help others as after much searching I have not found a readymade solution that can handle the wattage required.
lipoprotectection-jpg.13180
 

Attachments

  • LiPoProtectection.jpg
    LiPoProtectection.jpg
    71.4 KB · Views: 2,960
Last edited:

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Anybody any comments please.

What I am especially interested in is whether the readings on ADC pins(GP0, GP1 and GP2) will be linear - I suspect so, but have not found any mention of this in the datasheet or application notes. More specifically, will the following work:

SETUP:
1. Adjust VR1 until the reading on the 'Measure min voltage' test pin is the minimum voltage you want to allow.
2. In the program:
a. Take the reading from ADC pin GP4 and store in MINVOLT.
b. Take the reading from GP2 and store. Also compare against MINVOLT and if lower switch off relay.
c. Take reading from GP1, subtract GP2 reading and compare result against MINVOLT; if lower switch off relay.
d. Take reading from GP0, subtract GP1 reading and compare against MINVOLT; if lower switch off relay.
e. Keep doing b to e.

The above assumes that the differences (GP1-GP2), (GP0-GP1) and the reading from GP2 will be the approximately same for the same voltage and comparable against the reading from GP4.
I have to build and setup my PIC environment again to test this, so anybody with experience would be of great help if they would indicate whether the approach is good or whether there are design flaws.
When the relay is shut down the PIC will also get shut down.

Notes: The 7805 will have caps on input and output as per datasheet = this is a simplified circuit diagram. The PIC will decoupling caps on supply.
GP4 will most likely have an LED and piezo buzzer connected.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
The readings on the ADC should be linear within some small tolerance.

If all the voltage dividers are the same then the voltage on each cell will be proportional to the ADC reading for the lowest cell, or the difference between 2 ADC readings for the other cells.

The 7805 should be powered from the entire battery, not from just 2 of the cells..

Your method of calibration might work, however you can probably just convert the ADC count to volts using math. If it's an 8 bit ADC, then each count represents (5/256)* (15.6/5.6) volts at the battery. That's 0.0544V. So a count of 255 would represent 13.87V If your minimum voltage is (say) 3.25V, then you would not want to see a count (or count difference) less than 60 (which is actually 3.26V).

Note that the accuracy will probably be determined mostly by the accuracy of the 5V regulator and the tolerance of the resistors. You might want to use a regulator that is either more accurate, or which can be trimmed to a more exact voltage.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
@Steve:
Thanks for your response, much appreciated. I will now set up my PIC environment again and get a PIC12F675-ICD header for debugging so I can make sure I get the ADC working correctly.
 
Top