Critique my lipo cutout design

[Riotpack]

Hi,

I am designing a portable bluetooth speaker and have finished the lipo battery management. I have attached the schematic to get your views on my design.

https://onedrive.live.com/redir?resid=1AB6590A9FB4749D!16757&authkey=!AO6UNHe3x904PMY&ithint=file%2cms11

[hamdi.tn]

1- why use multi-cells ? i think most bluetooth speakers on the market just use one cell. and i don't think 3 or 4 rechargeable battery will be any cheap.

2- i think there is way too much resistors that will contentiously keep draining current out of the cell, i don't think it will be any good on terms of autonomy and run time.

[janoc]

Well, that's not much of a protection when you have a resistors across the input that will keep slowly draining the cells regardless of whether or not the circuit actually trips. R13 and the voltage dividers using R1 you don't want there.

What's the deal with the R5? That one is just wasting energy.

Also I am not sure why do you need both of those FETs. Only the second one is actually used to turn the power on and off. Is the first one supposed to act as a voltage dropper to bring the 4S battery voltage down to 3S so that you can use the second part of the circuit for both? If yes, that's really terrible, because you are just blowing all that extra energy as heat in the first FET. What's the point of that?

Why such complicated design? Why both 3S and 4S configs? Are you going to have replaceable battery packs with different number of cells in that speaker? Again, why?

I would rather decide on one battery config only (e.g. 3S) and build the safety cut-off using one P-FET and a simple comparator circuit. If the battery voltage drops below the threshold, it will turn off.

[Riotpack]

The cell configuration is 3 cells in series and 6 of these 3S in parallel.

So 11.1v@13200ma/h - When the battery is at 9v (totally safely flat) the current draw is half a milliamp.

so in theory 26400 hours in a sleep state.

The packs power a Tripath 2024 amp.

[ovnr]

500 µA is still rather a lot for a battery protection circuit, and once your battery is already flat from usage, it's going to continue draining it.

I'd strongly suggest replacing the TL431 with a comparator. Linear has some nice dual comparators (so you can easily do a window comparator) with built-in ref.

Also, as has been stated already: There is no point in having the two MOSFETs with the configuration you've chosen (but I'd like to point out that having back-to-back MOSFETs is commonly used to be able to cut off power both ways, as otherwise the body diode would allow current to flow one way all the time).

[hamdi.tn]

i think you can find some good battery protection IC from taxas instrument and you can sample that. This can seriously simplify your design and make it more efficient.

[Riotpack]

I changed the design around a bit, now when the battery is depleted the current draw is 1.7 microamps.


I wanted to make ie using simple easy to get components.

https://onedrive.live.com/redir?resid=1AB6590A9FB4749D!16761&authkey=!AO3qPdmmvOW0BcY&ithint=file%2cms11

[ovnr]

Please post a screenshot. Not everyone has Multisim, and those that do may not be bothered with downloading your file.

[Riotpack]

I have attached an image. The idea was for it to use easy to find parts and be easy to build.
https://onedrive.live.com/redir?resid=1AB6590A9FB4749D!16761&authkey=!AO3qPdmmvOW0BcY&ithint=file%2cms11

[Riotpack]

3rd revision
https://onedrive.live.com/redir?resid=1AB6590A9FB4749D!16762&authkey=!AG5_XuVIHkHPgeI&ithint=file%2cms11

[janoc]

Is this even going to work?

What is going to keep the Q3 (and thus Q1B) energized once you release the ON button? The base of Q3 will be floating when both switches are off. That will make the Q3 stop conducting (no base current) and Q1B will stop conducting as well because of that R3 resistor.

Also why R9 and R10? What is the purpose of those two resistors? First, they could be combined into one, second, they don't do anything but dump about 300uA to ground when the device is on.

I also can't say that I like the design with the TL431CP. That is a voltage reference, but there is no comparator anywhere. All you have is the threshold voltage of Q1A - which varies from part to part, changes with temperature, etc. That doesn't make for repeatable and robust design. There is no hysteresis neither, so your circuit is likely to oscillate right around the threshold voltage (the first FET will NOT turn off, so everything is powered, the battery recovers a bit, the second FET turns load on, the voltage drops again, the load is turned off, lather-rinse-repeat ...)

The calibration with the trimpot - what if someone drops your gizmo (it is meant to be portable, right?) and the pot moves? Boom goes the battery when the thing doesn't turn off ... You don't want this to be adjustable.

Also exceeding that threshold voltage does not ensure that the transistor will be fully conducting (it only starts to conduct - you need a lot more Vgs to open it fully!), so you are likely going to be producing a lot of heat in that FET.

I suggest you take a step back and do a bit of research on how others have implemented these things. I would also completely separate the on/off circuit and the safety cutoff. Their role is different and like this you are only likely to compromise the function of one or the other. Or both. Especially considering that the battery cutoff circuit could be easily part of the battery pack itself, as it is commonly done.

BTW, this topic has been discussed here several times already and the threads did include links to both circuit ideas and parts you may want to use:
https://www.eevblog.com/forum/projects/3-7-lipo-undervoltage-protection/
https://www.eevblog.com/forum/beginners/i-need-to-find-a-good-low-voltage-cutoff-circuit-for-lipo-batteries/
https://www.eevblog.com/forum/beginners/latching-'off'-circuit-lipo-cutoff/
 etc ..

[Riotpack]

If the battery level is sufficient then current flows through R9 and latches Q3 on. The off button pulls the base of Q3 down. R10 pulls the base of Q3 low when the battery is to low to latch.

There is no hysteresis, but the circuit will simply latch off 

[Alex Eisenhut]

If the battery level is sufficient then current flows through R9 and latches Q3 on.

Not without a dot it won't.

[Riotpack]

Fixed. I know it will drift with temp etc, setting it at 3.2v cell and have added a cap across the divider. The idea was to make something using easy to access components. Replacing the pot with a divider made of two 1% low tempco resistors would be ideal, its not something you would find laying around. Secondly, the cells have their own protection, this is just a first line solution.

[Riotpack]

https://onedrive.live.com/redir?resid=1AB6590A9FB4749D!16772&authkey=!AI5IvNJswsuT6vs&ithint=file%2cms14


Version 4 adds fixed voltage divider, capacitor on divider to prevent high load transients tripping the lv cut.

This could be done with a comparator, v ref or a dedicated one chip solution. This was designed to use garden variety through hole components. It works in the simulator, will update when prototype is complete. 

[janoc]

Um, you seem to love adding random components to the circuit. Why the diode now? Just give Q3 base current when the load is turned on through a resistor.

BTW, the cap across the divider doesn't fix the oscillation failure mode I have described. That happens when your battery is drained to near the turn off threshold of your circuit. The circuit turns off, the battery voltage recovers a little when there is no load and the circuit turns on again. The voltage drops below the threshold and the circuit turns off again. Repeat until the battery is sufficiently discharged to not recover anymore ...

The proper solution is either a circuit with hysteresis or a latching setup that will turn off and stay off no matter what happens to the battery until reset somehow (e.g. by pressing the ON button).

Duh, when the circuit turns off it will actually stay off now, because Q3 loses base current once the second FET turns off. That should fix the oscillation issue then.

BTW, a simple way to create a comparator with a hysteresis using only jellybean parts is a Schmitt trigger circuit. 2 transistors and few passives driving a P-FET will likely give you better performance than what you have here with the shunt regulator and what not.

Also the dots should be on the intersection of the wires, not in semi-random places. Like the connection from the base of Q3. Or the dot in the emitter of Q3. Or the odd one between the two switches.  I don't know the software you are using, but I am starting to suspect that the circuit you are actually simulating doesn't actually match the schematics, thanks to the overlapping wires and the dots in odd places.

[Riotpack]

Thank you for your reply.

Your honesty helps immensely. I was trying to make do with parts on hand and using simple through hole. The obvious solution is to do it properly as you have stated.
 
Do it once, do it right. 

[janoc]

Thank you for your reply.

Your honesty helps immensely. I was trying to make do with parts on hand and using simple through hole. The obvious solution is to do it properly as you have stated.
 
Do it once, do it right. 

You are welcome. There is nothing wrong with doing it with jellybean parts you have on hand and through-hole tech. I do the same when building something for myself. However, your circuit was overcomplicated and not working all that well, that's why the comments.