Reverse Polarity and Low Voltage Protection Intermediate Board

[Boltar]

Hi everyone and thanks for reading my post.

I'm trying to design an intermediary board to place in-between a 3 x series Li-Ion battery array and whatever the batteries are supposed to be powering (in my case an electronic cigarette but it's not limited to that). I could of course build all of this onto the e-cig board itself, but the daughter board I think is better as it could be used on other existing devices that don't have such protection on them. It's quite a high current application, so I'm using big 27A P-Mosfets (because common earth has to be battery negative and I don't want to get into charge-pumps and stuff in order to use N-Channels).

VIN = 8.4V to 12.6V depending on battery charge.

Q3 is configured for reverse polarity protection. As I understand that concept, the internal diode pulls the source to just below VIN (due to the drop of the diode), but that's enough to turn the mosfet on enough which then pulls the source to VIN and the mosfet turns fully on completing the circuit. In reverse the gate is at VIN so the mosfet is off. I've read about this configuration on a lot of sites, so I'm reasonably confident it's correct, but of course I'm only really a beginner so confirmation from experts would be reassuring.

The other circuit is what I'm not too sure about as I've come up with it myself. My theory is that when VCC is above ZD1's BV the bleed through will turn on Q1, syncing the current through R2 to ground and pulling Q2's gate low turning Q2 on and allowing power to VOUT. When the voltage is below ZD1's BV, there's nothing bleeding through it so Q1's base is pulled to ground turning it off, this stops Q2's gate being pulled low and allows it to be pulled up by R2 turning the mosfet off and cutting power to VOUT. I'd appreciate some feedback if I've got that correct or not. Also, I've used very high value resistors in an attempt to minimize the current drain on the battery as the device is normally in a continually on state. My concern is that using such high value resistors might cause the circuit not to work. It works in a simulator but that's not the same as real life, so again, I'd very much appreciate advice on that point too. Or for that matter any other approaches I could use.

Many thanks
Marcus.

[mij59]

Hi,

I think the transit of Q2 from on to off state will be to slow as the input voltage drops.
Also you're measuring the voltage of all 3 cells, its safer to monitor the voltage of the individual cells.

Its better to use a single chips solution, e.g. search for under voltage protection, or protection board.
 
http://www.ebay.nl/itm/Protection-Board-for-3-Packs-12-6V-Li-ion-Lithium-18650-Battery-charger-4A-w-LED-/261193182206?pt=LH_DefaultDomain_0&hash=item3cd053abfe&_uhb=1

[Boltar]

Hi mij thanks for replying.

With regard to the transient time it doesn't really matter too much. The circuit is not really to protect the powered device from low voltages. The purpose is to protect the battery pack from getting too low if the device is left unused for several days. If LiPos discharge too much it's bad news. I want to stop the pack getting much lower than 9V. I see where you are coming from with regard to monitoring each individual cell, I deemed this unneeded as the finished device will have an external port for a balance charger, so unless an individual cell fails they should all be pretty much the same charge at all times. Since under regular use, the battery would need charging every couple of days or so it should be fine to assume each cell is balanced. If any fail, the balance charger would pick it up anyway. Of course that's just my newbie conclusions, please let me know if individual cell monitoring would still be required in light of that new information and if the circuit above would be suitable for that specific purpose.

I had no idea there were dedicated boards out there for this kind of thing until you mentioned it. I did some searching but having trouble finding one with the current capability I need at a reasonable price. The ones capable of higher currents all seem to be boards for up to 10S packs expensive, and a quite large as well, which would also give me issues fitting it, but I'll keep looking, I don't give up after the first couple of searches

Again, thanks for taking the time to help me.
Marcus.

Edit Reason: Typo Correction

[David Hess]

Your circuit idea has merit but the zener diode is not going to be accurate at such a low current level.

Enough hysteresis needs to be added because at low voltage when shutoff occurs, the battery voltage will recover somewhat turning the circuit back on.

[Boltar]

Hi David, thanks for the reply.

I see your point but the draw from the device itself as well as the circuit shown above (bearing in mind the device will be in a low power sleep mode if it's left unattended) will be maybe 50 microamps. I don't think that would equate to much of a voltage drop from the battery but I could be wrong. All it's got to do is prevent power getting to VOUT if the battery level is below a threshold. That threshold doesn't have to be accurate or dead on, anything around 9V is fine. It's just to stop the battery getting too low that's all. Whatever it does at "around" the threshold level is pretty much unimportant. All I need to know is if the circuit will definitely prevent the battery draining to below 9V or there abouts. When you say the zener won't be accurate, is that because of the high resistances I'm using? Would it be significantly inaccurate?

I'm sorry but I had to do some Googling to understand what you meant by Hysteresis. I'm only a beginner. This is kind of an "overlap" like a schmitt trigger? Meaning if the battery drops below say 9V then a latch would trigger and prevent the circuit switching on again unless it rises to like 9.5V (just as an example)? If so then I really don't need it to be that complex.

Thanks you for the help, at least I learned what hysteresis was I tried mucking around with some Schmitt Triggers in a couple of emulators but I just can't get them to work. I have only tried both basic op-amp types, I'll give the dual transistor version a go as well.

Cheers
Marcus.


EDIT: Actually no scratch the idea. I've just realised after you mentioned battery voltages recovering when the load is removed. When the device is actually in operation, it's drawing big current. It is very possible that normal usage with a mid-charged battery could cause a drop to below 9V which would cut the device's power. Making the device only usable with a battery that's fully charged or close to fully charged. So no this is a bust. darn.  I suppose I could have an NPN transistor on the fet gate to ground which is switched on by the main output of the device itself so that when the device is fired, the fet gate is pulled low so it can't turn off.


I'll try to explain what the device is and the protection I need a little better. The device is an e-cig, it sits around for most of the time doing nothing but monitoring the state of a fire button. When the fire button is pressed it powers the atomising element (a heater coil) which produces the vapour. The atomising element can be as low as 0.4 ohms or sometimes lower depending on the voltage it's driven with. It is possible that the device may be left unattended or unused for a few days, since it is "always on" monitoring for user input, it's a constant drain on the battery albeit very small. Should the battery fall to a very low voltage if it's left long enough, the LiPo cells get buggered and cannot be recharged (not without removing the pack and messing with it anyway). All I need is a circuit to prevent that from happening. I don;t need Mr Spock accuracy on the voltages involved.

This is a video of my previous attempt at this kind of thing. Although the one in this video is mains powered. Imagine the same thing but smaller and battery powered.

There's a tiny bit of bad language in the video.

[David Hess]

I see your point but the draw from the device itself as well as the circuit shown above (bearing in mind the device will be in a low power sleep mode if it's left unattended) will be maybe 50 microamps. I don't think that would equate to much of a voltage drop from the battery but I could be wrong. All it's got to do is prevent power getting to VOUT if the battery level is below a threshold. That threshold doesn't have to be accurate or dead on, anything around 9V is fine. It's just to stop the battery getting too low that's all. Whatever it does at "around" the threshold level is pretty much unimportant. All I need to know is if the circuit will definitely prevent the battery draining to below 9V or there abouts. When you say the zener won't be accurate, is that because of the high resistances I'm using? Would it be significantly inaccurate?

Zener diodes do not operate well below their design current.  They become both inaccurate and unreliable.  A typical low voltage zener diode, 12 volts and below, is going to be very unhappy below 250 microamps.  The operating currents for a 9.1 volt zener diode is going to be in the 1 to 20 milliamp range.  Even their leakage before their breakdown is going to be on the order of microamps to 10s of microamps.

I'm sorry but I had to do some Googling to understand what you meant by Hysteresis. I'm only a beginner. This is kind of an "overlap" like a schmitt trigger? Meaning if the battery drops below say 9V then a latch would trigger and prevent the circuit switching on again unless it rises to like 9.5V (just as an example)? If so then I really don't need it to be that complex.

Schmitt triggers are a good example.  The reason to add at least a little bit of hysteresis is to prevent the circuit from oscillating.

EDIT: Actually no scratch the idea. I've just realised after you mentioned battery voltages recovering when the load is removed. When the device is actually in operation, it's drawing big current. It is very possible that normal usage with a mid-charged battery could cause a drop to below 9V which would cut the device's power. Making the device only usable with a battery that's fully charged or close to fully charged. So no this is a bust. darn.  I suppose I could have an NPN transistor on the fet gate to ground which is switched on by the main output of the device itself so that when the device is fired, the fet gate is pulled low so it can't turn off.

So it just needs to latch off when the battery voltage is low.

[Boltar]

I'll have to come up with another way. 1mA even as an absolute minimum, to get reasonable accuracy from the Zener is too much constant drain. Now I know the Zener diode approach is no good. Perhaps I can use a Schmitt Trigger with (say for example) half the VIN (from a voltage divider) into the trigger. And have the magic region at 4.4V to 4.6V ? I'm playing with a simulator at the moment but I cannot seem to get even a simple schmitt trigger to work. I'll keep at it.

The problem with it now, after you reminded me that batteries drop voltage when under load is; imagine the device with a battery at 10V, no problem should work at that, The device is fired and the battery voltage drops to 8.7V due to the high current load. The low voltage protection kicks in and cuts the power to the device. The battery immediately recovers to 10V again and the devices powers back up. In effect what's going to happen is users will complain that the device is resetting itself when the battery is at a lowish but still usable charge. I'd have to setup some kind of bypass, so the low voltage detection isn't active when the device is actually being fired.

Thanks for the tips, I can't see any kind of thanks system in use on the forum so I'll just say it

EDIT: Can't seem to obtain a stable reference voltage from the supply without draining too much current. To be on the safe side and ensure a zener clamp is going to work well enough i'd need say 2mA at 8.5V which is about 3mA at 12.6. With a 800mAh cell of which maybe 2/3 of it will be usable, that's 533, draining with an average of 2.5mA that gives me about 9 days of stand by time. Hmmm, that's not actually too bad.

[Boltar]

Is this along the right lines? I know an LM358 probably isn't the best thing to use. The signal generator on the left is generating a sine wave simulating the battery level rising and falling, the +5V is just obtained from a voltage regulator. It needs fine tuning but as it stands, the output drops low when the battery charge rises above mid way, and goes high when the battery charge falls below half way. I just mess with one of the resistors to adjust that point?

http://s13.postimg.org/ex0ga1mc6/schmitt.jpg

[mij59]

Hi,

Try things like LM285, LM385, TLV3701, MCP6541 to build your circuit.

[Boltar]

Thanks for the help everyone. I still don't really know if anything I'm testing here is actually going to work, but I guess I'll just try it and see. I have a fire extinguisher.

[David Hess]

The TLV3701 when combined with a micropower reference would be a good choice.  Linear technology has similar micropower parts.  An LM358 is going to draw too much power by itself but an LP358 at 85 microamps might be acceptable.

[Boltar]

Well Ive come up with a basic circuit to test and build just to make sure I'm on the right track. A little more complex than my first go It's power usage would be too high as it stands, but I could tweak it. I just need to see something actually working in real life first. Thanks for the tips on parts and stuff. Using those low power voltage references is a good idea as I could use the non-inverting version of the trigger, eliminating the need for the second npn transistor at the end stage.

Cheers all.



EDIT: Lol, only problem here is it's still a continuous drain on the battery, that is what I'm trying to avoid. I want to have the battery drain totally stop when it gets too low. But it's not too bad. I've learned something new and that's always good. Going back to the original circuit here, if I replace R1 with a 4.2K, i'll get 2mA through ZD1, that should make it reliable enough I hope. The device would oscillate on and off at the threshold but that's not really an issue for me, let it oscillate.