Overdischarge protection for SLABs, help choosing switch and making circuit

Thread Starter

-live wire-

Joined Dec 22, 2017
912
I am going to be getting some SLABs to use in various portable, high power applications. This includes RC stuff, portable heating/cooling, and other projects. The SLABs are 12V 35Ah. I may discharge them at 100A in certain applications. While that may not be ideal, they should be able to handle it. They may be configured in series or parallel (still no more than 100A will be drawn). That means 12V and 24V. The batteries have an internal resistance of 10mOhms, meaning I can only really have a few milliohms of resistance in my protection circuits and wiring.

I need a circuit breaker in case of a short. It would be very dangerous if there was not one. Additionally, I need overdischarge protection to avoid damaging the battery. For overcurrent protection, there are many commercial solutions that are economical. But for overdischarge protection, I could not really find anything that can handle the current and is economical. Ideally, a board would integrate both protections and allow for 12/24V operation. But that is not really available. So I think I will have to go with a DIY solution for overdischarge. Given how catastrophic shorts can be, I will go with this reliable commercial breaker. Having two switches is certainly undesirable due to added resistance and cost, but seems like the only option here.

So here is my idea. I could use help finding the parts for certain things. I could also use feedback on it. This is a rough schematic.
WIN_20180614_15_25_38_Pro.jpg
I would first have the breaker. Then there is a linear reference of 6V. It would use some 27uF filtering caps, which may be overkill but should not hurt. But it can vary from 5-6V, based on a trimmer/divider. This represents an undervoltage of 10-12V, a reasonable range. If it is not adjusted properly, it will not destroy it completely. Then this is then compared to half or a quarter of the battery voltage. Two in series would have twice the cutoff voltage. If it is under that reference voltage, it turns on its output. This resets a s/r latch. It can only be set by a no pushbutton. When it is on, it turns a mosfet on that turns the main switch on (probably a relay). I did not put the mosfet in the schematic because it was getting kind of crowded.

I have no idea what to look for in a comparator or s/r latch. If people could give specific suggestions, from mouser, that would be very helpful.

For the switch, here are the requirements. I have looked around but the relays I found consumed too much power when on. The mosfets did not look like they could handle the current (puny leads) and had too high on resistances. Additionally, SMD would be hard and annoying to solder (and might overheat it dangerously), so leaded or another package is ideal.
-less than 800uOhm on resistance
-actually rated at least 140A, continuous
-less than 1/2 a watt consumed when on (meaning a large coil resistance for a relay). A latching relay would also be good, but I could not find any viable options. This is one of the major limiting factors I have found. The problem with NC is it would consume too much power from a dying battery. And it should fail open, not short, if something happens to the control circuitry.
-less than $20. Ideally cheaper.

This thread stems off of this one. It seems that after a few days, no one pays attention to your post anymore.

Thanks in advance for your help,
Live Wire
 

Picbuster

Joined Dec 2, 2013
1,027
Slab battery 12V nominal is charged to 14.4V maximum save level is 13.6V.
Overcharge is not possible when loader is limited to 13.6V.
( when both are the same voltage no current is flowing).
under voltage defined @ 10.5V ( slab considered as empty internal resistance high).

Picbuster
 

Thread Starter

-live wire-

Joined Dec 22, 2017
912
I know what is standard for overdischarge, but if you have a 100A load and 10mOhm ESR that's a 1 volt drop. So it would probably be safe to discharge it to a little less. I am asking about the specific circuit, not the exact voltage to use.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
912
schemeit-project (3).png
Ok, so this is what I have. I know the schematic is not the best and is kind of chaotic. But hopefully you can understand. The goal is to shut off once the battery voltage gets below a certain threshold. It can only be reset manually, by a switch. It will trigger again if it is below the adjustable threshold.

S2 and S3 are the same switch. I will make sure to remove the load before turning it on. All the resistors are 10k, except the variable one is 2k. The capacitors are 27uF. The center tapped coil and switch represents the latching relay (it has 3 pins, one common, a set and reset). The fuse is really the circuit breaker. I will add flyback diodes for the relay and breaker (I suspect the breaker is an inductor, as it relies on EM stuff). I will add another 10k resistor, as a pulldown for the mosfet, and I will make sure it can handle that kind of voltage on the gate.

Do I need any extra components? Is the op amp configured correctly, as a comparator? The goal is that it is on if the voltage is below the reference. And could someone suggest an op amp from mouser here that would be well suited to this application? That would be really helpful. This is all designed to operate in 12/24 volts, with the only thing that I must change being the divider switch. I decided to go with this latching relay.
 

Hymie

Joined Mar 30, 2018
970
I think that your idea of measuring the battery voltage drop will not function as you want.

Once the battery becomes slightly discharged, your circuit will react as though this voltage drop is due to an excessive current draw - when it is actually just the battery voltage dropping as the battery looses its charge.

A better solution is to include a low resistance value (say 10m ohm) in the return path and use the voltage across this resistance to switch out your relay at a predetermined voltage.
Then you are guaranteeing that the voltage across the resistor is due to the circuit current.
 

Hymie

Joined Mar 30, 2018
970
Here is the basic circuit of using a resistor in the return path.

But at 10m ohm, 100A will dissipate 100W in this resistor, even at 1m ohm the power is 10W.

I would suggest you form the low value resistor from a short length of wire and ‘calibrate’ its value by passing 10s of amps through it to determine the value – and hence the voltage at which the op-amp switches.
 

Attachments

AnalogKid

Joined Aug 1, 2013
9,165
Linear Tech makes several electronic circuit breaker controller chips that use an external shunt to control an external power MOSFET. I used them in a couple of MIL projects with success back in the 2000's; they now have newer parts. Some have an undervoltage lockout function that should work as a voltage-based over-discharge detector.

Here is an older model:
http://www.analog.com/media/en/technical-documentation/data-sheets/425612fa.pm6.pdf

ak
 
Last edited:

Thread Starter

-live wire-

Joined Dec 22, 2017
912
I am not looking to make my own overcurrent protection. It is too vital. I am looking to make my own overdischarge protection. If you look at the schematic, you can see a circuit breaker before all my other stuff, as a precaution (it is modeled as a fuse). It would trip at 120A. My main concern with the commercial options is that they do not allow for the desired current, nor do they allow both 24V and 12V operation. The ones that do are probably WAY too expensive.

The total cost for the protection circuitry will come to $50. A $20 breaker and $30 (including shipping) of parts from mouser. It will consume essentially nothing when off. It is adjustable between a wide range of voltages. It probably has only 2 or 3 milliohms, max, in it. Can you show me a better commercial option? There will be some things to work out, but it should be far more economical and customizable.

So could you suggest an op amp here that would work well? Do you have any suggestions to improve the whole design? Or do you think DIY is completely not doable?
 

AnalogKid

Joined Aug 1, 2013
9,165
From what I see, when the battery voltage drops below the trip point, the comparator output goes low, which turns *off* the MOSFET Q1. Is that the desired action?

The output of a 3-terminal regulator makes a poor reference voltage for a comparator. Better to use something built for the job, like an LM4040 or one of many newer parts.

When the relay opens, it beaks the power return (ground) connection to the circuit but not the V+. Given how little current it takes to turn on a FET, this could have unpredictable results.

Labels on the converter pins would help.

Just BTW, a fuse is slower and less precise than a magnetic breaker, which is slower and less precise than an electronic breaker.

ak
 

Thread Starter

-live wire-

Joined Dec 22, 2017
912
Thanks for the feedback. I am using a magnetic breaker that can be reset (because of low cost, convenience, and low resistance).

The linear regulator (link here) has the 27uF capacitors to stabilize the input and output. It says it should be accurate to +-.025 volts (.4%), which means +-.05 volts per battery. This is not ideal but is not terrible, and should be good enough. But are those kinds of regulators you mentioned far superior and not too much more expensive? The resistors are .1% and 10k, meaning no issues there, right?

The battery voltage (divided) goes into the - input, and the reference (also divided a little) goes to the + input. So doesn't that mean it is high when the + input, the reference, is greater? This would be when the battery is below the threshold voltage.

And if you take a look at the schematic, you'll see that the latching relay (switch with box + coil) cuts off power to everything. It can only be turned back on manually (by bypassing it and pressing the button). It needs one of the coils to be energized to turn off. So if there is a fault for long enough, the mosfet turns it off and nothing happens until you reset it. I do not see how the mosfet could do anything by turning on, once the relay is off. This is the link to the relay, FYI.

Can you recommend any particular op amps, from mouser, for using as a comparator?
 

AnalogKid

Joined Aug 1, 2013
9,165
Output accuracy and load regulation are two separate things. The variation with changes in load is what you cited, but your load is so small it isn't an issue. The output voltage tolerance is 1.5%, 2.0%, or 4.0%, depending on the cost. The output can wander around within that tolerance window based on temperature, aging, or nothing at all. This is the area where a true IC voltage reference excels.

ak
 
Last edited:

Thread Starter

-live wire-

Joined Dec 22, 2017
912
Output accuracy and load regulation are two separate things. The variation with changes in load is what you cited, but your load is so small it isn't an issue. The output voltage tolerance is 1.5%, 2.0%, or 4.0%, depending on the cost. The output can wander around within that tolerance window based on temperature, aging, or nothing at all. This is the area where a true IC voltage reference excels.

ak
I don't understand. It says "1.5% load regulation in the datasheet" but it says a load regulation of 25mV. .025/6=.004, or .4%. 1.5% is .2V at 12V! That is not acceptable! So what series do you suggest for .5% or better regulation?
 

AnalogKid

Joined Aug 1, 2013
9,165
What I see is 1.5% output voltage tolerance. That has nothing to do with the dynamic load regulation spec. The two add to form the total output error: one value at a single output current, and another value related to changes in the output current.

TI, Analog Devices, and Linear Tech all make lotsa voltage references. Attached is an older one that has served me well for decades.

ak
 

Attachments

Thread Starter

-live wire-

Joined Dec 22, 2017
912
Is that regulator just a zener? And what type of regulator is best here, for accuracy and cost? Would some alternative be better than a linear regulator?
 
I know what is standard for overdischarge, but if you have a 100A load and 10mOhm ESR that's a 1 volt drop. So it would probably be safe to discharge it to a little less. I am asking about the specific circuit, not the exact voltage to use.
I have done this before as well but can't give you the exact IP but can say, as you already know, the cut off voltage has to be a function of the load current. What you may not know is that it is also a function of the battery temperature (measured at a battery terminal) and the state of charge SoC) of the battery. It will be difficult to develop something that copes with more than a narrow range of battery capacities or one that copes with high discharge rates. I found a maximum discharge rate of C10 or so meant that I could predict these things fairly close. In my case the goal was to prevent a discharge below 30% SoC which gives a much higher cycle life for an SLA deep cycle battery. It is also critical to factor in temperature though for both charge and discharge.

Books such as Handbook of Batteries, McGraw Hill had some good information but it took several books to complete the picture enough for me to programme a uC to estimate with reasonable accuracy the SoC of the battery. My cut off was not a voltage function either. I cut off the load at a specific SoC instead which is much more to the point if battery longevity is your goal.

Wish I could be more helpful but what I think you are going to have to do is accept there are no easy solutions and a good deal of research is needed to get this right. Good luck, it is doable if you persevere.
 
Here is the basic circuit of using a resistor in the return path.

But at 10m ohm, 100A will dissipate 100W in this resistor, even at 1m ohm the power is 10W.

I would suggest you form the low value resistor from a short length of wire and ‘calibrate’ its value by passing 10s of amps through it to determine the value – and hence the voltage at which the op-amp switches.
The usual shunt element material is magnanin which has a zero temp coefficient. It comes in two forms, one as an alloy the other as three strands of wire of different metals. In any event, the resistance will lose power. A lower loss alternative is to use a hall effect device spliced into a toroidal core with many turns of wire on it and a single wire (the load wire) passed through the centre. A power op-amp taking the (linear) output of the hall effect device drives the mutlitturn winding so as to cancel the magnetic field in the core and the output of the op-amp, the series resistance of the winding and the turns ratio of the single turn primary (load wire) and the multiturn secondary (driven by the opamp) will be proportional to the load current. As this does not rely on the linearity of the hall effect the only error is DC offset by the hall effect and variation of the DC resistance in the secondary coil. Using a proper shunt in series with the secondary coil will solve that problem too.

This type of current sensor is commercially available (active current transformer) but they may be a bit expensive. Not sure, been a long time since I checked last.

These can also do AC currents up to a frequency dependent only on your op amp. If you do use one of these arrangements be sure to make the cut in the toroid (for the hall effect) complete and as uniform as possible to avoid magnetic effects such as remnant flux and coercive field issues. A high mhu ferrite with reasonable high frequency performance should mean narrow BH loop and good concentration of magnetic field energy in the gap with the hall effect device.
 
Slab battery 12V nominal is charged to 14.4V maximum save level is 13.6V.
Overcharge is not possible when loader is limited to 13.6V.
( when both are the same voltage no current is flowing).
under voltage defined @ 10.5V ( slab considered as empty internal resistance high).

Picbuster
An open circuit terminal voltage (allowing for dielectric relaxation) has an SLA at 100% SoD at 11.0V.
13.6V will charge an SLA in a week or two if you are lucky to about 80% maybe, depending on temperature.
An initial charge voltage of 14.7V is perfect for a lower SOC and then reduced as SoC improves. Limiting factor being charge current which is battery dependent.
All these voltages must be temperature corrected. Any charging above 50degC is very likely to kill the SLA in short order.
All temperatures are of the electrolyte, not ambient.
These 'one size fits all' numbers are bandied about in a most alarming fashion! ;-)
They're just not right without a truck load of caveats.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
912
The usual shunt element material is magnanin which has a zero temp coefficient. It comes in two forms, one as an alloy the other as three strands of wire of different metals. In any event, the resistance will lose power. A lower loss alternative is to use a hall effect device spliced into a toroidal core with many turns of wire on it and a single wire (the load wire) passed through the centre. A power op-amp taking the (linear) output of the hall effect device drives the mutlitturn winding so as to cancel the magnetic field in the core and the output of the op-amp, the series resistance of the winding and the turns ratio of the single turn primary (load wire) and the multiturn secondary (driven by the opamp) will be proportional to the load current. As this does not rely on the linearity of the hall effect the only error is DC offset by the hall effect and variation of the DC resistance in the secondary coil. Using a proper shunt in series with the secondary coil will solve that problem too.

This type of current sensor is commercially available (active current transformer) but they may be a bit expensive. Not sure, been a long time since I checked last.

These can also do AC currents up to a frequency dependent only on your op amp. If you do use one of these arrangements be sure to make the cut in the toroid (for the hall effect) complete and as uniform as possible to avoid magnetic effects such as remnant flux and coercive field issues. A high mhu ferrite with reasonable high frequency performance should mean narrow BH loop and good concentration of magnetic field energy in the gap with the hall effect device.
I am not looking to do DIY overcurrent protection, or advanced monitoring involving discharge current, etc. So I will just have to figure out the best one set point.
An open circuit terminal voltage (allowing for dielectric relaxation) has an SLA at 100% SoD at 11.0V.
13.6V will charge an SLA in a week or two if you are lucky to about 80% maybe, depending on temperature.
An initial charge voltage of 14.7V is perfect for a lower SOC and then reduced as SoC improves. Limiting factor being charge current which is battery dependent.
All these voltages must be temperature corrected. Any charging above 50degC is very likely to kill the SLA in short order.
All temperatures are of the electrolyte, not ambient.
These 'one size fits all' numbers are bandied about in a most alarming fashion! ;-)
They're just not right without a truck load of caveats.
So how do you fast charge them but charge them in a way that is good for lifetime? Is CC/CV possible and recommended? Is a commercial charger the way to go? I do not want to have to constantly change stuff around.
 
I am not looking to do DIY overcurrent protection, or advanced monitoring involving discharge current, etc. So I will just have to figure out the best one set point.

So how do you fast charge them but charge them in a way that is good for lifetime? Is CC/CV possible and recommended? Is a commercial charger the way to go? I do not want to have to constantly change stuff around.
There was discussion about a current sensing element in the form of a shunt and the power loss inherent with a load current of 100A. I mentioned manganin (which is the usual material a current shunt is made of) because if you use Cu for example, the temp co is significant even with modest temp rise. The hall effect option is not as complicated as it seems and has virtually no power losses.

Using CC/CV charging is fine and what everybody does :)
It is a question of how you decide when to switch modes and charge voltages that is the first critical factor. Beyond that, modifying the parameters based on battery temp is a very close secondary consideration.
A float charge may be 13.4V or 14.1V depending on temperature. A fixed 13.6V would be a disaster if you wanted a quick charge or long battery life if the battery was being cycled. Boost charging could easily be 14.7V even at 25degC and equalising charge at 14.4V but only if the SoC was under around 90%. to maintain a high charge voltage on a charged battery will quickly kill the battery by overdriving the chemistry.
I don't know if a standard product exists to do what you want. I have developed functionally similar products but not with the same specs. I did that because at the time, the products did not exist. The best around had a fixed cut off voltage to stop over discharge and that just does not do what it says it does. The battery is totally screwed by 10.5V and the cycle life down to 10% of its potential.
 
Top