efficient way to disconnect battery when voltage drops?

[ja421]

We have an off grid water pump that runs off a deep cycle battery and is charged by solar.
It receives a wireless signal from the thank when the tank is 1/2 way empty and to start pumping. (on 3/4 low, an alert inside the house is given so we know there's a problem)

This is all working well, except... When the battery is drained too much, the solar charger locks it out as a bad battery.
So what I want to do is just have something that monitors the battery voltage, and if it should drop (say below 11 or 10.4v) it will not engage the pump.  I think most of the time, we can go a day or two w/o it pumping, and that should give the solar panel (and sun) enough time to recharge the battery.

I have toyed with a zener diode & transistor to energize a relay to cut the battery in/out, but this seems to waste a lot of energy (I have not tried to use a mosfet instead of a relay).  I am thinking about using an attiny85 & sleeping as much as possible...

It seems there might be a better / lower power way to do this. Any thoughts?
thank you

[Seekonk]

11V is way too low.  I don't go below 12.6V.  You could just switch the control circuit on and off, that would be low current.  Even an opto isolator may be sufficient.  I use the little arduinos for everything at only $2 each in my solar applications.  A TL431 is an alternative for discrete.  Do you really need to pump at night?

[wraper]

11V is way too low.  I don't go below 12.6V.

Really??? What's the point in the battery which will get cut off even before discharged by any sensible amount?

[ja421]

11V is way too low.  I don't go below 12.6V.  You could just switch the control circuit on and off, that would be low current.  Even an opto isolator may be sufficient.  I use the little arduinos for everything at only $2 each in my solar applications.  A TL431 is an alternative for discrete.  Do you really need to pump at night?

I'm not sure I agree with 12.6 -- deep cycle batteries can provide a lot of current at lower volts, and 12.6 seems like a battery just sitting idle, so it might never get used... Either way, the cutoff/in voltage shouldn't be the primary focus; it's the method for monitoring the voltage I'm after a fairly effecient way.

"could just switch it on off" -- yes, I could; but that requires walking a fair bit up/down a steep hill to the pump house.
The pump does run at night if it needs to (drops to 3/4 low); no one likes waking up and finding out the hard way that there is only enough water for 1/2 a shower.

I guess I'll experiment with arduino more and see how power compares to the zener method.

[Ian.M]

If its a Lead Acid battery, don't worry too much about getting ultra-low power consumption.  The typical self-discharge is 5%/month, which for a 100AH battery is equivalent to about 7mA continuous load.  It wont even notice a 1 or 2 mA load.

You do however need to add some hysteresis to make the turnon threshold sufficiently higher than the cutoff threshold so that the rise in terminal voltage when the load is removed doesn't turn the pump back on causing it to short-cycle.

Probably the best option would be a TL431 driving a high side big P-channel MOSFET.  Stick a NPN emitter follower on the TL431 cathode resistor to boost the +ve going gate drive, with a reverse biassed Schottky diode B-E so the TL431 can pull the gate -ve quickly.   Put a high value resistor from the MOSFET drain to the tap on the potential divider feeding the TL431 ADJ terminal to add hysterisis.  You should be able to get the quiescent current down to 1.1mA without any problems.

[edavid]

A relay is fine, just don't energize it unless the level switch is calling for water.  The relay power will be negligible compared to the pump power.

But if I were you, I would adjust or disable the lockout on the charger, or get a different charger.

[LabSpokane]

11V is way too low.  I don't go below 12.6V.  You could just switch the control circuit on and off, that would be low current.  Even an opto isolator may be sufficient.  I use the little arduinos for everything at only $2 each in my solar applications.  A TL431 is an alternative for discrete.  Do you really need to pump at night?

10.5V = 0% charge with lead acid.

12.6 = fully charged.

11V-ish is a reasonable cutoff that will protect the battery while providing a reserve charge.

[Seekonk]

11V is way too low.  I don't go below 12.6V.  You could just switch the control circuit on and off, that would be low current.  Even an opto isolator may be sufficient.  I use the little arduinos for everything at only $2 each in my solar applications.  A TL431 is an alternative for discrete.  Do you really need to pump at night?

I'm not sure I agree with 12.6 -- deep cycle batteries can provide a lot of current at lower volts, and 12.6 seems like a battery just sitting idle, so it might never get used... Either way, the cutoff/in voltage shouldn't be the primary focus; it's the method for monitoring the voltage I'm after a fairly effecient way.

That's the way I run my system.  What is different is my system only runs in the day and uses car batteries (ones from vehicles not in use).  I don't allow anything to start unless I have 13.7V, quite easy to obtain with only a couple amps charge, even at 8am.  13.5V is the low battery warning and 12.5 is the desperate warning light.  These voltages are easy to obtain with just a couple amp charge.  I can always run the system manually.  Draining battery down is just running behind the 8 ball.  Easy to get in situations that can't get caught up.  This is what works for me living many years off the grid.  You are better off storing in tanks than a battery. Letting it go to 11V is only something someone would say who hasn't worked with batteries.

[LabSpokane]

11V is way too low.  I don't go below 12.6V.  You could just switch the control circuit on and off, that would be low current.  Even an opto isolator may be sufficient.  I use the little arduinos for everything at only $2 each in my solar applications.  A TL431 is an alternative for discrete.  Do you really need to pump at night?

I'm not sure I agree with 12.6 -- deep cycle batteries can provide a lot of current at lower volts, and 12.6 seems like a battery just sitting idle, so it might never get used... Either way, the cutoff/in voltage shouldn't be the primary focus; it's the method for monitoring the voltage I'm after a fairly effecient way.

That's the way I run my system.  What is different is my system only runs in the day and uses car batteries (ones from vehicles not in use).  I don't allow anything to start unless I have 13.7V, quite easy to obtain with only a couple amps charge, even at 8am.  13.5V is the low battery warning and 12.5 is the desperate warning light.  These voltages are easy to obtain with just a couple amp charge.  I can always run the system manually.  Draining battery down is just running behind the 8 ball.  Easy to get in situations that can't get caught up.  This is what works for me living many years off the grid.  You are better off storing in tanks than a battery. Letting it go to 11V is only something someone would say who hasn't worked with batteries.

Lead acid batteries only sit above 12.6V either while being charged or immediately being disconnected from a trickle charger.  Disconnect your batteries and let them sit idle for an hour and you'll watch the voltage drift down to 12.6V.

I've spent a fair bit of time with lead acid technology and while very discharged, 11V is not a damaging discharge level for a battery in good condition - particularly if it is discharged under a substantial load. Discharging to 11V at low loads is a very different thing and will cause the battery to be at a greater state of discharge. Discharging past 10.5V causes damage unequivocally.

If your lead acid batteries are weak at 12.6V after sitting idle for an hour after a full charge, they are likely damaged or weak.

[Seekonk]

My system doesn't even wake up till the battery is charging.  I make my decisions always under charge conditions.  If he uses a UNO he can eventually have a smart system that can make decisions based on what is happening.  If diligent he  will able to correct for all the bad advice he has been given.  If the system experiences occasional moderate discharges, it is a bad design.  It is pretty easy to get into a situation of heavy discharge and stay there.  That is not what batteries like.  My batteries have to go back into vehicles and I expect normal life expectancy from them.  My system is rather small but it supplies all my needs for 5 months, including hot water and it has to work every day without fail.  People love to have big dumb systems because that is all they can understand.

[Monkeh]

In other words you have a system of inadequate storage capacity using inappropriate cells.

[LabSpokane]

My system doesn't even wake up till the battery is charging.  I make my decisions always under charge conditions.

You *cannot* decide the health of a battery under charge conditions only. 

My batteries have to go back into vehicles and I expect normal life expectancy from them.

Prepare for disappointment.  Vehicle batteries are not designed for this application.

People love to have big dumb systems because that is all they can understand.

That's a pretty bold assertion considering you're ignoring the very fundamentals of lead-acid storage design.  Yes, you're getting away with it ... for now, but the numbers you cite indicate that your batteries are already failing.

[LabSpokane]

I guess I'll experiment with arduino more and see how power compares to the zener method.

An Arduino with a relay output would a simple way to handle this.  You could simply use a voltage divider to drop down the 12V side and use the internal 5V reference to measure the battery state.  Crude, but quick and simple and would get you up and running.  Then you can play with the parameters and not have to do any more plumbing.  The voltage regulator on the standard Arduino is pretty inefficient, but you can buy a DC-DC converter easily enough and power the system through the 5V pin if necessary. 

[DenzilPenberthy]

I wouldn't bother re-inventing the wheel. There are dozens of manufacturers who make 'battery protectors' which are designed for this exact application.

Here is an example:
http://www.windandsun.co.uk/media/303907/Battery-Protect-Data-sheet.pdf

Also I'm going to throw my opinion in too: 12.6v is fully charged for a lead-acid battery so it's dumb to use that as your cutoff voltage, you are only using up the tiny amount of surface charge from the charging process that way.

Also in my opinion, 11v is too low.  There's not a great deal of useful energy to be had once the discharge curve starts to turn downwards and it will affect the long term health of the battery going down that far.

 I'd regard 11.5 to be 'fully discharged' but if you want long life out of your batteries then I'd stay above 50% discharge.

[LabSpokane]

The battery disconnect Denzil posted is nice since the voltages are programmable and it has hysteresis designed in. The default disconnect voltage is 10.5V, which is fully discharged. The choice of a disconnect voltage above 10.5V is really down to what the user wants, be it fully depleting the batteries or leaving enough to start a car after leaving a load on too long. There is no "right" answer.

An arduino would be a fun, simple project where the user can control his/her destiny for better or worse. Otherwise, you can buy a solution. Be wary of prefab vehicle disconnects because many have the 10.5V cutoff hard coded as that may not be what you want.

[ja421]

Thanks for all your thoughts & reinforcing that my cut off values are reasonable.
I had looked at some of the pre-made units (ebay, reuk.co.uk, etc) but most missed a quiescent drain spec; as did almost every schematic I found.   I guess it's a matter of make a few & test them to find one that makes me happy.  Finding this one http://arduinomaster.myfreesites.net/battery-saver-cut-out-switch made me think; maybe just using a attiny & sleep it, test, sleep & use a mosfet might be a really easy way to go; and allow me to easily adjust the low/high values.
I really do appreciate your ideas & input. THANK YOU

[DenzilPenberthy]

I think you're over-thinking the quiescent current thing.  Unless your system is very very small or only operating very infrequently, I can't see it being a significant factor.  A Victron BP-50 will probably suit you. (unless your pump is bigger than 50A continuous/200A peak) It has a quiescent current draw of 1.5mA when on and 0.6mA when off. 

I don't know which country you're in but you can buy one here for £39.90.
http://www.kurandamarine.co.uk/battery-protect-bp-50?language=en&currency=GBP

If you can build a similar spec system with a lower Iq for that price just in parts (even if you work on it for 20 hours for free) I'll buy you a beer!

[ja421]

I think you're over-thinking the quiescent current thing.

You may be right... I do get hung up on thinking about the little stuff sometimes, for fear that the little stuff will add up somewhere else when I'm not looking... but you make a valid point.

If you can build a similar spec system with a lower Iq for that price just in parts (even if you work on it for 20 hours for free) I'll buy you a beer!

Well, now I have to work on doing it.

[Red Squirrel]

I built a small portable power system for camping/emergency use and wanted the same thing.  I ended up going with a latching relay as it only requires power to turn it on or off, then it latches in that position.  You connect the positive, then there are two grounds that can be completed by a transistor to turn it on or off depending on what ground you apply power to. This is a "dual coil" relay, there are some single coil ones but you have to reverse the polarity, which would be a bit more tricky to do with a MCU.  I think a H bridge is what you need.  Dual coil is just simpler to setup.

In my case the microcontroller has other tasks such as controlling LCD display (voltage and other messages).  I can then connect a laptop to set different parameters such as cut off voltage and calibrate voltage/amps readout.  I can't seem to get the amp readout to be accurate as it jumps around too much, but that's another story.  If I was to do it over again I would have used a high side current sense IC instead of shunt/op amp like I did.

11 volts is a decent voltage set point to go with.  You could do 11.5 to play it safer and extend life of battery.  12.6 is way too high, you'll barely get any use out of it.  That's practically a full charge sitting idle with no load.  It will go lower with a load.  I have seen some 48v systems that stick around 50 volts for hours though with a load on them (that would be like 12.5v on a 12v system) but batteries vary.

Idealy whatever system you design you should not hard code stuff like shut off value, so you can field adjust it as needed.  That's what I did for my camp solar system, I can plug a laptop and there is a text based interface to setup different parameters such as voltage calibration, shut off, shut off timer, etc...  Was fun to code.

[Seekonk]

I think that is the totally wrong concept.  A simple philosophy, Don't start what you can not finish.  Whatever switch you have to turn on the pump, supply that to a TL431 and a relay.  That relay is used to trigger the pump circuit.  This only powers the relay when there is pump demand.  No hysteresis is needed because it always checks at rest condition.  You don't get short cycling.  This assumes the pump generally operates for a fixed period. Otherwise use a timer.  This 10.5 & 11V comes from inverter manufacturers trying to protect inverters from excessive currents from under size wire power leads.  No one in their right mind would drag a battery down to that.

As Sheldon would say.  If I were wrong, don't you think I would know it.

[arlipscomb]

The lower the battery is discharged, and the longer it stays in that state, the shorter its life. If you take a lead acid battery charge down to the point where it is at 12 volts at rest (no load) and stays there for a day or so and you will start to get sulfation. Of course if you need water, you may not have a lot of choices. A couple of things come to mind:

Check the charge voltage against the recommendations of your specific battery's manufacturer. Many times the charge controllers are too conservative on charging voltages and you end up running out of daylight before your battery is fully charged. A half volt can make a lot more of a difference than you would think it would.

If you are building a programmable solution maybe start backing off on the pump run cycle earlier in the discharge cycle? For example if the battery is 75% discharged only run the pump for half the normal time. I am not sure what the water is for so you would have to figure out if something makes sense based on the need. 

Add a small wind turbine to supplement the solar. Often when the sun is not out you get some wind.

[Tibby]

We build chargers for SLA's and use a relay to disconnect load when battery drops to minimum acceptable. Since the load current is quite high we energize the relay while load is active and release when battery needs to be disconnected. Then its only our little sleeping PIC keeping an eye on battery that draws small current when it wakes up momentarily.

You can also use a latching relay, only uses short energy pulse when switching.

[KL27x]

My system doesn't even wake up till the battery is charging.  I make my decisions always under charge conditions.

Well, then. Perhaps this is why your numbers are so far off from the rest of the forum. Some of us may have assumed a low voltage cutoff would be more useful when the battery is at rest or under load. This is usually where the voltage is actually going down, in my experience.

[Seekonk]

Not responsible for peoples reading comprehension levels.  OP never said if he had L16's or a Wallmart deep cycle battery.  Voltages for various deep cycle batteries are not the same.  People just pop up a chart and think they know something.  You should have a smart system that can determine if you can even start and then monitor end current.  The OP may or may not figure out 2 years from now why his batteries have failed.  There are solar pumping systems that have no batteries.

[ja421]

The OP is on year 7 for these batteries, at least a majority of them; 1 is 10 year and a couple are less than 2 years...  Most of that longevity is a manual monitoring process.  My goal, before my health completely craps out, is to make the system smarter so that it is easier for my wife to maintain after I cannot.