I have a working AA-USB charger. It has a wide range of working vIn, from 2.2-4.5v, but it only delivers high current at high voltage, so I have to use three AA's. They're rechargeable NiMH's. It's not good to discharge a rechargeable battery below 1 volt. With 3 batteries, this means a vIn of 3v. Unfortunately, the circuit keeps on working to well below 3v, and overdischarges my batteries.

How can I stop the circuit from discharging below 3v? Preferably a very small implementation, as this all fits in an Altoids mint tin, with some room to spare. Thanks for any suggestions!

 

I would use a LM393 analog comparator and use 3V as the reference voltage. The other branch of the comparator would be connected to your battery. If the voltage falls below 3V it would switch off thus disconnecting the load from the battery.

The LM393 (twin) or LM339 (quad) can operate from 2V to 36V single supply voltage. So it might work in your conditions.

Allen

 

I've been struggling with this as well...

I identified a different comparator though - a LT1017 - it's lower power than the LM393. Also note that you need a precision reference voltage to compare against. The LT1634 is a good choice for this it has a minimum operating current of 7uA. Both of these parts combined have a worst case quiescent current of about 107uA over temperature... not bad.

Remember that any current not going into your USB charger is current that is hurting your efficiency and will continue to drain your batteries, albeit slowly. Of course, these parts are a bit more pricy than the LM393 too.

 

One final note:

YOu're probably using a buck-boost converter to supply your 5V regulated voltage... that's fine, but note that you will continue to deliver the unboosted voltage to the load, even if you shut down. It is advisable to also use a FET to remove all power from the regulator when your comparator sees your battery voltage has fallen below threshold. This will hurt your efficiency, though.

 

Here is a circuit.
It will still draw about a 0.7 ma. from the battery when disconnected, but maybe that is ok.
Also a candidate for the messy schematic contest.

Almost any logic level FET would work and a TL431 could also be used for the reference.

Or look at the LM10. Reference and comparator in one IC.

 

A few issues with your circuit. Please note that I'm not trying to offend, but I am trying to help educate.

1.5k seems like too much resistance on your 2.5V reference. The current at 3V supply is only 333uA if the 2.5V ref is regulating correctly, which it probably won't be. The LT1009 is only guaranteed functional between 400u and 10mA.

The common mode input voltage of the LM393 is 1.5V below the positive rail (at 25degC, 2V over temp), that would be in this case, 1.5V... so a 2.5V reference will be too high. A 1.25V reference (or lower) will work better, and you will be able to increase the resistance significantly because the current required to regulate will be lower. You could use a resistor divider on the 2.5V ref, but this will decrease the accuracy depending on your resistor matching. You could use a trim pot to remove any error (at nominal conditions).

You can probably get the LM393 to work at ambient temp with a bit more effort, but if you need to operate over temp I don't think it will work (it's going to be tough getting a good voltage reference below 1V.). Of course, the LM393 is ~$0.10 compared to ~$2.50 for the LT1017, so there might be an advantage to making the LM393 work if you need to build thousands of them.

I will again reiterate that at your low battery voltage, you want draw as little power as possible to keep from over-discharge. Hence my suggestion of the LT1017/LT1634 suggestion. Another benefit to the LT1017 is that you don't need the pull-up resistor, if you design it right, eliminating a part.

Another thing to think about is battery impedance, when your switcher is on and outputing 0.5A (max USB 2.0 spec) then you are pulling somewhere around 2-3A out of your battery when you are charging your inductor... this kind of current will lower the voltage out of your battery... when you stop charging your inductor, the voltage will go up because there is little current draw... this creates a very noisy input voltage, so you have to filter your comparator somehow and also provide adequate hysteresis so that you don't 'motor-boat' and turn your switcher on and off repeatedly.

 

Eeek , Right you are. Never noticed that spec. before, but then I guess I've never tried to run one that close to the + supply. Did cut the current a little close to the bone for the reference. So I guess we could use a reference like the LT1634-1.25 and solve both problems. I like the LT1017. Little pricey, but low current and voltage.

 

Just throwing this out there; setting a low threshold of 3V won't guarantee against overdischarge of your batteries. Depending on age, internal resistance, initial charge level, etc, they probably won't all discharge at the same rate and remain balanced in level. you could end up with one cell at 1V, one at 0.5V, and one still chugging along at 1.5V at the time that your 3V shutoff occurs.

But, it's still better than nothing .

 

wouldn't a lithium ion protection pcb do the trick? they shut off at 2.8v at the lowest. and cost is minimal- http://www.batteryjunction.com/prcimopfor3l.html

dont worry about the individual cells. just charge them all to the same voltage and if one goes bad over time you will probably lose all 3 cells but who cares, its just a few AA batteries.

 

http://www.linear.com/product/LTC1998

I found this today, and thought it would be just right for your application... hopefully you're still around, or someone else can benefit. You're not using Lion, but the voltage levels are similar... also it's $1.33 for the first 1000, and then less than $1 after that.