Low battery indicator circuits

[Excalibur]

If conserving battery current is a real issue, you may want to replace that NE555 (draws 7-15 mA) with a 7555 CMOS type (draws a few hundred uA). It will be a drop-in replacement for this application.
The LM311 draws about 5 ma. and even lower curent opamps are available.

 

[aos]

Blinking solutions using two transistors such as presented earlier in this thread take less space and are cheaper, though.

 

[tangent]

Quote:

The LM311 draws about 5 ma. and even lower curent opamps are available.

From my original post, "There's a low-power version of the comparator (LP311) that would be ideal here..."

 

[tangent]

Quote:

LP311...would be ideal here

Scratch that. The LM311 and LP311 are horrid comparators. I tested my circuit with LM339s, which is what I had on hand at the time, and that worked fine. I got some of the '311s in, tried them today, and their output voltage is extremely soggy. No good at all for the purpose. I dropped an OPA227 in and it worked perfectly.

Bottom line, if you build my circuit, you need a low-voltage op-amp that swings its output to somewhere near its rails.

 

[tangent]

Quote:

Note that circuit is somewhat sensitive to resistor values and choice of transistors

aos, I just tried your circuit with substantially the same resistor values, changing only to 2N3904s, and it worked. However, I found that there's a dead zone of over 1V where neither LED is on.

Thinking about it, it looks like this is normal: it seems to be due to the softness of the zener knee plus the fact that the high LED can still have current going through it yet not have enough current to light. So, the "current robber" transistor still has base current, keeping the low LED's transistor shut off.

I was right about the current requirements: aos's transistor circuit only takes a fraction of a milliamp above what the LEDs require. The comparator circuits are rather more hungry.

 

[tangent]

Hadron, I tried your second circuit, and found that zeners aren't appropriate with it. The zener voltage just isn't stiff enough at low voltages. Low-voltage zeners perform badly.

Instead of going back to the voltage regulator idea, because that's either expensive or has high Iq, I found a cheap voltage reference: LM385, $0.50 for the TI version at Digi-Key. It will run on as little as 20uA, and it's available in a 2.5V version. Low voltage gives more flexibility in setting up this circuit. Here's my cut at it:

I don't have that reference part on hand, so I don't know if it works any better. I'll get some in and report back.

 

[aos]

Yes, Zeners are rather non-ideal as I found out looking at datasheets while making this cricuit (I first tried with higher wattage Zenner and that didn't work - it's just too soft). You could increase transistor hFE to compensate (e.g. use Darlington) but I'll see what I can do otherwise.

 

[setmenu]

I recollect making some sort of 2or 3 led indicator a while back using schmitt
triggers set at varying trigger levels, it was had to be very precise as the level differences were small, worked a treat, and used little power.
Can't remember what I did with the circuit or schematic now though, so
not much use really

Setmenu

 

[aos]

Tangent, shouldn't Vbd be actually tied to the reference?

 

[tangent]

Quote:

You could increase transistor hFE to compensate

How would that help? The problem is that the robber transistor still has base current past the point where the LED turns off. Increasing hFE is fixing a non-problem.

What is needed is a more precise way of cutting off the LED power.

Quote:

shouldn't Vbd be actually tied to the reference?

I suppose it could be made to work either way. I like it this way.

 

[aos]

How can it work this way? The second comparator is comparing voltages from two voltage dividers. That means it's only comparing voltage ratio, not the absolute values. You cannot say it will trigger at 7V or any other voltage then.

As for the other problem, higher hFE means the circuit (which acts like a switch) will be more sensitive (comparator should always have high gain). It is quite possible though that you'd need to play with the values of resistors to make it work with a particular LED and particular transistor (depending on its turn-on point, Vbe, hFE etc.) which is obviously very much not desired. I'll try to make it better. But the other appealing solution is similar to yours, yet it uses a different chip. I ordered it from Digikey as well as your regulators so I'll try it tomorrow.

 

[tangent]

Quote:

That means it's only comparing voltage ratio, not the absolute values. You cannot say it will trigger at 7V or any other voltage then.

Sorry, some bad algebra here.

I'll get around to changing it later.

 

[aos]

I've just stumbled onto what looks like to be a simple, compact and cheap solution, yet high precision and uses only about 1.1mA or so (and that only until the battery is "empty"). I have parts at home so when I get there I'm going to try it and post it if it works...

 

[aos]

Here is a new circuit. I tested it and it works (with a three-pin bicolor as well). I don't know how big is the transition voltage range as I don't have a variable power supply (I tested by removing one 1.5V cell from a pack) but I doubt it will be very large, given that a regulator is used.

The circuit is based on the previous one, plus one of the examples from the regulator's datasheet. It's the already well-know TL431. Only $0.6 Canadian.
The rest is just transistors and resistors, dirt cheap.

If you want a tri-color (i.e. add orange color (red and green active together) for between red and green) you can double the circuit and remove the third transistor, then change the resistors to set up threshold voltages to overlap.

This circuit is set to detect a 10AA battery pack from falling below 11V (i.e. 1.1V per cell) - because that's what I needed. To change the voltage, set the voltage divider R1/R10 so that it's at 2.5V when the battery voltage is as desired. Set the R11 resistor so that the current through the regulator is at least 1mA when the voltage of the battery is minimal where you still want to have the LEDs operating properly (otherwise it'll shut down and it will revert back to what was "normal" power, though if you set it low enough there will be no power in the battery left to light the LED anyway). Set the R3 and R4 for desired LED intenstity.

Now if you'll excuse me, I'm going back to Star Ocean.

 

[hadron]

tangent, I am sure you are correct, the second version of the circuit I put out there was untested. I agree that a bandgap ref will work much better than a weak-in-the-knee zener; while most cost significantly more than standard zeners, the LM385 is one of the few competitively priced ones out there.