Visible low battery alarm

[porkluvr]

Here is something to do with an unused half of an LM358 - if you don't already have a Lo-Battery indicator. I am wanting to use rechargable batteries, where the voltage differential between a "good" battery and one that needs charging, is maybe only two tenths of a volt. I need a distinct "break point", a mushy audio tone won't do.

Hysteresis is not featured here, but I don't think it should be an annoyance. I originally started designing with a LM393 type comparator (or section "A", 1/2 LM392), but my simulation shows the LM358 to actually be better behaved - so far as not spiking during turn-on/off. Of course my simulation is not as good as a breadboard or prototype.

C2 and R4 values can be varied to give a fairly wide range of PPS/PW times. Not counting amplifier or comparator current, this circuit (with RC as shown) draws less than 100uA. I suggest staying with R values from 470 to 2k, and C between 10nF and 680nF. Short duration pulses may not be visible even at higher flash rates.

The original design (from Dave Johnson and Associates) uses 680 ohms and 680nF, and blinks slowly (about 1/2 Hz) with relatively long pulse duration. You shouldn't have much trouble finding an RC combination yielding a visible, low current alarm. Of course, different LEDs and R5 will affect results. Sorry I can't tell you precisely which LED to use, but LEDs are cheap, right?

Although you can't tell from the picture, the LED on time is about 260us.

[Carl-NC]

Nice little circuit!

[porkluvr]

updated info

Thanks, Carl - it's not perfected, but if so long as it doesn't make the speaker buzz then it's OK.

I got to looking back on my choice of C3: I think 2200u is a little bit of over-kill. I selected C3 for flat LED current peaks, and minimal sagging over a period of seconds - neither of which is particularly important. Good, but not extremely important. Using R6=1k and C3=470u may work smoother. There is room for improvement, I am sure.

Of course, anybody who uses this diddy will make sure that R5 is big enough to safely limit current through whatever LED is chosen, right? As an off-hand guess, short pulses of 20~30ma should be safe for most LEDS. Some want or need more. Some need LESS. But, we should not guess, and there is no need to.The last dozen pages of Digikey's OPTO section of catalog has abbreviated data on a wide variety of different devices. Here's their present opto link: http://pdfcatalog.digikey.com/T073/SectO.pdf (circa DEC 2007; the link will change with each seasonal catalog)

Alright, most folks will want to skip this next part, but I feel like it is only fair to include the following design example:
Different color LEDs have different forward operating voltages (and optimum range of current).
Choosing Rohm's SLI-343UR3F (red, on page 67 of TO73/O), it says typical Vf=1.9 and If=20mA. To calculate "typical" R5: (5-1.9)V/20ma = 155Ω. With a rated maximum (continuous?) current of 50ma, I could -safely- use down to a minimum of 62 ohms, or perhaps less with short pulses.

When I finish re-padding my PCB layout libraries (AGAIN) then I might be able to give this a practical go-around, and I can let you know how this works... . Meanwhile, all this teknikul math stuff's too hard for me, so I'm OUTA HERE, maybe cut some Zs, then re-pad my stupid parts!

[porkluvr]

erggh

There is a serious problem with my earlier submission:
All LED current is delivered through U1 (er, U2, whatever). If U1 is a dual, there is a high likelyhood of the OTHER section being adversely affected by the LED current spikes. If U1 is a single, then an RC filter would be an excellent addition to the V- power pin. And, in either case,C3 should be connected to the U1 output pin. This gets very messy.

I am ashamed. Some damage control will make me feel better.
Rather than just bury my head, here is the simplest, best thing I can figure out at this time:
Isolate the LED pulse generator with a transistor. This way, about 10ua. continuous Q3 turn on current is required through the opamp - instead of the previous 20 or 30ma pulses (eek).

Diodes D1 and D2 are needed to prevent Q3 Vebo from being exceeded. The C3 filter cap must not be made too large or else the blinker may not startup when needed.

Now, I come to realize that my simple alarm is not quite so simple as I originally thought. Maybe if I look hard enough I can find something else wrong - but is enough for now.

"Too much perfection is a mistake." I can feel better now.