Pulsing LED

[loogie]

Many months ago I asked for some help on a pulsing LED circuit for a friend in London.  Well, I finally got back over there and with the help of Maplin's in the Strand we were able to get it working.

The unit works great at room temperature, but when its actually used in performance it will be frozen into a grapefruit sized ball of ice.  There's a switch (not shown) and when you turn the ball over it begins to pulse.  Very nice.  But it doesn't last long.  I think the cold slows down the battery.  We started with a 6v and worked our way up to 12v.  At 12v the voltage across the LED doesn't drop enough to shut it off until the battery wears down a bit and then it works for a while until it finally quits.  We've installed a 2k7 resistor at V+ and that works, but I'm wondering if that just isn't burning away battery juice?

I was wondering, would it be possible using something like a zener diode to create a simple voltage regulator?  That maybe it could reduce the voltage without eating up as much juice as a resistor?

[PRR]

The cap MUST be bi-polar. It gets swung both ways.

Is this supposed to be on/off? It looks like it is wired fade-in/fade-out.

If it ever does pull up near 12V, the LED current is like 70mA, a LOT for most LEDs. However a 2K7 in line with the battery means 4mA max, unless you have a bypass cap on the rails (and you really should).

[loogie]

Its a slow, pulsing effect.  Slowly comes up, slowly dies away.  Very dramatic.

The original schematic showed a polarized cap.  We had a small plastic cap, but that would have been too big to fit into the container we were using at the time so I suggested a tantalum cap.  Well, we put it in backwards, but it still worked.  We also tried connecting two identical LEDs in parallel.  Along the way its quite possible we made every conceivable mistake, but we were encouraged by the fact that it worked just as we had hoped -- except when frozen -- and even then it worked for a bit.

At room temperature it will work for quite a while.  Long enough for the performance.  We are still using two LEDs in parallel.

So a cap across V+ and V-?  As big as we can squeeze in?  Should we go back to one LED?

[PRR]

> The original schematic showed a polarized cap.

Wrong. With 9V supply there is over 2V reverse voltage on the cap at one peak or the other. That may "work" with a new cap on the bench for a short while, but begs for a problem long-term or in severe conditions. And ice temp radically increases electrolytic leakage.

> I suggested a tantalum cap.

Still won't be "solid" with 2V reverse at low temp.

Try back-to-back e-caps. Tantalum may be better, but good modern Al electrolytics are pretty good.

I'm wondering about the BJT-input chip with the very large timing resistor. The LM358's nominal input current against 3M9 is an acceptable 0.22V error, and stable with temp, but a worst-case LM358 may have a 1V error, which might matter more at 9V than at 12V. It is a shame you could not use a socket; a TL072 swap would eliminate this possibility.

You can't bypass the signal frequency. Put a 0.1u ceramic across pins 8 and 4 just to be sure there are not MHz oscillations upsetting the action.

[loogie]

Thanks.  Fortunately we did use a socket.  Here at home I've been experimenting with a TLO72.  I'll check the Maplin's catalog.  Funny about the cap.  I'll update the schematic and alert my friend in London.  If you're right (what am I talking about?) you've got a free half of Timothy Taylor coming to you.  Down at the 'Harp'. 

[loogie]

Just got a report from London.  Five minutes of uninterrupted pulsing from a finished iceball.  A first!  My associate used a TL072, a non-polarized cap and 12vdc.  Very promising.  Still a few more units to construct, but we feel we're out of the woods.  Thanks!