Inductors and voltage and electro-destruction

I think your best option would be to isolate and regulate power to the LEDs using a LM7812 or 7805.  Most charging systems can have a pretty wide voltage swing during normal operation.  Flyback from fan motors, etc.. can be isolated with a mechanical or solid state relay.

IIRC, won't a capactitor work best to smooth out the spikes? Another question would be, shouldn't the battery smooth out the spikes?

Capacitor is what I was thinking. A smallish filter cap to ground in your supply circuit to the dash lighting would probably take care of it. You might also consider a larger value resistor to reduce the LED supply voltage further, maybe you're too close to their tolerance?

What value resistor are you using?  Have you measured the current, or just calculating?

Once an arc forms the air ionizes and the resistance (and so the required voltage) goes way, way down.  When you unplug something the gap is infinitely small when the two conductors first separate, so it takes very little voltage to form the arc.  The arc is then maintained even as you pull the conductors farther apart because the ionized air conducts well.  This will happen when you unplug almost anything, like even a simple light which has no significant inductance, and is mostly just a function of the amount of current the device is drawing, not inductance.  It's not because of a spike of 1000s of volts.

My guess is you just need a bigger resistor.  Remember, your alternator is putting out 14v (or more, if it's not being well regulated), not 12v, so size your resistor for that.  And check your alternator voltage at engine redline.  Many alternators are not that well regulated since they're made for cheap cars that aren't intended to run at high rpm for any length of time.  When you race it and run the rpms way up there you might find it's putting out 16 to 20 volts or more.

Also make sure you're using the proper voltage and current for your actual LEDs (they range from 1.8v to 3.5 v, and 10 to 30 mA), not just the values for a generic LED.

Different kind of capacitor, different purpose.

Google is your friend:
http://electronicrepairguide.com/filter … ction.html

You people with your "wisdom" and your "knowledge." You're letting the smoke out of your LED's somewhere along the way. You're probably running that high-fallutin' 100% synthetic blinker smoke when everybody on the blinker smoke forum knows that pure dino blinker smoke is best for LED operations on a Lemons car. Pure synthetic blinker smoke on cars this old just seeps past the worn seals...

As far as i know LED replacement bulbs only have a resistor in them.  http://www.superbrightleds.com/cgi-bin/ … 2F1157.htm  They also have a led / resistor calculator  http://www.superbrightleds.com/cgi-bin/ … d_info.htm

Most things meant to last in an automotive electrical system are designed to handle 80v spikes.  There are audio noise filters that help smooth out the 12v power, some use transistors, inductors, etc.

I think its being overthought a bit.  You could grab a potentiometer that will swing to 200 ohms, put it inline and slowly turn the resistance down until you think the light output is still acceptable.  Then get those resistors and add them in line to your leds.  If you have one, you could put a meter in line to measure how many ma they are actually being driven by and turn the potentiometer resistance up till the led is down to an acceptable ma level.  Then measure the potentiometers resistance out of the circuit without touching the knob and use that resistor per led.  Wiring LEDs in series means if one dies that group goes out so its a it works or it really doesnt kinda thing.

A bundle of LEDs, if wired in series, would be able to handle larger reverse voltage spikes than a single LED or LEDs in parallel.

I'm pretty sure at this point that the reverse voltage spikes are the problem. I'll test that by placing a 1N4005 diode in series and reversing the polarity to see if the LED survives. Its specs say if is only good for reverse voltages of 5 volts.

The Opel guy I mentioned earlier, the one reporting the -V spikes, confirmed that the spikes are absorbed at the battery. I guess my problem is that the battery, relocated to the back of the car, is about 12 feet away from the starter's power terminal and the LEDs are only 2 feet away and going in the other direction. Also, the solenoid terminal is where the biggest spikes were found, and that would feed right back up into my gauge panel where our starter switch is.