I have a 2000 civic which has a dome light that "fades" out when you shut the door. It takes about 6 or so seconds to dim out. I looked at this with my oscilliscope and when it dims out it starts flickering with PWM at 63.7hz. It goes from about 60% duty cycle down to about 5% then it shuts off. The problem is that when I use LEDs in there, you can see the flickering of the 63.7hz. With the regular factory LED bulb it will never flicker because the bulb can not heat up and glow or cool down that fast.

I made a video of this here: (sorry, it is sideways)
https://drive.google.com/open?id=149BNC7PA8Fv81TNguBFgFEfnkXbPAcyn

I was thinking of adding a capacitor to prevent it from showing the flicker so it has a some voltage during the PWM to prevent the flickering. I tried a 200 microfarad CAP, but this wasn't enough.

I measured the current draw of the original bulb and although it is 2.6ohms when it is off, when the bulb heats up it draws about 600 milli-amps. I am concerned that charging the capacitor when it is empty though. I don't know how to put a CAP big enough on there to accomplish this job without drawing too much current when the CAP isn't charged!

Can someone help suggest something?

 

What's the current draw of the LEDs?

You can use a larger cap and put a small resistor in series the limit the peak current.

 

I haven't figured it out yet- probably about the same as the factory. I have these awesome 60ma LEDs that are nice and bright. I may use 8 or 9 of these circuits at about 75% of rated current, so figure about about 400-500ma

 

This approach won't work well.

Without a resistor, it could damage the driver, by forcing it to charge the big capacitor almost instantly, effectively shorting the output.
With a resistor, it would alter the linearity of the dimming cycle, making it dim in a clunky way.

If this was a viable solution, you'd see it used everywhere, which it isn't.

 

I previously thought of (and worried about) the idea of the shorted output that you spoke of. I'm not so worried about the linearity of the dimming though. I mostly want to avoid the flickering symptom.

I'm open to suggestions on something different. I'm just looking for a rather simple solution rather than leaving the flickering.

 

Ha ..ha ... the video is a classic complete with dramatic music ... strongly recommended...

My first thought is , have you nothing else to worry about ? ...I have a dozen electrical things need fixing around the house much more urgent than that ..

I'll send you my address.

 

what about using regular power diodes to isolate the charging and draining of the capacitor.

I can use a large enough resistor to prevent overload during initial charging (to protect the driver)

during the drain- it will have the LEDs to power, so I don't need the resistor.

if I provide 2 paths and use diodes- should this be able to work? the diodes should prevent reverse flow right?

 

Imagine a bare LED with a big capacitor across it.

When charging, the cap takes all the current until it reaches the Vf of the LED, at which point current starts flowing into the LED, effectively clamping the charge voltage to exactly the Vf of the LED.

When you cut the power, the capacitor is charged to the Vf of the LED, any drop in voltage and the LED no longer conducts.
The cap will barely supply any energy to the LED, it will just hold some voltage just below the LED Vf.

You need some extra headroom to charge the cap to more than the LED Vf, additional current limiting resistors do the job.

If you added a limiting resistor before the cap, and limiting resistors AFTER the cap (which are probably already inside your LED strip?) it might work ok. The first resistor can be quite low in value, the dimmer already has to drive the cold resistance of the incandescent lamp.

The lamp resistance increases dramatically as the filament heats, a non-linear behavior of tungsten lamps.

 

Using LEDs could pose another problem. When the ignition is switched on, the car's computer(s) may do a check of the lights to detect a failed bulb and could interpret LEDs, or a large capacitor, as a dud bulb. I understand a way round that is to have a suitable resistor in parallel with the LED(s).

 

Yes the strip has many strings of 3 LEDs in series each with its own resistor. I'm not using a strip in my final design, it was just for testing.

This is what I am thinking of: