How to calculate ripple/smoothing cap for vintage motorcycle LED headlight conversion

Question

New here. Very little EE background but am tinkering with a project that has so far been nicely straightforward. However I've run into a real snag.

I have a few dozen links so it's just as well that I cannot post them as a new member or it might overwhelm.

1980 vintage motorcycle. Separate lighting and ignition coils. Nominal 6v system generates 4vac ~ 16vac in the RPM range.
Maximum power from lighting coil is 27watts. Ac cycles vary along with voltage @rpm. I cannot yet find a source for that ac cycle range. Headlight is presently 6v 21w incandescent; anemic orange at idle, and almost useful at full RPM.

What I'd like to do is take fullest advantage of what little is on tap here to power paralleled LEDS. So far, all from Digikey's site I've received....

rectifier GBU10A-BPMS-ND

regulator NE12S0A0V06PNFA

SMD LED (pair) XMLBWT-00-0000-000LT50E4CT-ND.

I'm now left with two calculations that are beyond my rudimentary electronic skills. How can I calculate the largest number of those LEDs, wired in parallel, before I run up against the 6A limit of that regulator? And, apparently, the toughest of all, the calculations for an optimum ripple/smoothing cap on the rectifier are so far beyond my abilities that I'm reduced to begging for someone to help me select one from digikey.

If discovery of the the ac cycle range is needed, and cannot be determined from the owner's manual itself (I can link to it), I will find a way to determine it if it's necessary. MY DMM doesn't do Hz, but can't it be calculated by
simply counting the number of coils in the lighting coil and multiplying them by the rpm range from lowest to highest? Maybe a little added algebra?

Thank you very much for any consideration of this.

Answer 1

It sounds like what you want is a LED power supply that has a wide input voltage range and drives the LEDs with a constant current, so long as that amount of power is available.

Since the input voltage can be 4-16 V AC, you probably want to use a boost converter. A buck wouldn't leave much room with 4 V input.

First full-wave rectify the AC to make DC. The peaks will be the square root of 2 higher than the AC voltage, minus two diode drops. At this low voltage, you can use Schottky diodes, which drop less voltage and full silicon diodes. To be pessimistic, assume the the Schottky full wave bridge drops 1 V. That leaves the peak voltage coming out of the bridge to be 4.7 to 21.6 V. Put a nice a big capacitor on that, and you have a rough DC power supply.

String enough LEDs in series so that the total needs more than 22 V to light up, then have a boost converter power the LED string from the 5-22 V input power. Instead of regulating the voltage to the LED string, have it regulate the current thru the string. To get a voltage proportional to the LED current, put a small current sense resistor in series between the bottom end of the LED string and ground. For example, if the LEDs are meant to take 200 mA, then a 2.7 Ω resistor will drop 540 mV when the LEDs are running at the intended current.

Answer 2

The Olin's solution is worthy, but I'd ask you why shooting at the mosquito by a bazooka? I don't know the specific led that you want, but I own some high-power led and it looks that it shines enough even with a pretty low current. A further current increment yields a minimal lighting improvement.

Said that, I'd try (note: try) to wire a led, along with its resistor tuned for the worst condition, which seems to be 16Vac. I mean that you should take in account the coil internal resistance: when the old bulb is connected, the coil voltage is still 16Vac? I hope not, otherwise the bulb should blow in a few (or living shortly).

BTW, it might be useful a normal (or schottky) diode in series to the coil, thus resistor and led. The meaning is to cut-off the negative half-wave. However, it's useful only if the reverse breakdown voltage of the leds is smaller than the generator peak. But it should be so: the led's reverse voltage is typically low enough.

A final idea could be to wire the two led chains (led+resistor+diode) as in anti-parallel pattern, so that each half-wave will energize at least one led branch.

Hope it's clear enough.