How can I get 4VDC from 12VDC?

[mikef187]

This is for a school bus and I'm trying to fix one of the flashing yellow lights. The buses used to use a incandescent light bulb, but has been converted to LEDs. The 12VDC power flashes (by design) at about 0.5 seconds on 0.5 seconds off.

I have a an LED array that had a potted built in power supply. A new costs $150. The potting cracked and made it work intermittently. I took it apart and while it was working I saw it was delivering 4VDC to the LED board. I chopped the old power supply off, and ran 5 VDC to the LEDs and they worked. I decided to solder a USB connector on the leads and plug in to a USB cigarette lighter. That worked fine, but here's the problem. The USB charger doesn't get the 5VDC started before the 12VDC shuts off again. The LEDs never light up with the short interval.

So how can I build a 4VDC power supply that runs off 12VDC that is instantly (or close to) on. The current draw is probably less than 1 amp. I can build whatever board and get whatever components needed. I have a full electronics workbench available including scope if needed.

I really appreciate any help.

[bdunham7]

An LM317 regulator plus two resistors, value to be determined by reading the datasheet.

https://www.mouser.com/datasheet/2/389/cd00000455-1795522.pdf

I think 100 and 220 ohms might be the correct values.

Or, if 5 volts will work, you can use the LM7805 type with no external resistors needed.

[mikef187]

Great, Thank you. I will look into this. Will there be a delay in the output voltage?

[Mr.B]

...Will there be a delay in the output voltage?

If using a linear regulator as @bdunham7 suggests, the delay will be minimal.
For an indicator flasher you are unlikely to perceive the delay with the human eye.
Don't use too big a filter capacitor on the output of the regulator...

[mariush]

If you want something instant, another option would to simply connect multiple diodes in series.

A basic diode like 1n400x (1n4001, 1n4002... 1n4007 - all the same just maximum voltage they can handle changes) will have a voltage drop of around 0.7v so if you want to drop 12v - 4..5v =7..8v, then around 10 diodes should do it.

You can put them on a small stripboard alternating cathode and anode ends to make it easy to solder the leads and keep it small and they'll use about as much space as a to-220 regulator and you won't need input and output capacitors and feedback resistors (if you use an adjustable linear regulator)

Example : 1n4004 (20 cents each if you buy 10, 10 cents if you get 100) : https://www.digikey.com/en/products/detail/micro-commercial-co/1N4004GP-TP/2233923

[Zero999]

Are these bare LEDs, or do they have series resistors?

4V sounds a lot like they're probably strings of two amber LEDs in series. You probably want a constant current, rather then constant voltage supply. If you run it off a constant voltage, it may be fine under lab conditions, but fail in the field. The forward voltage of LEDs has a negative temperature coefficient, which means the current will drastically increase, at higher temperatures, if they're run off a constant voltage supply.

I'd recommend the standard LM317 constant current circuit. Firstly, measure the current, to the LEDs, at 4V. Now use that value to calculate the value of R1, in the circuit below.

[Ian.M]

12V and 24V automotive electrical environments are particularly harsh with significant spikes and transients.  I would suggest using a  LM317HV, which is rated to 60V input to output differential, 20V more than the regular LM317, which on a reasonably modern vehicle with central load dump protection either in the alternator or ECU, should be capable of withstanding the likely positive going surges.    You'll probably still need a beefy unidirectional TVS diode with a clamping voltage under 60V shunting the input to  protect against brief positive spikes exceeding 60V, and a diode in series with the regulator input, to prevent reverse conduction through the regulator during negative spikes.

[NiHaoMike]

For how little time those lights operate, it makes little sense to optimize efficiency. Keep it simple with a resistor for current limiting. Add a capacitor across the LEDs to shunt out switching spikes.

[mikef187]

This is literally the first circuit i have ever built. I got my order from the parts house today, a few LM317 and a few LM7805. I did the math and figured 160 for R1 and 360 for R2. I got a handful of PCBs with 2 holes connected.

When I wired up with jumpers the LM317 circuit, it had an output voltage of 3.5v instead of my calculated 4. I fooled around for A LONG TIME and found that an 88 R1 with a 240 R2 got me 3.9v. Rigged the whole mess back in the lens housing, and I just about passed out    when I saw it ACTUALLY worked. I couldn't see any delays.  I only used 2 resisters and the LM317, no caps or anything else. If I blow it up, I'll build another one.

I really enjoyed this and this will open up a whole new can of worms for me. \

You guys really helped me and I learned a lot. Still trying to learn schematics.

[Terry Bites]

Well done. For reliabilty add a couple of cheap diodes http://www.acoustica.org.uk/t/3pin_reg_notes1.html. Dont bother with C2 in your LED application- its only added for noise redcution. Not needed here.