Hi everyone, I'm currently working on an LED array of around 80 leds to be used as a spot light (starting small, would like to use this on a 200+ led array if all works out).

I primarily want to run this off a car battery, both when the car is running or not and therein lies the problem. Once that works, I'd like to be able to power this from 240VAC as well.

Originally I was going to use resistors, but then brightness is affected when powered from the car. I then tried voltage regulating with an lm317t followed by resistors, but that was far too inefficient and made the regulator too hot even with a big heatsink (which I didn't want anyway).

So that brings me here, I stumbled accross a thread mentioning constant current buck converters and it sounded good, so thought i'd give that a shot. I've done quite alot of searching but still don't quite understand the whole concept yet and alot of what I found was either relating to high voltages/currents, or low current circuits well under the 1.6A+ that I need.

I'd also like this to be as physically small as possible and as efficient as possible, without going overcomplicated.

By the way i'll be using white 3.6v 20ma 5mm leds.

Thanks everyone.

 

You'll need to read something like this...

LEDs, 555s, Flashers, and Light Chasers

You can have 2 or 3 LEDs in series, each leg with its own current regulation resistor.

First, a car's 12V electrical system is notoriously unregulated. With the engine running it is around 13.7VDC, with it off and strictly running off batteries it is 12V. This shift in voltage has major consequences in the design of something like this, which is explained in detail in the article.

Let us say you have the 12V stability issue licked, and have 3 LEDs per chain, each with it's own resistor. The resistor is a fundimental requirement of LEDs that bothers a lot of people new to the devices. You can make it more complicated, as in adding a current regulator per chain, but you can not make it simplier, and the end results are not noticable. Assuming each leg of 3 draws 20ma, 81 such devices (multiple of 3) is going to pull 0.54 amps, not too bad, and 201 devices (another multiple of 3) is going to pull 1.34 amps, again not too bad.

There are some minor efficiency increases to be had by using higher power supply voltages, which is where a buck boost convertor comes in. A buck boost convertor can convert 12VDC (unregulated) to a regulated 30VDC, for example. If you have a higher DC voltage you can use more LEDs per chain, which reduces the number of resistors and will get rid of some of the heating, which is minor to begin with.

Running such an arrangment off 240VAC is minor, as this is an off the shelf commercial power supply.

 

As Bill said, the Buck boost converter is a good idea, though it may need more devices and space.

Another option would be to use 2 7809 regulator(1A) or 1 along with a power transistor (~2A) to totally regulate the voltage (~1.5A required) to 9V irrespective of the engine state. Then you can use 3 LED's in a series chain with 1 resistor each. It wont be that efficient, but can't get it simpler.

The same can be done with LM317 to regulate it to 11V and run 4 LED's per chain rather than 3 and LM317 can handle 1.5A with no power transistors.

But the best option would be as suggested by Bill, to run them using a high voltage (~30-40V) very low current (3V x 13 LED =~40V so 6 strings to give 78 led's and 120mA only) Buck booster to run several LED's in one chain with 20mA current.

So if you can get a buck booster at about 40V 150mA, you can do it easily and more efficiently than a voltage regulator which is easy on component count.

Also Bill, I have seen circuits running white LED's from a 1.5V using CMOS 555, cant we do that here??

 

9V is 2 LEDs per chain, which will bump the current up quite a bit because more chains will be needed. This may not be a bad thing, as the predictablility goes way up, and the difference between a running engine and battery only is minimized. Given this senario, a regulator wouldn't even be needed.

You always have to add the Vf, which can not exceed the power supply voltage. You should also allow at least 1V above the added Vf. This is because the Vf is not exact, there is a slight variance between several devices. So the math looks something like this, 3.6Vf X 3 is 10.8V, and the 1V clearance is 11.8VDC.

You can use a Joule Thief approach, but usually simplicity is best (unless there is a technical reason for other), which the OP has specified.

 

Why not consider some of those 1W (or higher) LEDs? Just one of those would replace 14 or more of your little 3.6v@20mA super bright LEDs. It certainly would save a good bit of space, not to mention assembly time.

 

People keep asking about whether a constant current source is needed with the 1W high intensity LEDs? What is your opinion Wookie?

I suspect one 1W LED is worth around 20 of the other LEDs, though I don't have the math or inclination to back that up. They are pretty much like regular LEDs, except they take around from 350ma to 700ma to light up. It is possible one of these 700ma series could replace 80 of the other, but they are expensive.

Google phillips "Rebel" series.

 

People keep asking about whether a constant current source is needed with the 1W high intensity LEDs? What is your opinion Wookie?

Yes.
The "simple" way is a linear regulator. The regulators can get mighty toasty and it's not very efficient.
More efficient is a "buck" type switching regulator which uses an inductor and a MOSFET switch. Even more efficient is a synchronous buck, which uses a 2nd MOSFET as an "ideal diode". However, it gets pretty complicated and/or expensive for a hobbyist to try to design and build such things when products are on the market at fairly reasonable prices.

I suspect one 1W LED is worth around 20 of the other LEDs, though I don't have the math or inclination to back that up. They are pretty much like regular LEDs, except they take around from 350ma to 700ma to light up. It is possible one of these 700ma series could replace 80 of the other, but they are expensive.

Well, one's time is certainly worth something. I'd rather string together a dozen or so 1w LEDs than a couple hundred of the little ones. 200 of the standard T3 or T5 LEDs would mean at least 400 solder connections. That represents a fair amount of time just to solder & trim the leads.

 

thanks for the input so far everyone. Maybe I should be more specific about my needs.

Most important is efficiency, as I'd like to run these for as long as possible off the car battery.

Second of importance is size, I don't want an atx psu size box to run these off, but if I have to to achieve my main goal of efficiency then so be it.

And last of all simplicity, this I'm not too worried about as I wont be building anymore than 2 200 led lights, it's only got to be simple enough to build on perfboard or stripboard.

Now, before I go on, i'll show you what I have just recently completed on single hole perfboard

 

That is what I call my "Pimeta Breadboard" its my own homebrew pimeta headphone amp but with my own virtual ground and powersupply circuit. There are profesional PCBs for this (well v2 actually) coming soon but I couldn't wait, and with this I can change the values of everything I would want to.

Anyway, the point is, I didn't have any problems what so ever building this (except for my low light work area). The worst part I found is the board layout, which I didn't put enough time into (as you can tell from the blue jumper wires).

I would like to steer away from using the leds in series, if 1 blows I could loose the whole lot meaning i've then got to resolder 200 leds. The reason I am using 5mm leds is because I recently bought 10,000 of these for a steal at 1.8c each from an online auction.

I really just need to know the most efficient way of going about this, I'd like to know more about these constant current buck? converters, and how to go about building one to suit my needs.

Thanks again everyone.

 

also , out of interest, when looking at the specs 3w leds the brightness is given in lumens eg 250. My leds are 8000mcd, how do i calculate how many of these to create the same light as the 3w star?

 

There's not an exact conversion. You need to know the mcd and beam angle.
Here's an online calculator: http://led.linear1.org/lumen.wiz
Here's an explanation: http://led.linear1.org/how-do-i-convert-between-candelas-and-lumens/1/
You may very well discover that the bargain LEDs you purchased don't emit as much light as claimed; at least not for very long.

If you wired up 200 of your white LEDs in parallel, you would need at least a current limiting resistor for each one, and you would need a 200 x 20mA = 4A current supply at somewhere around 4.2v to 4.6v.

If you built an array of two LEDs in a string with one resistor per string, 100 strings, you would require 2A current at around 8v. Etcetera.

Automotive batteries are not designed to be used for deep discharge. They're designed to quickly start an engine (brief, heavy current) and then to be promptly recharged fully. If you discharge an automotive battery very significantly, it will have a very short life.

 

Added to the above, you will waste a lot of energy as heat on the resistors with least efficiency. The more LED's you need to power, series is the way to go.

 

If you crowd 200 3.6V LEDs together at 20mA each then they dissipate a total of 14.4W and the white phosphor on them will dissappear.

I made a night light with 36 Luxeon superFlux LEDs at 53mA each (their max is 70mA).
It gets extremely hot and a white LED quickly turns a dim blue when its phosphor dissappears due to the heat.

 

There's not an exact conversion. You need to know the mcd and beam angle.
Here's an online calculator: http://led.linear1.org/lumen.wiz
Here's an explanation: http://led.linear1.org/how-do-i-convert-between-candelas-and-lumens/1/
You may very well discover that the bargain LEDs you purchased don't emit as much light as claimed; at least not for very long.

If you wired up 200 of your white LEDs in parallel, you would need at least a current limiting resistor for each one, and you would need a 200 x 20mA = 4A current supply at somewhere around 4.2v to 4.6v.

If you built an array of two LEDs in a string with one resistor per string, 100 strings, you would require 2A current at around 8v. Etcetera.

Automotive batteries are not designed to be used for deep discharge. They're designed to quickly start an engine (brief, heavy current) and then to be promptly recharged fully. If you discharge an automotive battery very significantly, it will have a very short life.

Ok so series is the way to go then. Don't I still need some form of constant current to avoid the brightness fluctuating with the voltage?

And as far as batteries go, I do realise this, but my 4x4 has a dual battery system with a deep cycle battery, which I can isolate.

 

I'm starting to wonder if I should just use a couple of 3w luxeon stars. I really wanted to use some of the cheap leds I bought more than anything.

Wont I run into these same problems with the stars?

 

Lets see, .02AX3.6 is 72mw. Not very much power there. If you were to pack them at max density (hexigon layout) it still wouldn't be that hot. The problem is the resistors, when can be a bit of distance from the LEDs. One watt would take 14 LEDs. The key in my opinion is to allow ventilation, maybe even a fan. The main thing with heat disappation is to not lock it in.

Heat is always a problem with electronics, and yes, you would run into this problem with high wattage LEDs.

Another possibility, drill the holes that will mount the LEDs in a sheet of metal. The metal will help move the heat away from the parts.

 

Normal 3mm and 5mm LEDs dissipate almost all their heat through the cathode leg. That's the one with the little metal cup that holds the silicon, you can see it clearly on clear LEDs (sorry for the pun).

Often in remote controls (that run a 5mm LED at high pulsed currents) they heatsink the cathode leg to a large copper plane on the PCB. If you were soldering the LEDs into veroboard (stripboard) then the strip that connects to the cathodes could have a lot of solder mass and even a thin metal plate or something attached?

 

also , out of interest, when looking at the specs 3w leds the brightness is given in lumens eg 250. My leds are 8000mcd, how do i calculate how many of these to create the same light as the 3w star?

Compare MCD, it is simplier.

What I would do is use the small LEDs, plan on using 13.7V, and go from there. It is simple, relatively cheap, and efficient.

If this has to use 12.0V, then the complexity notches up a step. This is when I would plan on a buck boost convertor to take them up to 27V, or some other higher regulated voltage (emphasis on regulated). The end circuit (besides the buck boost covertor) is still simple and efficient.

Think 3 LEDs in series, each with 3.6Vf. This adds up to 10.8VDC. 13.7VDC - 10.8VDC works out to 1.9V. To calculate the resistor 1.9V/0.02A = 95Ω, which ≈ 100Ω. Going backwards, 1.9V / 100Ω = 19ma.

Repeat this circuit as many times as needed.

Here is the example I gave in LEDs, 555s, Flashers, and Light Chasers.

The components are different, but the schematic is the same.