LED's in Series Calculation

Here's a link from about a year ago on a similar topic.

Thanks Bigg, I will scroll through it for some clarity. Your help is much appreciated!

the math is easy, I'm thinking of performance and function. you need to keep in mind a diode is pretty directional in its light pattern. your existing bulbs Emit light in every direction. you might want to experiment a little before you end up with 500 tiny spotlights

morehttp://www.whichledlight.com/led-beamangle

The big drag with putting all those LEDs in series is that if one fails they all go out (or at least the group in series).

How much fun will it be testing 24 LEDs to find the bad one in a dead set?

Be that as it may, if I understand your requirements; you have 520 sets or strings of 24 LEDs, each set of 24 LEDs is in series.

The current is 30 mA for any one of the strings of 24 LEDs (they are in series, so the current for any one string is still 30 mA).

If the voltage is 48 VDC across the string and the current is 30 mA (for each string of 24), then the resistor is R = E/I...

R = 48 VDC / .03 A = 1.6 K Ω.

The power is P = I * V = .030 A * 48 VDC = 1.44 Watts per each string of 24 LEDs

Use one resistor for each series string. That means 520 1.6K resistors at 2 Watts, not one giant resistor.

You may want to run the LEDs at something less than their full-current rating, so a 1.8 K Ω resistor will reduce the current to 27 mA. A 2.2 K resistor will reduce the current to 22 mA.

Your total draw of current for all 520 strings is 520 * 24 * 0.030 A = 15.6 Amps @ 48 VDC.

Thanks for the useful info on this, sorry forgot to mention that the PSU will be powering 1 unit divided into equal loads that will have fuses soldered in-line so if one LED fails only a tube of that segment will fail (regardless of the fuse one tube will still fail lol).

I can see your point with the Ohms law triangle, however I used this formula for the calculation;

R = (V_s - V_f) / I_f

Is this correct, I am a bit rusty with this? As you have mentioned I will use one load resistor per tube of 24 LED's (520 tubes for the whole circuit). Is it safe if I went the full wave rectification route and fed the circuit with 120VAC - DC by voltage division and load resistors, mayby throw in a cap to smooth it out a bit, I am worried though about voltage spikes as they are quite common in the place where I have to install this? Thanks again, I hope you don't mind me picking your brain a bit.

-Rob

I may be missing something here, so correct me if I am wrong, but I see it differently.

30 mA through a 1.6k resistor will drop the entire 48 volts across it. This leaves nothing for the LEDS which need 2v each.

By my calcs, 24 x 30mA 2v LEDS in series across a 48v supply will need only a 1 Ohm resistor which will dissipate just 0.9 mW, so a 1/4 watt one will suffice for each series string.

The 24 LEDs and resistor will draw 30mA and dissipate 1.449 watts.

The total draw for the array will be 15.6 amps with a power dissipation of 748.8 watts.

Good Answer for the first part.

But for the second part there's a problem: look at this part of a data sheet for a diode which approximately meets the requirements

Forward Voltage at 20mA: minimum 1.8V; maximum 2.4V.

If you happen to get a string of LEDs that are close to the max then they will hardly light at all; on the other hand if you get a string which are close to the min then you will end up with a couple of amps. Both of these situations are partially self correcting, but not enough to make for a practical solution.

....and you feel that he will pay for such a LED?

Well I don't.

Here a few pointers to bring you into the real world for a moment or two at least:-

Remember, he has well over 12,000 LEDS, if he replaces each tiny lamp with a single LED.

At single prices of almost $3.00 each for the LED you appear to recommend, even if the quantity price seriously reduces that, may still be too pricey and certainly far too powerful for the application....

He will need 5 watts for each of those LEDs, multiply that by the almost 12,480 = 62,400 Watts, not an inconsiderable amount of power for a light fitting!!!

Plus he will also probably need to supply sunglasses for everyone at switch on, a further cost!!

I think he should go for a standard LED type, hopefully not the cheapest Chinese ones, but if he drives them at say 30ma, in one year or less, they will start failing.

Repairs will be a nightmare.

I feel that he needs reasonable quality standard LED and to slightly "under-drive" them, if he wants to keep repairs to a sane level. I would suggest 3mm versions as they are as easy to use/handle as a 5mm (and far easier than SMD ones) and are available everywhere, well almost!!

I personally would not have more than say 4 LEDs in series, though more LEDs in series = higher voltage, but a single failure puts more LEDs "out".

Then if a string fails with only say 4 LEDs, he has a good chance that it will be far less noticeable and repairs can be undertaken much later when it becomes obvious.

Using RGB LEDs will bring a further plus point as even if a color fails, the other colors are usually available, so further working using different colors, especially with some form of "program" will probably cover any failures quite dramatically....

Though of course, RGB LEDs are not cheap either.....but as he hasn't told us what his budget is, we cannot say exactly...

Actually, we have simply not been told enough detail to let us make comments with any degree of accuracy, so if he really wants to make an artificial indoor REDFRED Sun using LEDs, your idea to use 5 watt versions may still be the best one here!!!

Where did I say that the OP should use that LED?

The OP does not seem to understand Kirchoff's voltage law, let alone how to calculate the power dissipated by a resistor. Yet they have funding and authority to re-lamp a massive chandelier. [We've all had brain farts, too.]

There are more contradictions than firm facts in this thread. Ranting about a misunderstanding helps no one.

I was being funny, just like you were!!!

I thought that my indoor Sun idea was pretty good! Even if it would have made a dent in the electricity bill!!

Give it but cannot take it?

Stay cool.....

1.53 ohms would not be enough resistance to do much current limiting. LEDs need a series resistor or other current limiting device because once they're turned on, the current increases a lot without much increase in voltage. If the difference between the combined voltage drop of the LEDs (24x2) and the supply voltage increased by a mere 0.153 volts, the current would increase by 100 ma.

You're going to have to go with some more serious current regulation. One possibility is current regulating diodes, although I don't know if they are available with the current you want. In any case, if the combined forward voltage of the LEDs is 48 volts and a 48 volt supply, you have very little room left for regulation. You probably need fewer LEDs or more volts.

You might want to experiment with a string of the LEDs and see how much light you get at lower currents and what the combined voltage is before you decide on a design.

I suspect that somewhere your information is inaccurate. I'm not sure if the inaccuracy is the data or your desired design configuration. 24 anythings wired in series with a 2 volt drop across each is 48 volts. Your power supply produces only 48 volts. [Let's ignore the idea that you could tweak a pot for a higher voltage. Designing any system to work outside of the nominal value is a bad design approach.) This means that your wires, connections and any current limiting device must have zero resistance. In other words, no current limitation at all.

My first design iteration change would be to consider fewer LED in the design, say 20 or 22. The remaining 8V or 4V drop will then be handled by your current limiting resistor of 260 ohms or 130 ohms respectively.

If your design must have 24 lamps then the simple option is to reduce the current flow. This will drop the voltage across each diode in an exponential fashion along with the total light produced. Then a current limiting resistance in series can be chosen. If you wish to do a nice full blown analysis of an LED I-V curve for calculation you can look into this link. The problem with both of these approaches becomes apparent when you look at the diode equation. As the junction temperature increases from self heating the current draw will change with the same voltage.

Your other more complicated option that avoids the current change is a boost-buck SMPS in current source mode but that gets much more complicated.

http://ww1.microchip.com/downloads/en/AppNotes/01047a.pdf

As I said, that is a much more complicated technique. There certainly are rewards for a boost buck topology but there are also drawbacks with hidden pitfalls.

Another error in calculation 30 milliamps (0.03 A) through a 1.53 k Ω (1,530 Ω) resistor is 1.38 watts. The voltage drop required across a 1.53 k Ω resistor to push 0.03 A of current is about 46 Volts. Your numbers are off.

Are you sure the LEDs are rated at 30mA.

Lots are rated at 20mA with a maximum of 30mA.

In any case all LEDs effectively have a "half life" which is inversely (but non linearly) proportional to the current you drive through them: so for a lighting project you're much better to "under drive" them.

Also in any case 2V x 24 = 48V, so you have no "room" to accurately control the current.

Let's say you went for 2 strings of 12 LEDs in each tube: and you wanted 15mA current. (You need one resistor in series with each string of series LEDs.)

2V x 12 = 24V leaving 24V for the resistor.

R = 24/0.015 = 1K6

resistor power = V x I = 24 x 0.015 = 0.36W

The trouble with that is that you are using up half the power in the resistors so you'd be better off going for more LEDs in each string. (Is that going to ruin the aesthetics of the chandelier?).

Choose LEDs with a very wide viewing angle.

You really need to find the forward voltage for the LEDs in this situation rather than the operating current. As already pointed out, your series string already uses the complete 48V, so no resistor is necessary to consume the unnecessary voltage.

Neglecting the trouble shooting challenges and such.

You will really also need to get "matched" LEDs from the supplier (Bin sequence for colour temperature, forward voltage and operating current) otherwise they will appear to have different intensities and seem to be different colour in the same string.

Using bin matched LEDS will also mean that minor differences in forward voltage will be shared across the string. Unmatched LEDs would see the difference focussed into one device that would then fail.

I am concerned though that your system will be unreliable unless you have very good voltage regulation on the source.

You might want to investigate either constant current sources or constant voltage sources for this project. If you have 48V input and a 48V load, then you will be dumping very little power in that device (Much less than the resistor designs) and achieve a higher reliability.

Hi Guys,

Thank you all for your very useful information, I am going to draw up my circuit and post it on this thread so you can see my potential problems I may have, I am very grateful for the assistance, I hope you are all well, take care.

-Rob.

"Unmatched LEDs..."

You mean the bargain Chinese flavor?

No, I mean manufacturer graded and sorted for colour temperature, forward voltage and actual light output (apparent intensity).

Most manufacturers will "bin" their production into maybe 20 grades based on these parameters. When you use items from the same "bin" they appear consistent to the human eye and also perform as balanced electrical components.

We were using around 60,000 per day through hole LEDs in amber or red and had to run bin control in our facility.

Apparent colour might not be an issue in this case as the apparent separation there would be and the "cloud" effect of so many in the light fitting might mask that, but the forward voltage at operating current would be important to make sure that one in the string doesn't overload.

(Note that we were buying our bare PWB materials from China at the time and found them to be at least equivalent to the USA alternatives for product quality and durability, far superior for delivery timeframe and responsiveness, while needing a little more management on "first article" development. First article prototypes delivered to our site in Australia within 5 days of their receipt of files, production quantities within 14 days once approved.)

We also sourced around 1,000,000 components per day from China and south East Asia, though I must admit our LEDs came from Poland.

Our products were exported back to South Korea and a few other places. Heck, we even sent a few bits to Dettroit.

Its going to be a mammoth task........

May I suggest a few changes:-

1) Reduce the number of LEDs in each tube. As you haven't shown us any pictures, we cannot offer much help there....but replacing each lamp with a LED will really test you out, no more than 12 LEDs per tube, 8 may even be better. Construction will be then far easier....but not easy!!

2) Consider buying some LED controllers from ebay, preferably RGB ones so that you can dial in almost any color you wish. As a starting point look here, but you will either need several of them or one with a higher current control for your new design. These work using pulse width electronics and they are all finished and ready to use.

Do pick one with a far higher current drive, this is just as an example only....

RGB LED Controller

3) This will save you a lot of time in designing and will give you different colors and many different programs with finger tip control......even if at the end of the day you only actually use one color!!

4) You may want to go with tiny SMD LEDs, they will give a better "Star" effect....but are fiddly to solder.....you may need to buy some special tools to get the job done. Your choice.

5) 30ma per LED is far too much anyway. High intensity LEDs of even 3mm size can be found that will be bright enough and very long lived, with only 5ma of current....some may even take less and still be bright enough - test - test and test!!

6) Low current means generally a longer life or less repairs.

7) Think the finances through as well....good quality LEDs are not cheap in the sheer numbers you are looking at....you are looking at 12,480 plus spares......

8) Buy only ones with clear lenses, that way it will still look good even switched off. They are available in all colors....

You have already been given a lot of good tips from other members. Think long and hard.....

May I suggest 3mm LEDs, either single color or RGB as being the easiest to solder, but not as big as the 5mm, so a smaller size of light source.

Now you need to sit down, do some simple testing, make a design for one unit and you will probably need to get some PCBs made to hold the LEDs, then a paper tube or similar with holes in it to allow the light out.....!!! Your own ideas count more.

Test everything.....

Just a thought!! Its the way I would go if needed....I wish you luck.....

Do you know what the light output was of the original incandescent bulbs?
I said before to choose LEDs with a wide viewing angle. Bear in mind that: say, a 100mcd 20° will output nearly twice as much light as a 200mcd 10° one. The calculations get a bit more difficult for larger angles but it's worth doing, and I think you should be looking at viewing angles greater that 120°.
Are you restricted to 24 lights per tube?
Also bear in mind that two identical diodes with 10 mA going through them emit slightly more light than either one with 20 mA. This effect is very small but, it's in the "right direction" if you want to reduce the drive current per LED.
Have you considered putting all the LEDs on a long thin PCB (about 20 surface mount LEDs on either side seems reasonable.)
Have you considered using a diffuser, just a small tube to cover the string of LEDs to "disguise" their directionality a bit.
If you are going to use a PCB then you might as well include a simple current regulator. If you just use a resistor to regulate the current then either you will need to waste a lot of the power in the resistors, or, you will need to do some kind of select on test to get the correct value of resistor for each string of LEDs.

I did this back in the 80's for lighting on the sides of runningboards that I manufactured and sold. We designed the LED for 12.5 volts and the LED's we used I think 5 in series worked well but the intensity of the light varied with the vehicle charging system. I had no problem with life but the variation in intensity was an issue to some on the other side if you were driving and the lights were not shining bright it indicated that your charging system was failing. We switched to a transistor control circuit that was much better in all functions and the product was a real winner once I solved the moisture issue. Our diode supplier claimed I was the first to use LED's successfully on the outside of a vehicle. Not bad for a farm kid at the time along with a friend that was electronics technician. Our inventors group has members that have a fantastic LED lighting conversion system from 4 & 8 foot fluorescent to screw in bulbs.

Roy:

Can you PM me on that flourescent to LED conversion system? I need to know prices in the US, for about 12 tubes, since I'm converting everything in my house to LED, and I have several 4 foot, 4-tube units in various spaces, including my workshop. The price I see from IKEA is USD40 per tube, and I can't afford that. But I'd rather relamp than install new fittings, if I can.

Micahd

1000bulbs.com has them for a lot less than that.

I'll check them, too. Thanks for the tip.

Why can't you use smt led strips that are already color matched?

Mount the 2 12 led strips back to back, in series with one current resistor at 20 ma max to reduce the heat. I just replaced a 24" flourescent lamp with a led strip that has 100 smt leds with matched output. It only draws 6w and puts out 3x the light. Use a led dimmer for the leds to adjust light out, as hard wiring a chandelier you will not be able to adjust light out to suit conditions, and if the leds are dimmed to any degree, they will run cooler and longer.

I think some is actually building a T.A.R.D.I.S. control panel.

Hi Guys,
This is what it looks like. The tubes are too small to put the LED strips in so it has to be soldered LED's. I am just looking for the cheapest method, at present there are no criteria's, the unit just needs to be converted to LED, I went with 24 tubes as they fit nicely in the tubes.
Thank you for all your suggestions so far!
- Rob.

That's an impressive chandelier. The last thing you need are retrofit lamps that make this look tacky or cheap. You must get a better understanding of the lamps mounting than what this image shows. Start by identifying the replacement lamp model number or style number. Once the incandescent lamps have been identified then and only then can anyone attempt to retrofit these lamps.

My first choice would be a suitable replacement LED bulb that takes the power distribution voltage of your location. [This is a remote possibility but if one does exist and you never looked...] My next choice would be to contact the LED manufacturers for the name of a custom bulb engineering firm to design and make suitable LED replacements with the manufacturer's parts. This custom design could use either the SMPS driven lower voltage or line voltage to the chandelier. I like the SMPS approach since the most unreliable part of most power supplies will be the capacitors not overhead. My last choice would be to design a replacement bulb myself and then try to find someone that can fabricate this many replacements.

Why you don't use "EL Wire"? (Electroluminescent wire)

These are just two pages, but you can find a lot of information. You just have to be sure that luminosity is enough for your purpose.

http://en.wikipedia.org/wiki/Electroluminescent_wire

http://enlighted.com/pages/customlightel.shtml