LED lighting on a budget!

Hi kcress,

I've been looking at all kinds of LED threads and I don't really think I'll be going your rought, but I just wanted to say that your opening post was one of the most well explained postings I have ever seen. I wish you had taught my physics class. Well done good sir.
 
Hey Humuhumunuku!

Physics?

F=ma?

x = x<SUB>o</SUB>+ v<SUB>o</SUB>t + 1/2at<SUP>2</SUP>?

KE = 1/2mv<SUP>2</SUP>?

E = mc<SUP>2</SUP>?

Great stuff. I loved physics. Was an EE major but had keys to all the physics stuff and hung out there. I did my homework in the center of a ten meter anechoic chamber. Nice and quiet.. Have a minor in physics.


It's always nice to hear positive feedback. Thanks neighbor (Chris Rene's town).
 
I have been quiet here waiting for someone to bring up some ideas and thoughts on this system that agree with mine. I personally have worked a lot with low powered LED's and just recently started working the high powered LED's. Initially my thoughts were exactly as yours are with resistor network however I have moved away from that train of thought on the high powered LED's.

From a cost perception one can simply run something like Lightteck LED driver to run 6 LED's for under $20.00. Fortunately these drivers have the AC/DC converter built right in them and they will handle up to 48 Volts meaning you can easily run 12 LEDs off one of these Drivers. Sure if you want to have an adjustable driver it will cost you more. But unless your driving many strings this is more cost effective than your method do to the cost of a good power supply alone.

Now if you are running multiple strings the better option in my mind is running your own Current Regulator. There are many current regulator chips out there and the ones that will handle the current and voltage we are looking at are well under $5.00 each.

Next is the issue of safety once you have your system up and running. Yes in the resistor method you can easily adjust the voltages. But we also need to keep in mind the resistance of the LEDs is not a 100% constant. As temperature increases on LED's there effective resistance does drop. If each LED is not identically, with identical solder connections, and heat sinking one LED might hit 40 Centigrade while another might his hit 60 Centigrade. With the effective resistance changing changing the voltage distribution between the LED's will vary with the cooler LED's now drawing more voltage than the hotter ones. If a single LED were to die creating a short it would increase the voltage to the others possibly by even 10%.

Next you have the stability of the power supply. Sure you can get a low cost 24 volt power supply but how stable is the voltage. When you get a more expensive power supply they use precise voltage regulating circuits that will give you +- 1.0% as opposed to +- 10%. Look at your line voltage it also varies by nearly 10% from 110 Volts to 120 Volts will your power supply automatically correct for this? With a current supply this is not an issue.

My fear is when you have a run of 10 LED's in series costing $5.00+ each and one dies with the resistance system will you end up replacing all 10 of them? Looking at the spec sheets for most LED's you find that 10 % change in Voltage will cause the Current to Change by as much as 100%. With a target of 700ma we may find the ideal voltage but if it rises by 10% on any one LED we may be finding ourselves with 1,400 ma on that LED roughly 40% over its maximium safe limit. Controlling the current then becomes a much more exacting way of stabilizing the system.
 
In following along with this tread and learning all I could about parralell strings of LEDs I found a circut on another site called a "current mirror" that is supposed to match the strings curret with upto 99% accuracy and adjust for voltage drift automaticaly. I posted a schematic in a different thread: http://www.reefcentral.com/forums/showthread.php?t=2110208

Not sure that it really works any better as I haven't yet tried it but it also seems like it would help protect the strings as if an led in the 1st string goes out the circut shuts down the second string. You still need a fuse to protect the fist string though, in the event the second string fails.

Anyway, just another approach I found that goes along with the topic of this tread. Great thread by the way Kcress!
 
This may be getting a little off the original topic of this thread but in the spirit of LEDs on a budget I found this fantasic sounding driver while searching for speaker components for my "other" hobby. http://www.parts-express.com/pe/showdetl.cfm?Partnumber=073-050

It's by Sure Electronics and seems to function like a dimmable buck puck only it costs less than $5! and runs at ~800ma. It seems to require more than 2 volts of head room from the power supply though, unlike the buck pucks the data sheet says you need 27volts to run 6 led strings of white, blue, or green leds, but that may just assume cheap 4.x volt leds.......

Has anyone out there tried these? the price seems too good to be true so you likely get what you pay for, but they look like they'd be worth a try.
 
zachts; The current mirror gambit works. The problem is it's very messy to physically implement. The integrated circuits and resistors etc., etc. need to be built and heat-sinked somehow. The method described in the first post of this thread uses easily mountable parts made to be screwed down for mounting.
 
This may be getting a little off the original topic of this thread but in the spirit of LEDs on a budget I found this fantasic sounding driver while searching for speaker components for my "other" hobby. http://www.parts-express.com/pe/showdetl.cfm?Partnumber=073-050

It's by Sure Electronics and seems to function like a dimmable buck puck only it costs less than $5! and runs at ~800ma. It seems to require more than 2 volts of head room from the power supply though, unlike the buck pucks the data sheet says you need 27volts to run 6 led strings of white, blue, or green leds, but that may just assume cheap 4.x volt leds.......

Has anyone out there tried these? the price seems too good to be true so you likely get what you pay for, but they look like they'd be worth a try.

Looked at this driver and it will only drive 6 LED's plus needs a DC voltage source. If you add the cost of a DC voltage supply your cost is greater than one that converts drirectly from AC voltage. Now if you have multiple stngs of 6 that you want to control seperatly then it might be financialy feasable.
 
Looked at this driver and it will only drive 6 LED's plus needs a DC voltage source. If you add the cost of a DC voltage supply your cost is greater than one that converts drirectly from AC voltage. Now if you have multiple stngs of 6 that you want to control seperatly then it might be financialy feasable.

Well, that's true, but I happen to have a box full of various power supplies salvanged from all sort of electronics, mostly laptops and the like. I'm also thinking of using these for a small scale prototype build to play with color mixing and find what I like before going all out for display tank. Then once I scale everything up these could get recyled as moon light and fuge light drivers.
 
If a single LED were to die creating a short it would increase the voltage to the others possibly by even 10%.
My fear is when you have a run of 10 LED's in series costing $5.00+ each and one dies with the resistance system will you end up replacing all 10 of them? Looking at the spec sheets for most LED's you find that 10 % change in Voltage will cause the Current to Change by as much as 100%. With a target of 700ma we may find the ideal voltage but if it rises by 10% on any one LED we may be finding ourselves with 1,400 ma on that LED roughly 40% over its maximium safe limit.

TropTrea: I agree that more needs to be done to protect the LEDs. Now, my thought runs a little against the grain of the "budget build" thread, since it adds about $.60 per LED. Placing a shunt on each star will protect the series string against voltage/current spikes caused by failed LEDs, there-by protecting the other strings, in parallel systems.
I first saw this idea in Nuclearheli's build-V.2 [page 33 of "210 Upgrade Build with LED Build"] sorry, I don't yet see how to post links within RC. The shunt is made with a PLED6S (by Littlefuse) on each star. the PLED's are surface mount but can be modified. nuke did it and so did Mike Dani. Wish the suppliers would offer it right on the star....are you listening Cutter, Rapidled,et.al.??
Kcress pointed out the fact that DIY'ers can easily replace failed LEDs, blown fuses and the like- which by the way can cascade to multiple strings in big parallel systems. but not all canopies and fixtures are that easily removed for soldering,etc. Plus a multi-string failure won't wait long and the fuses are not cheap either. Just ask someone who has had a single blown LED take out the fuses in all the parallel strings on a big driver system. With shunts, a failed led is a tiny problem that will wait, or can be ignored. In a 48 or 72 LED system you will never notice one missing LED.
as far as the cost increase goes, look at the $$ spent on extra LEDs (to compensate for driving below max.), oversized expensive heatsinks, fans, temp controllers and the like, to extend the life of LEDs.

IMHO, I don't feel this shunt idea is CRITICAL for diy builds, but it is a relatively simple way to avoid trouble in the long haul. If these LEDs really do last 50,000 hours, then we could have years of trouble/maintainence free service from our creations. Note to Those companies charging thousands of dollars for their fixtures: YOU better be on top of this or there will be a backlash.:furious:
Kcress, I'm not dissing your opinion of PLEDs, I just want people to know the option is out there, cause it has not been talked about very often- so I will beat the drum just a little longer:deadhorse: I surely value your expertise in EE and your experience with DIY. The amount of help you provide and the patience you display is staggering. keep up the good work.:beer:
 
Hi Dirrk.

I see shunts as:

1) Twice as many solder joints - each a small liability.

2) A waste of money as demonstrated by probably half a million DIY builds happily existing without them.

3) Pointless in the common case of a shorted LED.

4) Pointless in a parallel build. (As matching will be trashed anyway.)

If you think shunts make sense I have to ask, "Are you wearing diapers because they 'add some measure of safety'"? :p

Do you wear a helmet while driving?;)

I see shunts as the same thing as the these two absurd examples.:D
 
I have had older style "100ma and lower" LED's that have lasted for 20 plus years. Sure the newer LED's are pushing the edge a little more but there should not be any reason that they fail outside of excessive heat or voltage applied to them. The big thing in my mind is not how much protection you run on them but how close to there known failure point you actually run them.

Most 3 Watt LED's are actually running at about 700 ma and 3.8 Volts. They have been tested at up to 4 volts and run a little over 1 Amp in that range pushing them to around 4 watts in that rang but there life span is probably drastically reduced. To me if you look at there efficiency charts they are not the most efficient when you max them out but simply increasing the risk factor drastically for just a little more light.

The LEDs I'm running are 3 Watt LED's at 350 ma and the Voltage they are drawing is just about 3.24 volts meaning I'm running them at 1.1 Watt. In experimentation I tried them at 700 ma and the voltage required was about 3.8 volts meaning they would have ran at about 2.7 Watts.

Where the issue comes in when your mixing and matching types and manufacturers. You might have that series of LED's set at 700 ma but one type might be consuming 4.2 Volts while another might only be using 3.5 volts. Perhaps there is a difference in the heat dispensation between some of these causing the effective resistance to vary. But if you plan for the worst case scenario in your design this should not be an issue.

One simple way would be if you could take a 700 ma driver that will remain accurate at low voltages for test purpose. Tank each individual LED and run it alone testing what its actual voltage draw would be at 700 ma, and mark them with the voltages. Now when you putting two string together mix and match the LED's so the total voltage number add up to as close to the same value as possible. When your done you will have a fairly matched pair of strings.

It is my belief that if you run those string of 3 watt LED's at 650 ma that they will not give you any issues provided your voltage is in the recommended operating range of the Driver. Yes voltage will increase from when you fired them up cold to after they were running for 12 hours but the change on a string will show up at less than 1 volt difference. Now if your running them at near max 100ma per LED's these changes may be much more noticeable.

For those that have already built large LED systems how often do you see failures? How often do you measure your voltages and current and actually find a measurable change?

I remember when I first looked at LED lighting. The average 1 Watt LED was selling for around $17.00 without a star mount. Today you can find 3 watt LED's that are pushing below the $5.00 point with star mounting and lenses. If this continues by the time we need to replace our LED's they will be under $1.00 and running at 10 Watts each. Our power supplies will be the biggest cost of our systems.

I have seen comparisons where 75 Watts of LED lights is the equivalent to a 400 Watt Metal Hide light. While a 120 gallon tank usually runs a pair of 250 Watt MH's it should only need about 95 watts of LED's then. With 3 watt LED's that are running at 2.5 Watts each that means 38 should be more than enough light. So in reality I wonder how many DIY fixtures are really being over built ?

My next fixture I'm looking at 10 Royal Blues, 8 Blues, and 6 Neutral White LED's for a 40 gallon breeder tank. All running off of three 700 ma Drivers with separate schedules for each Driver.
 
I just lover your comparisons. Could not agree with you any more. Remember to wear your galoshes every time you outside you never can tell when it will rain.


Hi Dirrk.

I see shunts as:

1) Twice as many solder joints - each a small liability.

2) A waste of money as demonstrated by probably half a million DIY builds happily existing without them.

3) Pointless in the common case of a shorted LED.

4) Pointless in a parallel build. (As matching will be trashed anyway.)

If you think shunts make sense I have to ask, "Are you wearing diapers because they 'add some measure of safety'"? :p

Do you wear a helmet while driving?;)

I see shunts as the same thing as the these two absurd examples.:D
 
I just love your comparisons. Could not agree with you any more. Remember to wear your galoshes every time you outside you never can tell when it will rain.

vpc46w.gif


I expect Dirrk's a good sport and won't take my examples personally.
 
If your really worried about loosing a large quantity of LEDs after a fuse blows than perhaps another option would be running fewer LED's an each driver and more Drivers.

Lets look at that 72 LED array. If we were to run this off of a single 48 volt power source and use a simple power transistor as a current regulator for each string of 6 LED's the most we would have would be a 1/12 loss in lighting if an entire string were to burn out. But chances are we would only loose a string until we ended up replacing that 1 amp fuse and one of the LED's.

Knowing the nature of LED's I still believe issues will show up the first few days of operation mostly from poor soldering and once the units are stable they will basically maintaince and trouble free, unless your fish room is struck by lightening.:blown::blown:

Having done life tests on electronic components I will say 50% of failures occur in the first few minutes of powering them up. After that the longer a unit is operating the less likely you will have any problems with it. Failures after 24 hours can usually be isolated to stressing the component beyond its design.:headwally:
 
If your really worried about loosing a large quantity of LEDs after a fuse blows than perhaps another option would be running fewer LED's an each driver and more Drivers.

Lets look at that 72 LED array. If we were to run this off of a single 48 volt power source and use a simple power transistor as a current regulator for each string of 6 LED's the most we would have would be a 1/12 loss in lighting if an entire string were to burn out. But chances are we would only loose a string until we ended up replacing that 1 amp fuse and one of the LED's.

Knowing the nature of LED's I still believe issues will show up the first few days of operation mostly from poor soldering and once the units are stable they will basically maintaince and trouble free, unless your fish room is struck by lightening.:blown::blown:

Having done life tests on electronic components I will say 50% of failures occur in the first few minutes of powering them up. After that the longer a unit is operating the less likely you will have any problems with it. Failures after 24 hours can usually be isolated to stressing the component beyond its design.:headwally:
 
Hi Keith,

Thanks for the reply. I will definitely start PMing lol. Now the problem I have right now is I don't remember what I actually asked since you deleted the post :headwally: lol.

http://www.mpja.com/3-12V-2A-Selectable-Output-Supply/productinfo/9902+PS/

so if I used the power supply in the link, by changing the voltage, the amp of the wires will change? So dimming is not a good idea for your design? If it isn't, I will do that non-dimming method you have written out.

I do have another question. The meanwell driver allows you to set a constant current. On your design, we used total of 10 strings so we need a power supply that can produce 7A. You used an 8A power supply. What happened to the last 1.3 amps? Are they just not used or the power adaptor have a way of controlling its current output?

From what I read elsewhere, it says whatever device is hook to the power adaptor, they will only draw amount of current it needs.

The lowest running current for the LEDs are 350mA (XP-G) at 3V. So I am assuming the systems knows to take .7A instead of .35mA because of the 3.2V?

Thanks,
Peter
 
Curious as to invest in these or not
http://www.meanwell.com/search/hlg-120h/default.htm
Seem like a good way to drive these yet run around 92% efficency

Thoughts?

This sounds like a nice 120 Watt Driver. On a large build one could run 40 LEDs off of it at 3 Watts per LED. If your planning a big build Would definatly consider it. Only thing is what is the price for these? Sometimes you can run 3 or even 4 seperate drivers for less money than one big driver. You need to weigh in the cost difference.
 
On some thoughts on comercial LED fixtures.

I have worked years in Reaserch and Development areas for several manufacturers and it is not a real simple or inexpensive route to devlop a product to manufacturing.

Step one is the initial idea which is probably the easiest step.

Once you have an idea you go to work in three basic areas.
1. patent search to see if someone else has the patents on this device preventing you from marketing it.
2. market research to get an idea of how many people would be interested in your product. Without a market is it worth building.
3. Design feasibility studies, to determine if it can be built and what it could cost to build.

If all these of these areas are in the product favor the next step is to come up with a prototype or initial working model along with trying to lock in patents so no one else will benifit from your design work. Often the first prototype will not work and repeated runs are made before a marketable product is produced.

Then when you have a model and rights to the product you design and produce the first pre production runs. These runs are then extremly tested and used to get any government certifications needed. You usualy also discover many unpredicted issues with the product on this first pre production run.

Second and third preproduction runns are usualy more the norm before you have satisfaction with your product. As these preproduction runs are in progress you also send out pre release products to select users to obtain unbiased opinions on there preformance. Review there reports and make design changes to future runs.

Finaly at about this point you get marketing involved to actualy start selling the product. Manufacturing is geared up for initial sales expectations.

for all the products I had worked on the corperations had a minimium of a million dollars and at least two years invested into the product before the first unit came off the regular production line. the initial investment needed to come back to company usualy in one or two years to justify production based on product demand as well as markerable price.

With LEDs we are seing new designs hitting the market roughly every 3 to 6 months right now. With that in mind most manufacturers are just introducing products today using LEDs that are three generation or more away from it is available for us to use on a DIY project.

The other big thing to consider is the money that these corperations invest before the product hits the market and the limited market of some of these items. A lighting system for the reef aquarium markets is something that every household does not buy every year. If they do have a million invested in the product how long will it take them to sell 10,000 units a year where they need to get $100 each to cover there R and D costs? If it costs them $300 to build the unit and they make $100 to cover there initial costs they can sell it for $400. But they sell it to a distributor that needs to make money so he sells it to the LFS for $600, and the LFS needs to make a profit so he sells it for $900. In most cases you yourself could build it yourself for well under $450.00.
 
Back
Top