EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Artlav on October 16, 2014, 08:59:36 pm
-
I have 4 50W LEDs that i want to drive at once from a 12V battery.
Each LED needs 1.4A at about 37V.
What sort of a converter and arrangement is better for it?
Two basic arrangements are possible - all LEDs in series, and one 150V driver, or each LED on it's own, with 4 separate 38V drivers.
Both times would need a step-up converter, so it makes sense to make one rather than 4.
On the other hand, i never made a converter like that for more than 100W, and it was constant voltage and getting too warm for comfort at full power.
Would 4 50W 38V converters be better than 1 200W, 150V one?
What kind of a converter could it be in each case, keeping in mind that they should be constant current?
50W version sounds like within the realm of a boost converter, but there is no such thing as a constant current boost converter, so it would have to be followed by a buck one, quadrupling the size.
If a transformer converter is to be used, then it would be about the same for either 50W or 200W, so what's the point?
With a transformer converter 200W is possible, but i never tried making one that is constant current.
Is that feasible?
Perhaps there are other solutions?
-
How can there be no such thing as a constant current boost..?
I'd be happy with a single boost, and current balancing resistors. The resistors don't need to be nearly as large and wasteful as when running strings from a fixed voltage supply, so it doesn't hurt much. If you must, you can do individual channels. You can then get independent control, which might be handy, I don't know.
Tim
-
If you wanted to drive each one individually, I'd look at
http://www.taskled.com/hboost.shtml (http://www.taskled.com/hboost.shtml) 7A max input current, 6A recommended, 8-50V input range, 80V or 3.2A max out (mind your input current limit though)
So it would cheerfully run your 50W LEDs. The nice macro shot of the PCB is also a good hint at where to start if you want to do your own.
http://www.linear.com/product/LT3756 (http://www.linear.com/product/LT3756)
If you want to mess around on your own for doing a 150V transformer based driver as a learning experience, I'd grab a transformer from an ATX PSU, and run it backwards(The primary will be wound for 380V not 150V, but running it as a flyback converter means that the ratio isnt *that* critical). Grab a TL-494, and a TL-431, current shunt resistor and an amplifier for it and youre off to a decent start. However I havent designed enough transformer based SMPS to give solid advice for that.
-
Really? I've got several constant current boost converters sitting here, disagreeing with you. ;)
200W boost is a bit much, but hardly a problem. Linear has several parts which should work for you. Consider running two LEDs in series, then each string in parallel with current limiting resistors (or whatever solution you prefer). Hell, you can just boost to a suitable voltage, and use linear CC drivers.
-
If you wanted to drive each one individually, I'd look at
http://www.taskled.com/hboost.shtml (http://www.taskled.com/hboost.shtml) 7A max input current, 6A recommended, 8-50V input range, 80V or 3.2A max out (mind your input current limit though)
That looks great.
Thanks, i can use some of them, maybe for this project, definitely for some other.
How can there be no such thing as a constant current boost..?
Really? I've got several constant current boost converters sitting here, disagreeing with you. ;)
More precisely, i have repeatedly failed to make a constant current boost converter.
http://www.edaboard.com/thread300483.html (http://www.edaboard.com/thread300483.html)
https://www.physicsforums.com/threads/boost-converter-switch-node-spiking.716486/page-2#post-4543001 (https://www.physicsforums.com/threads/boost-converter-switch-node-spiking.716486/page-2#post-4543001)
In the end, it is just absolutely unstable:
https://www.physicsforums.com/threads/constant-current-boost-converter-problems.749692/ (https://www.physicsforums.com/threads/constant-current-boost-converter-problems.749692/)
Maybe i just don't know how to make one?
I'd be happy with a single boost, and current balancing resistors. The resistors don't need to be nearly as large and wasteful as when running strings from a fixed voltage supply, so it doesn't hurt much.
What kind of resistors would it be?
How do i select them?
The LEDs change voltage as they heat up, from 37.4V to 37V on average. There will be unknown imbalance.
I can't quite visualise how to get them balanced with resistors, at constant current supply.
Hell, you can just boost to a suitable voltage, and use linear CC drivers.
Tried that before.
https://www.eevblog.com/forum/projects/linear-led-driver-avoiding-a-hot-resistor/ (https://www.eevblog.com/forum/projects/linear-led-driver-avoiding-a-hot-resistor/)
It gets hot even when the voltages are tuned tight.
But it works.
That is actually a viable solution as well - 50W CV boost, and an LDO to keep the current right.
-
TI have tools to develop these sorts of drivers.
http://www.ti.com/lsds/ti/analog/webench/overview.page (http://www.ti.com/lsds/ti/analog/webench/overview.page)
Though I found it easier to just buy their evaluation kit and work from there.
-
http://www.aliexpress.com/item/Freeshipping-2pcs-12V-50W-LED-Driver-for50W-30-36V-1500mA-high-Power-50w-led-chip-transformer/1843599268.html (http://www.aliexpress.com/item/Freeshipping-2pcs-12V-50W-LED-Driver-for50W-30-36V-1500mA-high-Power-50w-led-chip-transformer/1843599268.html)
-
I use these (http://www.ebay.com/itm/150W-DC-DC-Boost-Converter-10-32V-to-12-35V-6A-Step-Up-Power-supply-module-/141316770525?pt=LH_DefaultDomain_0&hash=item20e722cadd) for my 100W LED (http://www.ebay.com/itm/111299756188?_trksid=p2059210.m2749.l2649&var=410291578046&ssPageName=STRK%3AMEBIDX%3AIT)s:
Fan mod needed. And ~0.5 ohm resistor.
If you want constant current than this (http://www.ebay.com/itm/DC-DC-600W-10-60V-to-12-80V-Boost-Converter-Step-up-Module-Power-Supply-LS4G-/121425964478?pt=LH_DefaultDomain_0&hash=item1c458d2dbe) should be ok.
37V? All 20W - 100W leds are typically 30-35V.
-
My light (running right now, healthy) might be worth some thought:
http://seventransistorlabs.com/Images/LED_Light.png (http://seventransistorlabs.com/Images/LED_Light.png)
Tim
-
There should be no need for fans at all, if you are generating that much heat something is wrong with the design or component selection. Even if it's reliable it kind of defeats the purpose of efficient LEDs.
-
Maybe i just don't know how to make one?
It's easier to just use a dedicated LED driver IC, you don't have to worry about amplifying the signal from the current shunt with the added potential for instability.
PS. that Alixpress one for 8$ a driver is a pretty good deal (if they don't blow up after 6 months).
-
Just tried making a boost converter with MAX1771 - it works up to 25W, reasonably well, but it squeals and does not work well unless i'm touching the FET's tab :-//
Same circuit worked perfectly for getting 200V for tubes.
All in all, i guess i'll get something from China to run their Chineese LEDs. :)
Thanks all for the links.
37V? All 20W - 100W leds are typically 30-35V.
Madeinchi Industries, i presume.
At constant 35V one does 1.06A, second - 0.76A, third - 0.8A, fourth - 0.96A.
Not very consistent.
And quite a pain, being just out of the range of most ready-made boost modules (which go up to 32V or 35V).
Also, not all big LEDs are 30-35V, i got some Cree 100W ones that are 42V.
It's easier to just use a dedicated LED driver IC, you don't have to worry about amplifying the signal from the current shunt with the added potential for instability.
Any examples?
-
At constant 35V one does 1.06A, second - 0.76A, third - 0.8A, fourth - 0.96A.
Not very consistent.
LEDs are constant-current devices, not constant-voltage. Also the I-V curve is exponential and sensitive to temperature as well; 50W is the sort of power levels where tiny differences in temperature can result in huge differences in voltage/current, and thermal runaway becomes more of a concern.
-
And quite a pain, being just out of the range of most ready-made boost modules (which go up to 32V or 35V).
That's just a question of bodging the output overvoltage protection IF it's present at all. The FET, Schottky and output capacitor can likely take a few more volts.
Any examples?
Go to digikey, search for boost, select the results from LED drivers (the last two steps are a roundabout way of picking out all the boost topologies), throw away AC/DC topologies and charge pump, select for internal driver both&no, select quantity 1, sort for price ... spend some time looking for a nice part (pay attention to the feedback voltage, some use 0.5 volt which at 1.4 A will put 0.7 Watt into the shunt ... so preferably something lower).
AP3021 looks okay (no current protection for the FET though, but I'm sure there are plenty which have that as well).
PS. those aliexpress drivers use the XL6006 which seems a little more competetively priced than the standard integrated controllers (only available from China though AFAICS).
-
Also the I-V curve is exponential and sensitive to temperature as well; 50W is the sort of power levels where tiny differences in temperature can result in huge differences in voltage/current, and thermal runaway becomes more of a concern.
Somehow i doubt that LEDs on the same hunky radiator would be that different in temperature.
Go to digikey, search for boost, select the results from LED drivers
Nice, i haven't noticed that category before.
Much more searchable than "DC DC Switching Controllers".