Can I get by without a driver using a CC power supply with parrallel strings?

Hello-

First post here!!!

I was hoping that someone could help me out or provide some advice for me. I'm trying to build custom piece of equipment for my lab using 12 3-up Cree XPE2 stars. I was considering running 4 strings in parallel with 3 stars on each string. I want to run 1Amp on each leg and this will put the potential somewhere around 10.2V per star or 30.6V per leg and on the system as a whole as the strings are "balanced".

So, total voltage =30.6V and total current = 4Amp

I also plan to run those 13 volt LSP LED shunt protectors from Bourns on each star so that if one goes out the whole string will not be down.
Bourns P/N LSP1300

My questions are the following: would putting a 1.1A fuse on each leg and powering the system at 4A with a 60V 5A DC power supply in CC mode be OK or do you think the strings will be imbalanced and LEDs or fuses will start blowing? Anyone have experience with the shunt protectors?

I'm trying to avoid buying four or more driver modules and don't have reflow solder equipment to make my own.

Thanks in advance for any advice?

Welcome to the forum!

Yes, that's quite a reasonable plan. An LED driver is nothing but a power supply in CC mode, except that it has typically been optimized for that mode only.

I designed a driver for a commercial lamp that uses 30 LEDs in 10 strings of 3. The strings are in a group of 6 and a group of 4, each group driven with a single driver. The arrays I've seen powered up balance so well you can't even see any difference. That probably means better than 15%. After selling hundreds (maybe thousands) of them, the customer decided the lights were so reliable they could overdrive the LEDs, and asked for the current settings to be raised. With your longer strings, the self-balance characteristics should be even better. The key is to get LEDs from the same batch, which you would normally get just by buying all of them at the same place at the same time.

You might see 30.6V initially. But that will actually change noticeably (dropping) as the LEDs warm up.

If one star opens, and the shunt engages, that would cause the current in that string to drop quite a bit, and that current would be transferred to the other strings. I'm thinking each other string would see an increase of 15-25%. I imagine you made this same guess, which is where you came up with 1.1A for a fuse value. Unfortunately, fuses are not precision devices. At 35% overload, an ordinary 3AG (which I found in 1-1/8A, but not 1.1A) fuse is allowed to take up to an hour to open. And it's allowed to open after 4 hours at rated load.

Oh, sorry, I just looked at the LSP1300 data sheet, and it does not clamp at 13V. It TRIPS at 13-26V. Once it trips, the voltage across it drops to 1.2V or less. That means this string would suck virtually all the current from the other strings. That would cause a rapid destruction of something, the fuse if it's present, the LEDs if not. Sorry, that thing is a BAD idea in this case.

That's okay though, as I really don't think you need it. LEDs rarely fail catastrophically if not mistreated.

If you wanted to build some drivers, there's no need for reflow equipment. I build all my drivers by hand, using a soldering iron with a 1/16" chisel tip. The schematics have been posted, and I'd even help you adapt one for your needs. The LEDs need to be reflowed, but it doesn't sound like you're looking to do that. If you are, hunt around the forum a bit, and you can find out how to do it very inexpensively.

Welcome here, JMc80! :wave:

JMc80 said:

as the strings are "balanced".

Click to expand...

Ehm, yes, nice. Ehm... balanced how?

Personally I'd go for a smallish series resistor in each string/leg (0.1~0.22 Ohms or so). Small enough to not affect efficiency too much, large enough to even out current draw differences between the LEDs. Simple, cheap and reliable.

I also plan to run those 13 volt LSP LED shunt protectors from Bourns on each star so that if one goes out the whole string will not be down.
Bourns P/N LSP1300

Click to expand...

I wouldn't bother - just make sure each LED has adequate cooling.

would putting a 1.1A fuse on each leg

Click to expand...

Makes sense, but I'd pick their value a bit higher, say between 1.3~1.5x operating current, to prevent accidental blows. Remember fuses serve to protect against catastrophic failures like short circuits and such. Not against currents that are 'slightly' above normal.

and powering the system at 4A with a 60V 5A DC power supply in CC mode be OK

Click to expand...

What's wrong with a dedicated LED driver (so 4A, ~30V) for above setup? :thinking:

Hi guys - thanks for the quick reply. I've been out of town all weekend and was able to check for responses until this morning.

So, it sounds like the fuses may not be what I'm looking for and really need a 1 amp driver for each leg. I'm probably going to go ahead and build without the drivers, fuses, or shunts and test to make sure the equipment does what I want it to. The LED array is being used to induce a photochemical polymerization reaction.

If everything works, I'll either buy or build some drivers.

So what's the deal with these shunts? I don't understand how they are suppsed to work. It looks like even when they kick on the rest of your string is going to get too much voltage. Are you supposed to run resistors in series with these or something?

I'll look around the forum for some DIY driver designs.

Thanks again.

BTW - Does this forum act strange for everyone? When typing characters are often missed and scrolling can act weird also when posting. May be that I'm using IE.

If you get LEDs from the same batch, they will have sufficient resistance to balance themselves quite well in strings of 12. It is not necessary to use separate drivers or external resistors.

The shunts are designed for individual strings being driven by drivers with large voltage compliance. That's not what you have, so they aren't for you.

I haven't seen DIY switcher designs on the forum, but there are sample designs in almost every datasheet. Texas Instruments, Linear Technologies, and a raft of others make driver chips. You can have the schematics for my linear ones if you want. https://www.dropbox.com/s/3bvk16uwlq...011_draft1.zip and https://www.dropbox.com/s/wd5l5oaaorw9vnw/SCH.GIF

I think the forum doesn't like some installations of Wondoze 7 and 8. I have this problem at home, not at work. If you Switch Editor to Source Mode, it will work much better. That's the A/A button in the upper left of the edit window.

The main experience I have of failing LEDs was strings of 4-8 LEDs being driven from drivers which occasionally failed and put increased current through the string.
Almost without exception, what happened under over-current with those strings (R/G/B Luxeon Rebels) was that LEDs failed as a short circuit, and the entire string ended up burning out.
LEDS failing as a short is a failure mode with considering.

So are you saying that the LEDs got overdriven, then failed shorted?

And were those single strings, or several in parallel?

In that application they were single strings subjected to significant overload (buck driver failing short), and all the LEDs in a string would end up failed shorted.

In that case, I'm not sure it's possible to clearly distinguish between 'simultaneous' and 'serial' failure - likely from the point of overload starting, all the LEDs in a string would be heating rapidly, and by the time any had actually failed sorted and ceased dropping meaningful voltage, likely the rest would be well on the way to failure even if some might have survived to some extent if current had been cut off.

In a less severe case, but still one where a shorted LED could raise current for the rest (which would be the case with parallel strings with or without some added resistance) a true serial failure might be more feasible - one device failing possibly from some defect or particular sensitivity to current (poor heatsinking?) which then causes more current and more stress on the remainder in the string.

I only really brought it up to point out that failing short is not impossible, and may even be common, at least for some designs of LED.