How to make 3W red Luxeon flash+dimm ?

Trajkociklista

Newly Enlightened
Joined
Oct 14, 2007
Messages
9
Location
Vojvodina,SE Europe
Hi there :)



So yeah,I want to build some seriously bright tail lights. :devil:
I'll use two 3W Luxeons in series.

LED driver just like nFlex would be perfect (constant/flashing/brightness regulation)
but since
it can only put out 1000mA,and Luxeons need 1400mA,it's not suitable.

Is there such a thing,but can drive 1400mA ?
If not,what do You suggest ?


Tnx


Cheers!
 
Woah, seriously bright!
I don't think the difference between 1A and 1.4A is that noticable...certainly with crees there is not much real world difference between 1A and 0.7A. I'd go with the nflex.
 
Sp[eaking of the bright red LED. What is the current brightest red LED out there on the market that can be driven 1000mA? Any announced plans by companies and forecasts in the near future? I want a 2x 160lm taillight!!!! :kiss:
 
it can only put out 1000mA,and Luxeons need 1400mA,it's not suitable.
OK... I Have to know... Why and where do Luxeons "need" 1400 mA? That's well beyond spec, and I see no reason for it. My experience with these high currents into Luxeons tells me that the human eye would barely notice the difference in light output. The HEAT output at that current would be staggering, however. Your runtime would suffer tremendously

Is there such a thing, but can drive 1400mA ?
If not,what do You suggest ?
What I suggest is driving the thing at 750 or 1000 mA. Two benefits: Much more efficient (with about the same brightness) and the driver already exists. There's a reason that George stopped at 1000 mA.
 
OK... I Have to know... Why and where do Luxeons "need" 1400 mA? That's well beyond spec, and I see no reason for it.

Aaaa,thats what I've read in the spec sheet here ?:huh2:

Luminous Intensity typ. (with 10° collimating optics): 140 lm
Power Dissapation: 3W
Luminous Flux (typ.): 140 lm
Viewing Angle: 130°
Wavelength (peak): 627nm
Forward Voltage: 2,95V
LED Current (If): 1400mA


I'll be more then happy with 1000mA if it gives out 140lm from a 3w Luxeon.


`
 
Aaaa,thats what I've read in the spec sheet here ?:huh2:
Good lord. That's insane. First I've heard of the spec being set for 1.4A.

Anyway... you most DEFINITELY do not NEED 1.4A to get bright light. Calling 1A "not suitable because Luxeons need 1.4A" misses the mark. You won't get 140 lm out of it with less current, but you'd be hard pressed to tell the difference in brightness at 1A. I use these as automotive brake lights at 350 mA! They are brighter than the incandescents that they replace. 1.4A is overkill. Most of your energy is going to heat! You should get more light with two emitters at 500 mA than you will get with one at 1.4A. And WAY longer runtime.

I'm not saying this stuff to be contrary, or to rain on your parade. I just think it is a mistake to try and run today's emitters at a current that high. You'll need some HUGE heat sinking, and lots of battery. After about 750mA, your light gains relative to energy used and heat produced - just goes in the toilet quickly.
 
Last edited:
You're behind the times. The R/O Luxeon III has always had a spec of 1.4A. It's been that way since they were released in 2005.

The die on the R/O Luxeon III is 4x the size of the regular R/O luxeon. It can handle the high current levels just fine, and maintians its efficiency at high current levels (1A and above) without problem. It also has a much lower thermal resistance than the Lux I version (6C/W vs 18C/W for a bare emitter).

My own testing has shown that the R/O Lux III scales quite nicely all the way up to its rated 1.4A. Going from 700mA to 1.26A (1.8x current) results in an increase in output of 1.69. That's 93% scaling for that current increase. The R/O Lux III is definitely up to the task of operating at high currents.
 
You're behind the times.

You'll get no argument from me.

I still take issue with anybody thinking that the thing NEEDS 1.4A to be effective (and that a 1000 mA driver is not sufficient). Is there anybody that doesn't think these things are insanely bright at 1000 mA? Ah... here's an example right here at home! LINK

All that said, it is obvious I've got my foot in my mouth, so I'll shut up.
 
Last edited:
Although I trust evan9162's numbers, and am happy to find that there is in fact a much brighter red LED being made than the Cree that I have (not that I need brighter or am planning to switch - just nice to know), I also agree with Darrell's assessment of the actual visibility difference between 1A and 1.4A with these, and the consideration that you don't "need" the extra current. The brightness function of the eye is logarithmic, you need about 10x the actual lumens to look twice as bright, and the xFlex settings of doubling the current for each brightness step is about perfect. Twice the current is about twice the lumens in most cases, and in real life, that doubling of the lumens does not look like anything close to twice as bright - I'd say it looks 30-40% brighter for each brightness step. The light looks twice as bright when you go from L2 to L5, which is about 8x the actual lumens.

However, I also think the xFlex drivers should be able to output up to whatever the highest rated current level is for the LEDs on the market now. Isn't their 1A max directly because the majority of power LEDs in use now (Cree and SSC and older LuxIII) have a spec of 1A max drive current? If so, why not bump that max up to 1.4A for the newer LuxIII, K2, Rebel, etc? Users of those LEDs should be able to drive them at their max current, even though the brightness gains are negligible in comparison to their heat output. You don't set the max at 750mA just because the bump up to 1A isn't "worth it" from a light/heat standpoint, so you shouldn't make that argument against setting the max at 1.4A for Lux users.

The other really helpful effect a higher drive current would have is the ability to drive LEDs in parallel or series/parallel strings at closer to their rated currents. For my bFlex application, I'm driving 4 Crees in 2S2P, meaning they each only get 500mA max. That's enough for me right now, but if the bFlex could drive 1.4A max, I'd set it at that, and have an actual 700mA each max, and be happier with it. While you're at it, why not set the bFlex max current to 2A to drive 2 LEDs in parallel at 1A each? Driving 6 Crees at 1A each from a 4-cell Li-Ion pack would be very enticing to some users.

If the "why not" answer has something to do with the physical design constraints of the board, that's understandable, but you and George could consider making a higher power version in the future to satisfy those who want to run parallel strings of LEDs at their rated currents, and/or want to create heat-dissipation challenges for themselves. There are cheap 1.4A drivers on DX, assumedly for users of the new Luxeons that can handle that, so those who want the higher currents (and don't mind the lack of features or possible unreliability) will go there. Don't you want those people buying xFlex boards instead, even if they're wasting more power as heat?

Alex
 
Indeed - 1A should be plenty on the R/O Lux III - the only reason to crank it up to 1.4A is if you need to meet some minimum spec that 1A won't meet, such as minimum taillight brightness on a vehicle (there are DOT regulations that mandate minimum brightness for tail lights when the brakes are applied). Whatever the case, the R/O Lux III can handle and will still scale well up to it's maximum rated 1.4A given sufficient heat sinking.
 
.. but you and George could consider making a higher power version in the future to satisfy those who want to run parallel strings of LEDs at their rated currents, and/or want to create heat-dissipation challenges for themselves.

I've paged George to this thread, as he's the only one who can answer this.

For the record, these circuits are George's. I've got no stake in the sales of these things, and couldn't design or build one if my life depended on it. I'm the UI guy for the offered drivers. We work together to make these great drivers work for the end user. I'm the guy to complain to if you hate the operation. I'm the guy to whom you should send thank you cards with money in them if you love the user interface. But if you have technical electronic questions... George is your guy.
 
Ok, firstly, just because the spec for a LED says 1.4A rated, doesn't mean that a driver than can only output 500mA or 1A or ... won't work. When you dim a driver (unless it is PWM) it will lower the drive current to dim and the LED still works just fine.

Next, bFlex and nFlex use 1A rated switcher chips. So, bumping the output current spec to 1.4A is more than just tweaking something - it's a complete new redesign. Bigger switcher, bigger inductor, bigger diodes, bigger board, bigger cost.

Next, MOST switcher chips that can fit on a small board, typically max out at around 1A - just a fact of life. Factor in efficiency of the switcher chip, junction temperature and dissipating capability of the IC package and there's a good reason you don't see 1.4A or 2A or ... switcher chips that are fully integrated and conventional SOIC packages.

Eyes respond logarithmically (sorta/kinda) to intensity. So, increasing current to 1.4A will barely be brighter than 1A. That's a lot of extra heat and battery consumption for little Visible gain.

Answering one post - show me a DX driver that is a SWITCHER that is capable of 1.4A... All the ones I've seen are essentially linear regulators. All they do is stack more & more regulators in parallel.

Given the sweetspot of LEDs running in the 500-1000mA range, I don't see a compelling reason at this point to build a driver that goes to 1.4A and the added expense to everyone and the size increase to dissipate the extra heat for the 1.4A case.

You can't parallel a driver like maxFlex or bFlex/nFlex etc at the LEDs - the way they current regulate doesn't lend to that. If you want more current, then you need a different design.

Finally, there are LEDs that take several amps (multidie units) - just because they exist doesn't mean nFlex/bFlex should be able to output several amps :)
Use my CCHIPO driver if you want to do that - see my website. That's what happens when you want more power - things get BIGGER and more EXPENSIVE...

cheers,
george.
 
Most importantly, idiots like myself will toast their Q5 crees trying to squeeze out a wee bit more light. People always want a little bit more and a lot of people would try at 1.4A.
 
Thanks for a great reply, George. The design and thinking of your product ranges is much clearer now.

Next, bFlex and nFlex use 1A rated switcher chips. So, bumping the output current spec to 1.4A is more than just tweaking something - it's a complete new redesign. Bigger switcher, bigger inductor, bigger diodes, bigger board, bigger cost.

Next, MOST switcher chips that can fit on a small board, typically max out at around 1A - just a fact of life. Factor in efficiency of the switcher chip, junction temperature and dissipating capability of the IC package and there's a good reason you don't see 1.4A or 2A or ... switcher chips that are fully integrated and conventional SOIC packages.
This is what I had expected, that the 1A rating was not just because that's the datasheet limit for lots of popular LEDs, but because of hardware design limitations. However, don't the nightlightning lights from NZ use your bFlex board with their own UI, and they can drive 1.2A from it? MaxFlex drives at 1.2A also, but it's a different circuit. Not as though 1.2A is any kind of worthy increase from 1A.

Answering one post - show me a DX driver that is a SWITCHER that is capable of 1.4A... All the ones I've seen are essentially linear regulators. All they do is stack more & more regulators in parallel.
Yes, totally true. They're not $3 just because they're from China, they're $3 because they're poorly designed, inefficient, probably use substandard components, etc. TaskLED circuits are far better quality in addition to having a far better UI.

You can't parallel a driver like maxFlex or bFlex/nFlex etc at the LEDs - the way they current regulate doesn't lend to that. If you want more current, then you need a different design.
I'm not sure what you mean here. I have 4 Crees running in 2S2P on my bFlex, giving a Vf of ~7.4 and each getting half the current (500mA max) and in an email I asked if that was a good idea, and you said sure, it's fine, they'll just each get half the current the driver provides, and you should make sure the 2 strings have as similar a Vf as possible so one LED isn't much brighter than the others. Now you appear to be saying that I can't do that - are you saying something else here?

Yes, to get more current (than 500mA each with 2 in parallel) you need a different design - like your master/slave bFlexes! (Which are not yet available, right?) I was pretty excited to hear that they can do that; it means I could drive my 4 Crees at full 1000mA each with only 11.1V from the batteries, if I wanted to buy 2 bFlexes. For me, the small brightness increase is not worth the money and battery drain, but it's still a pretty cool idea.

Finally, there are LEDs that take several amps (multidie units) - just because they exist doesn't mean nFlex/bFlex should be able to output several amps :)
Use my CCHIPO driver if you want to do that - see my website. That's what happens when you want more power - things get BIGGER and more EXPENSIVE...
Yes, it's true that xFlex should not automatically have to output whatever the highest amperage multi-die LEDs can take at the moment, given the hardware limitations. However, it does seem that you'd want to maximize what your hardware can do (which you are doing).

I hadn't looked at your CCHIPO in detail before - that's a really cool (and big, expensive) driver! It can take 5A input - what's its max amperage output?

Say in a year or 2, Cree brings the 4A single-die 1,000 lumen LED they've announced in their lab to the market, but at well over 100lm/W, and it becomes the new cool LED everyone here wants. If you wanted to design a driver for it, would it necessarily be as big and need such heatsinking as the CCHIPO? If so, is that necessarily the future of high-power LED lighting - big drivers that need considerable heatsinking? Or is there some other way? It seems that a 85% efficient driver, pumping 100W to the LED or an array of them, would need to dissipate 15W - that's alot!

Alex
 
However, don't the nightlightning lights from NZ use your bFlex board with their own UI, and they can drive 1.2A from it? MaxFlex drives at 1.2A also, but it's a different circuit. Not as though 1.2A is any kind of worthy increase from 1A.

Yeah, the 1.2A range is pushing the envelope for bFlex. With my own sales I went the more conservative 1A limit.

I'm not sure what you mean here. I have 4 Crees running in 2S2P on my bFlex, giving a Vf of ~7.4 and each getting half the current (500mA max) and in an email I asked if that was a good idea, and you said sure, it's fine, they'll just each get half the current the driver provides, and you should make sure the 2 strings have as similar a Vf as possible so one LED isn't much brighter than the others. Now you appear to be saying that I can't do that - are you saying something else here?

What you are doing is fine. What I meant is that at the LED end you can't have parallel max/n/bflex drivers connected together to create more current drive capability.

I hadn't looked at your CCHIPO in detail before - that's a really cool (and big, expensive) driver! It can take 5A input - what's its max amperage output?

As long as the 5A input limit isn't exceeded, and power is under 45W, you can raise the output current as required. i.e. say your input voltage is 12V and your output voltage is 24V (24V worth of LEDs in series). With 45W at the output and 24V that would mean you could adjust the output current up to 2A. At 16V output you could have close to 3A at the output.

Say in a year or 2, Cree brings the 4A single-die 1,000 lumen LED they've announced in their lab to the market, but at well over 100lm/W, and it becomes the new cool LED everyone here wants. If you wanted to design a driver for it, would it necessarily be as big and need such heatsinking as the CCHIPO? If so, is that necessarily the future of high-power LED lighting - big drivers that need considerable heatsinking? Or is there some other way? It seems that a 85% efficient driver, pumping 100W to the LED or an array of them, would need to dissipate 15W - that's alot!

Alex

Yep, if LEDs get up to 4A or whatever, the drivers will get larger due to component sizes (external FETs etc) and heat dissipation requirements.

Larger drivers aren't that crazy if you already have a 20+ Watt LED to heatsink... These aren't going to be small portable lights - just the battery packs will be relatively huge....

Most of the high powered multidie leds are targeting automotive and more importantly fixed lighting applications where space is a lesser issue.

cheers,
george.
 
What you are doing is fine. What I meant is that at the LED end you can't have parallel max/n/bflex drivers connected together to create more current drive capability.
Right, that makes sense. You can't run 2 drivers in parallel to the same LED array. That's true for pretty much any driver, right? But if you have master/slave bFlexes, you can get the same effect.

As long as the 5A input limit isn't exceeded, and power is under 45W, you can raise the output current as required. i.e. say your input voltage is 12V and your output voltage is 24V (24V worth of LEDs in series). With 45W at the output and 24V that would mean you could adjust the output current up to 2A. At 16V output you could have close to 3A at the output.
Wow, that really is a high-power driver. It'd be great to drive an LED array for car/motorcycle headlight replacement - a 6S2P array of LEDs (multidie or multi-LED) would require 21-22V, could be driven at 2A each, and output over 2,400lm. Better than HID!

Alex
 

Latest posts

Top