LEDs in Series vs. Parallel vs. Series/Parallel

[mn velocity pilot]

In developing a high power aircraft strobe system, I've been looking hard at various current limiter topologies. Many of the LED controller supplies I've seen show the LEDs in series instead of parallel, or sometimes in series/parallel. My question is... is there an advantage to doing that?

Obviously, the LED forward (turn on) voltages become additive in series, which is why I haven't used that layout so far. But, other than controlling the forward voltage, is there a different advantage that I'm not seeing?

THANKS!

Dave

 

[alpg88]

it is not about advantage\disadvantage, leds are wired either way depending on many factors, available voltage\current, amount of leds\required output, driver availabuility wiring size...etc. Based on those an appropriate wiring set up is chosen.

 

[Steve K]

as a rule of thumb, running LEDs in parallel will require some method to ensure that they are sharing current equally. When LEDs are wired in series, they are all running at the same current.

A lot of the decision has to do with the range of the input voltage when compared to the LED forward voltage and what is the cheapest way to drive the LEDs while meeting all of the requirements for the light.

For an aircraft light, some of the requirements would include small size, low weight, high reliability, a wide operating temperature range, good EMC performance, and suitability for a small production volume. My guess is that it would be powered from the 28vdc bus. Lots of options, but my guess would be a buck converter design and probably LEDs wired in series, although some series/parallel combination may be needed to achieve the desired light output.

 

[mn velocity pilot]

Thanks, alpg88... that's kinda what I suspected. What was on my mind was that there may be other factors that aren't so obvious, such as any inherent current limiting effect of having the diodes in series (such as turn on voltage ramp time which would increase...).

I'm having a challenge finding an acceptable solution for a 5 - 10 amp switching current limiter... enough current to really drive 7 or 8 XM-L2 LEDs in series with a very short duty cycle (around 10%).

Thanks again - any other thoughts or ideas would be gratefully accepted!

Dave

 

[mn velocity pilot]

Thanks, Steve K. Here's a link to what I've got implemented now in my experimental Velocity aircraft:

http://www.instructables.com/id/LED-Landing-Light-for-Experimental-Aircraft/

I'm using a linear regulator for current limiting (an LM338). I'm trying to find an acceptable way to go from the 5A drive I've got now to something more like 8-10 amps. You are correct that size, weight, reliability, and EMC are issues. Heat is as well, which is one of the reasons I'd like to get away from a linear regulator.

This strobe system is really a "one of" design, although I do publish my results in the experimental aircraft forums. I doubt if many people are as willing as I am to deal with all the complexity of actually building/replicating what I've done, but then on the other hand, each of us did build the whole airplane, so we are (as a group) people that enjoy complex, stupid, expensive, and frustrating challenges.

Thanks again!

Dave

 

[alpg88]

if you could tell us in details what set up do you need, and what you have now,(power wise) may be we could come up with designes, and you'd pick, or get an idea how to build your circuit.

 

[mn velocity pilot]

Wow, thanks alpg88! I'll take you up on that offer!

If you've had a chance to look at the instructible link I posted above, you'll can see in gory detail what I implemented for the nose mounted landing light strobe system... it's basically a PIC controller driving a mosfet switch, and using an LM338 linear regulator to set the current to 5 amps. That current drives 7 Cree XM-L2 LEDs wired in parallel.

Using the same controller setup, I've implemented wingtip strobes as well. I used one PIC controller driving two separate mosfet switch/LM338 regulator boards, each of which drives 8 XM-L2 LEDs in parallel (one gang on each wing). To make this one work I had to add protection diodes to handle the ground bounce due to the long wires from the controller to each wingtip.

These are all ... satisfactory ... but I know that I can get much more brightness if I drive more current. As an example, in the nose installation, at 5 amps each LED is getting just over 700 ma (assuming perfect current sharing). That's just getting started for these LEDs, which can handle 3 amps each. Again, assuming perfect current sharing, and assuming I could get rid of the heat, in theory I could drive the nose light with 3*7 = 21 amps and be within the limit for the LEDs.

So... in a perfect world, I'd have:
- at least 10 amps drive current
- programmable controller (like I do now)
- very little heat (especially from the controller)
- VERY high brightness
- acceptably small size.

One thing that's been much on my mind as I've implemented all this is... my hangarmate (with a similar airplane) purchased an off the shelf wingtip strobe system for $600+ bucks. It's bright - at least as bright as what I've implemented - but, there is no external controller. How his strobe is doing current limitation is a mystery to me.... I've wondered if they are just keeping the duty cycle so low that the LEDs are not overheating...?

Any ideas are really welcome. Thanks again!

Dave

 

[DIWdiver]

First thing that strikes me is that the schematic doesn't match the text - it shows an IRF1501 and a CXA1304. I assume the text is correct?

If so, you could get way more light for the same power by running 6 LEDs in a 3S2P configuration. You'd be running them at 2.5A each, for a total of around 5000 lm.

In order to run the 3S configuration at full power, you might need a regulator with lower dropout voltage, as the LM338 probably needs 3.5V or more in current regulation mode. Depending on LED voltages, wiring losses, etc, you could be running in regulation or not.

If the plane is in the air, you've got some serious airflow - any chance of using that to cool the system?

If you want even higher current, run two regulators, each with 3 LEDs. You can still control them with one FET.

 

[DIWdiver]

Oh, and there's a plethora of better FETs available. An IRFB7540 would barely get warm at 10A, and that's just getting started.

 

[Steve K]

So what are your design requirements? i.e. what is the range of the input voltage? What sort of voltage transients are you designing for? What is the desired output power? What is the operating temperature range? anything like the mil temp range... -55C to +125C, IIRC.

As a general rule, you can get excellent efficiency from a switching power supply, although you will end up with a larger, heavier design. It may require designing your own switcher, though..

Once you've fully figured out what the specs are, you can go to manufacturer's web sites such as Texas Instruments (ti.com) or Linear Technology (linear.com), and they have selection tools to help you pick the best converter design. They also have simulation tools to work out some of the details.

 

[mn velocity pilot]

Thanks, DIWdiver -

You are correct, the schematic in the instructable is downlevel for sure. I have waited to change it only because I had to work through the issue of how to handle the ground bounce caused by the long wire length between the controller (mounted in the cockpit) and the wingtip mounted strobes. Along the way, in addition to adding the protection diodes, I changed to a XM-L2 LED (because the CXA LED has a 9.6V turn on voltage), and also moved to a IRL-2505 mosfet... which I surely should have used right from the start. I had a bin full of the old IRF-1501s to use up... that's my only excuse.

Boy do I wish there was a step-function improvement in the LM338 regulator the way there was for the mosfet!

Your point regarding moving to a series/parallel setup is exactly what I was hoping to learn about by starting this post. THANK YOU! But here's my question...

Here's the relevant chart from the Cree datasheet:

The specific LED I'm using is Cree's PN XMLBWT-02-0000-0000T5051, which according to the datasheet has an output of 260 lm @ 700ma and 85C. So, with 5A total current, assuming even current distribution, the landing light should be putting out about 260*7 or 1820 lm total.

Using the chart above, if I run 2.5 amps by using 6 LEDs in two strings of three, the math works out to approx. 2.8*260*6 = 4368 lm... well over 2x!

Outside of the potential electrical issues (such as the increase in forward voltage), the issue is going to be LED heat... There is very little I can do to dissapate the heat in the wingtip strobes. The only knob I've got to twist would be the duty cycle of the strobe.

So, here's the question... as I've cranked down the pulse width of the flash, I've not visually been able to see much difference. Right now, I've got it set to a six flash sequence with 16 ms flashes. How short of a flash can I use? What's the relationship between flash length and visibility?

THANKS AGAIN! That's a GREAT idea!!!

Dave

 

[mn velocity pilot]

Hi Steve K - thanks for responding!

My basic design requirements are to use ship power (14V), a nominal ambient temperature range of around 0 - 100F, and as high an output as I can reasonably get. I'm not sure what you mean by voltage transients as a design parameter. Obviously, electrical noise is an issue - so if that's what you're poking at, you are absolutely correct that there is a limit... but I'd have no idea what that limit would be without extensive testing.

Regarding switchers... I've spent a fair amount of time looking at Linear and TI's controllers, and also others (using Digikey's part selector). The LED controllers I've found generally have very limited current capabilities... the highest I've seen is around 4.5A (Linear LT3478). Again, I'm looking for something well north of 5A.

Any specific suggestions? Thanks again!

Dave

 

[Steve K]

just popping in for a quick comment... at Linear's site, there is a selection tool that can be helpful.
http://www.linear.com/products/step-down_(buck)_led_drivers
The LT3756 came up in a search that I ran. It uses an external mosfet, and the selection tool says that a design can be created that will drive 10A to a load of 4 LEDs in series (assuming that these are either BIG LEDs or a few wired in parallel).

The thought of using a switcher to convert roughly 100W immediately raises concerns about EMI. Not a huge problem for someone with experience and lab facilities, but for a home project for an aircraft... I don't have good feelings about it.