Portable LED Flash

As modern photography is revolutionized by LEDs, I'm applying my experience in this field to enhance my hobby

Similar projects worth following
I'm an amateur photographer ( ) and light is the essence of photography. I've always recycled leftovers from my work to enhance my artistic activities but lately I've hit a brick wall...
I need a cheap, versatile, rugged, convenient, compact and *practical* flash ! Or more generally a flexible light source that I can use both for lighting and flashing.
Ted Yapo has played with short pulses through LEDs and I'm building on this experience to create my own battery-operated design, using both the "short pulse" overdriving and "long, temperature-controlled" undervoting of 400W of LEDs.

LEDs are now well understood, I have worked with them since 2005 in power ranges from milliwatt to tens of kilowatts.

The "idea" is to use the cheap, large, 100W LED arrays (approx. 10$ a pop on eBay) both in under-voltage (for continuous lighting of scene without heating) or over-voltage (during a short flash, the wattage is high but time is too short for significant energy absortion and heat generation).

The more LED, the better : the energy (and heat) is scattered over a larger surface and photography rarely likes light point-sources. So let's mount several modules on a flat heatsink, the rest is "just a matter of electronics" right ?

The battery pack is made of LiPo cells (designed for RC models) which are flat and can sustain short high-current pulses.

The LEDs will switch from one unusual regime to another :

For the Flash regime : #Bullet Movies has tested microsecond pulses but the light output must be greater ! more current, longer time, in the millisecond range or more. This means incredible current spikes and even greater parasitic problems. Brace yourselves !

For the "lighting" regime, the only two considerations are the temperature and the energy source's autonomy.

  • I'd like to have about 1 or 2h of autonomy
  • Temperature should not rise above 40°C

Well, the second part is easy to deal with because I've already created a cheap, simple linear system for strong temperature feedback. But this system is battery operated so it needs a more energy-efficient regulator !

Once again, I repeat : the actual power will not be 400W ! Probably in the 10 to 20W range once the temperature has settled. Undervolting the LED makes them more efficient and the heat is spread across the whole heatsink, reducing strain on the parts. It's better to undervolt a 100W module than use a 20W module at rated power (less "hot spots" and much more headroom for overdriving during flashes). See the discussions in the logs.

1. Another ordering mistake
2. Energy source
3. Structure
4. More progress


  • 4 × KOKAM SLPB 486495P LiPo pack, 2S1P, 7.4V 2100mAh, 10C, max recharging current: 2A
  • 4 × 100W LED module, neutral white (4500K)
  • 2 × anodized aluminum dissipator 200×100×18mm

  • Unexpected (happy) semi-ending

    Yann Guidon / YGDES08/14/2019 at 23:33 0 comments

    Do you remember the log Another ordering mistake ? Well...

    I have a new indoor leafed companion and he feels very sad in the dark. He's used to a different climate with intense direct sunlight so he has a lot of troubles with growing :-(

    So I took the "full spectrum" LED modules from the archives and made a crude power supply. The modules require about 28V  but I didn't want to go full power so a dual single-wave rectifier transforms 12Vac (from a small wall transformer) into 26.6V (average). It's... very bright, though not overwhelming, which should be more than enough for my companion.

    2 diodes, 2 bulk capacitors, some wires and connectors... that's it. For now.

    The story is not over because I still need to solve a lot of practical issues, such as finding a long enough wire, or mount the LED module in the right height and position to target the leaves. Oh and some alu paper around the whole will reduce the waste of light, as well as keep me from waking up with the light :-D

    And if you really want to know, it's a baobab coming straight from Africa... He has already lost his leaves several times (due to accidents) but keeps regrowing (slowly) so I want to give him a chance :-D

  • More progress

    Yann Guidon / YGDES04/21/2017 at 23:00 0 comments

    Today has been a pretty big day !

    This project is being featured on HaD:

    Hackaday Prize Entry: Portable LED Flash For Photography

    Thanks again people !

    And I just received the heatsinks. I can finally have a better idea of the mechanical system. I only have 2 LED modules so I can try to complete with the pink duds...

    The 2 lower modules are the pink duds, with a shape different from the above 4500K modules. So I'm still stuck at this point...

    The mechanical work will be quite delicate but I can count on my neighbours :-)

    Now, I have to find suitable hinges...

  • Structure

    Yann Guidon / YGDES04/05/2017 at 03:00 0 comments

    I have estimated that a reasonable compromise for the weight and dissipation would be 10cm×10cm area behind each 100W LED. So for 4 modules I was searching a dissipator that measures about 20×20cm.

    Instead, I have found this :

    2 pieces of 20×10cm. Damn, it's not large enough but the surface is ok. What can I do ?

    The problem is actually a solution for something else : size !

    If I'm doing photography outside, I need to carry the thing and it takes some room. So each dissipator can hold 2 100W LED modules and then... it folds !

    Yes, the new idea is to attach them with hinges to make a more compact assembly, which also shields the fragile LED surfaces. The dimensions are also suitable for the batteries that can be attached behind one heatsink, with the electronics.

    There will be "a bit of mechanic work" but nothing too crazy, I just need a few hinges that leave some space for the LEDs once the sandwich is closed. The modules are less than 5mm thick (I haven't considered the screws) so I need a hinge with a large diameter (about 10mm). On the other side of the flap, a piece must keep the heatsinks from touching the LEDs.

    Overall, here is the unfolded system :

    I still have to figure out how to keep it open when I need it... and attach it somewhere so I don't have to hold it with my arm :-D

  • Energy source

    Yann Guidon / YGDES04/04/2017 at 18:12 0 comments

    This kind of project wouldn't be viable without a suitable source of electricity. I happen to have studied Lithium batteries, a decade ago :-)

    This project is a perfect fit for the stock of old batteries I had bought from a RC store, a while ago... 60€ for 6 2S1P KOKAM packs, that's a bargain :-)

    As far as I know, the low price (10€ vs 87€) is because at that time, the RC world was transitioning from 10C to 20C cells, and these are "old 10C" that wouldn't be great for flying engines. But still excellent for our purpose, with low current and short strong spikes.

    This battery is not heavy and the energy capacity is great : 7.4V×2.1Ah = 15.5Wh ! Give or take 20% due to aging (but not use because they are brand new).

    The peak current is also in the desired range : 10×2A=20A !

    I have 6 such packs, so I can make 6×8=48V easily. This is perfect because the LED modules require between 24 and 32V (there are 10×10 LED chips in the module). 4 packs in series will make up to 34V when fully charged (that's 8S1P...) but I have not taken the various voltage drops into consideration...

    But first, let's see if they are still alive ! The last measurements for one 2S1P pack is:

    • 20060608: 7.63V
    • 200712 : 7.57
    • Today 20170404 : 7.21V

    Looking good !!!

    So one 8S1P pack can drive short pulses of high current in the projected 4×100W modules, while also maintaining some decent light output. The pack could sustain short pulses of 20A, or about 5A/module if the 4 are wired in parallel.

    The 100w modules are made of 10×10 LEDs, 100W/100=1W/LED so we estimate that each LED is designed for 300mA. 0.3A×10=3A for 100W. 4 modules in parallel will draw 4×3=12A@400W (that's a lot).

    The voltages so far :

    1. 7.21
    2. 7.34
    3. 7.13
    4. 7.21
    5. 7.14
    6. Can't measure, bad contact ?

    The remaining 5 packs clearly need to be re-balanced. Fortunately, none seems to have swollen and they are all from the same batch so can be used in series. I'll parallel them (through resistors) and keep the 7.34 (healthier) for an eventual project.

  • Another ordering mistake

    Yann Guidon / YGDES04/04/2017 at 17:57 8 comments

    The project started a few weeks ago, with a little misunderstanding.

    The idea came from my unpractical 10"×10" 60W panels and I was looking at the new cheap 100W COB modules. My panels are nice because they mix both warm white and cold white, which makes them suitable for not-too-bad photography, but the 100W COB are single-temperature...

    I mostly find cold white (8000K) or warm white (3000K), both of them are unsuitable : I need something inbetween. And on eBay I find 100w modules with "full spectrum" light : Great !

    Of course, you suppose what is going to happen : I ordered 4 modules, got them after waiting, then tried them.

    The result is totally unsuitable for photography because it creates a weird pink/fuchsia tone !

    Then I understand why they are used and made in bulk : for growing plants. I don't. Anybody needs 400W of chlorophile-friendly light emitters ? I'm open to trades :-D (electronics-related)

    The happy ending (of this episode) is that I have already found 4500K modules. I'm waiting for the delivery of 2 others...

    (yes, the coating messed with my camera's white balance)

View all 5 project logs

Enjoy this project?



حزب مطنووووخ ✋️ wrote 07/10/2017 at 00:33 point


  Are you sure? yes | no

George Gardner wrote 04/21/2017 at 18:58 point

If it helps you, I've done a similar project in the past, testing the viability of using a LED array for photo flash. 

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/21/2017 at 19:02 point

Awesome ! I will look at it :-)

  Are you sure? yes | no

K.C. Lee wrote 04/21/2017 at 19:19 point

Nice.  I'll stick with the Xenon tube then.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/21/2017 at 21:58 point

"safe voltage" of 160V... I loled !

That's a great work that you did there, worth a HaD project page :-)

And it's great and reassuring that I am not the first to think about this kind of circuit ( @Ted Yapo inspired me BTW) but I'm not trying to make ultrashort flashes.

I'm more interested by pulses in the millisecond(s) range.

  Are you sure? yes | no

K.C. Lee wrote 04/05/2017 at 11:03 point

If you are going to be using natural convection, then you should rotate the heatsink by 90 degrees so that the warm air on the fins can rise long the fins.  At the current orientation, that air is trapped.  Make sure that stuff mounted on the fin side are not blocking air flow or preventing air to mix.  Natural convection is weak, so you have to do all you can for the efficiency.

Force air cooling for 100W is an alternative.  Even a tiny fan/blower can improve the cooling a few times.   Also look at CPU coolers.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/05/2017 at 16:19 point

I didn't mention enough that the system is temperature controlled. With a small temperature differential, the orientation matters less and the blackness of the anodisation helps radiate the heat away :-)

  Are you sure? yes | no

K.C. Lee wrote 04/05/2017 at 17:00 point

Radiation is a small fraction of the heat vs convection.  Your LED power will be thermally constraint, but you can improve the efficiency of the heatsink to push that limit higher.

See here:

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/05/2017 at 17:10 point

That's something to think about for the 2nd prototype :-)

Thanks for the picture, it's very clear :-)

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/05/2017 at 17:33 point

BTW if verticality is so important, nothing keeps you from rotating the whole assembly :-D

  Are you sure? yes | no

Hacker404 wrote 04/05/2017 at 10:47 point

The heat sink is a challenge. Aluminum is a very good heat absorber but not so good at transferring that heat to air. It has a high thermal mass and can store heat well so it makes a good heat absorber when the source heat fluctuates strongly but unfortunately being such a good heat absorber means that it is also quite good at re-absorbing the very heat that it just transferred to the air. 

The catch 22 is that some material re-absorb the very same heat they just expelled. 

Copper on the other hand is poor at storing heat but very good (conditionally) at transferring heat to the air. This would be most beneficial to your project as the flash is very low in consumed power compared to lighting. 

You could have a smaller heat sink if it were copper and fan forced on the condition (as mentioned above) that the copper is oxidized. Copper oxide is not better at expelling heat but it is however very poor at reabsorbing expelled heat so it's total efficiency is much higher. 

  Are you sure? yes | no

Ted Yapo wrote 04/05/2017 at 01:34 point

The nice thing about LEDs vs xenon tubes is being able to easily tune the light output (with current as well as duration - about the only option you have with tubes is quenching to shorten the discharge).  The downside is that the color temperature of white LEDs shift (toward blue) at high currents.  This complicates white balance, especially if you are just using the flash for fill along with (probably different color) ambient light.

I have wondered if you could mix in some other color LEDs to automatically tune the composite color temperature based on the white LED current.  You could either mix in some yellow LEDs that get used when the white ones are using high-current pulses, or blue ones that get turned off when the white ones are at high currents.

Getting crazy, maybe you could come up with a fill-in flash that tuned the color temperature to match the available light...

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/05/2017 at 01:36 point

Damnit, that's more complex than I imagined ! :-D I'll try to keep it simple though ...

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/05/2017 at 01:38 point

Anyway, the funny thing is : with 4 parallel modules, the individual LEDs will not be much overdriven so the color shift might not occur...

The only way to know is to try, right ?

  Are you sure? yes | no

K.C. Lee wrote 04/04/2017 at 16:47 point

I have considered doing this, but using discrete LED and running a few time the rate current as flash is low duty cycle.  I have also thought about using supercaps to supplement the battery.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/04/2017 at 17:12 point

I'm not the first nor the last one, I've used LEDs for photo lighting 10 years ago with some leftover bars from a KW project :-)

But I need something portable, compact, pratical, somewhat rugged, so I can shoot outside, and not burn my hands either...

  Are you sure? yes | no

K.C. Lee wrote 04/04/2017 at 17:37 point

I am going for a much much smaller flash.  Something as a slave that supplement my point and shoot digital camera.  I picked up a vintage "toy" xenon flash last year, but wasn't happy with the design at all.

I have just placed an order for 200 "5730" 1/2W LED (for $2) as spare parts for my LED lighting.  So I'll get to play with them.  Their peak current is 500mA (vs 120mA continuous). Not a whole lot, but having the flash from a large array of dimmer one instead of a bright spot isn't an disadvantage.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 04/04/2017 at 17:47 point

Well, it depends.

I've been using a 10"×10" array that can stand 60W but it needs 24V and a heatsink... which I don't have, so it burns your hands after 20 seconds :-D

And I don't like direct light from the camera, it makes the pictures look flat...

  Are you sure? yes | no

Does this project spark your interest?

Become a member to follow this project and never miss any updates