Project Logs

Collapse

• The box and the first solid lamp

  Enrico • 09/13/2017 at 21:31 • 0 comments

  After having a good idea on what heatsink to use, with a limited capacity on product aestethics and the will to get something done, I've end up to enclose anything in a box.
  

  I setup the enclosure starting from an ABS box found at the local hardware supply. The plan is to fix everything inside, by means of hot glue. Then place the heatsink on the box and fixing it with some threaded rods, since it provides some holes for mounting. Fix everything on the box by means of screws and put an acrylic square on top of it. All is kept very minimal, like the acrylic is just a square because was simpler to cut with a Dremel. Any kind of mechanical spacer is made by using nuts. Anything goes for just a prototype. The key: have something to put on a desk.
  

  Firstly, I placed the power LED drivers and the controller board in the box, connect them by means of wires and hot glue to strenghten the solder joints. Then I prepared the LED wires to be connected to the LED engine afterwards.
  

  Below, the box filled with the electronics and airwires ready to be connected to the LEDs.
  

  A detail of the controller board once in the box. On one side there is the power jack and USB connector for the remote/automatic control, on the other the encoder for the manual control.
  

  A detail of the power boards, just one hole made to let the wires pass through.
  

  The wires are then connected to the LED engine, which is connected to the heatsink by means of thermal tape. The heatsink provides space to use some threaded rods which I used as shown.
  

  So that I can apply a precutted and milled cover on the LEDs and fixing it with nuts.
  

  And the final result look like this:

  
  

• LED engine history

  Enrico • 09/10/2017 at 22:14 • 0 comments

  The issue

  When I started to breadoard the lamp, I was using this PCB, carrying on the LEDs (from now on called LED engine). The one shown in the breadboard was this one:

  Chosen to reach, according to thermal analysis of the LEDs (from datasheet) and the given suggested thermal pad on the PCB, no more than somewhere 70°C at full power.
  

  Making this in reality shows that my calculations were imperfect: the heatsink could reach almost 75°C and more, showing that this thermal pad
  

  
  wasn't probably conceived to be thermally closed within traces as shown in the picture. I say this because was the only variation from the specified layout: even the right amount of copper was used, according to the application engineers. The calculations were simply done in this way:

  Where the case-heatsink term is the thermal resistance given from the LED pad design provided from the application note. But what in the end make me realize that calculations were even more far, is that the LED power were overestimated, because I assumed that all the power is generated in heat, while in reality the LED have a given efficacy (around 30/40%), meaning that the 40% of power is generated in light and not in heat (well.. the heat is light per se, but not useful here). So in reality, it shall be even 40% cooler than it was.
  

  In the end, I also realized that if such heatsink shall be touched (I want to use those LED at desk height ideally), even 70°C were too much, forcing me to use either a fan (but I am against them were their usefulness is not proven) or a bigger heatsink.

  The monster

  When you get betrayed by a small heatsink, you go big. So I decided to look for some nice heatsinks available, and made a board for them. I came out with the third revision of the LED engine, as big and round as a coin, overcompensated with an heatsink beneath it.
  

  
  

  The big size has also the purpose of being helpful for supporting mechanically the lights somehow. As you can see from the picture below, designing the PCB was quite interesting too:

  The little brothers

  When I was taking a picture of the big one, I have also realized how many attempts I've made in the past. You can see them here one next to each other:

  On the left, the first ever board made was described in detail here. It was a first design trial implementing a linear LED driver, with firsts experiments with OSHPark and SMD soldering. Despite errors and imperfections, each LED was surprisingly cooled.
  Then the second attempt, still with a linear approach, with a different layout was described here and corresponds to the second from the left.
  The third one is the one used up to now and the biggest one is the final revision. Actually, there was a middle version of a bigger heatsink in the second revision of the LED engine, just to get things done, from a previous project:
  
  

  
  After this refreshing (with heatsinks.........) update, I can go forward with the assembly of my lamp, while improving the firmware. Stay tuned!
  

• Board first assembly

  Enrico • 08/15/2017 at 09:16 • 0 comments

  When I had started the project, I had (and usually have) no idea on how to integrate stuff. So as a natural consequence was to use my already developed LED drivers. But I had in my mind some use cases of this lamp.

  It must be connected to a PC, because I want it to be smart, but it need to deliver more power than a USB can do, so it must work also from a DC power supply - this already triggered the firmware generation in my mind, mind compilation has started.

  This is the board currently used for providing power to one or more series LED:
  
  

  
  And I need 4 of them, because I want my lamp to execute light algorithm with the white too, in order to be an improved RGBW system. That is why I was able to easily use my other board, initially started for general purpose "high power" Arduino equivalent, thought for 4 LED driving system. Here the board, Lino, with no tweaks, still a "vanilla" version:
  
  

  
  And I startede thinking to them. So I had this connections in mind:

  It turns out that I can directly put them on a breadboard when soldering some headers on the boards. So I have also crinched together an encoder recovered from an old car radio. So that I can do some functional tests and start working on the firmware, too. The LEDs that can be seen in the picture are the classic Luxeon Rebel series, soldered on a board, a satellite board. I was able to use a small heatsink which I bought, being sure that was thermally compatible with my calculations. But the satellite history is something that deserves a separate log.
  

  In the picture above, you can spot that the process of testing the Lino in various operating modes, debug and stresses, added to the board the signs of the mods.
  

  The overall effect with some basic firmware I have to say it is quite interesting in terms of light and its ambient effect.
  
  

View all 3 project logs