Project Logs

Collapse

• Painted lampshade, and some other upgrades

  t.oster92 • 01/28/2021 at 12:42 • 0 comments

  The lamp is still working after almost one year. But over the months, i noticed that there were some things which needed a little improvement. Yesterday, i finally decided to do so. 
  The first thing is: Sometimes, gnats would enter the lamp, without being able to leave. Therefore, i had to clean the lamp regularly, which is not that easy. It was simply not designed for that. The obvious solution is to seal the lamp better, and this is just what i did. I glued some white foam under the acrylic glass and on the inside of the lampshade, and this foam seals the lamp now. There are still some places where gnats can enter, but the risk of that is much lower now. 
  

  Also, i bent and painted the lightshade. The bend should make the light a little more focused, and the paint makes it look a little better.
  
  

  I still need to change something about the corners, since the aluminium foil doesn't really look good. But for now, it does its job. Also, some cable management would be nice. 
  
  Also, i decided i should change a little detail about the boost converters. 

  In one of my previous project logs, i wrote that i placed a 1MOhm resistor in front of the "run" pin of my boost converter ICs. I decided that this was not enough, and placed an additional 100k Resistor on each "run" pin to form a voltage divider with the 1M resistors. 
  
  The "run" pin can be used as an under-voltage lockout (UVLO). If the LTC1871 measures 1.25V or less at this pin, it will not run. When taking the voltage divider into account, this happens at an input voltage of 13.8V or less. This is important, because most switching regulators act as a constant-power load - the lower the input voltage, the higher the input current becomes. If, let's say, a 5V source was connected, and the lamp was set to a power of 75W, it would have to draw 15 Amps, and the fuses might blow because of this. 
  But because of the UVLO, the boost converters simply won't try to start at a voltage which is too low for reliable operation. 
  This can also improve the startup process of the lamp, since when turning the lamp on, the input voltage will ramp up, instead of being present instantly. If it ramps up too slowly (this depends on your power supply), the lamp could attempt to start at the beginning of the voltage ramp, with the consequences i described above. 
  In the case of my power supply, this never was a problem, but it's generally better to design electronics to be compatible with different power supplies. 
  
  

• Spectrum visualization tool

  t.oster92 • 10/25/2020 at 10:48 • 0 comments

  I wrote this python script while designing the lamp. It will calculate the output spectrum and efficiacy of the lamp, and some other things, depending on LED currents. For me, it was most important to get an idea of how well my plants will be able to utilize the light, so i chose to take the 'Yield Photon Flux' (YPF) as a primary measure of how efficient my lamp will be. For most other lamps, in contrast, only the 'Photosynthetic Photon Flux' (PPF) is given, but this won't take into account that a plant will use green light less efficiently than red and blue. 
  For the calculation of YPF i used the well-known McCree curve. McCree averaged the response of single leaves of different species, so this curve most likely doesn't tell the whole picture, but it is a good starting point. 
  
  I tried to make my lamp model as accurate as possible, so LED temperatures, forward voltages and current dependant droop are taken into account. 
  
  
  

  The tool

  You need to have Python 3 installed, and several other common libraries like matplotlib etc., and download light_spectrum.zip in my file section. After unpacking and starting light_spectrum.py, the blue sliders can be shifted with the mouse to change the LED string currents. The spectrum will change accordingly, and efficiacy, power and YPF are calculated from the currents. 

  
  I created models for Lumileds SunPlus20 series, but other LED models could be imported as well. For this, you need to change the csv files in the "CurrentVVoltage", "FluxVCurrent", "FluxVTemp" and "Spectra" folders to match your LED behaviour. You will also need to change the x0r,x0b,x0w factors in the "leds.py" script. They can be calculated with the 3 scripts lumen2powah, powah2lumen and PPF2powah.
  

  This way, you can compare how well different LEDs would do in this application. It won't matter if PPF or lumens are given - these tools will make all LEDs comparable. I used the tool for this purpose.

• Schematics

  t.oster92 • 10/06/2020 at 10:16 • 0 comments

  LED Channels

  The lamp uses three LED channels, whose currents can be controlled individually. Each LED channel has a LTC1871 switching controller, a power MOSFET, a shottky diode and a coil, which form a boost converter. The LEDs are wired in series, and two parallel 2Ohm resistors are used to measure LED current.

  NOTE: This schematic contains an error. Place a 1Mohm resistor (or even a high- impedance voltage divider) in front of the "RUN" pin, otherwise, the LTC1871 won't work. I overlooked this during the design, but it was an easy fix. 

  Overall Schematics

  All boost converters derive their clocks from a LTC6902 spread-spectrum clock generator. I use this chip in 4-phase configuration, of which 3 are utilized by the switching regulators. The free-running clocks of the switching regulator ICs must be matched closely to the LTC6902 clock, otherwise the synchronization may not work. 
  
  An operational amplifier is used to scale and shift the voltage accross the 1Ohm LED shunts, to match the FB pin voltages of the LTC1871 ICs. 

  NOTE: R22 should be 10R, not 10k. 

  I use an external controller to monitor the lamp. It is switched off in case of an overtemperature or fan failure. Those schematics may follow later. 

• Still working

  t.oster92 • 10/06/2020 at 09:54 • 0 comments

  The lamp is running for several months now, and so far, i did not notice any problems. Most plants are growing well, especially basil and lemon grass. I did not have too much luck with hot peppers though, but i don't know whether this is due to the lamp or just the room climate. 

  Enjoy my plant pictures! First schematics will follow in the next log. 

• New Fans, and a cool optical effect

  t.oster92 • 05/13/2020 at 20:24 • 0 comments

  The old fans were a noisy mess right from the start, so i exchanged them for some new ones. They run much quieter, and also they are white, which makes it possible to get a really cool effect by illuminating them with a LED. Have a look! The video is not slowed down; this really is what you can see when you view it with your own eyes. 
  

  Can you guess how i did this? ;)

• Protective Glass and Near UV LEDs

  t.oster92 • 04/28/2020 at 18:41 • 0 comments

  Recently, i added some protective acrylic glass in front of the LEDs. Before, i was afraid that some plants could grow into my lamp, if i don't pay enough attention - well, that's unlikely, but you never know. The glass doesn't seem to impact efficiency too much. 

  Acrylic glass does reflect about 8% of incident light (at 90° to surface), and almost 92% are transmitted. Of those 8% that are reflected, most will be reflected again by the white PCB, making a second attempt to pass the glass plane. This again will be 92% successful. In the end, the efficiency will most likely be reduced by 1-2%.
  On the plus side, the light distribution will be a bit more uniform, so shadows are not that harsh. 

  Also, i finally put some near UV LEDs onto the board. I don't think they will help too much in irradiating the plants, since the are really low-powered, but they're nice to have. 
  
  

  To have any effect on my plants, however, the UV LEDs probably should have some shorter wavelengths as well as more power. It has been shown that UVB radiation can alter the content of carotenoids in some plants: https://www.sciencedirect.com/science/article/pii/S1319562X10001063
  And those seem to work well as antioxidants in the human body: https://www.researchgate.net/publication/9035299_Antioxidant_activity_of_carotenoids

  So, UV light might be good to have, but i doubt that my weak UVA lights will have any chance of doing anything useful. But at least i can try to observe some differences.
  

• New Lampshade

  t.oster92 • 04/22/2020 at 19:20 • 0 comments

  The old lampshade was made from cardboard and aluminium foil. Just a quick and simple solution, but it didn't look too nice. So today, i finally cut out some aluminium sheets and drilled holes in them to make a new one - and it turned out quite nicely. 

  I might add some corner pieces in the future, but for now, it is good enough. At the corners, there won't escape too much light anyway, since it is blocked by the lamp frame. 
  Also, i might add one more bend to each of the shade pieces. This way, i could approxmiate a parabolic reflector a little better.
  Also, the shade needs a painting on the outside. Maybe black with a waterdrop effect. 

  
  First, however, i will work on the cooling system some more. I've got a really cool idea for a visual effect, which has something to do with the fans already being synchronized. But i won't tell too much, you have to be patient :P
  
  
  

View all 7 project logs