R.I.P. electro stuff
Let me start with the good news. The electronics are in! The LEDs are looking exactly as the other one so that shouldn't be any problem.
What actually IS a problem is that I fried some of the parts... So some mysterious reason one of the Xbee modules just stopped working after a week of continuous action. I have no clue why but I sure do hope it won’t do it again.
The LED drivers are a different story. I tried them in a breadboard with the example code and it worked very well without any problems.
Connecting the prepared PCB to the PSU, arduino and LEDs made a big difference. The chip got 100+ degrees and didn’t do anything anymore afterwards. Obviously something was not right. Checking the wires made clear that the TIP48 were not insulated from the metal plate and made a short circuit. After insulating them I gave it another shot with the second LED driver. Didn’t get very warm now, looks good. Lady luck wasn’t on my side this time either. It overheated in just a minute and again chip 2 was dead. Rechecking the wires showed that the ground pin was not connected to anything. How stupid… I think we have to order new ones and make sure everything works fine on the breadboard before using this PCB. Otherwise just make a new one.
Till we meet again.
'KnightRider' test with 16 LEDs
Electronics
The casing seems to fit tight even without gluing the parts together. The friction and construction keeps it nicely in position. The inside is what has to be created. The LEDs are ordered and should be here a week ago. Unfortunately there is something wrong with the supplier. He only has 4 LEDs in stock and I need 5 (of course). So he’ll send the first four and I reuse the one that’s in another prototype. Based on the other prototype I reused the schematics for the LEDs.
It is quite basic with 3 transistors used as gates. The LEDs drain too much power (350mA@3V) to connect them directly to the arduino or LED driver. It should be possible to switch 1W of power which makes the tiny BC550 transistors which I normally use for switching LEDs are not sufficient. Now I use the TIP48 which are a massive overkill but at least they will survive.
-
- 3W X-Power Kingbright RGB LED
-
- Control PCB with TIP48 transistors and resistors
Individually controlling 5 RGB LEDs takes up 15 PWM channels. The arduino just has 6 which means external electronics are needed to get more channels. Luckily somebody wrote a tutorial on the arduino forum how to do this with a TLC5940 chip. This chip has 16 PWM channels with 4096 levels each. This is more than sufficient for the lamp I would say. The downside of the chip is that it sinks the current into the chip. This means a channel is triggered by a low signal. In practice it turns the code upside down. Instead of turning a channel fully on by sending the high value (4095) it makes the channel low. Not a big problem, but it should be taken into account when programming.
Now we wait for the electronics to come in and see how it works out.
Building the demonstrator
The time was there to build the indoor climate lamp. Now it comes down to what materials should be used. One thing that is for sure is that the arduino is used to receive the input from the Xbee modules. The logical thing to do is use the arduino to control the output.
For the output a light is chosen that can be seen during the daylight and would give a nice fluent glow when lit. Here at the spire institute they had a 3W RGB LED that meets the requirements. The only downside is the heat it produces, together with the transistors to drive the 350 mA current.
The material for the diffusion of the light was also a coincidence. In the research center was a piece of rest material they used for a projector screen which breaks the light in a fluent way. The only challenge left was to find the right material for the base.
Another electronic challenge was the power supply. The LEDs need approximately 3V per channel at 350mA. This comes down to roughly 1A per LED. In this case 5 of these are used so a 5A power supply was necessary to only drive the LEDs. Not alone 3V was needed, but also 9~12V for the powering of the sensor boxes. The best way to go is an old power supply from a computer. It has all these power levels and can deliver enough amps per line, plus at a university are enough old computers lying around that can be scavenged.
The sensor boxes are made out of foamcore which gives them a neutral look from a shape and color point of view. People should not reflect on the object but should use the object to talk about the indoor climate. The lamp should follow this neutral language.
The first try was to use 3mm foamcore to create the base. This turned out to be a precise job and was taking too much time to cut it neatly. The other downside is the durability of the material. It dents easily which makes it not the ideal material for transportation.
Foamcore mockup
During the ITD course I worked with a lasercutter to cut the materials from flat material and glued the whole thing together as a big puzzle. The material was changed to something more durable but still easy to finish, MDF.
Denmark, or at least Sønderborg, was not the best place to find a lasercutter that could cut MDF. The decision was made to do this at a company in the Netherlands.
Overview of the components
With all the loose parts together the big fun of building could start.