Creating an 8x8x8 3D tri-color LED cube from the ground up involves a variety of tasks, one of which is burn-in-testing the eight 8x8 panels before assembling them into the final cube.
Just in case you haven't been following my earlier columns on EE Times -- augmented by videos on YouTube -- I'm in the process of building a 3D 8x8x8 RGB LED cube. This project was inspired by this article, which was written by our very own Max Maxfield.
In my previous column, I went into detail about the jigs I created and used for constructing the 8x8 RGB LED panels, eight of which are required to complete my cube. In the video associated with that column, I demonstrated some low-level testing of the LEDs to confirm that none of them had failed during panel construction. I also mentioned that I would be conducting a burn-in test on each panel prior to final assembly.
Picking up from where my previous column concluded, I would now like to provide more detail to accompany this video for the 8x8 panel testing.
The last thing I want is to carry out the cube's final assembly only to find a defective LED right in the middle. That would be disastrous and almost impossible to replace without having to dismantle something, possibly causing irreparable damage.
I'm sure there are some members of the EETimes community who could comment on component reliability. I don't proclaim to be any kind of expert with regard to this and my view is very simplistic. I have a vague memory from many years ago when someone told me about the concept of the "bathtub curve" and that components are more likely to fail at the beginning and at the end of their life span. With this notion in mind, I want to identify any LEDs that are likely to fail as early as possible.
I already discovered during the initial testing of 1,000 LEDs that a couple of these components were DOA (dead on arrival). Also, during the construction of the 8x8 panels, a few LEDs that were previously working subsequently failed. These additional failures can probably be attributed to one or more of the following causes:
- Weak components that were going to fail anyway
- LED leads being bent too close to the LED housing
- Excessive heat during soldering
- Not taking adequate anti-static precautions
There is nothing I can do about a component that's going to fail anyway -- this something I have no influence over -- so this is why I'm testing and re--testing my LEDs at all stages throughout the construction process so I can immediately eliminate or replace any failed LEDs, whatever the cause of their demise.
As you may have seen in my 8x8 panel creation blog and video, I bend the LED leads very close to the LED housing. I have since found out that best practice suggests leaving at least 3mm between the LED housing and any bends in the leads to reduce stresses on the diode junctions. This is something to remember for the future, but it's too late for me to change this now because all eight of my panels have been completed.
Another factor that can degrade an LED (or any component for that matter) is how much heat -- and how long that heat is applied -- during soldering. I'm never quite sure what the ideal soldering temperature is (if one exists) because there are multiple factors to consider (these factors are beyond the scope of this article). Personally, I've always liked a relatively hot soldering iron tip and I generally operate my iron at 400°C. I do use a fume extractor a lot of the time, and I think that has a cooling effect as it draws air over the iron and joint.
I read recently that a typical solder joint should take no longer than 3 seconds (less for surface mount components), I'm sure I have inadvertently taken longer than that for some of my joints and that may account for some of the LED failures. All I can do in the future is to try and improve my soldering skills and/or reduce the temperature of my iron.
Finally there is my blatant disregard for taking adequate anti-static precautions when handling LEDs. Until very recently (when Max mentioned this to me), I hadn't considered LEDs as being static-sensitive devices. But even if they are generally more tolerant to static discharge than say a microchip, it's still worth taking appropriate precautions so as to rule out static as the cause of failures (see also Zapping Things With ESD – Just One More Service I Offer). I have since improved my anti-static precautions and -- at a minimum -- I wear a suitably earthed wristband.
The bottom line is that, when all of these factors are combined, it's no surprise a few LEDs have failed. Next time around (if there is one) I'll know better.
During my original 1,000 LED test I compared one LED with another to identify any colour and brightness discrepancies. At the time of that testing, I didn't spot anything untoward. Having now lit up a whole 8x8 panel, however, I did identify one week red LED and had to replace it. It's much easier to carry out a repair now than after the cube is fully assembled -- hence, I came up with the idea to carry out a burn-in test for each panel. I plucked a period of 24 hours out of thin air, and decided to run each panel for that period cycling each LED through various colour changes. It's also an opportunity to verify that the colour and brightness of all the LEDs are even across the entire panel.
To Page 2 >