As is now more than obvious, we are behind schedule. Thank you very much for your patience and understanding while we continue to resolve some problems and start producing light bulbs at a fast pace once again. Here is a recap of what has happened since May:
We produced our first set of light bulbs, over 1000 pieces, and started shipping these out in June. Production was starting to go smoothly and although we were a month late with the first batch, we were hoping to ramp up production so that we could deliver the rest of the batches on time or early. However, a few weeks after the first set of light bulbs went out, we received a few reports of light bulbs that died soon after being received. We asked the Kickstarter backers to return the bulbs and we spent some time analyzing and identifying the cause of failure. In the meantime, we paused production. We discovered a problem with the controller IC behavior and we implemented a workaround that fixed the issue. Although this issue didn’t affect a large percentage of bulbs.
So about August time, we gave the go-ahead for the second, much bigger, batch of production. Thousands of bulbs were ordered. About a month later after bulbs were emerging from the pipeline and started their burn-in testing phase, again we noticed that bulbs were failing. This time, the rate of failure seemed higher. We continued burn-in to gauge a longer term estimate of the failure rate. 2 – 3 weeks on burn-in and bulbs were still failing. We couldn’t ship these bulbs knowing that they might not last all that long out in the field. Therefore, we halted the production again.
The clear symptom was that in a failed bulb we always found that one of the LEDs had failed and broke the current flow to the rest of the LEDs. No other failed components were found. We first looked closely at the power supply. Although the current to the LEDs is regulated very tightly to prevent any flickering of light, we saw that there is an overshoot of current at the time the bulb is turned on. We thought that this may be the culprit: a surge that overloads the LEDs and causes one to fail. We implemented a soft start feature so that current is ramped up more slowly with zero overshoot upon power up. We produced about 100 bulbs with the soft start to check if the problem was solved. However, the failure rate was not improved and we were still stuck and it was already October.
The next course of action was to look at the LEDs in detail. These types of LEDs are made on a mass scale and the methods are tried and true, or so we thought. We spent a solid week just looking at failed LEDs under a microscope. We set up several tests to isolate the problem. This is where science and well controlled experimental technique was essential. To make this story short, we discovered that the cavity inside the LED that contains silicone was slightly under filled. This caused an air void near one of the gold wires. The gold wire electrically connects the LED die to the package housing so that the light producing chip can get power. During operation, the light bulb heats up and the silicone inside the LED expands, causing pressure onto one side of the gold wire. Because of the air void, there is no counter pressure to keep the gold wire in place, and it bends on a microscopic scale. Upon repeated thermal cycling, the gold wire is bent back and forth until failure. With great sadness, we declared the approximately 150,000 LEDs plus the hundreds of light bulbs that we produced unusable. It was like being in the middle of a graveyard of LEDs and Nanoleaf bulbs. Yes, saddening, that is all to be said.
Having identified the problem, we asked for some changes to production in order to drastically improve the reliability of the LEDs. There were three changes that we made that are aimed to strengthen the region of the site of failure dramatically. We did another small production of LEDs, enough for about 60 bulbs and these bulbs have been in burn-in testing for about a week now and so far zero bulbs have failed, whereas before, bulbs started failing after only two days of burn-in. As you might guess, we are extremely happy to see this result. We actually do expect some non-zero failure rate. Like with any product on the market, each part has a certain failure probability and thus a perfectly reliable product is impossible to make. We hope you can see from our story that we care very much about the quality of our Nanoleaf bulbs and we will continue to study points of failure in order to improve long term reliability to the best it can be.
So, we hope that this Kickstarter update will be taken on a very positive note. We have already asked the facility to start production of new LEDs with our changes. There are a lot of LEDs to be re-made and this will take a bit of time. Production will start this week. The printed circuit boards are already made and ready to go and just waiting for new LEDs. We have a revised production schedule as follows:
Please note that shipping time is approximately 3 – 4 weeks depending on the destination.
We are quite embarrassed and sorry that we have had to delay these bulbs several times now. We want to get these Nanoleaf bulbs out to the backers desperately. We have been pushing all the production people to bump priorities and get things moving as fast as possible. Thanks so much for your support and patience.
The Nanoleaf Team