How to prepare to manufacture hardware – what to do when things go wrong
Wednesday, June 25, 2014/
Drawing from our experience bringing Holiday by MooresCloud to market, this series covers the whole journey: concept to design, preparation for mass production, compliance testing, manufacturing, and what to do when things break down – because they will break down.
When things go wrong
By early November in 2013, MooresCloud’s manufacturing partner in China had received all of the components needed to assemble Holiday. Kean Maizels and I developed, tested and delivered a suite of test jigs, code and procedures, so the manufacturer could know whether the finished product performed as intended. After a few weeks of drama, we’d finally passed compliance testing – with some additional component and assembly costs. Everything in place, we gave the order to begin the manufacturing run.
Within a day we knew we had a serious problem.
This wasn’t our first sign of trouble ahead. During the assembly process for the LEDs, there had been reports of some unusual behaviour, of globes generating the wrong colours, or no colours at all. We thought these failures were few and far between, and assumed they fell into the small number of component failures you might expect.
That was our second mistake.
We’d made our first mistake a few months earlier, when we test manufactured 10 Holidays in preparation for compliance testing. Kean had many of the necessary components in stock at his office, so he shipped them to the factory – faster than waiting for a component order to be filled.
We assumed that the components the factory would use to build our full order would be identical to the components used to build our test manufacturing run. That was our first mistake.
In September, the manufacturer of our LEDs had changed their design, but didn’t inform anyone of this change. We purchased one batch of LEDs in April – used in prototypes and that test manufacturing run – and assumed a second batch, purchased in October for the full manufacturing run, would be exactly the same. They were not.
As we didn’t know about this change, we had no explanation for the failures seen on the production line. Everything had gone smoothly in our test manufacturing run. What could be happening?
The problem analysis with our manufacturing partner moved painfully slowly, hampered by time zones and language barriers. We’d recommend a change in the morning; they’d send back photos of the results in the evening. All the while, our schedule was slipping.
Won’t somebody please think of the customers?
MooresCloud had promised our customers we’d begin shipping units early in November. By the middle of November, with no end in sight to our manufacturing woes, we knew we’d miss that window, and perhaps even miss the Christmas season all together.
We didn’t know what we could do to fix the problem, or even if the problem could be fixed.
At this point we could have gone silent, or dissembled as customers asked when they could expect to receive their orders. Instead, guided by MooresCloud’s policy of transparency, we emailed weekly updates, informing customers of the current status of our production problems, and our efforts to find a solution.
Our customers responded in kind: grateful for the updates, patiently waiting for our manufacturing fortunes to turn around. When things go wrong, honesty is the best policy.
Our man in China
On the 21st of November, during a lengthy call with the manufacturer, I realised that if we expected to solve this problem quickly, we had to have our own man on the ground. Three days later, Kean was on a flight to Hong Kong, to spend a week working directly with both our manufacturer and the manufacturer of the problematic LEDs.
On his third day overseas, Kean sent me an email – three letters repeated, over and over: “ESD ESD ESD ESD ESD.” Electrostatic discharge, which delivers shocks in cold, dry weather, can rip apart delicate semiconductors. A small design change in one element of the LED component made them incredibly sensitive to static discharge. Our LEDs were being destroyed during the manufacturing process!
Kean instituted much stricter anti-static measures on the production line, and immediately saw Holidays pass through assembly and testing with flying colours.
Our problems had been solved, in just a few days, because we had someone on the ground.
All of our customers received their Holidays before Christmas.
What have we learned?
When you start a manufacturing production line:
● Use the same batch of components for both test production runs and full production runs;
● Have someone on the ground when the full production run begins, to quickly identify and resolve any problems;
● Track key component manufacturers carefully for any changes in their designs;
● When problems arise, communicate transparently and frequently with your customers.
These articles have touched on much of what we’ve learned at MooresCloud as we brought Holiday to market. Hardware manufacturing is exacting detail work, requiring both patience and stamina. But the results – a smile breaking across a customer’s face as they have a play with Holiday – more than make the effort worthwhile.
Mark Pesce is co-founder of MooresCloud. His website is at www.markpesce.com.
From the frontlines
Startups, synagogues and soonicorns: Exploring the world’s most innovative ecosystem Charlotte Petris Timelio founder
Australia needs to follow the UK and introduce a flexible work bill Gemma Lloyd WORK180 founder
The ‘anti-startup’ story: How to turn $1,000 into $15 million with no investment Alex Georgiou ShineHub co-founder
New venture? How to decide who and what to bring along for the ride Colin Anson pixevety co-founder
Five critical questions: Are you listing your startup too soon? Lisa Schutz Verifier founder
Three massive influencer marketing fails businesses can learn from Anthony Richardson Q-83 founder