The topic of outsourcing is often a subject for debate by politicians, managers, engineers, business owners and even human rights groups. Depending on the product, it can be a significant cost benefit to outsource certain aspects of the production process. From an engineering perspective, outsourcing can have positive and negative effects. It can help to bridge short term labour shortages, add expertise or simply free up internal resources to focus on key developments. It can also slow down processes and increase costs when something goes wrong, i.e. it’s much more challenging to debug an issue an ocean away than from your own lab or factory. And costs can add up quickly when time to market is a priority.
I was recently confronted with an issue with one of our image sensors. One of the many on-chip ADCs we had sourced exhibited higher random noise when compared to its neighbours. While the layout and design are exactly the same for all ADCs on the die, we found that the supply routing in the package was slightly different for this particular channel. It seemed modifications to the package would be needed to address the issue. What if a re-spin still resulted in higher random noise? We would still lose days – maybe weeks, not to mention the additional cost for a quick turnaround. No one wants to risk that, especially not before a mid-year review! 🙁
For an issue like random noise, a lot of thought and brainstorming might help to make a point, but at the end of the day, it’s more effective to go to the lab and start experimenting. What if you discover that a modification to the wire bond process would prove your analysis and the wire bonder is still a few time zones away? You could draw up a new wire bond diagram and ask the manufacturer to build 5 parts for you. And he will respond by asking for the drawing and a small fortune in cash, and you will get your parts two weeks later… (after the mid-year review, and with a slip in your schedule).
Thankfully, in this case, the wire bonder was in-house and all that was needed was a sketch on a piece of paper and an explanation to go with it. And a day later, modified parts are on my desk and a hardware fix is in place. Phew.
I was blown away by the quick turnaround! I can see that the development cost of a successful product can disappear quickly in the financial “noise floor”, but in this case I would argue that the in-house wire bonding delivered both improved quality and kept our schedule on track!
This event was a good reminder that manufacturing outsourcing can be a double edged sword – on the one hand, it can provide real value to our final customers by virtue of lower production costs and access to technology that would be too expensive to implement on a large scale. On the other hand, it necessarily leads to a loss of agility at the R&D stage – that agility has real value in terms of time to market in the world of R&D where one of the few certainties is that the unexpected will happen and design or process revisions are the norm.
In the end, the decision to outsource or in-house requires a delicate balance between often competing demands – cost, technology access, and time to market.