Issue link: https://iconnect007.uberflip.com/i/1108006
30 FLEX007 MAGAZINE I APRIL 2019 The second foundry quoted us a much higher number for NRE to get our first prototype out. But they had a plan to get us to 1,000 wafers a month at a much lower cost. It's tempting to say, "Let's save that $100,000 because we're a startup and we are short on cash. We'll worry about tran - sitioning later." But at that point, you're already trying to commercialize, and your burn rate has gone up. You have customer engagement now, and in the middle of that ramp, you must change foundries, test, package, and requalify the line. That kind of thinking has happened more often than it should. If you're prudent, you'll recog - nize that by investing in a good partnership early on, you'll reap the benefits of that later tenfold. It's the right way to do it. Matties: In terms of the assembly process, what challenges did you face at Lenthor? You mentioned the solder mask, but were there any other concerns in the assembly? Was it all done by hand or was it automated? Clemente: It did go through a standard reflow assembly process for the most part, but the key for us was the board construction; again, it wasn't a typical rigid-flex, so it had the advantage of rigid materials. To keep the com- ponent area flat, solid, and stable for compo- nent attachment, we needed to come up with assembly fixturing that aided in keeping the product in a state that allowed for standard reflow automated pick-and-place assembly attachment. That was our primary mechanical challenge from an assembly standpoint. Tony can maybe discuss some of the other aspects of it, but from the daily mechanical work aspect, we needed to come up with some internal tool- ing and fixturing to hold the product in place. Flannery: Yes, I had a lot of meetings with Matt Kan, the assembly manager at Lenthor. We have a lot of sensors; the optical sensors have a sensitive window, the microphone has a port, and the accelerometer is sensitive to certain types of vibration and shock. I had experience in the past where people used the accelerom- eter that I had built, put it into a smartphone, and their pick-and-place machine was like a jackhammer. They used stainless steel tips, super-fast pick-and-place, and it was hammer- ing the accelerometer and destroying these delicate, inertial navigation elements inside of the accelerometer. Lenthor had done some MEMS work, but we were pushing the envelope on their experience in terms of all the sensors we were putting into it. We went through each sensor and explained their sensitivities, such as shock or vibration, and gave white papers and app notes on their assembly. For example, the microphone had an open port. Most people use a no-clean flux, but we said, "You have to use a no-clean flux. You can't spray liquid on a microphone port, or it will destroy the part." We also explained, "The optical windows can't be scratched, bro - ken, or have a residue left on them or it will affect our photodetector, LEDs, and the effi- ciency of their coupling in the skin." We went through all these details even before design and assembly. I sat down with Matt and said, "My concern is there's a solution for every single one of these, so let's make sure that we're conscious of this and don't build these with zero yield." By carefully stepping through the challenges early on, they were able to program their pick-and- place machine a little bit slower—a little softer hit and land. We also talked about different tip materials and reflow profiles. In the wearables market, almost everyone is dependent on sen- sors, which have certain challenges. If you're making a solid gel electrode array and trying to couple that into your circuit, you have to take that into account when you think about your assembly and how you're going to build your final product. In the wearables market, almost everyone is dependent on sensors, which have certain challenges.