“What’s wrong?” I asked Jeroen, my cofounder and VP Engineering.
“I don’t know,” he said. He looked confused at the results we were looking at. We had just gotten the wafers back for our first product and the results were not what we expected.
“Something’s probably not connected right to probe the wafers,” I said confidently. I’d been through hundreds of product launches, and usually the mistakes were something easily correctable.
“No, I already checked, and the connections are right.” Jeroen looked like he was getting frustrated.
“We shouldn’t be seeing these type of results,” he said.
“Okay. I’ll leave the lab. I know you’ll figure it out,” I said.
About an hour later, Jeroen came into my office. He had figured out what was wrong. He asked me to come into the lab with him.
I walked back to the lab, and he showed me what he had found on the oscilloscope.
“Even though the output looks like a short (it looked like a flat line), if I increase the resolution, you can see the part is functioning properly.”
I squinted at the screen. Jeroen was right. The output was following the input like it should, but the voltages were clipped.
“You’re right, but why is the output clipped?”
“I don’t know, but I think the foundry made a mistake. We need to talk with them.”
The foundry was a multi-billion dollar company. They had a great reputation for quality. That’s why we went with them, and paid more, instead of going with a cheaper alternative.
We knew when we told customers they would feel comfortable working with us when we told them our products were made at this foundry. Yet, here we were suspecting they had made a mistake.
We arranged the call for later that evening. Jeroen explained what he was seeing and the foundry agreed to quickly look into the issue.
Two days later, the foundry told us the answer. They had inadvertently flipped two mask layers, and they had created what looked like a dead short in the process.
Your biggest enemy when you are starting a company is time.
I’ve spent virtually my whole career in a world where a product launch cost over $1M. When you “press the button” to tape out a product, you need to know that everything is right.
That’s why we had a rigorous review process before we would release a product to tape out. We knew the risks.
The foundry agreed to respin the wafers for free and expedite them through their foundry. The foundry’s screwup would cost us three months.
Three months might not seem like a long time, but it’s an eternity when you’re burning through your cash. Worse yet, delays are cumulative.
You are not only delaying the initial product you are working on, you are delaying all the follow on products you are working on. So now you are delaying revenue on multiple streams of products.
You are risking running out of cash sooner than you expected when you are delayed.
We actually assumed that half of our products would require respins. We based this on our previous experience, so the delay wasn’t unexpected.
However, we didn’t expect the fab to screw up. So these type of delays hurt because they weren’t in our plan.
We could see that we were in danger of running out of funding earlier than we wanted. So Tina, our excellent controller, worked to secure a loan with Silicon Valley Bank. The loan would actually play a key, and unexpected role, in saving the company a couple of years later.
You need precision and speed to win as a startup.
Every business is different. You have to look at what a mistake will cost you when you push the button.
In the semiconductor world, a mistake costs you time and money. In other businesses, where the precision/time trade off is different, you can release products that are flawed and fix them later.
The reality is, regardless of the business you are running, that the clock is ticking. Every day, you have less funding than the previous day.
You’ll get killed if you don’t have a sense of urgency. And you’ll get killed if you aren’t focused on the details too. You need a proper understanding of both to win.
For more, read:
Do You Want To Grow Your Business? Maybe I Can Help. Click Here.
View original post on Quora.