Any tips from the past 3 years?
I’d say even in three years, single board computers or very tiny Linux-type things have gotten so good, that unless you need really extreme battery performance, it’s really hard for me to recommend trying to go for a microcontroller.
I’m sure people are going to have opinions about this, but from my perspective, the biggest thing I’ve seen happen to projects that really can cause issues is when you engineer yourself into a corner.
So you spec microcontroller X. And it has X amount of memory, and you can only connect to Bluetooth. For what you think the customer wants at that time - it’s perfect.
And then you start trying to sell it and you realize, we need to add feature X. And it’s like “well gonna need to rebuild everything”. Whereas if you can go for a slightly more powerful, but very miniaturized thing, we can actually get away from that.
It becomes more like you only have to iterate software to add features, right? You don’t have to try to re-engineer your entire hardware stack.
The worst thing is if you have to redo your board, then you have to redo tooling on your enclosure parts, and all of sudden: $100,000 and non recurring engineering costs.