How Fast Can I Iterate?

A project log for An IOT Device That Tells Dad the Stove is Off.

An Internet of Things Device to Let My Aging, Disabled Father Know that the Stove is Off, and Ease His Paranoia...

Casual CyborgCasual Cyborg 05/09/2015 at 04:200 Comments

So, I'm finally coming off of the high of winning the Twilio sponsorship prize of the all expenses paid round trip flight to Twilio's Signal Conference out in San Francisco, and I'm now getting slammed with the sobering reality.

Yes, it's great that the stove top monitor got awarded. It's great that my $#!ts and giggles idea actually has some public interest. The publicity is great too. But you know what would be even better? Having a much more viable product out.

So, today, I recuperated some sleep, went online and started shopping for sensors and evaluating microcontroller boards. I could just go with the Teensy LC, which I'll need to get another one to make a back up prototype, and to do some science experiments with the stove with. I'm also trying to find out if the Teensy board is open source, so that we can use the electrical schema and integrate that into an overall PCB board with the sensors and Wifi. I'm also on the hunt to figure out how to program the Freescale arm chip in just straight C on the teensy.

I'm also looking at the MSP430, and the MSP432, but I'm not that familiar with programming it, unless I stick to just Energia. I did do a crap load of C a long time ago though, so I've recently been giving myself a refresher in just straight C on the MSP430. I also can't believe there isn't really a decent tutorial on the MSP430, Linux, and MSPGCC! That's going to be a side project in the future for educational resources.

The MSP430's nice because it's low power. So is the 432, but the the problem is price point. I have no idea how much chips are for the 432 should we decide to make, say, a thousand of these.

I also want to get rid of the CC3000 wifi break out board, and see if I can instead integrate the ESP8266 module, because it's smaller and cheaper. The CC3000, while it's nice, is a lot of hardware and expense. And I want to build something that's manufacturable and cheap component wise. Plus I can include the ESP8266 information as a tutorial in how to use it.

But the most important thing is I'm picking up a thermopile.

While Marcin,, and I were hammering out the functionality of the sensor at the hackathon, this dude, John, came over to our table, and as soon as he saw what I was doing, he immediately said he had just what I was looking for. He brought over this Texas Instruments bluetooth enabled sensor cluster package that also had a really sensitive temperature sensor, far better than the clunky DHT11 we were using that was really slow, and it wasn't that great at a distance. For a proof of concept, the DHT11 worked really well.

Hey, I can't complain, I'm going to San Francisco because of it!

But John's sensor, which unfortunately was already integrated with an ipad would've set us back several hours, and it was already 3AM Sunday, so we went ahead with our bread board and phone work. We finished the code for it at around 4AM, at which point, I took a nap. At 5AM we did several more tests on the proof of concept, and added a few code modifications. At 7AM, we declared victory, and chilled out/grabbed quick naps. 9:30AM, the hackathon was over.

All during that time, I kept the thermopile in mind.

Now, winning that prize at the NYC Disrupt hackathon was great. Part of the prize also included two free tickets to the TechCrunch Disrupt conference, and each ticket sells for 2995 bucks! That's almost 6k worth of tickets! Which of course, I took full advantage of. But it was while I walked around the conference, bumped into startup peeps and investors that I realized something.

I was woefully, terribly unprepared for that incredible opportunity. Here I was, walking around at least 20 billion worth of venture capital, and I couldn't do jack freaking squat, because I didn't have a business plan, a pitch deck, and I didn't have a product. What I had was a nice proof of concept.

Well, for this upcoming Twilio Signal conference, there's no way in hell I'll be that unprepared! So I bought the sensors today, and I've already started revising the code. By the time I get out to San Fran, I'll have something that's much more impressionable, and I'll have the pitch ready.

Oh, and we got profiled by Temboo, whose API we used because, we needed SSL for text SMS!

Texting to Victory at TechCrunch Disrupt

"David was aiming to build something to give peace of mind to his elderly father, who often worries that the stove has been left on when he leaves the house."

Um, it's actually about giving back some sanity to me, my mom, and my brothers. But I'm not one to quibble.

Gotta give kudos to Temboo though. Part of the problem with a straight IOT device text sms is the requirement with Twilio to do it with encryption using SSL. But there isn't really a good, fast SSL library that will run with a tiny memory footprint on most microcontrollers, and when you go up that high in microcontroller power, you've got a very overpowered device doing a very simple task. Budget wise, that's not smart. Enter Temboo, a NYC IOT API based startup, whose rep's Claire and John were walking around the floor. And I think we're the only ones who used their API...

Anyway, they came to our rescue, handling all of our encryption. These guys are generous with their API and access.

Up this week, getting the parts in, and then heading to Fog City!