Ask HN: Why No Smart Phone Esque IoT Device?

Buying a phone and replacing the software will be cheaper unless you're making truckloads of these? (There are a dozen brands of point-of-sale systems that are an iPod Touch or similar with a magnet attached to the headphone jack)

Not quite sure what you mean by "Smartphone-esque". There is a bunch of devkits with multiple sensors integrated, systems with standard connectors and a range of matching sensors and outputs, ..., which make prototypes and one-offs quicker to make. Of course more expensive than starting with individual parts. And sometimes using a phone, possibly with a small custom board, can indeed work.

Phybox... it's basically just a programming workspace to use your phone as a sensor.

Not sure what you mean exactly but coming from the IoT end there are plenty of options to connect existing platforms to mobile networks (ArduinoGSMShield for example). Coming from the smartphone end there are several hackable projects too that enable you to build a smartphone like IoT device. Look at pine64.org for example they basically have all the parts to build a smartphone like device with sensors.

There are dev devices like the M5StickC & M5Stack Fire that comes with a bunch of sensor bundled.

Are you asking about a small tablet or ipod-like device, or a phone with cellular?

I haven't closely followed tablet/ipod type devices, but pinephone and purism librem 5 are (not-quite-out-yet) linux-based phones.

They're called development boards

Unfortunately, even if the BOM costs them maybe £100 at volume for a chunky FPGA and some sensors and signal parts - much more for something specialized like an rfsoc board (like easily enough to buy a house), they're intended to be bought with people's bosses money so they're not really within budget for a hobbyist. That's also partly why the documentation can be quite poor - the bargain chips (think Chinese companies with funny names on LCSC and taobao) kind of assume you'll either reverse engineer it as a team, someone speaks chinese, or you'll ring them up and ask for support. The more mainstream parts in higher margin products have much better documentation, i.e. the big American silicon companies will usually document how their chips are programmed in depth rather than some obscure-ish Chinese parts I've seen where you either work it out yourself or use their slightly wonky freertos fork.

What kind of sensors do you have in mind? Phones do have sensors but the software isn't really optimized for high speed data acquisition (I think using camera interfaces is a hack used in industry for getting data into high level embedded systems)

I'm guessing the economy of scale isn't quite there, as there's much less demand for DIY IOT devices than phones. But there's some products out there that approach this idea. E.g., https://www.seeedstudio.com/Sipeed-Maix-Cube-p-4553.html (incredible value), or https://www.seeedstudio.com/Wio-Terminal-p-4509.html.

What's your actual requirements?

Platforms are a fraught industry; Apple has done amazingly at making a closed platform that nonetheless manages to take a cut of third party apps. Generally either a platform is totally closed to extract all the value or totally commoditized so nobody makes much profit from it.

I would say that ESP32 is the "smartphone" for Wifi platforms; the SimCom and Sierra Wireless devices are the equivalent for GSM.

> I still haven't managed to get a consistent working GSM connection.

Pay very close attention to the power supply specs and antenna routing.

Cellular modules, they come in a lot of flavors.

ideally something like this http://janos.io/ but failed!

Hope https://postmarketos.org/ will be more successful.

I think you're missing that most IoT products are designed to be put into products sold at scale. There are exceptions like devkits and hobbyist products but generally, the expectation is that everyone interested in IoT hardware is developing a hardware product and needs little more than a convenient interface to prototype and evaluate with before buying millions of chips to send to a PCB fab.

In cases like this, you want to pay only for what you need, monetarily but also thermally and in PCB real estate. You can't afford to have a bunch of random unnecessary sensors on board.

That's why you don't see some smartphone-esque everything-included devkit.

If you want a nice composable ecosystem to built things with, I'd look at Arduino and Raspberry Pi. They have a variety of "hats", which are extensions that you can use to interface with things, like a GSM modem or a sensor suite.

I'd also look at the TinkerForge ecosystem, which has a dizzying array of high-quality, well-integrated, pluggable blocks that you can use to construct various things. It is expensive however.

You could check out the pine phone with ubuntu port or postmarket os. It has all the standard sensors, gsm, and runs mainline linux kernel. It's basically a phone for developers.

https://www.pine64.org/pinephone/