Hardware startups and pound shop chic…

I spent yesterday lunchtime fettling a ping pong ball – and I strongly recommend that other hardware startups do much the same!

OK, here’s what happened. I was doing some tests on a well-known Aptina image sensor (basically, I’d noticed that its analogue gain circuitry wasn’t quite as linear as its datasheet claimed, and hence wanted to build up a calibration table to correct for that non-linearity): and the way to do that was to create an diffuse, evenly-lit captured image. Hence I decided to take half a ping-pong ball (as per the ones psychologists use in Ganzfeld experiments) but placed over the lens rather than over the subject’s eye.

Having previously cut the ball in half with my Swiss Army knife for a similar hardware hack, all I had to do was to fettle down the half-ball’s rim so as to avoid any light leakage around the edge (because I was testing with a fisheye lens). All of which added to my puerile linguistic amusement: really, it’s not every day you get paid to fettle the rim off your balls. And yes, it worked really well: I even have a nice scatter graph in Excel to prove it.

And then to stress test the same sensor in low light conditions, I took a video camera neutral density lens filter that my co-worker Piers H had kindly brought in a few days before, and duct-taped it onto a clear lid to create a pretty good night-time simulator for the same fisheye lens. Two ultra-cheap hardware hacks in one day!

The more substantial point is that if you are in a hardware startup, and – as is often the case – find yourself needing to quickly hack together some kind of super-temporary test rig for some feature or other, what you typically need is a bit of cheap plastic in a certain shape, size and transparency. And where better to go to find odd-shaped bits of plastic than a Pound Shop? A quick wander round Poundland (or even shops like Wilkinsons if you find yourself really stuck) and you should find something for a pound or two that will do the job. Subtractive high-street hacking rocks!

Ultimately, why buy a 3d printer when you can buy in cheap imported plastic rubbish and fettle it down until it fits? For hardware startups, the strategic answer is clear – don’t move to a tech cluster, move to a neighbourhood full of pound shops. By that measure, it would seem that Redhill ought to be full of hardware startups, though I’ve yet to see any others… perhaps I’m just ahead of the curve, yet again. 🙂

Hardware is apparently the new software (at long last!)

The New York Times just ran an article discussing all manner of neat hardware startups, a fair few of which I already knew well (Nest, Lytro, Raspberry Pi), but quite a few I hadn’t heard of before (Pebble, Bilibot, Electric Imp, LittleBits, Ouya, etc). The conclusion, of course, is that Hardware Is The New Software, and that Venture Capitalists are getting superexcited about this trend.

Well… it’s a great theory, but unless you’ve got an unbelievably sci-fi pitch (like Pebble, whose core idea of a totally programmable watch I remember first proposing to a VC friend a whole decade ago), building your company up to the stage that you can sensible go looking for scale-up funding [which is what almost all VC investment has now become] is a particularly hard trawl, with very few angels wanting to take that road with you.

However… it struck me while reading the article that even though the hardware development cost curve would appear from the outside to be trending towards zero, this is almost entirely as a result of a systemic realignment within the development / design world away from traditional custom dev systems and towards low-cost tools. For example, I shudder to think how many USB dev tools lie scattered around my workspace – oscilloscopes, logic probes, Bus Pirates, RS485 interfaces, wireless adapters, phone interfaces, etc. Hence this is not a zero-trend, this is merely shifting from an older development paradigm to another newer one. Hardware development will remain resolutely non-zero for a loooong time (and let’s not get into the issue of CE & UL testing, ok?)

As for 3d printing, people have been using this for prototyping for well over a decade now, but the big difference these days lies in the scale of the usage and the wider range of materials that are available to print in (i.e. not just compromised low-end ABS). Yet even so, the real world of plastic manufacturing continues to move ever further away – a typical industrial device (such as my security camera) uses a whole symphony of different plastics to achieve both functionality and reliability, and new materials are introduced all the time.

The real hardware revolution will start when we can print injection moulds in our garages… but though I famously pitched that to the Tech City LaunchPad1 competition, nobody seems interested (as my ‘Zoe’ avatar says at the end, “no chance – next time stick to social media, ok?”) It’s only a trillion dollar industry, why should anyone want to invest in anything so pathetically small? 🙂 But once again, the chances would seem high that I’ve arrived at a great party seven years too soon, as per bl&^dy normal… oh well. 😦

Incidentally, one of the NYT article’s authors is the very affable John Markoff, who also writes about historical cryptography, one of my parallel passions (in case you don’t already know).