Ask HN: hardware startups, where to begin?

Pick an embedded platform (SDK) suited to your problem (in terms of price versus computational power needed), and learn how to think about code and write C in terms of the limitations of an embedded environment. (HAHA filesystems. what's that?)

Prototype, make it efficient, make it cheap (fewer components), make the spec, find someone who can do a test-market batch, and if it catches...SCALE.

On this note, any blogs/learning sites/videos that teach working with hardware? Computer and Electrical engineering basics...

I am investigating the possibility of a tiny tablet the size of iPod Touch, with 3G and Android. Either touch-screen or a folding keyboard; all that matters is it has to be pocket sized.

Any ideas where to start investigating.

i'm based in shanghai, china, there are lots of factories in china that can produce a prototype for sophisticated hardware on the cheap, i mean, they have all the machinery and processes in place to assemble all sorts of today's most sophisticated products (ipad, iphone, android phones, etc). all it takes is some negotiation to get the price down, but you can accomplish this by selling them on the potential of creating a large order or if you have some connections. only danger is they could take your designs and run, but we all know you need a marketing machine to crank these things out, so at the end of the day, it's more worth it for them to just manufacture.

You will get much better advice if you can be more specific about your ideas. Some examples would be helpful.

I've built a few gadgets that collect data, analyze it, and then send it to a data store on the web. Think carefully about your needs for each of those 3 steps, as well as your readiness to get into low-level technical details. And then tell us more about what you're trying to do.

In the mean time, here is some general advice:

In my experience, the connection to the web is the most challenging part of the project, and the one that has the most impact on the total cost of parts. Most of the microcontroller kits that appeal to casual developers (such as me) do not include ethernet or wifi connectivity. They assume you will be connecting via USB to a PC. Adding wifi or ethernet to an arduino significantly increases the cost and complexity. For stationary (non-mobile, non-battery-powered) gadgets that need to communicate with the web, I can recommend using a wifi router as a development platform. Many of these can run the OpenWRT version of linux, and collect information from a USB port. Here is a good description of working with linux and USB peripherals on a wifi router: http://mightyohm.com/blog/2008/10/building-a-wifi-radio-part... As to cost, the router used in this blog post costs about $30 (http://www.newegg.com/Product/Product.aspx?Item=N82E16833320...).

Another thing to consider is the analysis step. Some devices will have no need for fancy data analysis, but for gadgets that need to work with data in complex ways (learning/compressing/transforming) you may run out of processing power and/or memory on some of the microcontroller platforms. Again, a wifi router is quite capable here - as long as you stick to fixed-point operations.

Once you have an idea of a suitable hardware platform, then it's off to the races. There are many great places to see examples of hardware projects, as mentioned by fragmede. And you may want to consider working with a hackerspace (NYCresistor, noisebridge, etc) - I've found that having an oscilloscope is very helpful for troubleshooting my gadget project.