Ask HN: How to get simple gadget prototype built? | Hacker News Reader

Ask HN: How to get simple gadget prototype built?

7 pointsJun813 comments

There are tons of sites where you can list a software problem and people build to write code for you, similar ones for logo and graphic design, etc. Is there a similar site where you can specify the main idea of a simple electronic gadget and have a prototype built (cheaply)?

What I need is a stand-alone, self powered Bluetooth module that will continuously transmit a simple code (to be installed in large numbers in a store). I am a software person so don't have the slightest idea how to build a prototype.

View on HN

Doing accurate triangulation (localization) of a bluetooth receiver from a bunch of active bluetooth nodes is maybe a little more difficult than it seems.

Certainly a great idea in theory, but the accuracy of your localization will depend on the signal propagation model you apply to the received signal strength (RSS) that the phone provides. I would bet that it's difficult to find a phone that will give you signal strength in dBm at a high update rate.

I worked on a localization project using passive RFID tags and an RFID reader (a more power efficient setup than having active bluetooth nodes). Especially in a store/mall type environment with many people moving around, I think it becomes hard to get a decent model of expected signal propagation (mainly due to constantly changing multi-path signal amplitude).

All of that said, for your application you might be able to get by with the reduced accuracy (within 10 feet might work?).

With bluetooth, you also have to consider the number of connections a bluetooth receiver can handle. That will probably greatly restrict your density of nodes. RFID again holds an advantage here, in that RFID is a very low-level interaction, while bluetooth is much higher level. This problem just screams RFID. Too bad everybody's phone doesn't have a built-in RFID reader... or a software radio!

I can help you with that. Email me if you like.

Hardware prototyping is a little different than software proto's, because you mostly incur real costs to develop things, which also makes mistakes costly.

Depending on what you want, there is a high probability you can use COTS dev kits and a little bit of a microprocessor (PIC, Atmel, Stamp, etc.) to make a fully functional (albeit somewhat larger and more expensive than production version) prototype.

Thanks brk, I'll email you tonight to discuss.

If you want to have everything built you are going to have to use multiple people: 1 for the enclosure and then another for the electronics and software. Look on elance.com and guru.com use people ONLY with recommendations and experience in what you want to do. Also in the req doc place a code or phrase they need in the subject line so you know they read the req doc.

To buy parts: digikey, mouser, arrow

There are a ton of good prototyping firms in the US that will make you a prototype within a week, once you have a design.

There was a post not to long ago about a guy who made drum sticks that make drum sounds as you beat them in the air. His first prototype was sensors strapped to cardboard tubes. I would suggest doing something similar have your first prototype be anything you can get to work regardless of form factor. After you prove the concept then work on more realistic prototypes.

This sounds like it could be better to use RFID. Depending on the application of course. Could you go into more detail?

The idea is the basic indoor mall/store navigation thing: Each BT transmitter will send out a unique code tied to its location. An app on the phone detects the nearest n of these and determines location by triangulation.

If the sensors are cheap, they can placed dense enough to provide pretty good localization. The mall can pay for them. Say, if the transmitter is ~$5, we can place 200 of these in a mall for $1K, which isn't that much, I think. The transmitters should be low-power to minimize maintenance, e.g. go for a month or more. They should also be small to be unnoticeable.

What do you think?

Hmm. I figured you were depending on people's phones to do the reading. I assume you're doing this in order to help the customer navigate the store?

If you could have the customer pick up a handset RFID reader (with your app on it I suppose?) that they return when they leave, RFID could work. I suggested RFID because the tags don't need any battery or power supply. The readable range might be a little less than with Bluetooth transmitters, but I have seen read distances of 50 or so feet. And tags are cheap and readily available, too.

That is actually a good idea in settings where customers can/want to borrow a device from the store or mall, e.g. the RFID reader can be incorporated into the self checkout barcode readers that some supermarkets handout.

For a typical mall scenario, though, I think people will be reluctant to deal with the hassle of another device.

Right, another device is always a hassle. Anyway, food for thought.

A month of BT is going to require a pretty decent sized battery. (I seem to recall 4-5 mA idle, 40-50 mA while transmitting through one of those $40 BT modules.) Call it 10 mA average--you're looking at >7 Ah over a month. Assuming primary cells, that would be two C or D cells (actually, that's not too bad).

I think $5 is low. I'd guess more like $50 each for small quantity (<500) hand assembled in the US, using an off the shelf enclosure.

Does Apple even let you have access to the Bluetooth serial port profile?

It's a good idea, I was musing something similar recently actually.

So, you want to use a sensor type that a common smartphone can read, BT and Wifi seem to be the best choices.

Power wouldn't be a major issue, they would be plugged in to a wall outlet, or possibly even powered from a PoE line.

My mind is thinking about triangulation and the like. If the BT devices were networked, I believe you could have them transmitting an ID and a medium resolution time stamp. The phone would also have to be using network time and you could get a fairly accurate triangulation (again, I think, this part is far from my speciality).

There is a lot of merit to this idea. You could roll this into something to do with digital signage and advertising as well. As stores in a mall come and go, they have to update directories. You could build a system to let a mall auto-update it's layout maps as tenants change.

To support iPhones and iPods, you'll very likely need to join the MFi program.

Why not use GPS? The main problem to solve would be creating a database of the location of each store in the mall.

Did you find that there isn't enough GPS coverage in malls?

Reply on news.ycombinator.com