Playing catchup
So much stuff is happening every day, and I haven't been finding the time to write about it:
Choosing an idea
This is something I was going to write about extensively, but I don't think I'll get the time. The summary:
- I started with a list of about 200 ideas that I'd collected over several years
- I grabbed the ones that looked like they could be feasible right now
- I performed a very quick SWOT analysis on them (lame ideas didn't even get it written down)
- I culled down to half a dozen, then started doing some basic market research (Google searches for competitors, market sizes, etc)
- Based on my research, I chose an idea
I wound up with five good ideas that I believe could be popular. Two were software products, two were hardware products, and one was a service. The problem with the software products is monetisation; I believe they're useful products, but just not easy to sell. The service requires a lot more experience and time than I have right now. I've gone with my chosen idea (1.5A driver for Luxeon K2 LEDs) because it's relatively low-risk, it's interesting to me, and the more research I do, the more I'm convinced that there's a gaping market gap. I'd love to go full-time on this so that I can have a product selling within the next month (which is about when the Luxeon K2 starts becoming generally available), but I'd have to give a month's notice to my day job.
GoYoungBiz course
Last night, I went to an information night for the GoYoungBiz course, a 'grow your business' course run by Sydney Community College. They're targeting people with established businesses, so it's not for me, but they were very enthusiastic about my chosen industry. I didn't even get a chance to tell them my idea! The conversation went something like:
"So, what's your business?"
"I'm planning to do some electronics manufacturing."
"OMG! Will you be exporting?"
"Uh, yeah."
"OMG! You have to come to this course! We want you! It's what Australia needs."
There was a lot less OMG in reality, but you get the idea. The "it's what Australia needs" part is a direct quote. I personally think that Australia can do without the nasty chemicals produced by electronics manufacturing.
Chips vs. designing by hand
Originally, I was going to build my regulator out of op-amps and an Atmel ATMini microcontroller. It's be fun, and I'd learn all about switchmode power supply design. Right?
Well, I found two chips that will do what I want. Except that one of them has too low a current capacity in the worst case (but is OK in the typical case), so I probably won't use it. The alternative is twice as expensive, twice as hard to solder, and difficult to obtain. Can you see how much I love my customers already?
It does worry me a little that out of all of the semiconductor manufacturers, out of all of the chips they make, there is one part that can do what I want. They don't have strong local distribution here, but they do have a decent online ordering system. Here's hoping they don't charge a bundle for shipping.
It's a useful part for all sort of projects (I still want a good bench power supply, and this chip can do it all), so while buying a whole bunch of stock will be painful, it'll get used eventually. It's a moderately expensive chip.
LTSpice
I found LTSpice and, well, wow. Wow. It's like a usability designer attacked some engineering software. You draw a circuit, and it tells you what the circuit will do. It's not a lot more complicated than that. I'm really amazed. These tools are usually made to help sell chips, and damnit, it's working. I can twiddle circuit parameters and see - visually - what that will do to the output of the circuit.
Different products
I'm still playing with this idea of doing multiple products. There are two main classes: a fixed current board, and a 'flex' board. The fixed board has the output current limiting set in advance, and has a simple voltage in/voltage out interface (maybe dimming control if there's board space). It'll be as small as possible.
The flex board will have dimming control, power control, maybe input current limiting, maybe a 5V logic output. It could be Buckpuck compatible, but then I don't know if Buckpucks are getting integrated into anything on a large scale (thus justifying direct competition). If I can make my board smaller and more powerful, that's more compelling. I presume. For all I know, there are a dozen buckpucks in every new TV that's made and having price/performance competition is the best thing I could do. I'll evaluate down the track when I do the board design; if I can't make my board much smaller, I'll make it Buckpuck compatible. If I can improve on it, I'll do that, and maybe offer a 'Buckpuck adapter' that matches the pinout and layout. Already it looks like my design will be less tall; I have no idea what's taking so much space in their casing.
Distributors
The more I think about it, the more it makes sense to have a strong distributor network. I reduce overseas shipping costs; I have higher and more predictable volumes; I move some of the sales and marketing cost to other people. I think Jaycar would jump to redistribute these in a second, because they sell Luxeon LEDs but don't have a driver for any of them. Their customers match very closely my target customers.
Sales has never been my strong suit. This is obviously problematic for someone running a one-person business. I've found that I have less trouble and less anxiety if I have a reason to be selling or marketing - specifically, making a profit. So while I'll have noticeable anxiety when I'm making calls and talking to people, I should cope well enough. Online marketing doesn't require me to talk to people at all, which is great, but it'll be difficult to sign up distributors based on some Google ads.
PCB tinning at home
For a while, I've been wondering what to do about PCB manufacturing. I don't want to pay for it, but I worry that if I don't, my product will be seen as low quality.
Now that I've priced out the parts that I'll be buying, I'm less concerned. I've accepted that there will be upfront costs and the risk inherent in that. I'm also confident that I won't have trouble moving stock. My initial production run was going to be about 100 boards (or less if the parts shipping from the US isn't too expensive), and the PCB is going to wind up being a relatively small portion of that. If I can sell that initial batch, I'll probably be making a small profit overall (ignoring my labour cost). A lot of the upfront expense is stuff that I can use for the next board I design or for my own projects (hope the tax office doesn't see that!)
I'll still be doing my prototypes by hand (laser toner method, ammonium persulfate and spray sealant). I'll be assembling them with solder paste. I'll hand-solder the first one to make sure the board design is correct, because it'll be damn near impossible to remove the main chip by hand if I solder its underside solder pad. Later ones will be using the reflow skillet method, and doing this I estimate I can produce about 100 boards per day at home, on my own.
To deal with the board quality issue, and to make soldering a bit easier, I research electroplating, roller tinning and tinning crystals. All but the last are infeasibly expensive for me to do at home. Even the crystals are pretty expensive; Farnell wants $180ish for them (pfft!). Computronics has them at a far more reasonable $38, but it's still a lot of cash to spend considering that a professional production run of 100 board will probably cost me about $120 shipped. They'll also look a lot less nasty and have ever-so-useful solder mask.
I came up with an idea of my own for cheap home tinning, inspired by the SparkFun stencil paste tutorial. Take your blank PCB and smear a thin layer of solder paste over the entire surface. Attach a wire or something so that you can move it without having to touch it. Somehow, this has to not interfere with heat transfer to the board - I haven't worked this part out yet (I never said I tried this, just that it was an idea I had!) Whack it on the hotplate so that the solder paste melts all over your tracks. Pick it up with the wire, and while the solder is still molten, tap the edge on the hotplate to knock the excess solder off. Insanity? Probably. Without solder mask, close solder pads will probably just bridge. If you have solder mask, you probably already have tinning. But I'll give it a shot anyway. PCB material is cheap.