Re: Thought experiment:: creating great hardware assuming software will follow Message #7 Posted by DaveJ on 4 Oct 2009, 8:47 p.m., in response to message #6 by Bill Wiese
Quote:
Sorry, Dave, a 'real' watch does not have an alum. case. Good watches (which excludes most all Seiko, Citizen, Skagen, Pulsar, Casio, Swatch, etc) have at least 316 stainlees for case, and they also have (near-)scrtachproof sapphire crystals.
If you're gonna produce something attempting to be sellable, don't fight for the low-end. No margin for time & effort. Sell 'em for $500 - $1K, not $125 or less. Also, higher-end watch cases and bands are avaiable for custom watch builders from sources like [url]http://www.ofrei.com[/url]
Yes, I really meant aluminium as a minimum, not plastic. Stainless is a more expensive option, just gotta specify it. Easy.
Sapphire crystal is another expensive custom option. Anyone know where you can get such things custom machined?
I'm realistic enough to know that I perhaps don't quite have the skills on my own to design a watch worthy of a $1000 price tag.
At that sort of price people expect fancy presentation cases, a really elegant case design, 5 year warranties etc. This tends to push the risk aspect (financial and time) into the stratosphere.
The uWatch was butt ugly, but it made a lot of people very happy. I can't begin to imagine what impact a slightly more refined version would have, one that looks like you could actually wear it every day. And I can do that with very little financial outlay or risk.
I know enough about various industries to know that price-point is an important factor. It can make or break your product.
Quote:
Please, don't overdo graphics. While graphix LCD might be required to do better-than-character-LCDs don't do graphics per se on the calc. Get RPN + memory/transcendentals and maybe some 12C PV/FV/PMT/i/n running and you're done.
Think something like Tritium capsule or other watch illumation like Indiglo.
For small volume production rape Timex or Casio calc watches.
I haven't overdone graphics. I've pretty much settled on a 128x32 display, much to the disdain of the community who wanted a 128x64
Backlight is integral white LED and is very nice. Not the lowest power solution, but backlights are not used often, and it's integral, so no development effort or risk.
Raping parts from other products is fraught with danger. If I sold a couple of hundred uWatch's, then I'd be crazy to think in design terms for anything less than 1000 Mk2's.
Quote:
Please O Please No! A low power 80c51 etc is fine - or even better a watch CPU. Find the BEST lowest-power CPU you can. Don't build a "computer". Battery life needs to be 1-2 years just like Casio calc watches.
Given touch sense (below) a Cypress or Atmel is likely best sol'n.
I've speced in an Atmel touch device.
A low power micro is fine, but remember the other stuff takes power too, so it ain't just about the micro.
The uWatch used a 64KB/8K 16 bit processor, and we ran into the limitations very quickly. No one is going to want to have to code in assembler...
You can have your cake and eat it too here. A big processor can be clocked at 32KHz or shut down and draw adequately little power in time/date mode, but then has the grunt when and if you decide to use it. One can also argue a 32 bit processor is actually more efficient power-wise than an optimised low power 8 bit micro. There ain't much in it.
Quote:
I agree. I also had the CFX20 wathc and CFX400. Capsense touch is the way to go and can work thru a watch crystal. A side button and timeout reversion should shift in/out of calc mode so capsense disabled and not burning power etc. Do not try use big LCD legends
for keypad, use printed logo. Capsense touch processing can solve 'fat finger' issue and can even do stroke recognition.
Actually, no. Cap sense creates the "big finger" problem. You can get higher density using tact keys because the finger can "feel" before pushing. Cap sense has to be spaced big enough to absolutely ensure it can be used by anyone. Although admittedly I have not experimented in this area yet, mainly going by what the informed are telling me. You think say reliable 6mm pitch keys with touch sense is possible? I'd love to see some foolproof implementation examples, please.
Quote:
Nobody wants a friggin' watch that has to recharge. 1+ yr battery life, or stay home.
If a stylish, $1000 stainless+sapphire QUALITY {too many nerds here who think Seiko=quality) watch that has an RPN calc with 1+ yr battery life hits the market, enough folks will buy, and money is made to recoup costs.
Frankly, Plan B is to work with Casio on a custom run to reprogram one of their touch calc watch variants and have it dropped in a quality case + sapphire face w/quality band.
Unfortunately I'm being forced down the rechargable path.
The display takes around 100uA continuous, so even if I had a decent 500mAh capacity coin cell that's only around 200 days. And I have not factored in a DC-DC, touch sense, processor, occasional backlight use etc. Then the case design is more complex because you have to allow easy battery access. Yes, it can be done, but it means a custom LCD display and the entire project explodes into complexity from there. Complex development means more money, more time, more risk, and will ultimately never get done because I'll lose the enthusiasm. BTW, I'm already looking at around 1 month battery life for the rechargable.
Find me a suitable lower power LCD solution off-the-shelf and I'll happily consider giving you a year battery life. Really.
People who have not actually done in-depth investigation into designs like this can't really appreciate the countless trade-offs involved. I don't think I could even begin to list all the trade-offs I've encountered in this project. Not as many for pocket calc to be sure, but still, many gut-wrenching decisions need to be made. And I've said this before, this never works well by-committee or by-community. It need one smart person to do the legwork and make the big decisions for such a project.
That's why Hugh needs to simply make decisions, do the work and SHOW us something tangible. Then people will get excited and want to help out. It's a real shame nothing but software has come out of OpenRPN.
I believe OpenRPN never got off the ground because perfection was aimed for. I like being a tad more realistic with my projects. Perfection and gilding the lily is nice, but not needed for a successful product.
Dave.
Edited: 4 Oct 2009, 10:32 p.m.
|