Building a cheaper handheld (moved from the pyra cost thread)

[Chip]

Making your hardware is expensive but the results are beyond worth it IMO. The quality surpases most chinese crap, and the tailor fit os is a huge step up over andriod, which I do not like.

The quality of the system board is binary - either it functions or it doesn't.  I suppose the argument could be made that a shoddy board might work initially and then fail.  But if they're cheap enough...

The case and the controls on the other hand, those are definitely subject to manufacturing standards.  The case you would have to make yourself regardless of whatever other components you could glom.  

The analog sticks are going to be as off-the-shelf as the ones in the pyra insofar as they're a singular subcomonent that you're going to buy from somebody.  You just need to find somebody selling decent ones - the sticks in the GPD XD are reportedly quite good.

I honestly think the dpad and buttons are better off coming from an existing controller.  I'd call up 8bitdo and try to buy the dpad, buttons, and mats that they use in their (incredibly good) gamepads.  It's going to be a hell of a lot cheaper than tooling up to make your own, and you know for a fact that they'll feel good and respond properly.

The keypad can definitely come from an existing device.  As ED has pointed out, they're quite expensive to have custom-made.  Far cheaper and "good enough" to try to buy a mat from something like this and make it work with your case/PCB.  It also sidesteps the thousand-post threads about keymat layouts.

And as far as the OS, at no point was I suggesting sticking with android.  I love android on my phone, but it's not suitable for a proper computer.  If nothing else, the inherent input lag for the gaming controls makes it a non-starter.  No, I was assuming a custom linux-based OS would be used.  That's probably the trickiest part of the whole shebang, as the SoCs in most of these phones are less than amenable to aftermarket OSs.  In a perfect world, somebody would figure out how to get linux onto one of the atom-based phones from asus or lenovo and you'd use that as your donor.  Then you could run any number of PC games in WINE.  Failing that... I don't know how possible it is to get linux on a mediatek or rockchip without assistance from the manufacturer.

Just take the example of switching ED's custom CPU board with a SODIMM slot SoM. You could still keep all your peripherals, just wired up onto a different connector. You'd only have to design the main board, rather than the main board and the CPU board - this would save you some cost. ED said he could only get OMAP SoCs, but with SODIMMs, he could have picked a Pi Compute, or several other ARM SoMs - including a Tegra 3. So he would have had more choice.

Sadly, this is not the case.  While there are several SoMs using the SODIMM connector, there is no standard pinout.  None that I am aware of are intercompatible.

While the SODIMM connector is great for certain systems in that it is cheap, readily available, and comparatively robust, it is not ideal for a handheld device.  It takes up more space vertically than is necessary, and will add thickness to the device that could be avoided with a more svelt interconnect.  ED chose wisely for the pyra.

 

[FaeMinx]

I challenge anyone to put together a list of readily available parts, as well as costs for case manufacture, and the not insignificant costs of getting all the pieces to play nicely together, os development... in small quantities - I.e. 5000

Then compare that to Pyra, and I'm almost positive the reduced cost will be almost trivial, whereas the reduced features will be huge.

*community and support are features.

 

[pmprog]

Sadly, this is not the case.  While there are several SoMs using the SODIMM connector, there is no standard pinout.  None that I am aware of are intercompatible.

They wouldn't have to be. You don't have to support all SODIMMs, just one (or maybe a family). The idea for that is cost - not upgradability
 

It takes up more space vertically than is necessary, and will add thickness to the device that could be avoided with a more svelt interconnect.  ED chose wisely for the pyra.

I was thinking about this, and there are ways to reduce the vertical height. For example, rather than having the connector mounted on top of the main board, you cut a SODIMM shape out of the board, and have the port mounted across the edge of the PCB. PCBs are usually a couple of mm thick, which means you'd probably only see a couple of extra mm each side of the PCB. Imagine the PCB on the same plane sat in the centre recess at the bottom of this.

But yes, ultimately it'll probably add a bit more - but as I said, if you aiming for something that's not exactly the same, this could probably still get you quite close

 

[rygD]

Today I was in the german "Saturn" Electronics Market, damn, I must admit these sleak full-metal-unibody "Ultrabooks" are very interesting, I may get one of these (no Apple of course   ) I  imagine a Pyra with such a sleak case (just in small), still 10h battery time and still cheaper.  I'm sure it  can be done, I'm not sure how.  h34r:

Or the other way round, how about an rugged outdoor-like handheld for the Adventurers of  us?  Basicly a Pyra in an tough case or so. I always liked these military designs.

I like both of these ideas, but I would not give up the ability to easily get inside my Pyra just so I can have a sleek metal body.  All for a ruggedized one as long as it isn't too huge.


I think SoMs might be an option for a future Pyra-like device.  I don't know if it will cut costs, but it could.  More important is that it should give more options when buying in the quantities needed for such a project.  I don't know why anyone in that market would say "no we won't sell you a few thousand and provide you support" or whatever the issues were in the quest for a Pyra SoC (I read it all, just forgot most of it).
 

 

[Chip]

I challenge anyone to put together a list of readily available parts, as well as costs for case manufacture, and the not insignificant costs of getting all the pieces to play nicely together, os development... in small quantities - I.e. 5000

Pretty much impossible as a mere thought experiment.  Even if I had half a million bucks and wanted to do this for real, it would be difficult to get started.  Based on my comprehensive understanding of Chinese manufacturing business practices (kidding!), you really have to have someone on the inside or you're going to get screwed.  Despite my oversimplified wording in my initial post, it's not as simple as just calling up a factory and getting a price on a few thousand boards.  You need someone in China that speaks Chinese (preferably someone who is Chinese) to do some wheeling and dealing.  If Johnny Westerner just called up from across the sea and tried to do this, it would end up like the buying-a-caravan-from-the-pikeys scene in Snatch.  More accurately, it would end up like the very first attempt at the pandora where the Chinese factory just made off with the money.

Just to get a semiaccurate number, you'd need to hire a sourcing agent.  But if you want some pulled-from-my-butt numbers based on uneducated guesses:

Mainboard & display: ~$100 (probably less, based on $120 phone)
Dpad & buttons (plastic caps + mats): $4
Keymat: $2
Analog sticks: $5 each?  I honestly don't know, but if they can't cost much more than that if the XD is only $160 total.
Battery: ~$12

That's probably it for the things you can buy pre-made.  The case, lower PCB (including connectors and components), and interconnect between upper and lower sections will need to be custom made.  That's less than $130 in parts, and no money spent on developing/prototyping/tooling.  ED would know far better than I what the rest would cost.  Bear in mind that the lower PCB would be enormously simpler than what's in the pandora/pyra - 4 layers at most, and possibly only 2.

 

[fusion_power]

Pretty much impossible as a mere thought experiment.  Even if I had half a million bucks and wanted to do this for real, it would be difficult to get started.  Based on my comprehensive understanding of Chinese manufacturing business practices (kidding!), you really have to have someone on the inside or you're going to get screwed.  Despite my oversimplified wording in my initial post, it's not as simple as just calling up a factory and getting a price on a few thousand boards.  You need someone in China that speaks Chinese (preferably someone who is Chinese) to do some wheeling and dealing.  If Johnny Westerner just called up from across the sea and tried to do this, it would end up like the buying-a-caravan-from-the-pikeys scene in Snatch.  More accurately, it would end up like the very first attempt at the pandora where the Chinese factory just made off with the money.

Exactly this was the main problem with the Pandora, I'm pretty  sure. It's not wrong to make such an device in China but you have to have "your" people there that know what's going on and can check and communicate with the chinese people. many companies  do that with success. So it's much better  to work with these companies and let them manage their connections to china.
Open Pandora tried to communicate directly with a chinese Factory. din't turned out that well, A miracle there was an usable result at all. But it took ages  and ED could tell scary storys about their "communication" with China.

Maybe ED finds an western Company that produces in China under control. this for sure could make some  things cheaper than doing the entire production in Europe.

 

[comradekingu]

For the reasons western companies are doing it, i think making devices in china is wrong both in practice and theory.

China does not produce the same quality for the export market. That is not their business, nor their skill.
You can get around that by building your own factory and streamlining everything, but then its wrong for just another reason.

Either you let up on quality control and attention to detail, where local production and hands-on is key, or you are just dumping labour.

 

[fusion_power]

Even quality products are more and more made in China, see Apple & Co. Of course they can produce quality for the export market that fulfill western standards. But yes, you need to make sure that they reall DO this. ^^

 

[comradekingu]

What on earth does quality products have to do with Apple?

 

[Chip]

It's not just apple - a lot of "quality" products are made in China.  A lot of crap, too.  China is a really, really big country.  There are tens (hundreds?) of thousands of factories in China, making everything from Dell and Lenovo laptops to the mini polystation 3.  Saying "manufacturing quality in China sucks!" is like saying "food in the UK sucks!" - sure, most of the food is crap, but there are some good things to eat, you just have to know what and where to find it.

 

[fusion_power]

What on earth does quality products have to do with Apple?

No clue, I don't use any Apple products. Well it was just an example that basicly all big electronic companies build their products in asia, even the most expensive ones. So it must be really expensive not to build this stuff there. Which brings us back to the toppic here.

 

[comradekingu]

Biggest based on volume, or quality? Since planned obsolescence is the competitive landscape with mass-production, I would look at anything akin to a monopoly or exceeding capability with more scrutiny.

The most expensive ones build their products in Europe or Japan (or in some cases similarly positioned US) where expense relates to cost of production, which is a thing because of better quality, not as in just more markup.

It is fair to say a premium is charged for this extra quality and pleasantness of knowing where and how products are made.

However if you want to reap the benefits that big companies use, consider that they sub-contract out everything to not be directly responsible for everything. They also have the money to throw in, should something go wrong with what is essentially their production.

So where does that leave a portable laptop/gaming device. That is unique, but not to the components available.

You can end up saving a lot, in a similar manner, by using off-the shelf, or semi-off the shelf solutions, and get away with far less risk, accepting just as little, or less responsibility.

For example, Ipad3 screens (LG), omron switches, 3d printed case, off the shelf battery, ready made single board computer with cables reaching out from the sides not close to the corners (or negative space pluggable) is very low-volume/hacker- friendly.

If you want to get more advanced, there are plug in cpu-modules, where you only make the mainboard yourself.

The sweetspot for this is bigger, in terms of volume, than the pandora or pyra.

 

[Chip]

The sweetspot for this is bigger, in terms of volume, than the pandora or pyra.

Absolutely.  If you can sell 10-20k units, you'll not only end up with lower development and manufacturing startup costs per-unit, but you'll have a lot more pull with factories and suppliers who are more keen to get big orders than small ones.  The cheaper you can make the finished product the more potential customers you'd have.  If you can get the device down around $200, you might actually be able to sell 10k+ of them.

Or maybe not.  Probably worth doing some market research before you drop seven figures on the project if you're interested in actually doing this.  As much as I think this is all a very good and viable idea, it might be complete rubbish.

 

[comradekingu]

What I meant was the size of the unit increases.