Neat trick for 1.6mm thick Arduino device

[RupertGo]

Just came across Arduboy - http://www.bateske.com/arduboy/ - which is a DIY project for a 1.6mm total thickness credit-card format Tetris game with OLED display. The guy mills out press-fit holes for SMD components, which is a neat trick. As the site says -

"The primary trick of this design is having milled cutouts made for surface mount components to be press fit into, using the circuit board as a kind of frame. Components selected have a thickness near that of the circuit board (1.6mm). Furthermore, to minimize the board thickness, the Atmega328P is inverted so that the bulk of its height below the surface. The result of equal thickness and recessed installation provides a flush appearance. The primary benefit beyond the aesthetic quality is the device is easily slid from a wallet. The high quality boards and the excellent service from oshpark also makes this build possible."

He also points out that he's having to hand-assemble these, because he doesn't know of any production house that could make these in bulk - pick and place is no good here...

[EEVblog]

[amyk]

Can it really play Gameboy games?

No. I'm afraid the chip doesn't have quite enough horsepower to accomplish this.

A 4MHz Z80 and a 20MHz AVR are roughly in the same ballpark for performance... although emulation overhead is probably the reaosn. Maybe it's time for an ARM?

How do you change the battery?

Two quick touches with a soldering iron easily installs a new one. I should hope that someone dedicated enough to deplete the battery is ready to take on the challenge of a little soldering, even if they did not buy this in kit form. I can also offer this as a service if absolutely necessary.

It's not even rechargeable?

[RupertGo]

I don't think the project itself is anything other than a gimmick. The uCalc is much more fun - but the three-board 3.6mm thickness is proper porky by comparison. Practically a mainframe.

Still think it's a neat trick, embedding the SM components in the board. One of the tweaks I was taught when I was fixing old radios was to knock up a small sliver of double-sided PCB or Veroboard with wires soldered to both sides, used as a shim between two cells in a battery compartment to measure current without having to muck about. This sort of idea means you can slip something with some smarts into that sort of space, which is in very short supply in modern portable devices.

[bateske]

Hello, Kevin Bates here, creator of the Arduboy!

First off, love eevblog. I never get enough time to watch it, but I do when I can. Dave you've helped show me there is an audience for this kind of thing. Also I really enjoyed your smd soldering video, was very helpful on this project!

If anyone is aware of a manufacturing facility that can support this, please do contact me.

RE: AVR Speeds
The chip technically has more mips than the z80, but you the gameboy processor had some hardware graphics and software pipelines that the AVR simply does not have.

I am looking at creating a next version perhaps based on the Teensy 3.1 or Stellaris. Power consumption will then be a bigger issue.

RE: Charging

A rechargeable battery itself is going to be more expensive, but also I would have to include a charging circuit which would drive the price up. Also, some kind of interface to accept the charge (usb plug, or pads to hold a 9v battery) would add complexity to the design. I want to keep things simple for now.

Charging will be a feature in the next version for sure. Hopefully even solar.


Happy to answer any questions anyone has!

[FearTec]

Shutup and take my money

[lgbeno]

For manufacturing the big deal is flipping the parts over, they would need to come on a reel upside down.  Seems do able but would need high volumes to make it feasible

[LukeW]

I don't think it's economically viable to manufacture scalably.
What is this person going to do when he eventually puts it on Kickstarter or something and thousands of people order them?

Much better, I think, to use industry-standard assembly practices that are scalable using standard pick-and-place technology.
I would change the PCB to something thin like 0.8mm, change the TQFP AVR to a QFN package, change the passives to 0603 or 0402, mount all the components on the board in a standard way compatible with established manufacturing practices. And just accept that may add to overall thickness.

Maybe keep a cutout for something like the battery, because the assembler may be hand-soldering them anyway and they may not be pick-and-placeable, so you might be saving thickness and not gaining any manufacturability by keeping the cutout.

If you launch a Kickstarter campaign for an electronic hardware project and you're hoping for big results, backers will be looking for DFM, an understanding of manufacturability, and financially viable high scalability without the need for a redesign or a big blowout to the timetable or budget. (Or at least that's what they *should* be looking for, if Kickstarter backers had a decent sense of skepticism.)

I also don't like the questionable "let's get rid of all the decoupling capacitors and LED current limiting resistors" engineering approach. It doesn't add engineering credibility to the project IMO.
And a LED with no current limiting resistor is probably draining the finite charge capacity of the lithium coin cell very fast relative to what a LED current on the order of 1mA (through a 1k resistor) will.

Also, put a standard AVR ISP header on it (even if you just pogo-pin it and don't solder header in) because you'll want a rapid-programming jig to put the firmware on it for high-scale manufacturing. You can think about these high-volume manufacturing issues even right from a small start - it will save headaches later. (And a serial header with an FTDI-compatible pinout or whatever, even if you don't solder pins in, people will be able to program it and hack it later and develop their own code with the Arduino IDE.)

[charlespax]

I don't think it's economically viable to manufacture scalably.
What is this person going to do when he eventually puts it on Kickstarter or something and thousands of people order them?

Well, I guess we'll find out. Arduboy is fully funded on Kickstarter in just a few hours :-)

[MLXXXp]

I don't think it's economically viable to manufacture scalably.
What is this person going to do when he eventually puts it on Kickstarter or something and thousands of people order them?

Well, I guess we'll find out. Arduboy is fully funded on Kickstarter in just a few hours :-)

No, we won't find out. This a significant redesign of the product that this thread was started about. So what would have happened if the original had been offered on KickStarter will never be known.

[ez24]

Well, I guess we'll find out. Arduboy is fully funded on Kickstarter in just a few hours :-)

Seems to me to be a poster child of a successful KS project.   Good presentation.  I would not be surprised if they hit their one million mark with their marketing skills.  Where did I see pictures with others used for business before (hint SL)

[Muxr]

Can it really play Gameboy games?

No. I'm afraid the chip doesn't have quite enough horsepower to accomplish this.

A 4MHz Z80 and a 20MHz AVR are roughly in the same ballpark for performance... although emulation overhead is probably the reaosn. Maybe it's time for an ARM?

Yeah..

D-MIPS comparison is something like 0.43 MIPS for Z80 and 1 MIPS/Mhz for Atmega. So at 20Mhz AVR is about 40 times faster. But yeah due to emulation overhead it would probably struggle since it doesn't have dedicated hardware for sound synthesis either. They should have went with STM32 for this project. It's about the same price as AVR too.

Also looks like the display is the standard 128x64 OLED which isn't enough resolution for game boy which was 160 × 144.

[ez24]

He has SOLD by my count (today) about 5,000 of these!   The final product is NOT 1.6 mm.

https://www.kickstarter.com/projects/903888394/arduboy-card-sized-gaming/video_share