[dorkbotpdx-blabber] Femtoduino: an ultrasmall (20.7x15.2 mm) Arduino compatible

[Fabio Varesano]

Hi everybody,

I finally started working on a ultrasmall Arduino compatible board.

You can find all the details here:
http://www.varesano.net/blog/fabio/femtoduino-designing-ultrasmall-arduino-compatible-board-atmega-328p-qfnmlf32

I've been able to reach the unbelievable small size of 20.7x15.2 mm.
This one will really push the group PCB factory limits but it should
still respect the design rules.

Any comment, suggestion or idea warmly welcome.

Thanks for your time,

FV

_______________________________________________
dorkbotpdx-blabber mailing list
dorkbotpd...@dorkbot.org
http://music.columbia.edu/mailman/listinfo/dorkbotpdx-blabber

[Jim Larson]

One quick thought: do you have a way to load the Arduino Bootloader?
That is, is there a programming header included?

-jim

> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 10.0.1191 / Virus Database: 1435/3347 - Release Date: 12/29/10

[Fabio Varesano]

No, the programmer header is not included.
This would made the PCB too bigger.

Anyway, to my understanding of the original Arduino UNO schematics,
the programmer header is simply connected to D11, D12, D13, RST, VCC
and GND so it should be possible to use ad hoc programmer cables to
load the bootloader.

Is this a correct assumption? Do you have alternative ideas?

Thanks,

FV

On 12/30/2010 05:22 AM, Jim Larson wrote:
> One quick thought: do you have a way to load the Arduino Bootloader?
> That is, is there a programming header included?

[David Madden]

On 12/29/2010 05:44 PM, Fabio Varesano wrote:
> I've been able to reach the unbelievable small size of 20.7x15.2 mm.
> This one will really push the group PCB factory limits but it should
> still respect the design rules.

I'm not exactly sure what your intended use case is. If it's just to
have a REALLY SMALL Arduino, it seems like it'd be more useful to have
it fit breadboard spacing, at least. The 50-mil pitch of the pads is
pretty small for free-wire soldering (at least for my eyes!), the width
isn't right for breadboards, and the pads across the top are no good.

My guess is you could shuffle things around some and get everything 600
mil, with half the pins along each side at 100-mil pitch. Not a lot
different from a Dorkbord, really. Or, how 'bout this:

300 (or 600) mil wide, pads along the side at 50 mil, but interleaved so
you could put half the pins pointing down, and half pointing up. Then
you could put the device on a breadboard facing up or down. Easy
(easier, anyway) experimenting, and when you're done developing code,
you can solder it up with all the wires facing one way.
--
mersenne law llc -- www.mersenne.com -- IP law
patents, copyrights & trademarks for engineers

[Fabio Varesano]

Hi David, thanks for reviewing Femtoduino.

The idea behind Femtoduino, beside being a learning exercise and a
personal design/soldering challenge, is to be a very small arduino
compatible board which could be easily embedded into other PCB designs
to provide "arduino intelligence" in a simple way.

50-mil pitch is actually not that complex to solder. Lot of flux,
magnifying glass and a tiny solder bit are the keys to success.

What you suggest is actually what I imagined it at the beginning. But,
after some more evaluation I noticed that, in a 50 mils header, if you
solder odd pins up and even pins down you get exactly the same result
while keeping the size as smaller as possible.

The problem will probably be the pin male header diameter itself.
50mils male headers have a smaller diameter then 100mil one.. so I'm
not sure if that work.

Right now, the top and bottom headers are 500mils making it perfect to
place in the middle of a breadboard (given that one did the alternate
top/bottom soldering above). The only problem is the header on the
right.. unfortunately that was the only feasible place to put that but
with a 50mil header soldered there it should be pretty easy to drive
wires from the header to the breadboard.

Thanks for your insights.

FV

On 12/30/2010 05:18 PM, David Madden wrote:
> On 12/29/2010 05:44 PM, Fabio Varesano wrote:
>> I've been able to reach the unbelievable small size of 20.7x15.2 mm.
>> This one will really push the group PCB factory limits but it should
>> still respect the design rules.
> I'm not exactly sure what your intended use case is. If it's just to
> have a REALLY SMALL Arduino, it seems like it'd be more useful to have
> it fit breadboard spacing, at least. The 50-mil pitch of the pads is
> pretty small for free-wire soldering (at least for my eyes!), the
> width isn't right for breadboards, and the pads across the top are no
> good.
>
> My guess is you could shuffle things around some and get everything
> 600 mil, with half the pins along each side at 100-mil pitch. Not a
> lot different from a Dorkbord, really. Or, how 'bout this:
>
> 300 (or 600) mil wide, pads along the side at 50 mil, but interleaved
> so you could put half the pins pointing down, and half pointing up.
> Then you could put the device on a breadboard facing up or down. Easy
> (easier, anyway) experimenting, and when you're done developing code,
> you can solder it up with all the wires facing one way.

[Paul Stoffregen]

My first impression is there's a lot of unused space on the back side,
but more than zero parts, so why not make better use of the other side?
Perhaps this thing could be even smaller?

My other impression is pretty much the same as Dave's, that it really
won't work for many people without some way to use it on a standard
breadboard. Your comment below indicates soldering 100 mil spacing
headers on both sides. But even if the holes were large enough (38 mils
is needed for the standard square types, but smaller can work for the
spendy round machined pin types), how could you solder both sides?
Because every other hole isn't staggered somehow, after you've soldered
a 100 mil pitch header on the bottom, it'll cover up the pads where
you'd need to solder one on the top.

Perhaps another approach might be placing 5 pins at 100 mil spacing on
the left and right, perhaps lengthening the board somewhat? If you
could get the RX, TX, SPI signals, power and ground to the edges, that'd
be quite a bit more useful. The I2C pins (which are also analog inputs
if not used for I2C) would be the other widely used pins. Then again,
lots of projects use PWM.... so it's a difficult decision which subset
of pins would be useful on 100 mil headers. But even a small set would
be vastly more useful than none. 50 mils is really only useful for
connectors.

It's too bad we don't have a multi-layer process with blind and buried
and filled vias. Then you can put a via in the center of every SMT pad,
cramming all the parts as close as possible on both sides, and route
everything on the inner layers! But even with 2 layers, it looks like
you've got quite a bit of unused space on the back side.....