Skinny from SparkFun

Hi,

No spam intended. I am having a fairly in-depth conversation with David and team-arduino and we'd like to get your feedback.

I designed Skinny without really checking with people - that was my first mistake. Here's the board:

and link:

David Mellis has problems with a 3.3V board that is pin compatible with the 5V Diecimila (and shields). He's got a point.

Here are my goals for Skinny:

  1. Move USB off Board to save cost for users with multiple installations
  2. Make it low-cost
  3. Make it 3.3V to
    a) work on battery / portability
    b) directly interface to the increasing range of 3.3V devices

That was the goal of Skinny. I see the Arduino Diecimila as the 'gateway drug'. Every Arduino derivative after that is what a user would use as their 2nd board. Skinny is not meant for first-time users. It is meant to fill the void in portability/cost/3.3V.

SparkFun wants to move towards full Arduino support so any feedback is welcome.

Happy Friday!
-Nathan

i like the red pcb & the two LEDs (yay!) - sometimes people dont include them but i think they're essential. the high voltage 3.3v regulator is a good deal too. also digging the .1uF on Aref.

the 5v/3.3v shield thing is not a -huge- deal: avr's are (im like 99% sure) 5v compliant so, at worst, it should still work. (if using a r divider, then 1.8V is still higher than most CMOS Vih)

ironically, if you actually solder in female headers to put a shield on it, it will be exactly the same height as a diecimila :slight_smile:

theres no 1k series resistor on RX0, which means if you've got a GPS (or xbee, or xport, etc) connected to rx/tx - many tutorials/examples do this - and the serial link installed (for debugging/uploading) its going to perform very strangely. if its not included, it should be noted

the switch would be 108914981273 times awesomer if it was RA and on the board edge so you could turn it on/off when a shield is on. no big deal tho. (i am rooting for a right angle 3pole switch in the next arduino rev.)

is the 3.3v on the serial connection an input (to VCCIO) or an output (from the internal miniregulator in the 'RL). if its the latter, thats kinda sketchy. :-/

the fuse seems a bit odd - the diecimila fuse is there to protect the computer's internal usb hub. a reasonable 3.3v regulator (you dont have the part number anywhere so i cant check it) should have short circuit protection. and avr's are current limited on the pin. what would be -really- useful is a power diode, such as a Schottky or 1n400x so when they plug in a 9V backwards it doesnt kill the regulator (good ones have pol protection but most cheap ones dont) and then heaven forbid, if it fails short pass the -9V down to the avr and everything connected to it :frowning: good news is that you can get 1A diodes in supersmall packages so ya can reuse the pads!

and one last thing:
i dont really get the 8mhz/'lilypad' thing. i use internal oscillators all the time for cheapie kits, but i'd never do it for a project where customers are expecting to use a serial library where they cant tweak the TCCR values (and where the bootloader is pre-burned with a set value that is not calibrated). i've found that the int oscillator does vary a bit over large runs/temp/voltage. why not include (at least) a ceramic oscillator? you can get a skinny one for $0.20, a full-on crystal setup is maybe $0.35 or $0.40

ok i think im done now :slight_smile:

Oooh. I like it.

Ladyada comments dittoed.

I like - I had plans and a Digikey order forming to convert one of my serial board to 3.3V.

I would like to see at least holes/pads for a crystal and caps - as ladyada points out, timing can be important.

As to not getting the 3.3V lillypad thing, the 3.3V (actually you can go down to 2.7V or 1.8V with the V part, if you can sacrifice clock speed) it makes it easier to run directly off battery (look ma, no regulator!), and it makes it easy to interface to low voltage parts.

-j

I designed Skinny without really checking with people - that was my first mistake.

I think it's great that you built an Arduino variant outside the 'authorization process". The Arduino project is more than stable enough to accommodate different variants that aren't 'official'. Feel free to put the Freeduino mark on it!

D

I designed Skinny without really checking with people - that was my first mistake.

I think it's great that you built an Arduino variant outside the 'authorization process".

I took the comment as meaning "people" being "the community" for feedback, rather than "people" being "Arduino Inc." for authorisation.

--Phil.

the 5v/3.3v shield thing is not a -huge- deal: avr's are (im like 99% sure) 5v compliant so, at worst, it should still work. (if using a r divider, then 1.8V is still higher than most CMOS Vih)

Awesome feedback. Thanks! I am more worried about customers buying Skinny and getting frustrated when it doesn't work with their 5V XBee shield, etc. The ATmega168 should be able to clamp higher voltages without too many problems.

ironically, if you actually solder in female headers to put a shield on it, it will be exactly the same height as a diecimila :slight_smile:

The power jack and USB connectors are pretty tall aren't they? And if we're going for pocket-ability, I'm thinking peeps will want to solder in their own wires. The female connectors were a fantastic idea in the beginning. You ever try to desolder a through hold connector like that? :wink:

theres no 1k series resistor on RX0, which means if you've got a GPS (or xbee, or xport, etc) connected to rx/tx - many tutorials/examples do this - and the serial link installed (for debugging/uploading) its going to perform very strangely. if its not included, it should be noted

Good point. I will add.

the switch would be 108914981273 times awesomer if it was RA and on the board edge so you could turn it on/off when a shield is on. no big deal tho. (i am rooting for a right angle 3pole switch in the next arduino rev.)

Nuther good one. I will go scrounging.

is the 3.3v on the serial connection an input (to VCCIO) or an output (from the internal miniregulator in the 'RL). if its the latter, thats kinda sketchy. :-/

My designs are always a little sketchy :slight_smile: If it sparks, unplug it. If it heats up, you better check VCC/GND. No really - The 3.3V pin on the male header on skinny is an input. The USB board provides 3.3V. You can have the board 'on' and hook up USB without problems. It's a little sketchy to have two vregs on the 3.3V bus at the same time, but both the FTDI vreg and the on-board SOT-23 mini vreg are both pretty burly when it comes to abuse (misuse).

the fuse seems a bit odd - the diecimila fuse is there to protect the computer's internal usb hub. a reasonable 3.3v regulator (you dont have the part number anywhere so i cant check it) should have short circuit protection.

I like PTCs. If the novice user does anything silly, the fuse takes the brunt. I shorted 3.3V and GND on the board and the PTC kicked right in. The micrel v-reg on skinny is dang impressive though. I hooked up 10, 15, 20, 25V DC and loaded the board with 100mA load. The v-reg did great up to 22V when it started to auto-shut off. I then hooked up 10V, 15V, 20V backwards. The tantalum input cap popped at 20V. Tantalums fail unsafe (shorted VCC to GND with carbon) so I removed the popped cap, attached correct 15V and the board continued to work. I like parts that can stand up to my mistakes. The diode would be good but may cause problems with batteries (~0.5V forward drop).

i dont really get the 8mhz/'lilypad' thing. i use internal oscillators all the time for cheapie kits, but i'd never do it for a project where customers are expecting to use a serial library where they cant tweak the TCCR values (and where the bootloader is pre-burned with a set value that is not calibrated). i've found that the int oscillator does vary a bit over large runs/temp/voltage. why not include (at least) a ceramic oscillator? you can get a skinny one for $0.20, a full-on crystal setup is maybe $0.35 or $0.40

Yea, my bad. I was cutting it a bit thin. Next rev already has tiny 8MHz resonator.

I don't really see any use for this. There several ways to interface an "normal" Arduino to 3.3V peripherals. And since they made a "full-sized" Arduino board anyway, why didn't they use the PDIP-Version of the chip (which can be replaced in case of an "accident").
Eberhard

Cost. None of the components on the Skinny are PTH. It's expensive to solder a DIP package. We figured we could get the price down. Besides, anyone know of the accident/fry rate of the ATmega168? Like zombies, those things are hard to kill.

Thank you again for the review. We'll spin the next rev.

the 5v/3.3v shield thing is not a -huge- deal: avr's are (im like 99% sure) 5v compliant so, at worst, it should still work. (if using a r divider, then 1.8V is still higher than most CMOS Vih)

Awesome feedback. Thanks! I am more worried about customers buying Skinny and getting frustrated when it doesn't work with their 5V XBee shield, etc. The ATmega168 should be able to clamp higher voltages without too many problems.

Well, uh, you -did- test it with the XBee shield, right? :wink: does it work?

ironically, if you actually solder in female headers to put a shield on it, it will be exactly the same height as a diecimila :slight_smile:

The power jack and USB connectors are pretty tall aren't they? And if we're going for pocket-ability, I'm thinking peeps will want to solder in their own wires.

i guess it depends if people are not expecting to put any shields on

The female connectors were a fantastic idea in the beginning. You ever try to desolder a through hold connector like that? :wink:

crush/break the header first, using diagonal cutters, and it will easily come out.

is the 3.3v on the serial connection an input (to VCCIO) or an output (from the internal miniregulator in the 'RL). if its the latter, thats kinda sketchy. :-/

My designs are always a little sketchy :slight_smile: If it sparks, unplug it. If it heats up, you better check VCC/GND. No really - The 3.3V pin on the male header on skinny is an input. The USB board provides 3.3V. You can have the board 'on' and hook up USB without problems. It's a little sketchy to have two vregs on the 3.3V bus at the same time, but both the FTDI vreg and the on-board SOT-23 mini vreg are both pretty burly when it comes to abuse (misuse).

the regulators im not worried about, even a .2V difference they should be able to cope with. what would concern me more is that the internal usb vreg can only provide maybe 50mA. people may plug in the USB thing and notice that the LEDs come on and say "neat i can power it from USB just like an Arduino" and confuse the 3.3V miniregulator with the 500mA USB limit. Everything works fine till they try to connect a 40mA GPS and it just starts to flake and dip into BOD zone. These problems are the hardest to debug... :frowning:

About the shield issue... if its a concern that 5v shields would be used with this board and the solution is to change the pin spacing of the headers to force a new 3v shield design then I just have to ask whats the point of sticking with the arduino footprint?

If you plan to change the headers then just go ahead and shrink the board size and introduce a couple new smaller shield designs. I do like the idea of the lipo friendly factor. I think in the interest of ultraportability then just go ahead and make it smaller!

I have had this issue with the lilypad and now this board, your programming header doesnt really stick with any sort of standard. If you had the 6pin ftdi compatible pinout (look at lemore's boarduino or paul's bare bones) then your customers would have more options for programming adapters and you could enable the auto reset function in your designs.

Finally since this seems to be about options, I would add 2 pins at .1" spacing for + and - at the power connector to allow for some male pin headers to be soldered by the user instead of a barrel connector. This could be under the barrel connector or to its side somehow. If done well, the large holes for the bigger connector could be used as strain relief if soldering in wires.

And which vreg did you use? If it is indeed only good for 40mA then that really needs to be noted.

Make it smaller!

:smiley:

-Brian

Finally since this seems to be about options, I would add 2 pins at .1" spacing for + and - at the power connector to allow for some male pin headers to be soldered by the user instead of a barrel connector. This could be under the barrel connector or to its side somehow. If done well, the large holes for the bigger connector could be used as strain relief if soldering in wires.

I like this idea. Sparkfun already does this on their breadboard power supply. I soldered both a power jack and header pins on mine which makes it very versatile.