Nick Gammon's site really helped me to understand low power modes and to get the sleep function working with a chronodot RTC:

http://www.gammon.com.au/forum/?id=11497

The lack of a 3.3v line on TinyDuino makes it very difficult to develop code for sensors that require 3.3 volts because the levels shift to 5v when you plug into the USB port. However with the distributed regulator & level shifter components already in use by the tiny boards it would be easy to create boards specifically for I2C  sensor connection, that stabilized these voltage swings.

This would be really great for those of us trying to run from things like AA battery packs, that change voltage significantly over time, as we run into the same problem with sensors that can not withstand the voltage swing. Yes, I can go hunting for a voltage regulator, and a level shifter, and cobble something together. But if Tiny Circuits sold a $15 board just for this purpose, I would buy it in an instant, as I want to spend my time making the sensors do something useful, rather than spending my time just getting them running.

Essentially you could just take and existing product,  like the light sensor board, leave the other components in place, but remove the light sensor itself, and replace the pads with through hole connections near the edge of the board. I think the regulator you are already using would supply most of the I2C sensors out there, but you could also make one with a beefier regulator if people asked for it. My guess is that if you left the inbound VCC, GND, SCL and SDA ports already on that board, there would be plenty of customers for this regulated I2C sensor board for other battery run systems, outside of the tinyduino ecosystem, because it would save them a lot time and effort too.  And, no matter what sensors people are using (in my case high range MS5803 pressure sensors), Tiny circuits generates a sale.

Do similar modifications to the SD card board, and you would have an SPI sensor development board to sell as well, although that one would need the "inbound" lines broken out for the external market.

I really think this kind of generic sensor board could change the Tinyduino system from one with a 3.3v limitation, to one of the easiest systems to work with, and opens access to all the I2C/SPI sensors out there that have limited voltage tolerances.

And now that I think about it, it would not hurt to throw the logic level shifter on there as well... That way most available I2C sensor could be connected compatibly with the Tiny Duino system, even if the sensors own voltage range was quite restricted.

Just an update on the RC aircraft pushrods for 5+ board stacks:

I have been using HBZ7128 Pushrod and Clevis Set for the Cub aircraft, as extra long mounting pins.

They are about $1/pair on ebay, but unfortunately they are only threaded to 3/4 of an inch on one end of the rod.  I have been bending the rods into an angular "U" shape, and threading them up through a plastic platform, so that the two threaded ends are at the non-stack connector side of the boards. The rods "just" make it through the tiny board holes, so they are probably 1.45mm in diameter, and I have no idea what the thread dimensions are. The plastic clevis ends that come with them are easily cut down with an exacto knife into 3mm high hold down screws for those threaded ends. The metal in the rods is reasonably easy to bend and cut.  I am still using the spacers from the tiny board mounting kits. So its a workable solution, but probably not as elegant as just ordering those 00-90 brass rods
(which I also ordered from http://www.specialshapes.com/threadedrod.html for $14.95, they sell the 00-90 nuts 12/$3.60, shipping was $10)

Just posting an update on these RTC's. I have been testing 6 different units (from 2 suppliers) and they seem to be working Ok, including the 32k I2C eeprom.  I did some modifications to them: First was to de-solder the power led, and the second was to remove the charging circuit (by desoldering one resistor) , and replace the rechargeable lir2032 button cell with a non rechargeable cr2032.

I have taken them down to 2.8 volts, and not yet seen any problems with the RTC, or read/writes to the 32k eeprom, even though the units original spec was for 4-5 volts. More tests are ongoing, but so far they seem to be working. Of course, at this price, I am still a bit uncertain of the quality of that DS3231. Lots of talk on the web of ghost runs, or people grabbing out of spec chips, that were supposed to be destroyed, and sending them back to market in these discount boards from China... 

And one obvious choice for this new board,  would be a "unsoldered" version of the protoblocks shield ASD2005-R, as there is lots of free real estate available for the regulator, and the 3v pin is just sitting there ready to use. (And I am sure many of us are already forced to use that protoboard because of the need for a pass through connector.)

Many sensors provide usable readings only if they receive a stable input voltage, but the tiny architecture does not provide a 3.3 volt line, and has regulators distributed to the various shields. Since you guys at Tiny-circuits have already done all the leg work to source these boost&buck regulators which handle  swings all the way from vbat to the 5v from the FTDI, why not produce a protoboard that already has these little guys on it to provide a stable 3.3v line? That would save us from having to cobble together stable voltage solutions from other parts, just to get our temp/pressure/etc sensors running consistently.

I have tinyduino  loggers assembled from:
Processor board + accelerometer board + sdcard shield + protoboard connected to a DS3231 RTC

these units run fine from USB, and also from 2xAA unregulated power supplies connected to the external power pads on the processor boards

However when I try to power them (on the connections previously tested as good) using the Sparkfun's NCP1402 3.3V step up regulator (https://www.sparkfun.com/products/10967) All I see is one little blip from the led on the processor board, when the power supply is connected, and then no further operation. The output from the voltage regulator seems to be stable at 3.3v when the tinyduino is connected to it.

I am not sure if this is some kind of inrush current problem causing a low voltage blip from from the regulator (causing brownout on the tiny), or excessive ripple from the sparkfun board, etc? Any suggestions on things for me to try?

Has anyone else run the tinyduinos from a boost regulator?  I have sensors that need a regulated power supply for stable readings, so I am hunting around for solutions. 

Well... I can confirm the frustration that amdrake expressed about trying to turn down those brass rods. Gad!

So I went back to searching for another solution.  And I think I might have found one. RC aircraft push rods:

http://www.radicalrc.com/category/.032-Pushrods-469

As far as I can tell, many of them are threaded all the way down the shaft, so we can cut them to the length we need. I have no idea what the threads are, but the sets usually come with "clevis" ends, that ought to do well enough as hold down nuts. Might even be easier for my fat fingers to deal with.

I have some on order now, and will let you know if they work.

Well, at $18/ foot,  I might try to buy a die

http://www.microfasteners.com/die0090-00-90-dies-carbon-steel.html

and some generic brass rod from the local hobby shop, and see if I can thread my own.   

I've never tried that before though, anyone else?

Right , so if I jumper every wire on protoboard1 directly to its equivalent on another protoboard 1:

I can hook this to the main stack, and then pot my main stack in epoxy leaving one end of the two connected protoboards exposed, and then hook the ftdi board to the exposed end and have communications to the cpu, even though it is now encased in resin.

Of course, even the thought of soldering 40 individual wires beside each other at that scale is painful.

I was about to order some more chronodots for my datalogger project,  but came across this DS3231 AT24C32 combined RTC & Memory Module just about everywhere from new egg to ebay:

http://www.amazon.com/DS3231-AT24C32-module-precision-Arduino/dp/B00HCVNPVQ

I was also looking into separate eeproms, but this might allow me to kill two birds with one stone. Has anyone tried one of these yet?  I am thinking that this thing might be a power hog somehow.

(and of course, I am still eagerly awaiting the promised TinyRTC to make my unit more compact...nudge nudge)

Lots of folks have asked about ribbon connectors of various types in the forum, but I have some significant moisture and vibration issues to deal with so I suspect I will be forced to make them by hand before they become available. I would like to pot the entire stack in resin, and jumper across to another protoboard, to allow the FTDI communications after the main stack is turned into a solid lump.

Has anyone else done this yet? 

Do I have access to enough lines for the FTDI communication on those protoboards, or are there "hidden" connection lines in the stack connectors that I cant replicate that way? 

and finally:
Any other things I have not thought of yet, like "How long can those wires be before I induce other problems from things like line capacitance."

Well this is reasonably OK for me, as its the SD card writes that I am trying to warn the user about with the warning led... So I guess my real question is, can I somehow mess up the SD card writing by putting High/Low write statements in the code for the led pin? or will the SD clk always trump any previous settings to make sure the writes go ok....

If I load the good old "blink" sketch, my tiny duino behaves as expected. However, I seem to be getting odd behavior out of the indicator led on pin 13, once I start sd card operations.  Can anyone with more experience explain how those two functions get along on pin 13?

Half the time the indicator led works, and half the time it will not light up with the standard "digitalWrite(13, HIGH);" line.