If I have more than one device on an I2C bus, does each one need it's own pull-up resistors on SDA and SCL? I assume the answer is no since from my understanding the pull ups just keep the lines from floating since the MCU can only pull them low, but I'd like to make sure. Also, as a side question, what is a good value for the pull ups? IIRC it said to use 4k7 somewhere. I'm running at 3.3v if it makes a difference.

Has no one ever done this? It would be a pain to have to use an atmega instead of an xmega. I've also heard that you can program an atmega16u2 as a PDI programmer but I couldn't find any code.

I guess 650KHz would be the way to go, seeing as I don't know much about how to deal with noise and all the precautions necessary when running an MCU at high speed.

I thought of using the MCP1253 as well at one point but I'm not sure if it's worth it. According to some Li-ion discharge curves I've looked at, and the fact that even an LDO will only work down to 3.5v (3.3v + 200mV regulator dropout), it seems like I might be pissing away a fair bit of the battery capacity using an LDO, whereas the MCP1253 would continue to boost the voltage up until the Li-ion is completely discharged.

Also, if I do decide to use the MCP1252/3, should I use the MCP1252 (650KHz switching) or the MCP1253 (1MHz)? the MCU will likely run from 16 - 32 MHz.

It certainly looks better than a lm1117, I think I'll give it a try.

I'm planning on running my next project at 3.3v because the only batteries I have that are small enough are 3.7v Li-ion's. While the MCU should work fine directly from the battery, other components like the LCD and SD card might not be so happy with 3.7v rather than 3.3v. The obvious solution would be to use something like an lm1117 3.3v regulator, but they have >1v of dropout and would be useless. Is there anyway to regulate the battery to 3.3v without significant battery life losses due to dropout. Is there some sort of 3.3v regulator which can either step up or step down depending on input voltage. My project should consume less than 100mA absolute max.

So I've been looking at the atxmega256A3 recently and it seems that it would be the perfect MCU for my next project. Problem is that you can't program it via SPI based ISP, like you can with atmegas. So i'm wondering if anyone has written a library which can turn an arduino into a PDI programmer. If not, is there any software which can use my computers parallel port as PDI programmer?

Ok, problem(s) solved. It turns out that the relay was clicking in so loudly that it triggered itself! so I moved the microphone to a seperate board and increased the RC time constant of the 555 timer so that it wouldn't trigger so fast and now it works most of the time. It's not sensitive enough though; I need to clap so hard that it hurts my hands. Oh well, I'll fix that later.

It draws about 50 mA, so ~1mA to the transistors base should saturate it, but just in case I've tried it with as low as 470 ohms with the same results. The problem just seems to be that the 4017 resets itself for some reason.

yep, relay works fine with this power supply.

That was one of my initial thoughts as I'm only powering it from a 200 mA wall wart. however I disproved this by manually pulling transistors base high and seeing if the Q1 goes low when previously high, which it didn't.

So I've been trying to build a clap detector which switches a relay on and off when a clap is detected. I ended up with this schematic (modified from some website) . As far as I can tell, it amplifies whatever the microphone puts out, triggers the 555 which sends out a pulse to the clock pin of the 4017, making Q1 go high. When another clap is detected Q2 goes high, resetting the 4017 back to 0. If I disconnect Q1 from the resistor and check the state of Q1 with an led to ground then things work as expected, it turns on when I clap once and turns off when I clap again. However, if I attach Q1 back to the resistor - enabling the relay -, then whenever I clap Q1 goes high and within a few milliseconds it goes low again as if it had reset itself! I've messed around with the resistor value and adding a diode across the relay without success. Does anyone know what could be wrong?

Ok, I found the PCB's for both the wireless handset and the desk phone. Unfortunately, whatever company made this phone didn't feel it was necassary to label their PCB's with useful information such as company name and product name, so I can't even find a datasheet. I assume the handset had the same processor (before I ripped it of and threw it away, derp) as the desk phone but I can't find a datasheet for that either so I can't really follow the traces anywhere. Not sure how images could help at this stage but it wont hurt I guess:

http://imageshack.us/photo/my-images/20/p1090217.jpg/
http://imageshack.us/photo/my-images/850/p1090215w.jpg/
http://imageshack.us/photo/my-images/861/p1090214x.jpg/
http://imageshack.us/photo/my-images/831/p1090213v.jpg/
http://imageshack.us/photo/my-images/638/p1090212.jpg/
http://imageshack.us/photo/my-images/802/p1090211l.jpg/
http://imageshack.us/photo/my-images/191/p1090209v.jpg/
http://imageshack.us/photo/my-images/196/p1090207v.jpg/

Ok, I put 5 volts on the pink pin and got this:


I guess it came out of a phone...
Also this probably means that the pink pin is not the actually the contrast pin. Although lowering the voltage does make the display dimmer (until I get to 3.5v), my HD44780 needs between 0-1 volts on the contrast pin. Then again, my HD44780 is white letters on a black display, unlike this LCD.

Now that I think about it, this display probably came from the phone I garbage picked years ago. I threw away the case just a week ago though 
Maybe I still have the PCB. I'll go look for it.

Alright, so I tried comparing the LCD with one of my HD44780 displays and I believe that the pink pin is the contrast pin because my HD44780 has a very similar resistor bank connected to the contrast pin. Now I just need to find the main supply voltage pin and I should at least be able to get something on the display.