I have a handful of AVR chips that I'm using in various projects (including WOSP). You use the SPI interface to upload the software to the chip's flash memory. I wrote a custom program on the Arduino that would read in Intel S-records and push the program onto the chip. I hoped to simply send the programs a file at a time to the chip. Didn't work. There was no hardware handshaking between the USB Arduino and the computer, so it would fill the buffer and then hang. No fun. I tried adding xon-xoff control but couldn't get that working either. I could put the programs in using Hyperterminal, one line at a time. That got old before it was even new. I decided to clone the AVR910.

It wasn't hard to do, I just read through the published source from Atmel and converted it to C for the Arduino. Once I had it sending and receiving characters I got AVRDUDE and started debugging. I had some initial problems because I tried to tell avrdude I had an AVRISP. I didn't, I had AVR910. It took a few night's hacking but I got to where I could read and right EEPROM and Flash on the target chips. The last problem I fixed was a byte order thing with sending an address. Seems that avrdude sends the programs one byte at a time, preceeded by the address. So each time I would get the address but in the wrong byte order. I swapped it around and suddenly it worked.

My targets are ATtiny13, 15, and 2313. The tiny13 isn't officially supported by avr910, so I just assigned it an ID of 1 and it works. I haven't tried the tiny15 chip yet, but the programmer works for tiny13 and tiny2313.

So look for more AVR projects from me in the near future!

So you want to drive a 30mA load, but your PIC will only give you 20mA? You can use a BJT as a driver. The math is quite simple.

First consider the load: 3 super bright LED's, they want to run at 30mA and have a forward voltage drop of 3 volts. You have a +12V line you can use that will provide the current. This is a good start. The 2N3904 NPN transistor has a maximum current rating of 200mA, so it should work fine. The average gain (hFE) for the '3904 is 100, so with 20mA into the base we could drive 2000mA, which would fry the transistor. Better keep the base current under 2mA (2*100=200mA, the max current).

To get 30mA load current we would need just 0.3mA base current. Let's aim for 1.5mA. This is five times the minimum required current, and will help keep the transistor saturated if the supply voltage dips. So for 1.5mA on 5V from the PIC we need (5V/1.5ma)=3.333Kohms. Put that between the PIC output and the transistor base and it should work just fine.

Here is the math:
Collector current is the voltage across the resistor divided by the resistance (ignoring voltage across the transistor)

Ic = (V/Rl) = (12V-3V*3)/100 ohms = 30mA

We want the base current to saturate the transistor (5x minimum current), so we take

Ib = Ic/hFE * 5 = 30mA/100 * 5 = 1.5mA

Now to get the base resistor

Rb = Vpic / Ib = 5V/1.5mA = 3.33Kohms

If we use a 3.3Kohm resistor it gives about 1.52mA, way under the PIC output max of 20mA, so that works.

(response to uC meets BJT)