Hi All,

Sorry I havent been around lately but I've been playing around with the ATMEGA644 PDIL40, first in line is the integral 10 bit adc, it's basically crap, in differential free running interrupt mode, it will only work with a Vref of >2V and the input mux suffers badly from channel to channel crosstalk, only in free run mode though, if the mcu invokes a conversion then no crosstalk occurs however this mode ties up the mcu which waits for the conversion to complete and then returns a result.The manufactures claim that the adc differential mode is not guaranteed on the PDIL40?, and the input gain opamp is restricted to 4kHz bandwidth.

Later, I hooked up the MCP3302, programmed in differential mode and works beautifully, gives +-4095 counts with a Vref of 1.5V Vref works down to 1V!

Here's the kicker, the AVR ATMEGA644 works happily with a 30MHz crystal giving a 30 mips throughput, the chip does not crash and is stable.
Inconjunction with the MCP3302 adc, it can sample at >50kHz(single channel), thats 13 bits, eh? more than enough for some DSP applications.

The other point I would wish to raise is the bit handling capabilities of the AVR instruction set are crap, for example, with the '51 you can send the carry flag directly to any port pin, you can't do this with AVR which makes assembler bit banging routines complicated.
However, from a maths point of view, the AVR reigns supreme over the '51.

Cheers for now
JA