hi,


No, it is wrong. In main, the cost is formed not from number of pins but from real cost of core/die and the market requirements (first of all, the number of sold parts per time).

For example, low-pin-count AT89C4051 costs 1.57$ and AT89S51 with 40 pins does 1.16$. By the way, Atmega8 costs 2.24$ (the price is from my local Atmel distributor).



Sorry? What do you mean with it? When you have installed P-CAD or other SAP then I do not see difference - it is just question of knowledge.
Indeed, if you create something "hand-made" then DIP package may be useful. But I doubt that someone will create 1000+ by hands. For big quantity it is cheaper to create the SMD design once and then use SO/QFP components. And do not say me that it is Sooooo hard to soldering VQFP44 with pins` step 0.8mm or even SOIC28 with pins` step 1.27mm (!!!). If it is really hard for someone then he should drop all and go to 19xx years.



Well, if we talk about DIP40 and VQFP44 then I might agree. But somehow my projects utilize not only low-pin-counts parts but huge chips as well. For example, here we use 100-pins derivatives and even 208-pin ones. No way to use DIP/PLCC.



Hmm, I have opposite view: most new parts are 3.3V. If you do not wish to search obsolette derivatives in the near future then now is the time to switch from 5V to 3.3V. I should not suggest it for all because it depends from task to task. But at least for me this way is appropriate due it is based on parts we use in most projects: SiLabs 8051, Atmel DataFlash, Samsung NAND flash and their SRAM, SD/MMC, Ramtron FRAM, Altera and Actel FPGA/CPLD. They are 3.3V (FPGA/CPLD require core voltage 2,5V additionaly).
So what I try to say: it seems that 5V parts come to the past and at one day someone will be in trouble to obtain them.


Maybe, maybe... And is conversion from Keil to AVR Studio easy as well? Oh no, need buy pills once again (=

Regards,
Oleg