The ideal USB device Microprocessor

[menke]

Dear Forum,

It is obvious, there is no such thing as an ideal processor and this question was posted in several forums before but a) I think it has to be asked over and over again because every month new MCUs appear and other disappear and b) I'd rather like to start a discussion than want to have an answer.
For a rather small application it took me a pretty long time to find a MCU. I finally fell in love with the Silabs C8051F320 but I am still not sure whether it was the perfect choice. The reasons I picked this processor were:
 - Cheap development board available
 - Programmer included with development board
 - No crystal or reset circuit or other stuff needed
 - Offers I²C and SPI interface and several other features
 - Will be available for some more years
 - Can in fact be purchased e.g. from Farnell (I found plenty Cypress MCUs which couldn't be purchased anywhere!)

What I couldn't find was a chip that can be programmed via USB as the goodold AN2131.
 
So what are your favourite chips for small/medium/large projects and, more important, why?

Ma

[Ron Hemphill]

Hmmm... anybody else remember those days of massive spam on the old USB.org forum?

This is obviously a spam advertising message thinly disguised as a discussion prompt.

[Jan Axelson]

Ron, I'm not convinced it's spam, plus it's a good, open question. So I'll bite...

I've come to believe that for many projects, the best choice is a chip with an architecture that you have experience with and thus can get going on quickly. If you've worked with PICs, look at the PIC USB chips. If you're a .NET programmer, look at GHI Electronics' .NET Micro Framework products. I've heard good things about the SiLabs chips, and they sound like a good choice especially if you have experience with 8051s. For vendor-specific communications, FTDI's USB UARTs and similar chips can get you going with just about any CPU and no USB knowledge at all.

Jan

[menke]

Hi Ron,
hi Jan,

Ron: No, this is not spam and I don't know what makes you thinking this. As I said, this question has to be discussed over and over again.
Jan: Thanks for your support. And, of course, you'r right, development time is a big issue (if not the biggest). When I found the Silabs chip I was glad that it was an 8051. But the environment is a C compiler and so it actually didn't help. I'd prefer assembler for a chip that (EPROM-) size. My problem with PICs is the huge amount of different types where you never know which one will be available next year. And, AFAIK, they come without a boot loader or a programming device (the development boards come with pre-installed boot loader). Cannot say anything about the .NET thing. Not because I don't know .NET but because I know a lot about it (and I hate it, but this is another discussion and certainly OT here).

If I could say something about the FTDI USB UARTS:
AFAIK this chips are USB-to-COM adaptors which means the driver installs a kind of COM port on the PC. From the PC point of view data is then transferred via this COM port. The advantage is that no USB knowledge is needed to use these chips. But the downside is enormous: The PC software has to know the COM port number and this number changes with every boot or by all kind of events. This might be no problem when developing a more experimental project but for a real consumer product this is intolerable. Please correct me if I am wrong here.

MA


ME

[Jan Axelson]

If you don't want to use a COM port, you can access FTDI controllers via the vendor-specific API supported by their driver.

I think your comments support my point that one is often best off using what one is familiar with, which is likely to be what you've found usable, maybe even enjoyable, for past projects. For some, being able to program embedded systems in .NET is a great relief, for others, a nightmare. Some developers are happiest in assembler, others prefer C or even BASIC.

Of course the requirements and constraints of any particular project are also a factor, but for small- to medium-quantity projects, fast development time can be a major factor in keeping costs low.

Jan

[Guido Koerber]

I am using Cypress USB controllers since 1998 and their market availability has been quite stable, they just pulled the plug on a controller that had been around since 96. Anyway I was thinking about discontinuing the products based on that chip when the last-order notice came in. Though their development tools do come with a bit of a learning curve. Especially if you are like me and are not satisfied with libraries but want to dig deeper to adapt the code optimally for your application.

I am looking forward to try their new ARM based PSoC5 chips, going to skip the PSoC3 since I do have severe 8051 allergy...