USB Interface to Virtex-4

Jim,

Agreed. If you have to have a substantial core of soft logic to support that interface in the FPGA, it had better be worth it. If the FPGA has hardened cores for most, or part of that interface, it makes it a slightly better proposition.

It is all systems engineering.

I admit that when I sit down to do that job (very rarely now), I pick the components that will:

- do the best job

- cause me the least grief (both in software/coding and signal integrity/support)

- meet the cost objectives

When all the marketing hype is said and done, the hard work is just begun for the engineer.

Successful completion and testing of the prototype is an important step.

And then manufacturing can be another real trial.

So it ain't over till the customer is paying (and happy).

Anything that gets you to market faster is a real plus. The systems engineer can make, or break a project by their decisions.

Aust> Aust>

sitting

there's

I have to admit that I find some of the replies to what I thought was a dumb question a little odd. It's almost like some folks take personal offense at the notion of using a couple hundred slices and a couple block rams to implement a USB interface inside the FPGA. (The smallest V4fx has 12,000 lcs and 36 brams) I think most of the counterarguments could apply to any other peripheral: Why a uart? Why ethernet? Why even a PPC core when you can get a better processor discretely?

If I had a solution that:

1. Integrated into EDK such that it adds like any other peripheral. The IP should contain the micro, firmware, and the interface to the CPU.
2. Integrated into the board support package so that I can build my chip, carry a few files over to my linux source and be able to compile it into the linux kernel.
3. Stock linux and windows driver support.
4. Netusb compatibility such that I can take my usb master port and connect it to any other port to create a network link.
5. USB1.1 initially but eventually 2.0.

I would gladly use it over any discrete solution that I constantly have to worry about obscolecense or other supply problems, not to mention the extra size, cost(part plus ordering plus handling), and power. Who hasn't had a board build delayed because of a back order on some small part like a usb chip?

-Clark

That's because you are looking at a very simplified block diagram. The logic is a bit more complicated.

You would still need an external USB PHY no matter what and they are even more prone to supply issues because of the limited demand/volume of production.

Marc Reinig System Solutions