@ realiser,
The chip manufacturer provides tools to draw the schematic in, which will generate the VHDL, or the compiled binary file ready for download to the FPGA. Eagle not required and would be hard to use, as you need all the parts defined in VHDL, like flip-flops or And gates. Simpler to learn a new tool… 
This took much longer that expected but it is finally happening. See this http://www.tinyclr.com/forum/20/4808
I’ve been waiting for that! 
As a embedded hardware designer, I’ve just got a few things to add:
Is there a reason this board is using the older, slower LPC2387 instead of one of the newer STM32F4 processors that the Cerberus and other non-GHI projects are using? It comes in a smaller 12x12mm 64-pin LQFP package, as well as a small 10x10mm FBGA. The STM32 also has a lower-pin-count SWD (vs JTAG), and should blow the LPC2387 out of the water in floating-point operations.
Oh, and did I mention the part is 10-20% cheaper?
As for the rest of the design:
If you don’t mind where the pins are being broken out, a two-layer board wouldn’t be the end of the world to design for the LQFP-64 package. Just know that you’re going to get better signal integrity and potentially a smaller PCB area if you go with a 4-layer board. Power planes are your friend.
Keeping passives at 0603 would ensure the whole board could be hand-soldered quickly and easily, even by people with little to no soldering experience.
I’d oversize the length of the pads on the LQFP package; I’ve found doing this reduces the need for excessive wicking after hand-soldering, since you give the solder some room to flow. You also get better joints. This comes at the expense of the size of the landing pattern, of course.
Avoid excessive clearance behind the pad, though – novices who haven’t soldered a lot of SMD stuff before tend to get solder bridges coming out of the toe fillets.
If there’s not much action going on under the chip, you may want to put some JTAG pads (not pins) under the package – that way, if someone wants to do some native debugging, or they actually want to use this for a production run (for whatever reason), they could program the devices more systematically.
Also, never underestimate the power of unpopulated pads. Since there seems to be so much concern over a 15-cent microUSB connector, I’d DNI it, along with the 3V3 regulator. If you’re truly going to put this into a custom project (which sounds like the purpose of this project), obviously you’ll already have 3V3. Making use of 5V-tolerant pins on a 3V3 part is a hack, and causes excess current flow and poor signal quality. Use 3V3 parts and 3V3 signals. All 5V parts are 3V3 IO compliant, and many of them will run on 3V3 for power, too.
Just some thoughts!
There wasn’t an F4 port all those months ago - the last post was 6months ago ! Time moves fast
Please see cerb40 board!