Huh, anybody wants to play some NES???

Jacques, first of all, congratulations! That's a really amazing project! I have a question for you. You use 4 Virtex 2 FPGA chips, and from the picture, it looks like you use sockets to connect the FPGA to the surrounding components. Where did you buy the sockets? How did you wire wrap it? I mean, the distance between pins in Virtex 2 is very small, right? So, how do you manage to wire wrap it or solder it without using a machine? Does the distance between pins in the sockets wide enough for one to wirewrap without using a machine? You run the board at 40MHz, with all the soldering and wirewrapping, does the noise matter at all in your project?

Hendra

Hi, for your questions: we actually used 1 virtex2 xc2v1000 fg256 FPGA chip. We did not use any socket at all. We didnt use any lense, and everything was done within one week. We had a couple of team members with good eye sight and with some training, they master the art of soldering 1 mm pads! So no socket of any kind

The first serious issue that we had was actually the ground bounces that exist, because of inexistant ground plane. The way we resolved this was a good planning at the beginning of the actual board design. With a little bit of thought, it was decided to lay a sort of grid around the FPGA chip, using wire and solder. All the chip gnd pins were then connected, as well as those of the FPGA chip.

Initial test which was mostly static indicated no problem. Afterwards, when we downloaded our first SOC module, the nintendo audio unit, the

6502 cpu and some game cartridge, it was noted that the sound played fine, until you approach or touch some part of the circuitry. Then we realized that any piece of metal would destabilized the whole circuit, increasing the frequecy or decreasing it (the sound would be played either faster or slower)...

This was a serious issue with the DCM, but we found out later that one of the pins, from our sound dac, was left unconnected, and was supposed to be grounded. Also, some wires, where the actual symptoms were most noted, were cut to smaller length. Now it works without problem for days. The power consumption is about 180mA. The noise matter in general, but I think in our case, at 40MHZ, and max internal clock rate of 12.5MHZ, HS issues are not much of a concern.

Here are the specs for the board

Virtex2 FPGA chip, XC2V1000 -4 FG256 entire soldered with common blue kynar prototyping wire MODULAR DESIGN APPROACH AT THE BOARD LEVEL, EACH COMPONENT CAN BE PHYSICALLY DISCONNECTED FROM THE POWER SUPPLIES (GOOD FOR DEBUGGING PURPOSES) dual flash memories AM29F101, we did a core for a famous flash/eprom (starts with a W) programmer. We are using it as the main memory interface to do our system development. ADV7125 triple hs video dac is connected to the FPGA chip AD5332 dual dac are used for the two sound channel, as per the original NES

5 hc245 buffer chips, for voltage level translator one 40mhz can oscillator

The HDL code was developed using VHDL

We did two 6502 cpu core We also develop a four channel sound generator We also did a graphic unit, using reverse engineer data obtained from the net All the cores are tested in FPGA silicon.

That was alot, alot of work. It started last year, may2003 and was completed this week.

Jac

Very cool. Why are you selling it? Do you have any closeups of the Virtex. How did you route the pins out? How is the BGA attached to board. Did you use reflow?

We need money, because we were "self-funded".

The FGA chip was soldered with blue prototyping wires, that was attached to the FPGA pads directly.

We didnt use reflow, as it would have been too expensive.

Here is my website. Check it out for some of my projects, in particular, INES1M for the close back shots!

Jacques

here is my website.