Designers Corner with Prof. Mike Smith  Part 1  Part 2

Choosing parts

At this point we were almost ready to draw a schematic and fabricate a board. However, there is a rule that I have learned to follow over many board designs: “Never build a board without parts in hand.” So while we were designing, I spent some time finding out which parts were available. I’ll cover the major parts to give an idea of some of the issues involved in this stage of design.

We decided to use a Xilinx Virtex FPGA, the newest generation of the Xilinx FPGA family. We chose an HQ240 package since that gave us the ability to use the widest range of device sizes (from the V300 to the V800), if we used a zero-insertion force socket. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Xilinx XCV800 Virtex FPGA

We chose a XC9572XL CPLD, large enough to hold the nonvolatile configuration logic that we needed. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Xilinx XC9572XL CPLD

The PHY chip, the Level One LXT970A, was not readily available at the time we completed this design, with long lead times for delivery from all the distributors. We had to work directly with the Level One distributor, Wyle, to get chip samples. We also knew that in desperation we could probably get a few devices from Dave Van den Bout at XESS. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Level One PHY

Hamish managed to obtain some sample Seiko LCD panels, we used a 16- by 4-character display with built-in Samsung KS0066 controller chip. We found out later that these are fairly standard. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Seiko LCD Panel

Our Wyle contact also offered to supply us with SRAM from IDT. We used 4 Mbit (512k x 8-bit) 3.3V CMOS SRAM. We chose the fastest parts we could get at 10ns access times. This was asynchronous SRAM, and we perhaps should have thought about this a bit harder, but we were concentrating on finishing the schematic design and getting the board to fabrication. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm#SRAM

We selected the magnetics (the pulse transformers that isolate the cables from the logic) that Level One recommended for use with the LXT970A from Pulse. We were able to get these as samples. Later we ran into a shortage of these parts and switched to a local supplier, Halo, with whom we established a friendly relationship. See http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Pulse Transformer

We had rather a messy supply design. Because of the analog, memory and FPGA components we had chosen, we needed three power supplies: 5V, 3.3V and 2.5V. We decided not to worry about this at this stage, though it was certainly something that we realized would have to be faced at some later stage. We borrowed a power supply design from some previous boards built at Xilinx and used standard regulators. http://ilima.eng.hawaii.edu/XCoNET/Demonstation_parts.htm - Linear Technology

These were the major components that we used, other components we bought at Anchor Electronics, a local Silicon Valley supply store, ordered online from Digikey, or scrounged from the Xilinx FPGA lab. See http://ilima.eng.hawaii.edu/XCoNET/components.htm

Table of Contents   Previous   Next