Distributable fpga design

266 views Asked by At

I'm new to fpga programming, and I'm wondering how to make my fpga design distributable. Here's the scenario I have in mind. I have a network of computers, each deployed with an fpga based peripheral. I want to update the fpga design on the peripherals periodically. How do I accomplish this without spending a fortune on software licenses?

I have a small dev kit for an fpga that shipped with an executable to load example design files (it was an Altera fpga FYI). Does anyone know how I would create such an executable?

Some specifics: My fpgas are Xilinx Spartan 6Es. I'm using Xilinx ISE for fpga development. The host computers are running debian linux.

Thanks for any and all advice!

3

There are 3 answers

3
blueprint On

If youre dealing with Altera: one computer would have the software tools and licenses needed to synthesize the project. Assuming all the FPGAs are the same model on each station/node, Quartus will generate an .sof file which you can copy and open from station to station. All you would need to do is download the Altera programmer tool (I believe you can download it separately from Quartus II) on each station which is free. Then upload the .sof to the board using the programmer, where you can permanently store it on the fpga prom using a technique similar to the following:

https://m.youtube.com/watch?v=ZrMe8JS7Ktk

However if you have Xilinx and Altera mix, Xilinx has .bit/xdl files, and uses another tool (impact) to upload their bitstreams. They can't be converted to and from bit and sof. So it's recommended that you probably stick to one make (Xilinx or Altera) and model based on your plans.

0
Joshua On

It looks like what you are looking for is how to make your FPGA's field upgradable. Assuming your FPGA is loading from an external memory such as an SPI flash chip, then you need to modify your design so that it is capable of writing to the SPI chip (or whatever) itself. This is most simply done by putting a register in your design which maps to the individual pins on the flash chip, and then "bit bang" the register from a connected computer. Assuming your FPGAs feed data into your own software running on the computer, then you would modify this software to have the functionality of manipulating this register to reflash the flash device. Obviously, if this goes wrong you bricked your device until it can be flashed again with the JTAG, but it provides a way for all the devices to get updated in the systems they operate without needing to buy a JTAG cable for every single station.

0
Holminge On

If you have Ethernet on your board you can use the remote programming tool from fpga-cores.

Then you can remote login to the network and program the FPGAs or mail the new config file to you customer and they run the programmer. This is how we remotely updates our boards.

Spartan 6 is supported. As a bonus you can also do some remote debugging with the remote logic analyzer. Everything is free for non commercial use.