0
\$\begingroup\$

I have a possible project coming up that looks like it needs a small amount of digital logic (to generate some synchronous timing/control signals). Speed is not that high, in the low megahertz.

My background is a wide mix of analogue/low speed control/embedded microcontrollers and also application software - over quite a few decades. I did some digital design back in college (a lot of wire wrap 74LS doing really primitive digital audio) but that is long ago. If I took that approach to this project, it would probably be 9 or 10 chips - but I know how long that can take to put together, just wiring, testing, bug fixing.

I have a couple of Numato dev boards with Spartan 6 FPGA's on them, I'm sure that they could handle this, but the tool suite is a bit daunting right now. (Also I don't have the right programmer, the USB app they give is not that great.) I don't mind spending some time on getting up to speed, but it can't be too much. I probably don't have time to pick up an HDL for this one.

There is a lot of attraction to using a CPLD/FPGA - simulating and checking the design on-screen, being able to make bug fixes and changes - but I need a tool chain that I can get into reasonably quickly, and I need to get the thing doing something that starts to look like progress pretty fast.

Any advice? Recommendations? I might spend a little time with the Xilinx and see how it goes in the next days.

\$\endgroup\$
4
  • \$\begingroup\$ Find a tutorial for the family you are interested in. Follow along and use the simulator extensively. Due to the complexity of most modern FPGA devices the simulator will be your best friend in the long run, so good to learn it now. \$\endgroup\$
    – David
    May 9, 2018 at 21:49
  • \$\begingroup\$ Find a tutorial with a dev board that is in your price range. \$\endgroup\$
    – Voltage Spike
    May 9, 2018 at 22:14
  • \$\begingroup\$ The issue is that I don't have any familiarity with the families or toolchains that are common. I do know from experience (with microcontrollers) that there can be huge differences in usability and learning curve from family to family. Normally I would bite the bullet and dig in, but on this project time is a big factor. That's why I am asking for rec's. Thanks! \$\endgroup\$
    – danmcb
    May 9, 2018 at 22:22
  • 1
    \$\begingroup\$ @dmb - have a look here: joelw.id.au/FPGA/CheapFPGADevelopmentBoards \$\endgroup\$
    – awjlogan
    May 9, 2018 at 23:39

3 Answers 3

2
\$\begingroup\$

If what you need is simple i.e. definitely at the logic/cpld end of the spectrum and not the fpga-processing end, then have a look at Silego greenpak.

They are really fast to get going, with schematic design, 5V single supply operation, and analog functions (vref, comparators, oscillators)

If what you want to do is learn fpga's, VHDL, etc, they aren't what you want.

\$\endgroup\$
0
1
\$\begingroup\$

We don't really do product recommendations here, but FWIW, here's a recent experience I had.

I have been working with midrange Xilinx and Altera (now Intel) FPGAs for a long time — one of my specialties is real-time processing of HD video streams. But I had a need recently (for a different client) to implement a small amount custom logic under tight power and space restrictions, including the constraint that it could not be RAM-based (i.e., requiring a separate configuration memory device).

Of the two candidates that we considered, I had a much better experience with the Microsemi (formerly Actel) IGLOO Nano line. I got one of their evaluation boards, and I was quickly able to start customizing the demonstration code that they supplied to add my own functions. The board comes with a programmer, and it just requires two USB connections to your PC — one to power the board and one to control the programmer.

I was not successful in doing the same thing with the Altera MAX V board. It seems that this board was designed some 6-7 years ago, and it has not been updated since then to keep up with the current version of the Quartus tools. I was not able to get the tools to communicate with the board at all, even after down-reving to the version under which the board was developed.

\$\endgroup\$
1
  • \$\begingroup\$ thank you Dave. Sorry if asking for recs is unusual - but this is helpful, and I will check them out. there are so many things om the market, there is just no way to investigate them all - and anyway most of them, you don't know what you are dealing with until it is in your hands. \$\endgroup\$
    – danmcb
    May 9, 2018 at 22:56
1
\$\begingroup\$

Xilinx CPLDs (older 9500 series, and newer CoolRunner series) are also quite functional and powerful, and the package is manageable (QFP).

All are supported by free development tools (called ICE WebPack), and there are a couple of cheap development kits.

enter image description here

So with few lines in Verilog you can be up and running in no time.

As an example of capabilities, once I made a EPROM simulator that would copy all EEPROM content into a fast static RAM, providing then a 1-clock access to data, emulating super-fast ROM, all under XC9572 CPLD.

\$\endgroup\$
3
  • \$\begingroup\$ thank you! I am sketching H/W design right now. I have this dev board: numato.com/product/… - I will dig in over the w/e and see how it goes. Perhaps I just need to get my feet wet. I have the ISE s/w on here at the moment. \$\endgroup\$
    – danmcb
    May 9, 2018 at 23:46
  • 1
    \$\begingroup\$ @dmb Spartan-6 is a very capable and powerful family, beyond usual expectations of many. Which also comes in DIY manageable packaging (QFP-144). But it needs an external config EPROM, and it takes time to load the config after power-on. CPLDs have a built-in non-volatile config space, and are ready nearly instantly. So it is up to you. \$\endgroup\$ May 9, 2018 at 23:53
  • \$\begingroup\$ well, as I have it to hand, I can use it to start to develop a model if the design, learn the tools, and maybe spit out some signals to prove the idea. I doubt we would go with this chip for the final thing, the ones you post above look more sensible - but at the moment I guess it is about getting things going. If i have a decent model (hdl or whatever) it should be not so bad to move it to another device with enough capacity later, right? thx again. great info! \$\endgroup\$
    – danmcb
    May 9, 2018 at 23:56

Not the answer you're looking for? Browse other questions tagged or ask your own question.