Make: FPGAs anyone else have this?

[Howardlong]

http://www.amazon.co.uk/Make-Turning-Software-Hardware-Projects-ebook/dp/B01CDKN664/ref=sr_1_1?ie=UTF8&qid=1458081524&sr=8-1&keywords=Make+fpga

It looked promising when I skimmed through the book, but now I have some hardware to go with it I am in fear it's already going to be an expensive fail. In short, almost every page either lacks enough detail or has an error, mostly the former. Plus the odd warnings that this and that don't work properly in the Xilinx IDE, it hardly gives you much encouragement.

Disclosure: I am no more than a script monkey when it comes to FPGAs, but I am trying to change that... for the umpteenth time.

[uChip]

I don't have the book so I can't say if it's good or bad.  I will say that I've been using the Xilinx ISE toolset for the last couple of months and I have found it reasonably reliable.  I purchased a Digilent CoolRunner-II board and the HS2 programmer.  I was quickly able to code simple functions for the CoolRunner.  I made a several different flavors of display multiplexors and a clock generator.

Recently I was able to graduate from using the USB on the CoolRunner board to programming with the HS2.  This is a necessary precursor to building my own boards.  Also, I have stopped targeting the CPLD and am now focusing on a true FPGA the Spartan-3A series because it has an on-chip PLL for generating higher speed clocks, and it has many more flip-flops available.  I'm skipping the FPGA demo boards since they are very expensive.

My learning was all extracted from (in order of greater to lesser utility) YouTube tutorial videos (some from Xilinx as well), Xilinx documentation, Xilinx user forum, and ISE online documentation.  I did try the Altera iCEcube2 tool set but found it more opaque than the Xilinx tools.

My first test PCB has been ordered.  It should arrive in a couple weeks.  If I don't embarrass myself too badly I will write up my results and post them here somewhere.

Without trying to write something up now, what I found was a rather steep learning curve with little handholding compared to microcontrollers. Still, the information is out there.  You don't really need a book to learn it.

Good luck,
  - Chip

[ale500]

I started playing around with FPGAs in like 2009. That Xilinx ISE drove me crazy. If I had (many times) RTFManualS, it may have gone better ! Lattice ICEcube package is not a traditional FPGA-DE like ISE, no wonder you found it... different. For a (in my humble opinion) better than ISE experience, you may try Lattice Diamond. It (Diamond) does not (yet) work with the ICExx FPGA... sad... but the MachXO2 parts are great, you can get single supply TFQP-100 and -144 packages (and like 30$ breakout boards), sadly they do not have multipliers ... But to the rescue: there are Altera's MAX10: There are single supply models in TQFP-144 and they have multipliers .

I also do not have that book, but it sounds ok, for 14 quid, 8 projects ? quite a bit of fun

[Neilm]

I have been using Xilinx FPGAs for about 5 years and never had any real problems either with the chips or with ISE. Yes, you have to read the manuals, but you have to do that with whatever programmable silicone you are using.

When I used the Lattice system (over 10 years ago) I often found that it didn't optimise state machine coding very well and I was forced to manually select the coding type.

[Howardlong]

Regarding the MachXO2 and Diamond software... I've been there, then spent a couple of days figuring out that the current software at the time on their website had shit drivers for their demo boards on x64. I eventually fixed the problem (documented elsewhere on this forum) but the experience meant yet more dev boards collecting dust.

My three or four attempts at FPGAs have always met the same fate, flaky software, instructions that don't work or are incomplete, and/or telepathy/random button pushing failure. Actually I got a fair way with Lattice, it took me just so far, ISTR getting a blinky going, but then the instruction stopped abruptly, and I couldn't figure out what the next steps should be.

[uChip]

I won't try to argue that it's easy.  It's not.  As I said the learning curve is steep.

My suggestion would be to download the software (ISE, or whatever one you like) and try some simple designs.  Most of the tools have a simulator included so you don't need to commit to hardware to try things.  If you get stuck, ask questions here or in the forum of the tool vendor.  Unlike most companies I have found good participation in the Xilinx forum by their AEs.

One thing that really helped me was the support for design hierarchy.  Learn to use that early on.  With hierarchy I could design, for example, a timer without worrying about my entire circuit.  The timer was just a few lines of code making it easier to pick out the errors in arcane language and configuration.  Then once I had the timer working the way I want in the simulator I could use the timer like a separate chip just specifying the interconnects without worrying about what was inside the chip.  It kept me from having to think about thousands of flip-flops and LUTs all at the same time.  The simple circuits also gave me experience using the tools so that by the time I was doing something really complex I was not as lost as I would otherwise have been.

Alternatively, tell us what you want to create and see if someone will help you jump into the deep end of the pool. 

[Howardlong]

Thanks for the responses.

I don't actually have any particular plans, but I would like to at least be able to be confident that I could understand how to build something rudimentary from scratch.

My background is over 40 years in electronic design in digital design plus embedded, with plenty of RF if that's not a dichotomy. My age is a problem, getting the brain to engage in a different way of thinking ain't as easy as it used to be.

[tggzzz]

My age is a problem, getting the brain to engage in a different way of thinking ain't as easy as it used to be.

Don't fall into the ageist agenda: it was never easy. But the worthwhile things never are.

To some extent all that has happened is that you have forgotten how difficult it was, there are more distractions now, and you already know that you will never really get to grips with some topics.

Just don't ask me how I know all that.

[Howardlong]

My age is a problem, getting the brain to engage in a different way of thinking ain't as easy as it used to be.

Don't fall into the ageist agenda: it was never easy. But the worthwhile things never are.

To some extent all that has happened is that you have forgotten how difficult it was, there are more distractions now, and you already know that you will never really get to grips with some topics.

Just don't ask me how I know all that.

Indeed, when I learned to fly a very few years ago, I thought "How hard can it be, like driving with a few more knobs and dials". Watching 16 year olds going on their first solo after ten hours when I was well into the 20s wasn't great for the ego! But it did teach me that not everything's easy, and getting out of your comfort zone, while frustrating at times, gives you a much great feeling of accomplishment than just plodding on the same way you've always done. I'll get there, but FPGAs have been going on the back burner for a decade now for me.

[tggzzz]

I take some comfort from PJ O'Rourke's phrase "Age and build beat youth innocence and a bad haircut" .

As for dials - they all lie in various ways. Go gliding, and enjoy instructional flights where you instruments are all covered, and you have to fly by the seat of you pants. ? Pull 3.5g, enter spins at 1000ft, play with other aircraft in thermals, and just have fun!