How are FPGA designs typically transferred to PCB's?

cbmeeks

I really didn't know where else to ask this so I thought I would ask it here. :-)

I've been tinkering around with a small FPGA dev board I have (Mojo V3) and I'm starting to like it. But let's say I create some design with it for playing audio/video/whatever.

Eventually, I would like to transfer it to a real PCB. The problem with these development boards (even ones like Arduino) is that they make it "too easy" to get started...if that makes sense...lol

For example, the DIP40 Propeller, a small EEPROM and a PropPlug is all you need (minus passive components) to start building something.

FPGA is much harder. So, my question is, how have you guys transferred some of your FPGA designs to PCB's? Was it much more complicated than a Propeller circuit?

I read that you have to worry a lot about different voltages in FPGA designs. This almost makes me not want to use one because I don't want to get a college degree to build some cheesy FPGA designs. hehehe

Thanks for any tips.

Heater.

Never tried but all the FPGA's I have seen come in packages with millions of very small pins or ball grid arrays. I'd be worried about having to solder it down if I ever managed to get a PCB done for it.

Routing all the pins out of a ball grid array might need a many layer board. I would imagine you need a four layer board anyway.

Then you need those voltage regulators.

Then you need the configuration memory device. Although I gather there are FPGA's that don't neeed that now.

So, it's just a board design. With complications

Which FPGA do you have in mind? Some of the MAX 10 parts with on board config memory and EQFP packages might be easier.

cbmeeks

Yeah, that's been my fear all along about working with FPGA/CPLD. I'm a hobbyist and anything I build, I want other hobbyist to be able to build as well.

I would never be happy with a dev board as my "final product". Whatever I build, I would want to transfer to some permanent PCB.

Anyway, I have no idea what FPGA I should target...if any at all. I have the Mojo V3 board (Spartan 6 I think??). It's a lot of fun.

I could always build shields for it but where's the fun in that?

So, it's not really anything that's holding me back. The Propeller, 6502 and other micro-controllers are enough to keep me happy for a long time.

I was just curious on what you guys have done with FPGA beyond prototyping.

And yeah, messing with BGA devices doesn't sound like something I want to do! :-)

Seairth

I've been intrigued by the Altera MAX10 product line for homebrew designs. First, it's specifically designed to reduce external chip count (e.g. internal non-volatile memory), which can dramatically simplify board layout. Second, it comes in both FBGA and EQFP packages. The 144 pin EQFP package might be hand-solderable, but the pitch is still really small (0.5mm, I think).


I'd be curious to hear the lessons learned in making the 1-2-3 board.

cbmeeks

I've been reading about that as well. I did read, however, that it has 10k write operations? I assume the thinking is that you would prototype with a more "conventional" dev board and when you are finished, write the final code to the production builds.

But that doesn't quite seem right. Perhaps I'm not understanding that part. But seems like it wouldn't take long to brick a $8-$30 part.

Heater.

Do you mean the MAX10 parts can only be programmed 10k times.

Reconfiguring the thing every hour will take you over a year to reach that limit. So realistically it will live for some years. I'm betting it's probably good for twice that or more despite the spec.

At that price I would not worry about the 10K limit.

cbmeeks

Yeah, I guess that's true.

I'm pretty trigger happy when it comes to compiling/building, etc. I guess being a software developer by trade I'm used to compiling many many times in a day.

Heater.

I have no idea but presumably there is a compiler and simulator in the software dev package.

In my experience actually synthesising your design down to something that can be loaded to the device is very slow and tedious and not something you want to do so often.

Ariba

While you are developing you write the new design directly into the SRAM cells over JTAG. When you are happy with the design you can write it to the internal Flash - this takes quite some time. It's just like F10 / F11 on the Propeller.

Andy

cbmeeks

@Ariba

Ah, cool. That makes sense. Yeah, I could see running many many times during a day but only flashing it a very few.

Dr_Acula

Soldering fpga chips is a bit beyond me, so I tend to look for inexpensive pre-soldered boards and see them as large chips. Look for ones that have pins poking down eg http://www.smarthome.jigsy.com/fpga for a cyclone IV board. I've also done a design where the fpga goes in upside down for a cyclone II board http://searle.hostei.com/grant/Multicomp/index.html and these fpga boards are under $20 if you shop around.
Pre soldered boards are not much more than the bare fpga chip.

I also look for ones with two 10 pin sockets. One is jtag and one is AS, and as Ariba above points out, jtag is F10 and AS is F11.

Voltages are pretty easy if you just configure in software that all pins are 3V I/O.

MikeDYur

Check out tis kid from 2011, I am still trying to find out if he has them for sale.



http://www.thomasskibo.com/projects/pet2001fpga/

cbmeeks

Holy Smile! That's an awesome link! Thanks @MikeDYur

So I've been looking at the MAX 10 FPGA kits. Seems like a pretty decent one can be bought for $35 that has a lot of external goodies.

From what I understand about MAX 10 is that they are meant to reduce external chip counts and easier voltage supplies.

I think I could probably solder some of those chips too...if I were to build my own kit. I have a air reflow station so I should be able to handle anything that isn't crazy large...or BGA.