Hi, I'm very new to fpga, just came interrested in these things. The only problem I think I will have is the soldering. How to solder fpga's on the boards? I'm a student so I don't have money for very expensive machines. I want to be able to solder the whole range of packages of Xilinx spartan 3e, will that be possible with not to expensive tools?
Package types: Table 121: Xilinx Package Mechanical Drawings Package Web Link (URL) VQ100 / VQG100:
The QFP devices (VQ100, TQ144 and PQ208) are do-able with some practice with a standard soldering iron and some wick. Occe you get the corners soldered and all the pins aligned, you can solder the others easily. Dont worry about solder bridges, they are easy to pull off with the solder wick.
The BGA parts will probably not be within your capabilities. I have baked my own chips on before, but be prepared to ruin a few chips.
Your best best is to get a development board to experiment. If you need a standard alone module check out the Avnet Virtex 4 Mini module or the devices from Xess.
You might manage the flat pack stuff with a tiny soldering iron and a bit of luck, but the bgas are a different story. Even the 'professionals' mess up when soldering these.
- aligning the BGA almost perfectly to the grid on the PCB is tricky
- Soldering these needs an oven with a compatible heat-curve, which changes from lead-free to leaded components.
Just last week I had two s2 200s that had to go to be x-rayed, turned out they weren't soldered at the right temperature...
So I wonder what your motivation is for this? If you're starting out, try and get your hands on some starter kits, they're really good value, and you can learn loads from them.
If you're a student then go nag the electronics department, they should be pleased you're taking an interest, and might have some stuff lying around, plus if you could get some university-modules under your belt it would look nice on your CV.
Mind you, there's so much to be said for just doing things as you become interested.
I have to be honest and say that one of the big things that these Xilinx and Altera chaps are missing out on is the hobbyist engineer-type. If only they had that in mind when they built some of these starter kits, they'd sell them by the shed-load.
All VQ/PQ/TQ packages can be soldered by hand with decent equipment. Serveral Websites give sort of tutorials. You can also inspect all contacts easily.
All CP/FG package require much more expensive printed circuits boards, can only be solder with much more expensive equipment and are hard to inspect after solder. Nothing for the beginner.
But for the start, buy some development board, so you don't need to solder the FPGA at all.
or, a BGA part pre-soldered to a board with through-holes for attaching connectors, wires, etc?
spartan3 + power supply + SRAM on a stamp-sized board with breadboard type connectors on it? I know breadboard spartan3 dev boards are available, but it would be cool to get something tiny, cheap, and easily connected to. minus the switches, LEDs, 7segment LEDs, etc.
Contrary to others experience, I've done BGA parts pretty successfully in "hobby mode", as well as with professional rework equipment. I've developed a few "tricks" that help.
First, solder paste is very difficult to manage with hand placement. Instead, use water soluable flux on the board pads, and push a large solder ball over all the boards pads till you have a small shinny solder mound on each pad with a fairly uniform size to attach the BGA with. Clean the board well to remove the used flux, and apply fresh flux to both the board and the BGA balls.
Second, position the BGA part on top of the pcb pads. It's very useful at this point to have a silk screened outline of the package to center the part with, that is very close to the real package size.
Third, carefully place in a preheated convection oven, and bake. The correct time for this takes some practice and calibration of YOUR setup. Visually verify that the balls have a uniform "squat" on all four sides ... this generally means the parts wetted fine, and the part settled down on the balls at reflow temp.
Read the lit about recommended profiles for various packages ... you may not be able to exactly do those profiles, but you can come close with some practice.
I've had pretty good luck using several different table top forced air "convection ovens" with digital controls. These are fairly inexpensive, and typically can regulate temps within 10 degrees F or so. Their problem, is uneven air flow which may result in uneven board heating if you do not give this some thought. Two of the ovens I used, required setting the board on a soda can near the center of the oven, with the turn table removed. A third oven required fashioning a foil air duct to make sure the hot air flow evenly covered the board.
I suggest getting a non-contact IR thermometer with a spot capability ... these can be had fairly cheaply from a number of discount sources, including Harbor frieght. Using a salvage pcb of similar size and mass as your "test board" you can experiment with your "profile". I would suggest starting with the oven preheated to about 10-25F higher than your expected solder temp, quickly inserting your test board, and letting it bake for 2-3 minutes, open the door and quickly read several spots on the test board with the IR thermometer before the cool air drops the temp too much. Let the test board cool back to room temp, and repeat several times increasing the bake time about a minute each time. When you find the point where the board just reaches the solder temp, you can then program the oven to turn off 2 min after reaching that temp, and slowly cool the board back down without thermal shocking the BGA.
Another process, is to pickup some slavage BGA parts with a similar package, and make a test board with the on die thermal diode brought out to test points, so you can run the wires out the door and monitor the die temp as you develop your thermal profile. You can also do this with your own board and FPGA's. You can also epoxy attach several thermal diodes to a test board, and monitor the profile in real time at several points.
Pickup some Solderquik reballing preforms for the parts you plan to use. That way you can clean off your mistakes, and reball the parts to try again.
The one thing you do not want to do, is get the newer BGA's too hot, or thermal shock them with cold air. Older parts in BG432 and BG560 packages are much more forgiving. I would suggest learning this process with XC4K, Virtex, or Virtex-E parts in BG432/BG560 packages, and once mastered move on to newer high density packages.
routinely solder the QFPs by hand. Your best bet is to get a solder paste syringe from DigiKey and a hot air pencil. You can also use an iron, but the paste melts really well under hot air. Apply a very thin line (just enough to avoid separating the line as you draw it) just inside the outer edge of the pads, then plop the part down. Using an iron, solder a couple corners to make sure the part doesn't move. Then use hot air.
We do not do BGA by hand -- too costly. Although most contract manufacturers consider BGA more reliable and easier to do than QFP because they self-center better. Of course, with the new RoHS stuff, many will need to do costly temp profiling until they nail the processes.
Of course, you could also buy smallish integration modules like what we sell. They're more expensive than just a raw chip, but will save you a lot of time. Ours also include a software API and USB connectivity which is more than just the "chip on a board" eval boards have.
We have the Metcal irons and a drag-hoof tip. Although I've only had success with paste and the hoof tip, I've seen techs that are able to wet the tip and drag it on the pins nicely. Definitely something I don't have the hands for.
But the hot air pencils are pricey -- around $200-$400 on eBay for a Hakko 851.
Our XEM3010 product is similar to that, though using 0.8mm connectors to another board. We sell the BRK3010 as a breakout board to bring the pins out to lower-density breadboard sort of thing.
The flat packs are solderable by hand. I do various chips, including FPGAs with up to 0.5 mm lead pitch. I use solder paste, and a soldering iron with a VERY fine point. I work under a stereo zoom microscope. You may be able to use one at your school, and bring the rest of the gear to it. I specifically have done the TQ144, which is 0.5 or 0.65 mm spacing, I forget. The BGAs can't be soldered by hand, because the balls are between the board and the package. You pretty much need an X-ray system to check the soldering, too. Also, note that some of these larger BGA parts are REALLY expensive.
Weller has some micro-pencil irons like the EC1302 (vintage) and the WMP (current). They have conical pointed tips for all of these, that essentially go to a real point. You wipe the iron gently down the row of leads, touching both the end of the lead and the solder pad at the same time. If the right amount of solder paste is used (very sparingly) the surface tension will break the bridge of solder as you work down the side. There will be a bridge between the two leads that the soldering iron is on at any one time, but the iron pulls it along as you move. The alignment takes a couple of minutes, then the soldering goes very quickly.
It really won't work. You could put the necessary decoupling capacitors on the adaptor, but what DIP would you use? The original post was talking about
Maybe not quite what your are looking for but we have a new add-on module that can be used as simply as a XC9572XL CPLD holder. It is actually an IDE interface but can be made to do a number of things by programming your design into the CPLD. Pinned out on 0.1 inch pitch is relatively easy to use on a stripboard. Picture on our Raggedstone1 webpage if you are interested.
John Adair Enterpoint Ltd. - Home of Raggedstone1. The low Cost Spartan-3 Development Board.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.