Do you think is it possible to build at home a FPGA development board?

[Elektronman]

Due to the high cost of typical FPGA development board, I'm thinking about buying the FPGA die and build myself the needed board to program it with all the I/O. What should I need to do this beside the soldering station, copper, acid and a good PCB software ?

Thank you very much

 

[tsvetelin_velkov]

First find a FPGA chip that has package which you can solder. Big FPGAs are packaged in BGA, this package you will never solder at home. Then just found a development board that is using your FPGA chip and copy the schematic and maybe the layout. Add some additional components if you feel you may need them in the future. Don't forget to add some breadboard area on your PCB

I did one project like that, some time ago. It was for Xilinx CPLD, one of the "Cool Runner" CPLDs. Cool stuff, and need no external memory chip, because the memory is already in the chip

 

[alexan_e]

it would be very hard to route in a single side even if the board has minimum components, so have you made a double side PCB before?
You will also need very thin tracks, about 10mil, have you done anything similar?
Also have you soldered an SMD with a small pitch before, it is kind of tricky

I was only able to do a single side pcb with just the FPGA and all the pins going to headers so that cables can be wired, I assume that this is not what you want.




Alex

 

[jpanhalt]

There is a site dedicated to FPGA. I suggest it here only because of the difficulties for DIY outline by Alex. Maybe there are some solutions there.

fpga4fun.com - Site information

John

 

[mrflibble]

Possible? Yes.

Practical? No.

Economically sensible? No.


If you are only going to make a few of them for your own use, then you are far better of buying something.

Yeah yeah, all the cool boards are mucho $$$. Well, you are in luck then. Because the mucho $$$ boards with all the latest stuff on it is not what you will be capable of making at home anyway with a 2-layer no-bga-allowed setup. So you are really comparing "what can I do myself with a 2 layer board and non-bga packages" to "what can I cheaply get in terms of ready made boards".

Ready made boards can be had for $50 - $100, a bit depending on what you want. I challenge you to match that in materials used to get a working board. And that's not even counting the time.

Not trying to keep you away from a neat learning experience, just pointing out that you may want to reconsider building vs buying. The main thing you would be gaining is the experience in building it. if that is your main goal, go for it. If not, spend $50-$100 on a board, and redirect your creative time & energy into doing something with that.

There you go, $50:
Papilio One 250K Butterfly One (Papilio One) board fully assembled with a Xilinx XC3S250E and 4Mbit SPI Flash memory. [BPC3003-2.03-250k] - $49.99 : Gadget Factory

A nice list of boards:
Cheap FPGA Development Boards | Joel's Compendium of Total Knowledge

Another I recently learned about:
**broken link removed**


Personally I'd rather put my time into making things that are not so readily available, and buy the stuff that is readily available at a reasonable price. But whatever you choose, have fun!

 

[keith1200rs]

For around $50 I bought an FPGA development board WITH an imitation Altera USB blaster from ebay. It was from Hong Kong and works fine. In the UK the Altera USB Blaster costs £200 alone. Just check for good feedback on ebay before buying.

Keith

 

[tsvetelin_velkov]

Well, Elektronman,

By reading the posts I have to say that the suggestions really make sense. I am not going to say - making such board yourself is impossible, or so hard to do, or anything like this ...

There are different style of packages, and you certainly can get such that you can work with in your home lab. Two side PCB is not that hard to design, but keep in mind that you have to order it somewhere to be manufactured by professionals - such small tracks and small distance between tracks and vias cannot be made by hand or using printer and iron home brewing process.

I did one board my self, I know at least 2 or 3 similar boards made by my friends. All they work well. All are not professional, but are good / simple starter boards, however going to use it professionally is another story ...

Selecting a ready board is I think the best option. Then you will use time and energy really to play with the FPGA and to try to build the board. If I can share my opinion - sometimes it is so frustrating to try to make such a board yourself, so by the time and money you spend to get the board ready - you lose your will to play with it, which I think is your final goal.

If you don't have experience with FPGA, try to find a simple board, without much fancy options, just you need the FPGA, the memory, the programming port and some display option - leds, lcd display. By this you will make some experience and later when you know what exactly you want to do you will buy a board that suits your needs - it could be expensive, yes, but if you compare the price for parts, the know-how needed to make it, the efforts to route the board and to interface all components together and the manufacturing equipment that is needed to make it ... then the price of the ready board is actually good.

I was working with DSP boards combining fast ADC, DSP processor and FPGA custom logic, VGA port for interfacing - all interfaced together ... This is I think a good example of complex development platform. Making this at home really make no sense.

 

[Elektronman]

ok, thank you for your anwers. What I'm trying to build is a HDMI recorder that processes a video stream from HDMI source. I have thought about 2 possibilities to do this:
1- By using a FPGA board that comes with HDMI input. In this way I could configure the FPGA (for instance Spartan 3A) as a HDMI receiver using the xapp460
https://www.xilinx.com/support/documentation/application_notes/xapp460.pdf
This would probably be the easiest way as it is only a matter of programming the FPGA. The only problem is: such FPGA dev boards are expensive.

2- By using a separate HDMI receiver like AD9398 soldered on a homemade PCB and then connecting it to a FPGA board that receive the already decoded RGB signal. This solution is cheap but I may encounter some difficulties in handling the AD9398 chip, soldering and dealing with such high frequency signals (HDMI video).

3-Buying a FPGA board and trying to solder a HDMI port on it. Then configuring the FPGA to behave like a HDMI receiver assuming the FPGA is able to do this.
This is probably what I'd like to avoid because it is not guaranteed HDMI port connection will work due to the particular LVDS signals involved when dealing with HDMI, and I may as well finish up damaging the board...

so from your comments it is clear that I should give up the idea of homebuilding a FPGA dev board. Probably the best solution is using a separate HDMI receiver that decodes the signal and a FPGA board able to process such data throughput (8bit line at 150 Mhz).
Do you guys have some suggestions?
Thank you very much

 

[mrflibble]

... Then configuring the FPGA to behave like a HDMI receiver assuming the FPGA is able to do this.

How about reducing the set of assumptions? If you have no idea yet what your design is going to look like, and you want to do things on the cheap then I'd say first do a rough design. Do some prototype vhdl/verilog modules, do some simulations, see if it is feasible at all. Sticking with your spartan-3a example, you can download a free evaluation version of ISE to do all these initial tests. First get some idea of what size device you will be needing to do your HDMI recorder, and then buy it. That reduces the chances of being too cheap while buying that exta cheap board only to find out your design does not fit. And then having to spend more money to get that other board that will fit your design.

Regarding that HDMI recorder, you shortly described the input & decoding. What is the recording part? Stream to network for external recorder? Dump to attached USB disk? etc... Also, what bitrate hdmi do you intend to support?

Quick google to see if this had already been done turned up this one:

fpga4fun.com - FPGA Portable video recorder?

That includes some nice points to think about.

And stupid question du jour: are there no off the shelf items that already offer the functionality that you are looking for? It would seem to me that media recording is a high volume market. Read: cheap stuff!

 

[FvM]

I basically support the thoughts of mrflibble. In addition, I stumbled upon the HDMI recorder point. Record on what media? All usual harddisk recorders are using video compression. Did you also consider the FPGA resources and external memory required for it?

 

[lucbra]

when I'm facing problems like you describe, the first thing I check is ready made boards. Second, I'm looking to use an FPGA board (there are a lot!) and attach an interface to it (in your case the HDMI interface). If you can't connect the one to the other directly, you could build a conversion board on a double sided PCB with some voltage translation, interface translation, ...

This is usually the cheapest way to go.

With respect to the HDMI recorder, you will face a lot of problems that you will need to overcome. FvM and mrfibble allready pointed out some of them. You will soon find out that you'll need a lot of I/O to the FPGA to keep the frequency low, bandwidth of the storage media are also a big issue, then comes compression (ready made IP?), ...

 

[tsvetelin_velkov]

I also support the opinion that such application should be available already. If your final goal is to have HDMI recorder, you better go and buy it from somewhere, it will be definitely cheaper, 100% ready and working and you will have it now.
If you want to make it yourself for educational purpose or just for fun or because you have some business ideas, than I would say "Welcome", it will be a lot of fun. Such things are usually build using ready components, as mentioned already - ready made IP blocks. You have to integrate them together, to make the interfaces, configuration, etc. I assume it might not be possible to simulate the complete design, but once you have all blocks you will have an idea what kind of device you need ( clock speed and volume, i.e. number of gates ) and respectively what board to buy or build yourself. Keep in mind that you need to have some spare gates for interfacing and possible futures, why not some extensions in the future

 

[Elektronman]

OK guys.. there is a big new: at last I decided to buy an Atlys development board from Digilent, It costed me 300$ a little more than what I intended to pay for my project.. but after trying looking for different solutions, and thinking a lot about it and asking around.. I decided this would probably work best for me: this board has 4 HDMI ports and has basically everything I need for my project of my portable HDMI recorder.
Thank you for your help, I'm glad I will not risk become crazy trying to solder LQPF or BGA chips on homemade boards.. LOL

 

[FvM]

Atlys should basically work. But the only general interface of this board, that would be able to stream the video data (after compression), is the ethernet port I think.

 

[mrflibble]

Indeed. That's why in the other thread I posted this here link.

Check out the bit on gbit ethernet. I use that one myself and like it. Basically you will have to change the FSM to suit your own needs, but the verilog code as provided by Joel is a very good starting point IMO.