Following our recent Swinyard1 (Virtex-4) release we are now looking at the Swinyard2 module concept which will be based on a middle end Virtex-5 (initial XC5VLX50 and others) that will be supported by our Broaddown series of main development boards. Bearing in mind this a small module what features would you like us to put on this module?
and what did you all think of the general Swinyard concept?
This is you chance to influence what we deliver to the marketplace so do let us know.
John Adair Enterpoint Ltd. - Home of Broaddown2. The Ultimate Spartan3 Development Board.
I just checked out your Swinyard module. I like the smal formfactor. What are those connectors you use on the module? Do you have a Manufacturer/Part #? I like having the BGA parts on a PCB that I can fit on my PCB that I can hand solder. Are these connectors do-able by hand (I can generally do any QFP or QFN Package, but no BGA)?
The connector is a GFZ family from Samtec and are totally solderless. You will see a number of nylon bolts holding the board onto the host board that provides outside world services like supply voltages etc. The GFZ is a sprung contact that when the boards are tightened up together make contact. The simple version of the Moel-Bryn socket that we currently use has 2 GFZ connectors of 40x10 ways each so we get a high number of I/Os available and have a wide range of supporting features.
The Moel-Bryn socket pinout will be available under a license/NDA. The license won't necessarily cost anything unless you want to make your own modules. One of our intentions is that these modules could be used to bring high technology to what are relatively low technology host boards as well as the main development board use.
John Adair Enterpoint Ltd. - Home of Broaddown2. The Ultimate Spartan3 Development Board.
This is a truly excellent idea. For some while I've been frustrated by FPGA development boards that have a ragbag of low-tech functionality and connectors, pushing up the price just for me to get something I could easily provide for myself. What you're offering fixes that - for many users, even in small-scale production, the FPGA is the only part of the system that needs fine-line PCBs and non-trivial assembly techniques, and a small FPGA-only plug-in module is the right answer.
My only concern would be that you've gone for a fairly high-end FPGA so the entry cost is quite high. Fine for some purposes, but pricing itself out of a potentially useful market for others. Have you any plans for a significantly lower-cost product with a similar overall approach?
Thanks
--
Jonathan Bromley, Consultant
DOULOS - Developing Design Know-how
VHDL * Verilog * SystemC * e * Perl * Tcl/Tk * Project Services
Doulos Ltd., 22 Market Place, Ringwood, BH24 1AW, UK
jonathan.bromley@MYCOMPANY.com
http://www.MYCOMPANY.com
The contents of this message may contain personal views which
are not the views of Doulos Ltd., unless specifically stated.
The big issue with the GFZ connectors is that they themselves are not low cost and tend to set a base cost for the module that takes it out of the very low cost sector. However we have on our roadmap a FX12 module that is going to use our DIl style connectors at the lower end. We are also considering the same as a Spartan-3 solution. Other variations of the theme we are considering are a module that breaks out a BGA usable PGA on 2.54mm. We have already done that on a non-FPGA part for a very fine pitch BGA and in a reasonably economic way for one of our customers.
The Swinyard2 board will cover a cheaper sector than the Swinyard1 that you can see on the website. Also if you are considering a small production run there is room for discounting on Swinyard1. We are pricing on small batch assembly costs due to the price of some of the silicon. We don't plan on hold hugh stocks of these boards as there are 18 possible fits of FPGA on Swinyard1 by the time you consider all sizes and speed grades available. We are planning to hold stock of a small number of variants - the LX40 and SX55 initially more if we see the a steady market. As these FPGA are common with our Broaddown4 product we may also have a quick assemble to order capability but that depends highly on stock we have at any given time.
Putting a commercial spin on what we do we can usually come up with a cost effective solution for products in fairly low production numbers. I won't put exact numbers on it as I'm sure someone will disagree about what is considered cost effective but we commonly provide custom solutions for year product volumes of 10-25 units. By the time we hit 100 off batches we aren't China style prices but we can definately give good value in the European and N America context.
John Adair Enterpoint Ltd. - Home of Raggedstone1. The Low Cost Spartan-3 Development Board.
I would **LOVE** A module with a 2.54MM PGA form factor. Something with just the basics (decoupling caps, etc). (I am envsioning something that looks like an old socket 7 intel chip)
I work in research where we do alot of our own PCBs and hand-solding, but cannot afford BGA. I only do 2 or 3 boards for a particular project. This limits my ability to use FPGA technology. I would love it if Xilinx still made things in PGA packages...... Alot fo the develop boards are big and clunky. I just want a module. this also makes thngs a bit more modular so I can swap things out later.
Cost really isn' tthat big of an issue for me. I need prototyping ability.
I think you have the options pretty much on target. Other than DDR2 memory, the supporting board can pretty much fill in all of the required functionality.
By the way, do these GFZ connectors require gold plating on the supporting card to make proper contact?
Well hang on a bit and you may be pleasantly surprised. We have had this idea around for a while and done something similar for a 0.5mm pitch BGA already. It is just a matter of finding a spare day or so from my team to do it.
If we do this design ideally would like to line up with a common;y available ZIF PGA, or standard, socket allowing easy and low risk attachment to a board.
I think a flat board finish is good thing and one that does not have any nasty oxidising habits. We have done a reasonable amount of testing on some of Broaddown2 boards that have been around our lab for about 2 years. Those particular boards have a siliver finish and not the hard gold finish that would have been optional and whilst there is an obvious level of aging on the finish on the ones we have tested there don't appear to be any problems in any connections.
What I would not do with a GFZ c> I think you have the options pretty much on target. Other than DDR2
"John Adair" wrote in message news: snipped-for-privacy@iris.uk.clara.net...
Well, I would want a board that can mate with 0.1" drilled prototyping boards that I can build experimental circuits. I'd like to be able to connect a logic analyser with readily obtainable IDC cable instead of specialist probes. I'd like to be able to buy just the minimal FPGA board and buy plug-in modules for the features I choose. I'd like a board that can slot into racks for 100mm wide cards.
See:
formatting link
for a list of features and reasons for them.
Your Swinyard module is fine but seems like it needs a specially designed PCB to mate with the solderless connector. That cuts out customers who don't want to get such boards made up.
I would concur with some of the other posters. What would be really nice is a selection of decent size Spartan / Virtex parts on a PGA board that could be plugged into a socket 370 or socket 478 ZIF. These boards should have plenty of bypass caps on them - and it would be nice if these boards had a platform flash on them - making them, in essence, very nice CPLD's. If done properly, you could even strap a standard heatsink/fan unit on top!
There are companies out there that have BGA -> PGA adapters, but then you have the little trick of getting the BGA on the board - a non-trivial task at best. I would definitely be interested in a pre-fitted, tested, PGA module with a Spartan 3/3E or Virtex II/II Pro BGA mounted on it.
An ideal board would have:
1) Standard, commercially available socket pattern for the PGA.
2) Onboard 3.3 -> regulator
3) Bypass capacitors
4) Platform flash
5) Optional DDR SDRAM pads (or even devices). These could be on a "wing" that extends out past the socket like the Macintosh G3/G4 processor modules.
6) Reasonably high capacity XIlinx or Altera FPGA.
Actually, if you could just make a G3/G4 processor module with a Virtex II Pro, you might be able to use an old Macintosh AS a development board!
Have you looked at our main development board product range. Raggedstone1, Broaddown2, Broaddown4, Hollybush1 can support stripboard prototyping. Headers on there even have 3.3V and 0V. Hollybush1 goes better and even has
0.1 power header that is aligned on 0.1 inch with left and right headers (containing 115 I/O on 0.1 inch + power). The power header allows a user to set a power supply for 1.2-4.3V, 1Amp, by simply connecting a resistor and also has 5V available. 3.3V also available on left and right headers on Hollybush1.
Not to totally forget MINI-CAN that has a 40x2 0.1 inch header with both 5V and 3.3V for add-ons.
John Adair Enterpoint Ltd. - Home of Broaddown2. The Ultimate Spartan3 Development Board.
Hello John ! Are you related to the famous "Red Adair" ? He put out an oilwell fire here i Norway some 20 years ago or so. Anyway, I wish you luck with the new dev. board, I only hope is has a better life ecpentancy than the SP305, wich i purchased, only to realise that the board went out of production after a few weeks, and no support from Xilinx...
As to longevity I expect the boards you see on our website now will probably be available to buy still in 5 years time. OEM adoption and use of a number of these boards has virtually guaranteed small batches still being made in that timeframe. I don't expect they will be quite as popular by then having been replaced in the main by I dare say Virtex-5/6/7 and Spartan-4/5/6 replacements.
In another era I repaired boards from a main-frame computer. Some boards had modules that used a similar construction like the GFZ connector. Early modules used some sort of silicone with carbon traces, later on these where replace by silicon with gold wires. Still, both versions needed cleaning every now and then.
--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl
A very delayed projected (Broaddown3) will come close to what you want. This project is awaiting resource but those darned things called customers keep giving the team more and more work to keep them busy. Generally we aim to run our team here with on a 50% customer loading but for many months that has been closer to 100%. However some more heads are now allowing us some man time to do projects like Swinyard2 and Tarfessock1 and I expecting them to make an appearance in Q3 albeit at the end of this quarter. Broaddown3 is contending for a slot in early Q4 but that depends in some extent how well our current recruitment drive works out. We also have a number of interesting concepts floating around that could could take precidence. A refreshed Broaddown1 might even be back on the list for those that know of it's esoteric capabilities.
Swinyard3 has drawn the shortest straw in getting resources because we are not as yet sure of the economics to make it succeed as a saleable item. One of the current big barriers to PCI-E is the price and the size of FPGA core. There is not a lot of choice out there in cheap PCI-E cores unlike convertional PCI. We are considering the switch to something like a PLX solution or even a preprogrammed essentially fixed front end FPGA to resolve these cost issues so the concept may get moved around a little.
As to standard board interconnections have you looked at how we do the build option "OVERCOAT" available on our Raggedstone1, Broaddown2/4, Holybush1 and just maybe MINI-CAN. We can built parallel/serial arrays of boards as big as power supply will allow. It's a cheap and simple way to do it and we can do it in situe in a PC motherboard (subject to number of slots). We also have a dumb PCI backplane on our list of things to do for array processing.
But as always if you have an interesting and viable project do talk to us direct. We do announce things up front that we are doing but there are projects we don't talk about and just occasionally a customer requirement, and maybe some money, may tilt which project gets done first.
John Adair Enterpo> I would like a board designed from the ground up to do general software
Project now underway. I will post seperately when I have expected dates for something being really available for sale. It is Spartan-3E based this time.
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.