Finally first picture of Darnaw1 our PGA style FPGA board is here here
formatting link
More information on pricing and spec in the next couple of days will appear on the website. Those with eagle eyes can work it out the spec from the picture.
First shipments will have 16Mbit SPI flash to allow programming of the FPGA but also to act as a code store for processors like MicroBlaze implemented within the FPGA. There is also SDRAM on board. Small numbers of this product will be available to ship next week.
We would be interested to have feedback on this product and what you like, and what we could improve on this product and the related series of products we have planned.
Hi John, There was some bloke on here a while back asking about this sort of product, I pointed him at your DIP design. Craigsomething? So, you reckon it'll work in a wire-wrap system!? I wish you luck supporting that SI hellhole. :-) Although, can you buy wirewrap boards with ground planes now? Also, the capacitor packs you're using, are they interdigitated ones? I'm interested on what other folks are using for bypass networks these days. It'd be nice to find someone else who's had experience of the X2Y caps that were discussed here a few weeks back. Finally, I hope you ripped off Xilinx's sparse chevron thingy for your pinout. It'll be a big help for your SI. Cheers, Syms. p.s. 'ripped off' 'were inspired by' ;-)
I don't know about wirewrap boards with ground availability but one of our thoughts with this product was the user that likes 2-4 layer low technology pcb boards they can assemble themselves. With this product we allow them to stay with the board technology they like but they can have a high performance bga based FPGA in their system. I have seen wirewrap sockets I believe from Mill-Max or Precidip if someone wants to do that.
The capacitors are just conventional arrays as used throughout our product range. We have gone to the effort of making the board itself resiliant in terms of SSI but there is a limit given the target market and where it is likely to be used. If someone is after very high performance I would always integrate the FPGA into a complete board. We do often take one of our standard products as a starting point and produce a customer specific design from it.
The Craignell modules are sort of complimentary to the Darnaw range but are much smaller in I/O and differ in other ways. These are also now shipping. Darnaw1 can be a very serious processing capability with it's I/O resources and onboard SDRAM and FLASH. We may uprate the SDRAM to DDR2 in issue2 but that is still to be decided. For those familiar with our OVERCOAT series it is also possible to use this technique to stack DARNAW1's if you have a reason to do so.
I am hoping that a lot of products we have planned will make it to market this year given our slow output of our own product range last year. The engineering team has now grown very substantially and some resource is now going back to our own product developments. I think some of the things we are doing right now might even surprise a few people but I will leave the detail as something of tease and a surprise to come.
John Adair Enterpoint Ltd.
picture of Darnaw1 our PGA style FPGA board is here here
On Apr 20, 8:31 am, John Adair wrote: I don't know about wirewrap boards with ground availability but one of our thoughts with this product was the user that likes 2-4 layer low technology pcb boards they can assemble themselves. With this product we allow them to stay with the board technology they like but they can have a high performance bga based FPGA in their system. I have seen wirewrap sockets I believe from Mill-Max or Precidip if someone wants to do that.
John I've used wire-wrap boards that had plated-thru holes and copper planes on each side. They're made by Twin Industries and I bought them from DigiKey. I use the bottom plane for ground and the top plane for VCC. On the bottom I solder a 1206 cap from the power pin of a chip's socket to the gound plane. I run my wires between pins as if I was laying out a PCB and I DO NOT bundle the wire-wrap wires! I also make sure that I run the wires up against the ground plane to reduce wire inductance and ringing. I've had no problems running
40MHz clocks with 74HCT logic and Xilinx CPLDs (on home-made adapters) with this method. But I think that wire-wrapping a PGA device would be pushing it.
What you see on the current photo are the solder joints of a double ended pin so the pins come out the bottom as currently done. Given we don't need to meet a specific pinout we could take them out the top if someone need one that way or even have a socket + pin assembly like we use in out OVERCOAT arrays. We can also make these with an un- populated header and this could be a lot cheaper for high numbers used in production say in a low pin count application.
The PGA pin header is not cheap either to purchase or even to assemble so I would recommend anyone with a specific interest that would want a a few+ should come and talk to us. We can deal with Bill of of Materials variants for 5-10+ shipping units and can even be economic on a customised pcb variant of any of our products in fairly small numbers (say 25+ off on a small product like this). Our development board products really are only a demo of what we can do and supply ideas to be used in a mix and match fashion on customer specific designs.
The biggest issue with wire wrap is the time to do it reliably. The on- board regulators and decoupling make the board less prone to poor performance of input power supply. It most commercial applications it is probably cheaper and quicker to knock up a low tech pcb to support the module if you need a large percentage of the I/O. But for those that like wire wrapping or don't have an easy path to making a pcb wire wrap is a possibility. I have seen some very complex boards made in wire wrap but not in recent years due to the economics.
John Adair Enterpo> > I don't know about wirewrap boards with ground availability but one
A Virtex-4 based one is on our thought path. Probably a FX12 as we use this part in a number of places already although as yet not yet on our development boards. This is also a bit of a market tester to see how it takes and what people like, or not, about the concept.
Our first batch is relatively small but now it is proven we can more to large numbers very quickly subject to silicon purchase which is usually less than 8 weeks on Spartan (we already have some unallocated stock). If you have serious interests in this or a V4 version there are discounts for educational, or volume, usage.
Looks very interesting. First I was scared about the many pins which makes it difficult to put it on a simple double sided PCB where you can only route one wire between two pins. But I think the solution is, to not use a complete 21x21 socket on the PCB but put single socket pins only on that DARNAW pins which are used in the design. This way you only need holes in the PCP for the actual used pins (and you even can mix normal pins with wire wrap pins).
Do you plan to make a version with a SRAM instead of the SDRAM (something like the 1M x 16 Static RAM CY7C1061AV33 from CYPRESS)? Would be much easier to interface. And if there would be a Flash Memory on the same address and data bus as the SRAM, this would be an ideal platform for experimenting with simple processor designs.
It is our intention that that you only need to fit the pins you want or need. On the technology level we use on development boards we can squeeze 5 tracks between pins on this pitch but being more realistic for the likely usage you would get two maybe three between pins on a professionally manufactured, low tech, PCBs. So if you do only need say 60 I/O then you can probably get away with something like 100 pins fitted when you account for power etc.. Using say the out rows only would be an easy PCB layout and quite possibly a 2 layer implementation. As a benchmark I always like to point out that our product Raggedstone1 (low cost development board) has the entire I/O(264) used of a XC3S400-4FG456 and supporting 7-8 power segments on a 4 layer board and thats a 1mm ball grid that we dealing with.
At the moment we are not looking at SRAM because it would not support our design target of running MicroBlaze on the module very well. Also the Spartan-3E has some SRAM already available internally. If someone comes to us with a project that needs a build of a reasonable number of modules we can probably do a new version for that need. To do that would take my team less that a day to change the SDRAM for a SRAM in the design and then we just down to manufacturing time and cost.
Longer term there may be an intermediate module range that sits between Darnaw and Craignell ranges but we will wait and see how popular these new releases are. There are bigger Craignell's being planned already so these may be of interest.
Just got the boards together with a parallel port download cable. Is there any special download software or does this cable work with Xilinx software? Is there some ready to run demo for the Darnaw1 board available (just some output to the LED'S maybe). Didn't find anything on your Website.
There is no FPGA module on the market with the following characteristics :
- FPGA large enough to fit Microblaze and user cores (the Darnaw has it !) - Big, FAST RAM (the Darnaw has it !) so we can use xilinx cache links and DMA core and not be limited by a single slow SDRAM chip - CHEAP (your price for the Darnaw is good ! much better than competitors) - Easy to use connector : I think the PGA socket sucks. I would prefer a module shaped like a laptop SDRAM stick, sell the connector with it, much easier to plug and use, easier to route the PCB for it, better signal integrity. Plus, it's cheaper to manufacture for you ! - Ethernet PHY to use OpenCores MAC core (damn, Darnaw doesn't have it, but you sell some as extra modules, so it's OK)
So, here is the feedback ;) Can we get more info about the upcoming modules ?
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.