Site Home Archive Home FAQ Home How to search the Archive How to Navigate the Archive
Compare FPGA features and resources
Threads starting:
Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
If you allow for oscillators external to the FPGA, there are many available. Digikey shows DIP part at 80 MHz for $10 range (pg. 230). If you build your own xtal osc, use a third overtone crystal where third harmonic is the signal of interest. Me - I'd use the stock oscillator and move on. Make sure there is a decoupling cap on the power pin and a 51 ohm series res. on the signal line to avoid ringing/emi problems. Stephen Phillipson wrote: > > Martin Neilan wrote: > > > > Hi everyone, > > > > I have a need to implement a Digital Phase Locked Loop on and FPGA using > > V-System as the VHDL compiler and Xilinx as the synthesis software. If > > it is not possible, I will need to look at other ways to determine a > > suitable clock for an incoming serial stream of data (of undetermined > > frequency) and keep this clock in sync. > > > > However, in order to do this the second way (without a DPLL) I would > > require a quartz clock crystal running at preferably greater than 80 > > MHz. Does such a thing exist? If not, is there a way of creating a > > faster clock using some clever coding from a slower quartz crystal? > > eg, using a 20 MHz quartz crystal and creating a 40 Mhz clock from it? I > > have a feeling this might be possible but has anyone else done it? > > > > Any suggestions or help is greatly appreciated! > > > > Kind Regards, > > Steve Phillipson > > - A confused undergraduate engineer! > > Oopps! Had my details set wrong in Netscape :( > Correct details: > > Stephen Phillipson (hslcomp@iinet.net.au) > > Sorry!Article: 9701
In article <351FF25A.BBC9DBE0@eng.efi.com>, Gareth Baron <gareth.baron@eng.efi.com> wrote: > > IMO, the removal of the dongle license is irrelevant. What can you use > the Xilinx s/w for other than developing Xilinx FPGAs ? Xilinx can make > good revenues from the silicon itself. I have always mentioned to the EDA > companies this fact. What are they selling, Silicon or Software ? > > I'm not against using dongles, but I do disagree with the maintainence > deals that go along with them. You are tied in forever to buying their > "Upgrades" (especially when they re-release new silicon!). I agree. I have talked with Altera about this many times. It is not until more people refuse to pay a 15% annual maintenance that things will change. I am willing to pay for added features [that I want], but I see no reason I should pay more to add the ability to use parts I am going to buy from them anyway. Scott Taylor, DSP Systems Inc. -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 9702
In article <3521143D.30B7@e-technik.uni-rostock.de>, Hagen Ploog <hp@e-technik.uni-rostock.de> wrote: > > Andrew V. Nesterov wrote: > > > > In article <351813B4.9DE25395@computer.org>, Vitit Kantabutra <vkantabu@computer.org> wrote: > > > Thanks. Yes, I've written a paper about it and submitted it to ICCD '98 a few > > > days ago. I'll send you a copy via email. I just didn't want to send a paper > > > to a newsgroup. > > > > > I'd also like to take a look at it. Please send it to me too. > > Thanks, > > > > --Andrew > > Please send me a version of your paper too. > Me too. Might as well post it now, or at least put it somewhere we can get to it? Or else you'll end sending to it dozens of people. -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 9703
I'm working in Altera FPGA's.I'm want to make a board configurable from the computer, but I need the bitblaster or byteblaster. Somebody knows how this devices are make?? Felip Vicedo Roman feviro@eug.upv.esArticle: 9704
Daniel Alley wrote: > > If you allow for oscillators external to the FPGA, there are many > available. Digikey shows DIP part at 80 MHz for $10 range (pg. 230). > If you build your own xtal osc, use a third overtone crystal where > third harmonic is the signal of interest. Me - I'd use the stock > oscillator and move on. Make sure there is a decoupling cap on the > power pin and a 51 ohm series res. on the signal line to avoid > ringing/emi problems. > > Stephen Phillipson wrote: > > > > > Hi everyone, > > > > > > I have a need to implement a Digital Phase Locked Loop on and FPGA using > > > V-System as the VHDL compiler and Xilinx as the synthesis software. If > > > it is not possible, I will need to look at other ways to determine a > > > suitable clock for an incoming serial stream of data (of undetermined > > > frequency) and keep this clock in sync. Just HOW 'undetermined' is the frequency ? All PLL designs have a catch and hold range, and this is normally kept as small as possible to improve noise. If you can swing over > 2:1, you also need to ensure you do not harmonic lock. For PLL info, look at 74HC4046 (vco), and 74HC297 ( digital Filter ) > > > > > > However, in order to do this the second way (without a DPLL) I would > > > require a quartz clock crystal running at preferably greater than 80 > > > MHz. Does such a thing exist? If not, is there a way of creating a > > > faster clock using some clever coding from a slower quartz crystal? > > > eg, using a 20 MHz quartz crystal and creating a 40 Mhz clock from it? Clock Doubling is reasonably simple, an XOR gate and delay line give a pulse per XTAL edge. This CAN have phase jitter, which depends on how close to 50.00% the CLK duty cycle is. Going above that is harder, as you have to 'invent' time information, and you get even more phase jitter, and become very process dependant. If your regen CLK is low ( < 2MHz ), then a 80MHz clocked phase walking sync system is probably simplest, and gives reasonable noise immunity. - jg -- ======= Manufacturers of Serious Design Tools for uC and PLD ========= = Specialists in Development tools for C51 cored controllers = Leaders in Rapid Application Development SW for C51 uC = Ask for our Controller & Tools selector Guides = mailto:DesignTools@xtra.co.nz Subject : Selc51ToolsArticle: 9705
satish_me@hotmail.com wrote: > > Hello, > I am intersted in putting designing FFT and DCT on FPGA. The same I want to > implement just like Higher languages.Please help me how can I represent > Floating point on FPGA. > Thankyou for your valuable response > > -----== Posted via Deja News, The Leader in Internet Discussion ==----- > http://www.dejanews.com/ Now offering spam-free web-based newsreading Satish, While floating point processing is certainly doable in FPGAs, it comes at a considerable cost. A floating point macro is considerably larger and slower than it's fixed point counterpart. In most cases, doing your homework will allow you to use a fixed point design and obtain better performance, power,and device size numbers. If you must do floating point, you will want to look at a good text on computer arithmetic to learn the basics. Basically, for addition, you need to normalize one or both inputs to get to a common exponent. This is done with a barrel shifter and an alu to compute the required shift. After adding the normalized mantissas you may need to shift again. For multiplication, you can add the exponents and multiply the mantissas. again, you will usually want to normalize after the operation. THe shifters get expensive in terms of routing and layers of logic. In my practice, I do occasionally do a flaoting point design, but I try to avoid it whenever possible (so that the customer gets the best bang for the buck) -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randraka The Andraka Consulting Group is a digital hardware design firm specializing in high performance FPGA designs for digital signal processing, computing and control applications.Article: 9706
J.- pDSP all the way. Flexibility with compilers, cost of pDSP's are coming down rapidly, speed will be higher for FPGA but by the time you do anything of any complexity the next generation pDSP will have matched the speed. I can't speak to power it depends on your application and wich pDSP you use. If you choose a TMS320C54x or a Lucent 021 for GSM communication you will have a hard time beating it from a power consumption standpoint with any FPGA. Other applications may give different results. Regards, J. Khatib <jamil.khatib@pemail.net> wrote in article <35180577.878020@news.planet.edu>... > Please I want some hints and guidelines for selecting FPGA or pDSP > for some applications and the trade-offs involved ( ie power > consumption, speed, design flexibility ,etc) > > thanks in advance >Article: 9707
>However, in order to do this the second way (without a DPLL) I would >require a quartz clock crystal running at preferably greater than 80 >MHz. Does such a thing exist? If not, is there a way of creating a >faster clock using some clever coding from a slower quartz crystal? >eg, using a 20 MHz quartz crystal and creating a 40 Mhz clock from it? I >have a feeling this might be possible but has anyone else done it? I am sure you will get a few other replies here, but you can *buy* crystal oscillators of that sort of frequency, off the shelf. This is a far easier option. To generate a fast internal clock, using a DPLL, you will need a VCO and *that* will have to run at 80MHz. You cannot have the VCO internal to the FPGA, so you are looking at having an 80MHz signal outside the FPGA anyway, whatever you do. Unless I am missing something obvious. Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.Article: 9708
Zoltan Kocsi <root@127.0.0.1> wrote: [snip loads of interesting stuff] One of the most interesting threads for a long time. I have to agree that ultimately the silicon vendors must decide whether they are silicon vendors or software vendors. ISTR that the original PALs (remember MMI?) did not start to take off properly until PALASM was placed in the public domain. Having said that, I am currently on software maintenance with Xilinx and have been impressed by the helpfulness of both Xilinx and Micro-Call (here in the UK). I have few problems paying for maintenence / support if the service offered is genuine, rather than an excuse for blackmail. At least one other distributor of FPGAs here in the UK has lost my clients' purchase order because they seem unable even to find the reply button on their email, let alone compose a coherent answer. They must be semi-conscious (private joke). Richard ------------Richard Dungan------------- Radix Electronic Designs, Orpington, UK Email: Richard.Radix@BTinternet.com --------------------------------------- Unsolicited Commercial Email is not welcome. Kill filters are in operation.Article: 9709
In article <3521f340.6145175@news.btinternet.com>, Richard Dungan <Richard.Radix@BTinternet.com> writes >Zoltan Kocsi <root@127.0.0.1> wrote: > >[snip loads of interesting stuff] > >One of the most interesting threads for a long time. > >I have to agree that ultimately the silicon vendors must decide >whether they are silicon vendors or software vendors. ISTR that the >original PALs (remember MMI?) did not start to take off properly until >PALASM was placed in the public domain. While in reminiscence mode, don't forget Signetics FPLAs which appeared at about the same time, and their design tool - "H&L". I would say that it was the appearance and adoption of the "vendor independent" design tool ABEL (and also CUPL) in the mid-eighties that really allowed the PLD market to expand and diversify. These allowed users to migrate to exciting new devices like Lattice's GAL 20V8, Altera's EP300, which would probably have been out of the question if they had had to buy a new tool every time. By the end of the eighties, the great majority of PLD design was performed using vendor-independent tools. This allowed new suppliers to encroach on AMD/MMI's and Signetics' market share. The FPGA tool market seemed to be going the same way in '93 when NeoCAD came along, but it was not to be. The FPGA vendors have learnt the lessons of history, and know the importance of the design and implementation tools in determining market dynamics. So, IMHO, they're in the s/w business to stay. -- David Pashley < ------------------------ < < < ---------- Email: david@fpga.demon.co.uk | Direct Insight Ltd < < < < > Tel: +44 1280 700262 | | * The EDA Source * < < < Fax: +44 1280 700577 | --------------------------- < ------------------------------------------Article: 9710
>I agree. I have talked with Altera about this many times. It is not until more >people refuse to pay a 15% annual maintenance that things will change. 15%? You did well! Here in the UK, I was asked to pay $1600 for annual maintenance for XACT6, having purchased it for about $4000. Rather more than 15%, and the other thing is that this was not many months before Xilinx dropped XACT6 altogether... Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.Article: 9711
Hi Everyone, I have two questions concerning one time programmables such as Quicklogic and Actel. 1) Why do they seem to be lagging behind the SRAM based FPGAs so far in the number of available FLIP-FLOPs? Not counting using logic resources as latches the biggest of them seem only to have ~2500 flip-flops or so in their logic elements. 2) It also seems that they don't have the claimed savings in price. I think an Altera 10k40 is around $90 or so (maybe even cheaper by now). where as Actel had the following price which I grabbed from their web site. Device Time Frame 100 pc Percent High Percent Resale Reduction Volume Reduction A32200DX-FPQ208C March 1 $176.00 48% $97.00 54% 4Q97 $87.50 74% $47.60 77% It seems to me that $87.00 verses $90.00 is not much of a savings when you consider that the Altera device is reprogrammable. And both Altera and Xilinx have very low cost FPGAs (spartan and flex6k) that are around $10.00 - $20.00 for a 36k gate device. I'm I missing something? Since they use so much less silicon I would think the one time programmable would be much less expensive. So why aren't they? I would go with the one time programmables only if I needed their speed. But it suprises me that they are so far behind in every other way it seems. If anyone can comment I'd like to get a thread going to expose the issues here a little more. Sorry if this is a rehash of something that appears here a lot. It seems that it is the kind of thing that would be talk about to the point of annoyance like "which is better VHDL or Verilog". Thanks in advance, Joshua -- Joshua Schwartz ICC Design 972-8-9287375 Josh@gezernet.co.ilArticle: 9712
Martin Neilan wrote in message <3520F2E0.7590@hotmail.com>... >Hi everyone, > >I have a need to implement a Digital Phase Locked Loop on and FPGA using >V-System as the VHDL compiler and Xilinx as the synthesis software. If >it is not possible, I will need to look at other ways to determine a >suitable clock for an incoming serial stream of data (of undetermined >frequency) and keep this clock in sync. > >However, in order to do this the second way (without a DPLL) I would >require a quartz clock crystal running at preferably greater than 80 >MHz. Does such a thing exist? If not, is there a way of creating a >faster clock using some clever coding from a slower quartz crystal? >eg, using a 20 MHz quartz crystal and creating a 40 Mhz clock from it? I >have a feeling this might be possible but has anyone else done it? > >Any suggestions or help is greatly appreciated! > >Kind Regards, >Steve Phillipson >- A confused undergraduate engineer! Of coarse dubbling is possible here is pies of AHDL code how to do this subdesign double ( clkin :input; clkout :output; ) variable flipflop :dffe; doub :node; begin flipflop.d=!flipflop.q; flipflop.clk=doub; doub=lcell(clkin $ flipflop); -- incoming clock is XORed with output of flipflop clkout=doub; end; this design gives short spickes on each edge of incoming clock LevArticle: 9713
how can i do to use the bootstrap loader of the microcontroller SAB80C167 using a PCI interface as AMCC? excuse-me for my english : it's not good!! thanks chrisArticle: 9714
This is a multi-part message in MIME format. --------------583F823DD7132D1A3F3CF1EE Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I think your answer is good, but I have one more question: in the present Dual port, one LUT's write address and read address are the same , the other LUT's write address is the same as the first, but the read address is dual port address, I think some time just make the two LUTs' write and read address both seperated, that means, both the write address are the same and both the read address are the same, this scheme can double the RAM capacity, why not? --------------583F823DD7132D1A3F3CF1EE Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Zhiping Yao Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Zhiping Yao n: Yao;Zhiping org: SanCom Technology, inc email;internet: zpyao@spoke.ece.utexas.edu title: Engineer x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------583F823DD7132D1A3F3CF1EE--Article: 9715
Has anyone experienced problems trying to do post-routed simulations with Xilinx VHDL netlists with SDF timing files? I am simulating an XC4062 which takes close to 30 minutes just to load into the Modeltech simulator. The equivalent netlist in an ASIC format (Chip Express models from Synopsys Design Compiler) loads in less then 5 minutes. What the h*ll is Xilinx doing in their net list to screw things up so badly? When the netlist is compiled or loaded, I'm afraid my hard drive is going to come flying out the side of my PC it's thrashing so much. The results are even worse with Viewlogic's Vantage simulator. Does anyone have any clues or suggestions? Xilinx's website is quite on this subject. ToddArticle: 9716
I was trying to build a fifo 512X8 on XC4020E and i have a macro of 16X8 fifo which has push and pop alongwith full, empty and last logic. I have thinking of different ways to cascade it. I would appreciate if someone could show me a smart idea to cascade it. samir ------------------------------------------------------ Samir Khericha Graduate Research Assistant Department Of Computer Engineering Residence: 2383 duncan drive apt #8 fairborn OH 45324 PH No: 937-426-8076 _______________________________________________________Article: 9717
Felip Vicedo Roman <fvicedo@eug.upv.es> wrote: >I'm working in Altera FPGA's.I'm want to make a board configurable from >the computer, but I need the bitblaster or byteblaster. Somebody knows >how this devices are make?? > > >Felip Vicedo Roman >feviro@eug.upv.es Check the freecore library pages at http://193.215.128.3/freecore/ They have a schematic for a byteblaster -- richard_damon@iname.com (Redirector to my current best Mailbox) rdamon@beltronicsInspection.com (Work Adddress) Richad_Damon@msn.com (Just for Fun)Article: 9718
April 1, 1998 Are you tired of fighting with Xilinx, Actel, Altera, or some other FPGA tools I have an idea for a new technology. I call it HSI (Huge Scale Integration). HSI is based on TI's latest advance in Logic devices, PicoGate Logic. You can now get a single NAND gate in a SOT5 package that is about 3 X 3 mm in size. See http://www.ti.com/sc/docs/asl/sin_gate.htm I figure you could make an array of 100 X 100 of these PicoGate NAND gates on a board. The board would be about 1 square foot in area. This would give you 10,000 gates. You could could use a dependable PCB layout tool such as PADS to route the board to implement your logic. Every gate in the array could be probed for debuging. What do you think? Bill Seiler 3255-4 Scott Blvd, Suite 105 Santa Clara, CA 95054 408 982 5420 Direct 408 982 5430 FAX ccwest@ix.netcom.comArticle: 9719
In article <6fu6bu$jhf@sjx-ixn7.ix.netcom.com> ccwest@ix.netcom.com (Bill Seiler) writes: >April 1, 1998 > >Are you tired of fighting with Xilinx, Actel, Altera, or >some other FPGA tools > >I have an idea for a new technology. > >I call it HSI (Huge Scale Integration). > >HSI is based on TI's latest advance in Logic devices, >PicoGate Logic. You can now get a single NAND gate in >a SOT5 package that is about 3 X 3 mm in size. >See http://www.ti.com/sc/docs/asl/sin_gate.htm >I figure you could make an array of 100 X 100 of these >PicoGate NAND gates on a board. The board would be about >1 square foot in area. This would give you 10,000 gates. >You could could use a dependable PCB layout tool such as >PADS to route the board to implement your logic. >Every gate in the array could be probed for debuging. > >What do you think? > >Bill Seiler >3255-4 Scott Blvd, Suite 105 >Santa Clara, CA 95054 >408 982 5420 Direct >408 982 5430 FAX >ccwest@ix.netcom.com An excelent idea, and no need for dongles, dual port can be implemented with connectors at each end of the board, but best of all, the marketing guys/gals would be able to EXACTLY give the gate count. I would recomend using the PCB sw from Protel as it would fit so well in with the other advantages of this idea. Philip.Article: 9720
Felip Vicedo Roman (fvicedo@eug.upv.es) wrote: > I'm working in Altera FPGA's.I'm want to make a board configurable from > the computer, but I need the bitblaster or byteblaster. Somebody knows > how this devices are make?? > Felip Vicedo Roman > feviro@eug.upv.es * Look at www.altera.com * Visit the literature * Download the bit/byteblaster datasheet * .... ArminArticle: 9721
In article <3520F2E0.7590@hotmail.com>, Martin Neilan <s_tretch@hotmail.com> writes: > However, in order to do this the second way (without a DPLL) I would > require a quartz clock crystal running at preferably greater than 80 > MHz. Does such a thing exist? If not, is there a way of creating a > faster clock using some clever coding from a slower quartz crystal? > eg, using a 20 MHz quartz crystal and creating a 40 Mhz clock from it? I > have a feeling this might be possible but has anyone else done it? 80 MHz osc packages should not hard to find/get. They will be harder to find than slower speeds but they are available. It is possible to implement Digital PLL designs by processing several data samples in parallel. This makes more complicated logic that runs at a slower cycle time. You do need a fast front end (or delay line?) to grab the input samples. A 2 bit wide implementation could use the other clock edge to grab the other data sample. -- These are my opinions, not necessarily my employers.Article: 9722
Third party tools do solve some problems, but I have found that vendor specific tools provide support as much as 6-12 months before third party tools. I have also had better results from vendor tech support when trying to solve a compiler bug. In article <6e0M$TAR1gI1MAr1@fpga.demon.co.uk>, David Pashley <david@fpga.demon.co.uk> wrote: > > In article <3521f340.6145175@news.btinternet.com>, Richard Dungan > <Richard.Radix@BTinternet.com> writes > >Zoltan Kocsi <root@127.0.0.1> wrote: > > > >[snip loads of interesting stuff] > > > >One of the most interesting threads for a long time. > > > >I have to agree that ultimately the silicon vendors must decide > >whether they are silicon vendors or software vendors. ISTR that the > >original PALs (remember MMI?) did not start to take off properly until > >PALASM was placed in the public domain. > > While in reminiscence mode, don't forget Signetics FPLAs which appeared > at about the same time, and their design tool - "H&L". > > I would say that it was the appearance and adoption of the "vendor > independent" design tool ABEL (and also CUPL) in the mid-eighties that > really allowed the PLD market to expand and diversify. These allowed > users to migrate to exciting new devices like Lattice's GAL 20V8, > Altera's EP300, which would probably have been out of the question if > they had had to buy a new tool every time. > > By the end of the eighties, the great majority of PLD design was > performed using vendor-independent tools. This allowed new suppliers to > encroach on AMD/MMI's and Signetics' market share. > > The FPGA tool market seemed to be going the same way in '93 when NeoCAD > came along, but it was not to be. > > The FPGA vendors have learnt the lessons of history, and know the > importance of the design and implementation tools in determining market > dynamics. So, IMHO, they're in the s/w business to stay. > > -- > David Pashley < > ------------------------ < < < ---------- Email: david@fpga.demon.co.uk > | Direct Insight Ltd < < < < > Tel: +44 1280 700262 | > | * The EDA Source * < < < Fax: +44 1280 700577 | > --------------------------- < ------------------------------------------ > -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 9723
Hi Bill, Yeah, this is a good ideal! No limit on routing resource. The problems is the production and routing of the PCB. Inaddition, it's size, emission and power consumption......maybe the price is much higher than a FPGA......:) ____________________________________________________________________ Roger Yau CASIL R & D Co., Ltd.Article: 9724
In article <35215640.5CE9757B@eug.upv.es> you wrote: : I'm working in Altera FPGA's.I'm want to make a board configurable from : the computer, but I need the bitblaster or byteblaster. Somebody knows : how this devices are make?? : Felip Vicedo Roman : feviro@eug.upv.es Check out the site http://www.acte.no/freecore/didnt.htm Regards, Ho Voon Yee
Site Home Archive Home FAQ Home How to search the Archive How to Navigate the Archive
Compare FPGA features and resources
Threads starting:
Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z