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
Hello all, I have 2 Virtex-EM XCV800 (speed grade:6, BG560 package) FPGAs that I want to interconnect. They have 4 buses in their interface at 155MHz, do you think that using LVTTL pins is OK or must I use LVDS pins.. their connector cable is pretty small (<10 cm). Best Regards, HarrisArticle: 43201
I have a problem with Nios (2.0) simulation in Modelsim Altera 5.5e. After the SOPC builder configuration (and simulation contents setup of peripherals) and generation I tried to simulate the original Altera 32bit Nios vhdl example (standard_32). Modelsim seems to do the initialization ok but when I tried to execute run command it terminates with the message: # ** Fatal: Value -1742495648 is out of range 0 to 2147483647 # Time: 0 ps Iteration: 0 Instance: /test_bench/dut/the_boot_monitor_rom/boot_monitor_rom_lane0 # Fatal error at C:/vhdl/excalibur/nios2/examples/vhdl/standard_32/ref_32_system_sim/boot_mon itor_rom_lane0_module.vhd line 50 # I did everything step by step as required in the Altera's app note (app189 - simulating nios embedded processor designs) but it worked only with 16bit Nios configurations. None of 32bit functioned. I used the same settings in both 16 and 32bit configurations: no flash, uart with divisor 2, RAM (int or ext) with the same build file (e.g. hello_world.c), ... In all cases just the equivalent 16bit version could be simulated. What to do? Is it Nios or Modelsim problem? Please help. My OS: Win98 (I know Nios isn't 100% compatible but it fuctions ok with some minor workarounds - but is it relevant in this case?) Regards, MatjazArticle: 43202
I'd try messing with the pullup/pulldown settings for TDI and TCLK and maybe TMS. Good Luck; "Joze Dedic" <joze.dedic@email.si> wrote in message news:ac0163$dva$1@planja.arnes.si... > I am trying to configure XC2S200 (FG456) 5C in master serial mode (M(2:0) > all 0) from PROM XC18V02 - but the problem is that configuration does not > succeed every time. > FPGA DONE signal stays low and CCLK is running at 5 MHz whenever the cofig > fails - but nPROGRAM and nINIT signals are high. I have even monitored the > bitstream (DIN) of succeded and failed config - it started in the same way. > FPGA is located on PCI evaluation board (6 layers - one for 3.3 and one for > GND). On this EVB fits platform with SA1110. Thing is that config succeeds > every time whenever I hold SA1110 in reset. This indicates on some noisy > enviroment problem. I think... > I have monitored core supply (2.5) and it is clean and stabile. There are 8x > 10n, 8x 100n, 2x 1u, and 3x 10u F capacitators around FPGA. > > Are there any config setings that I should change - or it is just lack of > blocking C and poor supply condition. > > I would like to mention that one board (luckily the first one) works fine. I > have no ideas more. Thanks for any advice. > > jOc > >Article: 43203
155 MHz is pushing it for LVTTL for on-board connections, and pretty much out of the question for going off board through a cable. If this is a connection on the board, you might look at some of the other single ended flavors, some of them are a bit faster. Also, consider signal integrity modeling software to verify your layout. For off-board, you probably should be using a differential signal. "H.L" wrote: > Hello all, > I have 2 Virtex-EM XCV800 (speed grade:6, BG560 package) FPGAs that I want > to interconnect. They have 4 buses in their interface at 155MHz, do you > think that using LVTTL pins is OK or must I use LVDS pins.. their connector > cable is pretty small (<10 cm). > > Best Regards, > Harris -- --Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email ray@andraka.com http://www.andraka.com "They that give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety." -Benjamin Franklin, 1759Article: 43204
Hello, HAs anyone ever extended a jtag download cable up to 1-2 metres. What I want to archieve is to doewload at a reasonable speed but without using any buffers in the cable. I was thinking of using a "better" cable than the standard flatcable. My interface is a JTAG Technologies BV USB adapter JT3710. My device is an Altera 3128. What i have now is a clabel of approx. 1 metre but I have to adjes the download speed from 25kHz to 5KHz. My target is to go back to 25kHz by using a better quality cable. Thanks in advance for your advise! Wim PeetersArticle: 43205
Thank you all, the information you gave is very valuable. The only reason I am dealing with the 4K line is economical (at least by now, I hope so...). The XS40 eval boards use the XC4010XL. Other question: To get the FGPA configured, I will need to wire in the board a PROM with the bitstream, right? Thank you very much again, the promptness in the answers have surprised me! (Excuse my English, I am trying to improve it) Mauricio LangeArticle: 43206
Hi, Few years ago, we designed a board with an XC95108 and an XC4036ex. On that board, the CPLD sends data to the FPGA. At that time the board was working very well and we have never had any problems since. We have recently built some more boards (nothing changed). The CPLD is from a different batch. The boards, tested the same way, fail from time to time. It sounds like some timings are too close to the tolerance and sometimes goes over it and makes the system fail. Does anyone know if the die of these devices have recently changed and what influence it has on their behaviour ? Thanks. Philippe.Article: 43207
Oops. Datasheet is right, my math is wrong. 150 mAHr button cell will last its lifetime (~15 years) at 100 nA per the datasheet. It turns out that we have even less current drain, but we can test 100 nA! So the 1F supercap doesn't last as long as I (mis)calculated, but it does let you change batteries, or sit around for a little while (231 days -- obviously less due to leakage of the supercap, and the pcb). Austin Austin Lesea wrote: > Rick, > > I'll see why the datasheet has this number.... > > Austin > > Rick Filipkiewicz wrote: > > > Austin Lesea wrote: > > > > > The current drain is 0.1 nA, max (was not tested on early ES ... <snip> > > > > The Jan 2002 V-2 data sheet shows IBATT (max) = 100nA ?Article: 43208
Hmm I see, thanks Ray for you help :) I will check the other single-ended pins, is there a free signal integrity modeling software that I can download from the web? Greetings, Harris "Ray Andraka" <ray@andraka.com> wrote in message news:3CE3BB32.59217FF1@andraka.com... > 155 MHz is pushing it for LVTTL for on-board connections, and pretty much out > of the question for going off board through a cable. If this is a connection > on the board, you might look at some of the other single ended flavors, some of > them are a bit faster. Also, consider signal integrity modeling software to > verify your layout. For off-board, you probably should be using a differential > signal. > > "H.L" wrote: > > > Hello all, > > I have 2 Virtex-EM XCV800 (speed grade:6, BG560 package) FPGAs that I want > > to interconnect. They have 4 buses in their interface at 155MHz, do you > > think that using LVTTL pins is OK or must I use LVDS pins.. their connector > > cable is pretty small (<10 cm). > > > > Best Regards, > > Harris > > -- > --Ray Andraka, P.E. > President, the Andraka Consulting Group, Inc. > 401/884-7930 Fax 401/884-7950 > email ray@andraka.com > http://www.andraka.com > > "They that give up essential liberty to obtain a little > temporary safety deserve neither liberty nor safety." > -Benjamin Franklin, 1759 > >Article: 43209
Theron, I won't then. Austin PS: regular customers who need a "what if" can open a case on the hotline, and request an IBIS simulation be run for just such a case. University customers are supported through a different system and program. I would hope that the university sponsering this would be at the cutting edge, and have IBIS simulation tools available. I will check on that. I have run the simulation, and the overshoot and undershoot is terrible, and the rise and fall times are 1.4 ns for the fast corner, and 3.7 ns for the slow corner. Email me directly if you wish to receive the .jpg of the simulation. "Theron Hicks (Terry)" wrote: > Hi, > I am considering using a Spartan2E to generate a clock signal > (amongst other things). The clock will probably be driven by a slow > LVTTL signal with a 12 mA drive current. I am trying to decide whether > to terminate this output and if so how to do so (source vs load > termination, etc.) To do so, I need to know what the rise and fall > times of the output signal are. Please do not tell me to use IBIS. I > am quite competent with SPICE and PSPICE but I have no experience with > IBIS models and no availability to the software. For this simple > calculation I really don't need to use such tools anyway. FYI the > transmission line will be less than 0.3 meters > (> 1 foot). Load is one Spartan2E LVTTL input. Frequency is 100MHz. > Alternatively, I could use the differential LVPECL capacity of the > Spartan2E, but it appears to be overkill and would add an additional > connector pair to the system. > > Theron HicksArticle: 43210
I using one with 2.5m and it works ok. Had no special care, just plug the cable and hit the button... (On altera MAX 7000s using MAXIIplus and byteblaster, can't say about others... but would imagine it to be ok also) LC. "Wim Peeters" <peeters_wim@hotmail.com> wrote in message news:8f00e675.0205160608.771e20f9@posting.google.com... > Hello, > > HAs anyone ever extended a jtag download cable up to 1-2 metres. What > I want to archieve is to doewload at a reasonable speed but without > using any buffers in the cable. I was thinking of using a "better" > cable than the standard flatcable. > My interface is a JTAG Technologies BV USB adapter JT3710. My device > is an Altera 3128. > > What i have now is a clabel of approx. 1 metre but I have to adjes the > download speed from 25kHz to 5KHz. My target is to go back to 25kHz by > using a better quality cable. > > Thanks in advance for your advise! > > Wim PeetersArticle: 43211
www.srccomp.com "glen herrmannsfeldt" <gah@ugcs.caltech.edu> wrote in message news:abug8b$4al@gap.cco.caltech.edu... > "Steve Casselman" <sc.nospam@vcc.com> writes: > > >Yes the other way in C is to have functions that run in parallel like > >threads. The compiler must be smart enough to infer real parallel hardware. > >Also if you take the list operator and define the behavior as fully parallel > >then C does not seem sequential. > > Say I take an FFT algorithm from Numerical Recipes in C. > > What is the chance that I will find a compiler that can generate > a reasonable FPGA implementation of that algorithm, say in the > next 5 years to give someone a chance? > > -- glen >Article: 43212
FPGA does not work until it is fully configured (i.e. until DONE is high) - until then all the output drivers are off, and all FFs are held in reset...This pin is used (probably only) because of (programmed) timing sequence after DONE is high (enable outputs, release set or reset...) have a nice day, jOc "Stephanie McBader" <mcbader@neuricam.com> wrote in message news:3CE3B080.CBAB2AEC@neuricam.com... > Hi, > > Xilinx FPGA datasheets (both Virtex & Spartan) indicate that the DONE > pin is bidirectional. Does that mean that the logic configured inside > the FPGA will be able to read it? > > i.e., would a controller implemented in the FPGA be able to read the > value of DONE & say, remains in reset until DONE is asserted? > > Thanks! > > Stephanie McBader > Researcher/Design Engineer > NeuriCam S.p.A > Via S M Maddalena 12 > 38100 Trento TN, Italy > Tel: +39-0461-260552 > Fax: +39-0461-260617 > >Article: 43213
Stephanie, Internally, when in chains of FPGAs, the DONE pin of the master waits to see the DONE pin of the last slave go high (hence it is an input in that mode). Basically, no user logic can operate until DONE goes high, so I don't see why you need this. As soon as DONE goes high, all initial register values, FF states, BRAM contents, etc are all present, and on the next user supplied system logic clock, everything starts marching along. Austin Stephanie McBader wrote: > Hi, > > Xilinx FPGA datasheets (both Virtex & Spartan) indicate that the DONE > pin is bidirectional. Does that mean that the logic configured inside > the FPGA will be able to read it? > > i.e., would a controller implemented in the FPGA be able to read the > value of DONE & say, remains in reset until DONE is asserted? > > Thanks! > > Stephanie McBader > Researcher/Design Engineer > NeuriCam S.p.A > Via S M Maddalena 12 > 38100 Trento TN, Italy > Tel: +39-0461-260552 > Fax: +39-0461-260617Article: 43214
Hi, > I am trying to configure XC2S200 (FG456) 5C in master serial mode (M(2:0) > all 0) from PROM XC18V02 - but the problem is that configuration does not > succeed every time. have a look at the cclk. Try to terminate it with 100 Ohm - 150 Ohm. Best regards Thorsten Trenz -- Dipl.-Ing. Thorsten Trenz, Trenz Electronic Tel.: +49 (0) 5223 41652, Fax.: +49 (0) 5223 48945 Mailto:ng@trenz-electronic.de http://www.trenz-electronic.deArticle: 43215
Hi. I would like to know what exact properties a programmable array has. More precissely, I want to know whether it allows to represent any algorithm, that is whether a programmable array includes: 1. causality (i.e. conditional execution in general) 2. stability (i.e. some sort of memory) I was searching for this information in the internet but have found nothing. I would like to know whether such PA exists. Thanks in advance.Article: 43216
John_H wrote: > I'll see if I can find my book on digital frequency synthesis for a reference The book I bought in '96 proved very interesting at the time when I was doing some DDS design. I referred back to the text when I was doing Fractional-N investigations though the treatment of Fractional-N wasn't as clean as I hoped. Most of the Fractional-N work I did was from a rare data sheet or two, some papers on Delta-Sigma D/A converters, and a bunch of Mathcad analysis of spurs. Digital Techniques in Frequency Synthesis Bar-Giora Goldberg McGraw-Hill 1996 ISBN 0-07-024166-X Having thought more about your envisioned application, I've guessed you're trying to get the VCXO (or VCO) to tune precisely by dithering between two DAC values since the values of X and X+1 you mentioned would be too fast or too slow for a phase lock. This is where the Delta-Sigma converters came in to my development. Very high resolution converters with excellent linearity can be realized with reasonable frequencies (often below 50kHz) by using what some early papers referred to as a "1-bit DAC" controlled by very fast dithering, the output of which is lowpassed. By using more than just an accumulator output which would provide strong beat frequencies showing up as jitter in a synthesizer, the Delta-Sigma technique supresses the error by adding digital filtering to explicitly remove the error in the low frequency band. While ther have been "multi-bit DACs" used in Delta-Sigma converters, the linearity issues limited the results in the work I saw in the early '90s. The beauty of the frequency synthesizer is that the output of a phase comparator is the rough equivalent of the DAC in a Delta-Sigma converter. As long as you figure out where your spurs will be for the reference and output frequency combination and do your filtering right, the analog phase comparator in a Fractional-N synthesizer can produce exceptional results. The increased signal agility allowed by the technique comes at a small cost of a slightly increased phase noise floor - still very good compared to the alternatives - in the frequencies near and a bit above the PLL filter cutoff frequency.Article: 43217
"Mauricio Lange" <weirdo@bbs.frc.utn.edu.ar> schrieb im Newsbeitrag news:2f938098.0205160624.1b86ef5a@posting.google.com... > Thank you all, the information you gave is very valuable. > The only reason I am dealing with the 4K line is economical (at least > by now, I hope so...). The XS40 eval boards use the XC4010XL. Its NOT more economical. Go to www.nuhorizons.com and compare the price of Spartan-II againgst 4K (or Spartan(XL), you will be scared! -- MfG FalkArticle: 43218
Validation of VLSI designs using FPGA-based hardware is pretty common. Anyone know of a not-too-old article on the industry as a whole? A web search turns up a lot of companies that do this (Quickturn etc.), but I'd prefer not to use any one company as my source. I need a sense of the scale of the industry, is it growing, how's it changing etc. from an objective source. This is not for internal use but for research of a personal nature. Quick replies appreciated. Thanks! -- Don Teeter * Not speaking for Intel Corp. *Article: 43219
I had a similar problem with a Spartan2 (xc2s150), this is what I did to fix the problem: - pull up "DONE" to 3.3V with ~6.8K resistor - replace 4.7K pull up resistor on "INIT" with 10K res. - connect a 10uF Cap between "INIT" and GND The problem was that the fpga powered up and asserted INIT before the PROM finished it's startup sequence. ( a pullup on DONE was also missing) jakab C.W. THomas <cwthomas@bittware.com> wrote in message news:ue7ciq320ddd13@corp.supernews.com... > > > I'd try messing with the pullup/pulldown settings for TDI and TCLK and maybe > TMS. > > Good Luck; > > > > "Joze Dedic" <joze.dedic@email.si> wrote in message > news:ac0163$dva$1@planja.arnes.si... > > I am trying to configure XC2S200 (FG456) 5C in master serial mode (M(2:0) > > all 0) from PROM XC18V02 - but the problem is that configuration does not > > succeed every time. > > FPGA DONE signal stays low and CCLK is running at 5 MHz whenever the cofig > > fails - but nPROGRAM and nINIT signals are high. I have even monitored the > > bitstream (DIN) of succeded and failed config - it started in the same > way. > > FPGA is located on PCI evaluation board (6 layers - one for 3.3 and one > for > > GND). On this EVB fits platform with SA1110. Thing is that config succeeds > > every time whenever I hold SA1110 in reset. This indicates on some noisy > > enviroment problem. I think... > > I have monitored core supply (2.5) and it is clean and stabile. There are > 8x > > 10n, 8x 100n, 2x 1u, and 3x 10u F capacitators around FPGA. > > > > Are there any config setings that I should change - or it is just lack of > > blocking C and poor supply condition. > > > > I would like to mention that one board (luckily the first one) works fine. > I > > have no ideas more. Thanks for any advice. > > > > jOc > > > > > >Article: 43220
Hi, I am researching the suitability of FPGA technology for the real-time simulation of spiking or pulsed neurons in a feedforward network. The system I envision will have up to 200,000 neurons organized in several specialized layers. At least one of the layers will use massive recurrent feedback. The system must make it easy to attach an indefinite number of synaptic (input and output) connections to a neuron, and to disconnect them when necessary. Some of my neurons have as little as two inputs while others may have hundreds of inputs. The efficacy of the connections must be adjustable in real time. Good timing capability is essential although not stringent. I need to resolve signal arrival times to within 1 millisecond. I need to be able to add precise signal delays to specific inputs. Thanks in advance for any relevant info. Nemesis Temporal Intelligence: http://home1.gte.net/res02khr/AI/Temporal_Intelligence.htmArticle: 43221
Austin, My appolgies for what, as I re-read what I wrote last night, appears to be somewhat rude at best. I will try to find out what might be readily available for IBIS simulation on campus. You most likely are right, there probably is something available. By the way, has Xilinx considered providing PSPICE models for the output devices in the FPGA? (Perhaps I am just behind the times, but PSPICE is simple and readily available as a demo version.) Thanks for the answer. Austin Lesea wrote: > Theron, > > I won't then. > > Austin > > PS: regular customers who need a "what if" can open a case on the hotline, > and request an IBIS simulation be run for just such a case. University > customers are supported through a different system and program. I would > hope that the university sponsering this would be at the cutting edge, and > have IBIS simulation tools available. I will check on that. I have run > the simulation, and the overshoot and undershoot is terrible, and the rise > and fall times are 1.4 ns for the fast corner, and 3.7 ns for the slow > corner. > > Email me directly if you wish to receive the .jpg of the simulation. > > "Theron Hicks (Terry)" wrote: > > > Hi, > > I am considering using a Spartan2E to generate a clock signal > > (amongst other things). The clock will probably be driven by a slow > > LVTTL signal with a 12 mA drive current. I am trying to decide whether > > to terminate this output and if so how to do so (source vs load > > termination, etc.) To do so, I need to know what the rise and fall > > times of the output signal are. Please do not tell me to use IBIS. I > > am quite competent with SPICE and PSPICE but I have no experience with > > IBIS models and no availability to the software. For this simple > > calculation I really don't need to use such tools anyway. FYI the > > transmission line will be less than 0.3 meters > > (> 1 foot). Load is one Spartan2E LVTTL input. Frequency is 100MHz. > > Alternatively, I could use the differential LVPECL capacity of the > > Spartan2E, but it appears to be overkill and would add an additional > > connector pair to the system. > > > > Theron HicksArticle: 43223
I'm a newby trying to use the Xilinx ISE constraints editor to put timing constraints on internal signals (i.e. signals that don't go to output pads). The internal signals consist of a bus being reclocked. Any guidance on how this is done? (For discussion purposes, assume the clks are clk1 and clk2 and the signal busses are bus1 and bus2) DougArticle: 43224
Theron, Unfortunately, wew can not encrypt the technology files with psice, so we only support hspice under NDA for IO simulations (and only then under some duress as the there is no gaurantee of security in this method of encryption). One can appreciate that all of the device models are proprietary to our foundries, and to us, and their disclosure is absolutely prohibited. Austin "Theron Hicks (Terry)" wrote: > Austin, > My appolgies for what, as I re-read what I wrote last night, appears to be > somewhat rude at best. I will try to find out what might be readily available > for IBIS simulation on campus. You most likely are right, there probably is > something available. > By the way, has Xilinx considered providing PSPICE models for the output > devices in the FPGA? (Perhaps I am just behind the times, but PSPICE is simple > and readily available as a demo version.) > > Thanks for the answer. > > Austin Lesea wrote: > > > Theron, > > > > I won't then. > > > > Austin > > > > PS: regular customers who need a "what if" can open a case on the hotline, > > and request an IBIS simulation be run for just such a case. University > > customers are supported through a different system and program. I would > > hope that the university sponsering this would be at the cutting edge, and > > have IBIS simulation tools available. I will check on that. I have run > > the simulation, and the overshoot and undershoot is terrible, and the rise > > and fall times are 1.4 ns for the fast corner, and 3.7 ns for the slow > > corner. > > > > Email me directly if you wish to receive the .jpg of the simulation. > > > > "Theron Hicks (Terry)" wrote: > > > > > Hi, > > > I am considering using a Spartan2E to generate a clock signal > > > (amongst other things). The clock will probably be driven by a slow > > > LVTTL signal with a 12 mA drive current. I am trying to decide whether > > > to terminate this output and if so how to do so (source vs load > > > termination, etc.) To do so, I need to know what the rise and fall > > > times of the output signal are. Please do not tell me to use IBIS. I > > > am quite competent with SPICE and PSPICE but I have no experience with > > > IBIS models and no availability to the software. For this simple > > > calculation I really don't need to use such tools anyway. FYI the > > > transmission line will be less than 0.3 meters > > > (> 1 foot). Load is one Spartan2E LVTTL input. Frequency is 100MHz. > > > Alternatively, I could use the differential LVPECL capacity of the > > > Spartan2E, but it appears to be overkill and would add an additional > > > connector pair to the system. > > > > > > Theron Hicks
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