Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarApr2017

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

Custom Search

Messages from 4175

Article: 4175
Subject: Re: manchester clock recovery
From: murray@pa.dec.com (Hal Murray)
Date: 21 Sep 1996 22:06:45 GMT
Links: << >>  << T >>  << A >>
Reposting article removed by rogue canceller.

In article <3242C3DB.7EB8@xilinx.com>, Peter Alfke <peter@xilinx.com> writes:

> Please, never, never use monostables, and use an analog PLL only as the
> last resort. The world is going digital !

I think the analog guys still own the high frequency end.  Consider
doing clock recovery at 622 MHz.

Another way of looking at things is that a small counter is an easy
way to implement a monostable.
Article: 4176
Subject: ISPD-97 Call for Papers
From: Workshop - Gabe Robins <ispd97@cs.virginia.edu>
Date: Sat, 21 Sep 1996 23:30:28 GMT
Links: << >>  << T >>  << A >>
                             Call for Papers

               1997 International Symposium on Physical Design
                             April 14-16, 1997
                          Napa Valley, California

              Sponsored by the ACM SIGDA in cooperation with 
                   IEEE Circuits and Systems Society

The International Symposium on Physical Design provides a forum to exchange
ideas and promote research on critical areas related to the physical design
of VLSI systems.     All aspects of physical design, from interactions with
behavior- and logic-level synthesis, to back-end performance analysis and
verification, are within the scope of the Symposium.    Target domains include 
semi-custom and full-custom IC, MCM and FPGA based systems.
 
The Symposium is an outgrowth of the ACM/SIGDA Physical Design Workshop.  
Following its five predecessors, the symposium will highlight key new 
directions and leading-edge theoretical and experimental contributions to 
the field. Accepted papers will be published by ACM Press in the Symposium 
proceedings. Topics of interest include but are not limited to:

       1. Management of design data and constraints 
       2. Interactions with behavior-level synthesis flows 
       3. Interactions with logic-level (re-)synthesis flows 
       4. Analysis and management of power dissipation 
       5. Techniques for high-performance design 
       6. Floorplanning and building-block assembly 
       7. Estimation and point-tool modeling 
       8. Partitioning, placement and routing 
       9. Special structures for clock, power, or test
      10. Compaction and layout verification
      11. Performance analysis and physical verification 
      12. Physical design for manufacturability and yield 
      13. Mixed-signal and system-level issues.
      
IMPORTANT DATES:    Submission deadline:              December 20, 1996
                    Acceptance notification:          February 1, 1997
                    Camera-ready (6 page limit) due:  March 1, 1997

SUBMISSION OF PAPERS:
    Authors should submit full-length, original, unpublished papers 
    (maximum 20 pages double spaced) along with an abstract of at most 
    200 words and contact author information (name, street/mailing address, 
    telephone/fax, e-mail).

    Electronic submission via uuencoded e-mail is encouraged (single 
    postscript file, formatted for 8 1/2" x 11" paper, compressed with 
    Unix "compress" or "gzip''). Email to:

                        ispd97@ece.nwu.edu

    Alternatively, send ten (10) copies of the paper to:

                        Prof. Majid Sarrafzadeh
                        Technical Program Chair, ISPD-97
                        Dept. of ECE, Northwestern University
                        2145 Sheridan Road, Evanston, IL 60208 USA
                        Tel 847-491-7378 / Fax 847-467-4144 

SYMPOSIUM INFORMATION:
    To obtain information regarding the Symposium or to be added to the
    Symposium mailing list, please send e-mail to ispd97@cs.virginia.edu. 
    Information can also be found on the ISPD-97 web page:   

                         http://www.cs.virginia.edu/~ispd97/

SYMPOSIUM ORGANIZATION:

Steering Committee:          J. Cohoon (Virginia), S. Dasgupta (Sematech),
                             S. M. Kang (Illinois), B. Preas (Xerox PARC) 
Past Chair:                  G. Robins (Virginia)
General Chair:               A. B. Kahng (UCLA and Cadence)
Program Chair:               M. Sarrafzadeh (Northwestern)
Keynote Address:             T. C. Hu (UC San Diego) and E. S. Kuh (UC Berkeley)
Special Address:             R. Camposano (Synopsys)
Publicity Chair:             M. J. Alexander (Washington State)
Local Arrangements Chair:    J. Lillis (UC Berkeley)
Technical Program Committee: C. K. Cheng (UC San Diego),
                             W. W.-M. Dai (UC Santa Cruz), 
                             D. D. Hill (Synopsys), 
                             M. A. B. Jackson (Motorola), 
                             J. A. G. Jess (Eindhoven),  
                             Y.-L. Lin (Tsing Hua), 
                             C. L. Liu (Illinois),
                             M. Marek-Sadowska (UC Santa Barbara),
                             M. Sarrafzadeh (Northwestern), Chair, 
                             C. Sechen (Washington), 
                             K. Takamizawa (NEC),
                             S. Trimberger (Xilinx), 
                             M. Wiesel (Intel), 
                             D. F. Wong (Texas-Austin), 
                             E. Yoffa (IBM)
Article: 4177
Subject: Re: Are there any FPGA Starter Kits?
From: Richard Schwarz <aaps@erols.com>
Date: Sun, 22 Sep 1996 14:31:00 -0400
Links: << >>  << T >>  << A >>
Ray Heasman wrote:
> 
> Hi All,
> 
> I was just wondering - is there an equivalent of the DSP Starter Kit for
> FPGA's? I would love to fiddle with something like that.
> 
> Cheerio,
> Ray
> --
>              _/_/_/   ""\ ""\    ""\ ""\ """""""\ ""\   _/_/_/_/_/_/_/_/_/_/_/
>            _/_/_/   """"\ """\  """\ ""\ ""\      """"\   _/ StarWriter  /  _/
>          _/_/_/   ""\ ""\ ""\"""\""\ ""\ ""\ """\ ""\ ""\   _/    Genisys   _/
> _/_/_/ _/_/_/   """"""""\ ""\ "\ ""\ ""\ ""\  ""\ """"""""\   _/            _/
>  _/_/_/_/_/   ""\     ""\ ""\    ""\ ""\ """""""\ ""\     ""\   _/_/_/_/_/_/_/
>    _/_/_/             Amiga - The canvas of the Gods.


Ray,

APS sells an inexpesive ($250.00) starter FPGA board for work with 
XILINX chips. You will still need access to a synthesis tool and XILINX 
XACT software however. You can dowmload the boards brochure at 
http://www.erols.com/aaps

Richard Schwarz
Article: 4178
Subject: Re: FPGAs design tools for PC
From: Richard Schwarz <aaps@erols.com>
Date: Sun, 22 Sep 1996 14:40:05 -0400
Links: << >>  << T >>  << A >>
Vincent Rowley wrote:
> 
> Hi,
> 
> I'm looking for FPGAs design tools for PC including HDL and gate-level
> simulator, synthesis tool and also FPGA development system to implement
> Xilinx or Altera FPGAs.
> 
> I will appreciate all suggestion from experimented users.
> 
> Thanks,
> 
> Vincent RowleyHey Vince,

I have used both EXEMPLAR and SYNPLICITY synthesis tools, and found them 
both to be excellent products. APS sells an FPGA test board for XILINX 
boards which only costs $250.00.  The APS X-84 FPGA board takes any 
XILINX 4000/5000 84 pin PLCC chip and is a PC ISA board which provides 
real time IO via an 8255 chip, oscillator sockets, timer circuits and 
status LEDs. All IO pins arew brought out via ribbon connectors. The 
board has power connectors to let it be used outside the PC. The board 
comes with download and IO software. Brochures can be downloaded at 
http://www.erols.com/aaps

Richard Schwarz
Article: 4179
Subject: Inexpensive XILINX FPGA STARTER BOARDS
From: Richard Schwarz <aaps@erols.com>
Date: Sun, 22 Sep 1996 14:56:00 -0400
Links: << >>  << T >>  << A >>
APS offers a low cost ($250.00) XILINX FPGA board -the APS-X84- which 
can take any 4000/5000 84 pin PLCC chip. The board has an on board timer 
circuit, PC decode PAL, 8255 chip for real time control and IO, a 
prototype area, on board prom, and a power connector to allow the boarxd 
to be used externally.The board comes with download and control software 
and VHDL examples, as well as hardware interface schematics. APS is now 
developing a new board -the APS-X208- which will socket any of the 
4000/5000  208 pin QFP plcc chips. The board will sell for $350.00.

contact aaps@erols.com or see APS LOGIC page at 
http://www.erols.com/aaps
Article: 4180
Subject: Re: manchester clock recovery
From: scanner@dial.pipex.com (Peter)
Date: Sun, 22 Sep 1996 22:24:43 GMT
Links: << >>  << T >>  << A >>
Peter,

If the receiver did not know the bit rate quite accurately, and had to
use some sort of adaptive scheme, it would have to throw away an awful
lot of data before it would lock on. So, in most cases, one knows the
data rate exactly.

Regarding monostables, I have done lots of those in FPGAs: digitally
with a counter. Nothing wrong with that, provided one avoids things
like async resets (derived from decoding the outputs) which might
reset some stages and not others.

You could post your schematic here. A *compressed* TIFF would be only
a few k - about the same length as the average "make $50,000 legally"
spam :) A zipped postscript version would be even smaller.

Peter.
Article: 4181
Subject: test
From: smith@q.continuum.net
Date: Mon, 23 Sep 1996 01:48:54 GMT
Links: << >>  << T >>  << A >>
testing...

Article: 4182
Subject: Re: query: C to FPGA
From: Maya Gokhale <maya@sarnoff.com>
Date: Mon, 23 Sep 1996 08:18:14 -0400
Links: << >>  << T >>  << A >>
>There's a bunch of work on translating software languages (especially C)
>to hardware.
...
>If anyone knows of any others, I'd appreciate pointers to them.

I have been working for several years on mapping a bit-oriented parallel 
C variant to parallel FPGA arrays. The compiler generates VHDL and 
pre-placed, pre-routed technology-specific macros. Target architectures
are 
the Splash-2 (16 Xilinx 4010's) and NAPA (4 CLAy MCMs manufactured by 
National Semiconductor) parallel arrays. Unfortunately, I haven't been
able to release the compiler to the general public due to proprietary
restrictions. I am presently porting the work to the SUIF toolset so
that it can be made more widely available. 

References follow.

@Inproceedings{Gok95a,
  author =       "M. B. Gokhale and A. Marks",
  title =        "Automatic Synthesis of Parallel Programs Targeted
                  to Dynamically Reconfigurable Logic Arrays",
  booktitle =    "Proceedings of the 1995 International Workshop
                  on Field-Programmable Logic and Applications,
                  Oxford, England",
  year =         "1995",
  pages =        "399--408",
  month =        sep}

@Incollection{Gok96,
  author =       "M. B. Gokhale and B. Schott",
  title =        "A Data-Parallel Programming Model",
  booktitle =    "{Splash-2}: {FPGAs} in a Custom Computing Machine",
  year =         "1996",
  publisher =    "{IEEE} Press",
  pages =        "76--78"}

(the following will appear in SPIE's Photonics East conference this fall 
in Boston)

@Inproceedings{Gok96b,
  author =       "M. B. Gokhale and J. Kaba and A. Marks and J. Kim",
  title =        "A Malleable Architecture Generator for FPGA
Computing",
  booktitle =    "High-Speed Computing, Digital Siganl Processing
                and Filtering Using Reconfigurable Logic, Photonics East
'96",
  place =       "Boston",
  year =         "1996"}


As far as the original question on floating point, Nabeel Shirazi got 
some good results implementing floating point on Xilinx. His paper on
the 
subject appeared in a recent FCCM or FPL. It is probably also available 
from Virginia Tech as his Master's Thesis.

Maya
Article: 4183
Subject: Re: *** finding datasheets and chipmakers on the web ***
From: Frank Adlam <fadlam@ml.petech.ac.za>
Date: Mon, 23 Sep 1996 14:21:38 +0200
Links: << >>  << T >>  << A >>
Gray Creager wrote:
> 
> Just a reminder,
> 
> I maintain a well-used, but not well-publicized, listing of chipmaker
> website URLs. I have recently
> revamped it so that you can view just a brief listing of URLs, or a
> listing URLs with a summary of the chips made by each company.
> 
> http://www.scruznet.com/~gcreager/hello5.htm
> 
> Use the above URL to choose between the two sites. You can print out
> either listing in hardcopy format (I've made versions with colors that
> are "printer friendly") and post it on your wall, or just bookmark
> whichever webpage you like best.
> 
> I've added a growing engineer humor page as well (some funny stuff there
> actually).
> 
> I am constantly updating this listing and it is the most up-to-date that
> you'll find anywhere. I will soon be adding some investor links as well
> (for those who invest in semi stocks).
> 
> Use it as much as you like and tell others about it.
> 
> --
> 
> Gray Creager


Good stuff !
Thanks
-- 
Frank Adlam
Faculty of Electrical Engineering
Port Elizabeth Technikon
PO Box X6011
Port Elizabeth
6000
East Cape
South Africa

Tel: 2741 504 3520
Fax: 2741 531 864
E-Mail:  fadlam@ml.petech.ac.za

Article: 4184
Subject: Re: manchester clock recovery
From: Peter Alfke <peter@xilinx.com>
Date: Mon, 23 Sep 1996 11:24:10 -0700
Links: << >>  << T >>  << A >>
When I argued against monostables, I had the real, unstable,
noise-sensitive, #$%@& analog monostable in mind. A counter is the
legitimate digital equivalent of such a time-out device, and that's what
my design uses also.

I will struggle with a graphic representation, because there have been
13 requests for fax so far. This newsgroup has many quiet readers, a
good forum for a little app note !

Peter Alfke, Xilinx Applications
Article: 4185
Subject: Source for FPGA and PCI prototype board ???
From: ccwest@ix.netcom.com (Bill Seiler)
Date: Mon, 23 Sep 1996 18:25:36 GMT
Links: << >>  << T >>  << A >>
CALL FOR HELP !!

I need a PCI card with FPGA's (hopefully Xilinix).
The PCI interface could be in the FPGA'a or a custom chip.
I need about 5000 gates free to do my thing.

Thanks
 
Bill Seiler
Circuit City / Patapsco West  
3255-4 Scott Blvd, Suite 105
Santa Clara, CA 95054
408 982 5420 Direct
408 982 5430 FAX
ccwest@ix.netcom.com 



Article: 4186
Subject: Re: manchester clock recovery
From: kenneth.a.becker@att.com (Kenneth A. Becker)
Date: Mon, 23 Sep 1996 21:58:04 GMT
Links: << >>  << T >>  << A >>
Peter Alfke <peter@xilinx.com> wrote:

>When I argued against monostables, I had the real, unstable,
>noise-sensitive, #$%@& analog monostable in mind. A counter is the
>legitimate digital equivalent of such a time-out device, and that's what
>my design uses also.

>I will struggle with a graphic representation, because there have been
>13 requests for fax so far. This newsgroup has many quiet readers, a
>good forum for a little app note !

>Peter Alfke, Xilinx Applications

Well, I guess I should weigh in here: I've done a manchester clock
recovery scheme or two, and more analysis of the design than I care to
think about.

The basic point here is that whatever one is using for clock recovery
(which I agree must be done), it is, essentially, a phase locked loop
system. In my mind I don't see much basic difference between the
digital PLL of Peter's design or a one-shot; both of them dump out the
appropriate clock that one can then use to recover the incoming data.

However, there is a point that I think the current owners of this
thread have missed - noise performance. In a previous lifetime I was
the designer of an AMPS (Advanced Moblie Phone System, now known as
cellular telephone) digital receiver. AMPS uses Manchester encoded
data. I had to design the clock recovery circuit for this beast. The
data in an AMPS system has to survive some pretty nasty noise; in
fact, the data itself is CRC-checked-and-corrected, with repeats, so
that with a little luck one can recover the incoming data, even with
numerous data hits, in a 0 dB signal/noise enviroment. The noise was
molded as Rayleigh - i.e., the >>envelope<< of the noise had a
Gaussian distribution. It was also affected by the speed of the car
(dropouts while the car was moving due to cancellations of the signal
by reflections).

What this meant for me was that the signal would drop into the noise
for anywhere between one and 15 bit periods. And it would be
exceedingly more than nice if the receive clock (which, by the way,
had a sequence of 255 alternating 1's and 0's to get started with)
would not drift too far during this period of no signal. When one is
faced with an environment like this, one pulls out the PLL design
issues and stares at them.

For one thing, you need to avoid any type of edge-detector for clock
recovery in the incoming bit stream. This rules out one-shots.
Multiplying phase detectors (XOR) are much preferred since noise hits
tend not to affect them too adversely.

For another, the stability (read: bandwidth) of the PLL becomes
critical. I have a nasty suspicion that any PLL that can lock over a 0
to 200 MHz range (or whatever it was that Pete said) has a large
bandwidth. In a noisy environment, what you want is the narrowest
bandwidth that you can get away with. Further, you want the lock range
to be narrow, as well; it takes a lot less time to lock to the
incoming clock when the PLL is more-or-less on frequency in the first
place.

As I vaugely remember, the digital PLL that I implemented had both a
wide bandwidth for initial lock and a narrow bandwidth thereafter,
with interesting algorithms to detect when lock had been gained and
lost. Admittedly, all this was done in the dark ages - it was
implemented on a custom gate array, and modeled in 4000 series logic! 
This might also tell you that the data rate wasn't exactly high,
either.

So, you designers out there - watch out! The nasty analog world,
complete with noise, is just beyond your I/O transformer.


		Ken Becker
		kab1@lucent.com
		DACS Hardware Engineering


Article: 4187
Subject: Re: How to Begin with FPGA design?
From: Scott Schlachter <scotts@actel.com>
Date: Mon, 23 Sep 1996 15:43:32 -0700
Links: << >>  << T >>  << A >>
Xiangdong:
There is actually a very large amount of information available 
to the internet community regarding FPGA's at the various
vendor sites.  Before I came to work here at Actel, I spent some
time researching the general subject of "Programmable Logic 
Devices" for a report I did in school.  I found that the World
Wide Web sites for Actel, Altera, and Xilinx alone had fairly large
amounts of information on them regarding not just their 
particular products, but programmable logic devices in general.
If you want to learn more about FPGAs in general, as well as
getting an idea of what the various vendors have available, 
you might try checking out their Web sites.  
Actel's URL is http://www.actel.com  

(I forget what Xilinx or Altera's URLs are... ;-) )
 
Regards,
-Scott 

 Scott Schlachter    
 Design Engineer
 Actel Corporation

Janos Szamosfalvi wrote:
> 
> Xiangdong Li (lxd@cpre1.ee.iastate.edu) wrote:
> : Hi, everyone:
> : I am a graduate student in ISU. I will design control circuits using
> : FPGA. I have no experience of utilizing any programmable IC. Can you
> : give me any advice? Thank you.
> 
> You may want to start with simple PLD's and Abel before getting into
> FPGA's and their associated tools because the formers are a lot simpler
> to learn and use.

-- 
                                Scott Schlachter
      \/                        scotts@actel.com
      @\  ,,
        \( o)  
         / \
       ./___-+-
      _  \ /_ \_
\    /.\ /_*  /\\  
 \   \_/      \_/
~~~\                       ^
 ^ ~~\                ^   / \  
/ \   \              / \  / \
/|\   ~~\            /|\  /|\
 |       ~~^~^^~^^~\__|____|___
Article: 4188
Subject: FIR filter using FPGAs??
From: venkat@chaos.ee.vill.edu (K. S. Venkatraman)
Date: 23 Sep 1996 23:29:42 GMT
Links: << >>  << T >>  << A >>
Hi:

I am interested in writing VHDL code for the AT&T ORCA FPGA in
order to implement an FIR filter. I have the ORCA Foundry software,
but do not have Exemplar.. instead we have AutoLogic. I need to know if
I can generate the appropriate EDIF files using AutoLogic.

Any tips?

Thanks,
Venkat.
 


Article: 4189
Subject: Re: Source for FPGA and PCI prototype board ???
From: pak@cse.ucsc.edu (Pak K. Chan)
Date: 24 Sep 1996 04:26:55 GMT
Links: << >>  << T >>  << A >>
>
>I need a PCI card with FPGA's (hopefully Xilinix).
>The PCI interface could be in the FPGA'a or a custom chip.
>I need about 5000 gates free to do my thing.
>

I did a wirewrap PCI prototyping board with a PGA XC4013E and its
works up to ~30MHz.
You can buy the PCI protoboards from frys electronics.

Article: 4190
Subject: XilinX XC5200 address pointer based FIFO
From: vmaler@rasips1.rasip.etf.hr (Vedran Maler (RIP pa konst. Zagar))
Date: 24 Sep 1996 16:17:58 GMT
Links: << >>  << T >>  << A >>
Hi everyone,

has anyone developed (or has some links to) XilinX XC5200 
optimised circuit for calculation of FIFO half-full flag ? 
FIFO has two address pointers, read and write, for example 4 bits long.

I would appreciate all the useful advice, schematic, or description.

We are opened for cooperation anytime !

Vedran Maler, BSEE
xylon@alf.tel.hr
XYLON d.o.o. - Croatia - Europe 
Article: 4191
Subject: FS: Data IO MESA-I
From: crabill@leland.Stanford.EDU (Eric Jay Crabill)
Date: 24 Sep 1996 16:19:10 GMT
Links: << >>  << T >>  << A >>
Hi,
     I have a Data I/O MESA-I system I'd like to sell.  It is new (sealed in
box, still).  Here's a more exact description:

     MESA-I version 1.40
     Pode and probe (68 pin PLCC)
     Controller board for PC (ISA card)
     System software/manuals

If you are interested, make me an offer.  I'm also looking to trade for a
Data I/O Chiplab 48 or similar (whatever flavor they are offering now!)

Thanks,
Eric Crabill
crabill@leland.stanford.edu
Article: 4192
Subject: FPGA Routing Question
From: Kiran Kumar Bondalapati <kiran@halcyon.usc.edu>
Date: Tue, 24 Sep 1996 20:08:32 -0700
Links: << >>  << T >>  << A >>
Hi,
  I have not used any FPGAs as such, but when I read the data sheets I
came up with the questions -
 
  In the FPGAs with global interconnect and neighbour connects(Xilinx,
Altera Flex etc.)
 
 1. Is it possible for any cell to connect to global connect. If yes
then
    what is the delay compared to the delay in neighbour communication.
 
 2. what is the limit on the number of cells which can read the signal
on
    a global wire onto which one cell writes the output.
 
 3. Can two cells output be connected to the global wire and the result
    read as a wired-or of the two signals ? 
 
 4. If NO to Q3 is it possible for two cells to write the SAME value(=1
e.g.)
    onto a global interconnect ?
 
  I have not done any handson work with FPGAs and knowledge limited to
only datasheets and papers so pardon me if questions are too silly :)

Thanx,
-- Kiran
kiran@halcyon.usc.edu


-- 
-------------------------------------------------------------------------
Kiran Kumar Bondalapati(BOND)     kiran@halcyon.usc.edu,
bondalap@usc.edu
Ph.D. Student, Comp.Engg.,USC.    http://www-scf.usc.edu/~bondalap
Res: 935,W.30th St.,#210,LA,CA-90007, Off: EEB 232, USC,LA,CA-90089
Ph : 213-745-6885                        : 213-740-9127 
-------------------------------------------------------------------------
Article: 4193
Subject: Re: manchester clock recovery
From: Ray Andraka <randraka@ids.net>
Date: Tue, 24 Sep 1996 22:02:39 -0700
Links: << >>  << T >>  << A >>
Now this is getting interesting.  Sure you can use a digital 'one-shot' to 
recover the clock, but even a little noise is going to blow you out of the 
water.  A much better solution is to use a digital PLL with an XOR type 
phase detector. The logic required to do this really isn't that large.   I 
did a manchester decoder for recovery of clocking on a project related to 
GPS (global positioning system) a few years back that had to recover the 
clock in a noisy signal. The desired frequency was known within about 5%.  
At start up, the DPLL was set up for the nominal frequency, but with a wide 
bandwidth to help capture the signal.  Once the signal was acquired, the 
bandwidth was narrowed down to a very small window using noise gating. I 
guess the biggest concern is the ratio of the master clock to the signal 
clock.  The bigger the better, but that starts pushing the limits of the 
FPGA for high data rates.  My design was for relatively low data rates and 
was done in a fraction of an Altera 7000 series CPLD when they were new.  
(been there, done that, got the T-shirt mode).

-Ray Andraka, P.E.
Chairman, the Andraka Consulting Group
401/884-7930   FAX 401/884-7950
mailto:randraka@ids.net
http://www.ids.net/~randraka/
 
The Andraka Consulting Group is a digital hardware design firm 
specializing in high performance FPGA designs.  Services include 
complete design, development, simulation, and integration of these 
devices and the surrounding circuits.  We also evaluate,troubleshoot, 
and improve existing designs. Please call or write for a free 
brochure or visit our web site.
Article: 4194
Subject: Re: XilinX XC5200 address pointer based FIFO
From: Ray Andraka <randraka@ids.net>
Date: Tue, 24 Sep 1996 22:11:01 -0700
Links: << >>  << T >>  << A >>
Vedran Maler (RIP pa konst. Zagar) wrote:
> 
> Hi everyone,
> 
> has anyone developed (or has some links to) XilinX XC5200
> optimised circuit for calculation of FIFO half-full flag ?
> FIFO has two address pointers, read and write, for example 4 bits long.
> 

Check out the Xilinx App notes on FIFOs for the 4000 series parts.  The 
bit about using the LFSRs for addressing and the method of determining 
the flags applies equally well the the 5200 series, or any other FPGA for 
that matter.  Basically, the LFSR makes it extra easy to obtain the next 
and previous addresses (just take the current one out with a shift 
applied).  The next, current, and previous addresses for the two pointers 
can be compared using an identity comparator and the results used to 
generate the flags.

The Xilinx app note should be available on the Xilinx web site at 
http://www.xilinx.com/.  If not, check out the data book.

-Ray Andraka, P.E.
Chairman, the Andraka Consulting Group
401/884-7930   FAX 401/884-7950
mailto:randraka@ids.net
http://www.ids.net/~randraka/
 
The Andraka Consulting Group is a digital hardware design firm 
specializing in high performance FPGA designs.  Services include 
complete design, development, simulation, and integration of these 
devices and the surrounding circuits.  We also evaluate,troubleshoot, 
and improve existing designs. Please call or write for a free 
brochure or visit our web site.
Article: 4195
Subject: Re: FPGA Routing Question
From: fliptron@netcom.com (Philip Freidin)
Date: Wed, 25 Sep 1996 08:05:42 GMT
Links: << >>  << T >>  << A >>
In article <3248A22F.4F69@halcyon.usc.edu> Kiran Kumar Bondalapati <kiran@halcyon.usc.edu> writes:
> 1. Is it possible for any cell to connect to global connect. If yes then
>    what is the delay compared to the delay in neighbour communication.

There is always a way to get to non-local interconnect (there is a heirarchy
of interconnect, of which local and global are two of them). The issue is
how much routing is used to make this connection. As the device gets 
routed, the resources to make these connections may get used up, so 
depending on design complexity (and architecture of course) the delay is
variable. Neighbour connections are almost always faster but also can
be influenced by existing routing. I.e. block A is to the left of block B,
and the route goes from left side of A to right side of B. This could be 4
times slower than from the right side of A to the left side of B. The real
entertainment happens when you have multiple loads. Non-local interconnect
is usually far less sensitive to multiple loads compared to the local
interconnect equivalent.

> 
> 2. what is the limit on the number of cells which can read the signal on
>    a global wire onto which one cell writes the output.

No limit. May effect delays though.

> 
> 3. Can two cells output be connected to the global wire and the result
>    read as a wired-or of the two signals ?

Yes, but only on the Xilinx XC3000 and XC4000 family products, through
the tristateable buffers (in wired-and (not wired-or) mode) on the 
horizontal longlines.
 
> 
> 4. If NO to Q3 is it possible for two cells to write the SAME value
>    (=1 e.g.) onto a global interconnect ?

The software DRC should prohibit you from doing such a thing.

>  I have not done any handson work with FPGAs and knowledge limited to
>only datasheets and papers so pardon me if questions are too silly :)

Not too silly

>
>Thanx,
>-- Kiran
>kiran@halcyon.usc.edu


Your welcome.
Philip Freidin


Article: 4196
Subject: Xilinx X-blox Bidir_IO padnames?
From: Wouter.Suverkropp@jet.uk (Wouter Suverkropp)
Date: Wed, 25 Sep 96 13:09:52 GMT
Links: << >>  << T >>  << A >>
Hi,

Is it possible to define X-blox BIDIR_IO padnames in a separate file?

I am in the process of building an interface to the CAMAC
bus (yes, I know...) and would like to define 24 padnames
for one BIDIR_IO symbol. Rather than having to type
LOC[0]=P123 etc. in the symbol definition, I would like
to have a file somewhere with the names in. That way the
schematics stay 'clean'. We use Orcad for design entry, and
Xact 5.n for compilation etc.

Any hints most welcome.

Wouter

====================================================================
For direct E-Mail please respond to Wouter.Suverkropp@jet.uk

===============================================================================
    The above article is the personal view of the poster and should not be
       considered as an official comment from the JET Joint Undertaking
===============================================================================
Article: 4197
Subject: Re: Xilinx X-blox Bidir_IO padnames?
From: Andy Gulliver <andy.gulliver@crossprod.co.uk>
Date: Wed, 25 Sep 1996 14:31:56 +0100
Links: << >>  << T >>  << A >>
You should be able to define these in the .cst constraints file as an 
alternative to assigning attributes on the schematics.

-- 
Regards

AndyG

**************************************************
*Any opinions expressed herein are entirely mine,*
*unless expressly stated otherwise.              *
*(as if anybody else would admit to them.....)   *
**************************************************
Article: 4198
Subject: 4800 baud serial input to xc4000
From: curtis@tigger.stcloud.msus.edu (Curt Schibonski)
Date: 25 Sep 1996 13:34:28 GMT
Links: << >>  << T >>  << A >>
Help!

I am working on a project that requires a 4800 baud serial input to an xc4000.
I am using a Maxim202 to convert from RS232 to TTL...works great.

I have contacted Xilinx, checked out their homepage, and data books, but I can't find any examples or ideas to help me input this data.

The problem that I see is how to input the ascii data serially.

I want to input the character strings, manipulate them, and them pass them out the "other side".

How can I sample the serial input as not to lose bits of data?

Any ideas?

Thanks

Curt Schibonski
Electrical Engineering
St. Cloud State University
curtis@tigger.stcloud.msus.edu
Article: 4199
Subject: XUMA #16
From: ecla@world.std.com (alain arnaud)
Date: Wed, 25 Sep 1996 14:03:08 GMT
Links: << >>  << T >>  << A >>
Xilinx Users Mailing List
Digest 16
9/25/96

This digest contains articles regarding synthesis and Xilinx, and specifically
view synthesis.

There's also a query on Windows NT

	1. Windows NT query
	2. Subject: Re: View Synthesis
====================================
From: Abdellah Touhafi 
Subject: XACT on winNT

Hello,
I would like to know if it is possible to install the XACT step software
for Windows (ver. 6.0.0) unther Windows-NT. Are there some particular
things to look at or problems that can occure?
If it is possible to do the installation on Windows-NT, which version of 
WIN(NT)32 should be installed?

=======================================================
From: Francois Fremaux 
Organization: COMPASS Design Automation
Subject: ViewSynthesis and Xilinx

I do not know on which paltform you are designing fpga, but Compass
support Hp, Sun and Ibm software solution which includes schematic captur=
e
and Hdl synthesis to target Xilinx (xnf 4 and Xnf 5). It includes also=20
the use of F/Hmap and timing information.

For any furher informations ...
=================================================
From: Steve Wiseman 
Subject: Re: ViewSynthesis and Xilinx

Hi, 
after 3 yars of fighting Viewlogic / Xilinx tools, I've abandoned to Altera, 
where the tools work, the silicon is cheaper and the tech support works. 
This may not be the approach for you, but I can now turn round designs in a 
fraction of the time, without a lot of the grief. 
Altera swapped my Xilinx toolset for an Altera equivalent, so I had to 
surrender the Xilinx dongle, but I felt the deal was acceptable.
I still use Modeltec for simulation, both for source level and for 
post-place/route. Basically, I'm now happy, after 3 years of intense 
frustration. 

  All the best, 
   
  Steve Wiseman, 
=======================================================================
From: David Gesswein
Subject: Re: ViewSynthesis and Xilinx


>My questions:
>	1. Has anyone used ViewSynthesis?
>		Likes, dislikes
>
I have used it and it seems ok.  I haven't used any other package so I
can't compare.  It doesn't always generate the best logic and only supports
its subset of VHDL.  It can also generate incorrect logic if you give it
strange enough VHDL.

>	2. Is there a workaround to the FMAP? 
>	   Obviously Viewlogic tech support is completely useless on that 
>	   subject.
>
The old PRO 6.0 release we have does generate them. It had an option in
the synthesis menu for tech options with a check box for pack into CLB.  That
made it use the FMAP etc.  The version is viewsynthesis 5.0.  Before this
release it was terrible and didn't generate FMAP etc.  We have the
workview office version but I haven't had time to try it.
 
>	5. What size devices have you implemented in VHDL?
>
None, the project got delayed after I had started testing the implementation
of parts of the circuits.

========================================================
From: David J. Matthews
Subject: XUMA Digest 15

Yes.  We have used Viewsynthesis for both gate array and CPLD jobs.
My opinion is that it is an inferior tool compared to others on the
market. 

 > 
 > 	2. Is there a workaround to the FMAP? 
 > 	   Obviously Viewlogic tech support is completely useless on that 
 > 	   subject.
I don't know one, but there are some customer support people at Xilinx
that are specific to synthesis issues.  You might want to try there.
 > 
 > 	3. Has anyone used other synthesis products:
 > 		Synopsys FPGA Compiler, 
 > 		Synplicity Synplify, 
 > 		Xilinx Foundation (Metamore),
 > 		Synopsys FPGA Express,
 > 		Data IO Synario,
 > 	        Minc,
 > 		any other?
We use the Galileo tool suite from Exemplar Logic (now a part of
Antares).  It not only provides excellent support for Xilinx, but all
other families of FPGAs and some gate arrays.  It is also much more
reasonably priced than the Synopsys tool - especially if you consider
the additional cost of target libraries (which can run $10K each for
gate arrays).  I am currently doing a XC4013E design and found it a
good match with that architecture.

 > 
 > 	4. What do you like about the above products?
See above.
 > 	   What do you dislike?
I would like to see improvements in: gate array support, resource
sharing, and module generation.
 > 	   Limitations, bugs?
Like ALL CAE tools, one occasionally experiences a glitch.  However
they are still a small company and quite responsive.
 > 	   State machine support
One can use user defined data types for state bits and then select
between binary, gray, one-hot, or random state encoding.
 > 	   On what basis did you select it. cost, functionality, availability,
 > 	   tech support, platform, other?
I used it while consulting at a client's site.
 > 	
 > 	5. What size devices have you implemented in VHDL?
I have implemented FGPAs/CPLDs from Xilinx, Altera, Actel, and
Quicklogic, and small gate arrays.

 > 
 > 	6. Do you use schematic and VHDL or only VHDL?
Since we are an engineering services company, we are often forced to
use schematics if the original source is in schematic form and we are
enhancing or migrating the design to a different foundry.  For new
designs, we use VHDL or Verilog exclusively.

Hope this helps.

  Dave Matthews               
                                                                             
==================================================================
From: Lasha Aponso
Subject: re: XUMA Digest 15

The following points/answers may be of some assistance:

1.We started our latest project, which includes 4x5206 & 1x5210 part 
using Workview Office and viewsynthesis.
These designs were done completely in vhdl using the .fpn file for pin 
allocation, global buffer allocation and other
attribute selections like fast outputs etc.

Eventhough Viewsynthesis provides a reasonably good user interface it 
would be more user friendly if it did generate a single XNF file rather 
than multiple files although this can be worked around using the wir2xnf 
and xnfmerge/xnfprep functions from xilinx.

The biggest problem with viewsynthesis however is the relatively large 
number of bugs that we uncovered compared to other products from Exemplar 
and Altera. Eventhough the design passed the functional simulation (using 
Speedwave, which is very good although I found it difficult/impossible to 
use the text_io functions in my test beds) after synthesis, it failed. We 
uncovered 4 such instances which cost our project at least 1 month (3 man 
months) in terms of lost time.

2.	I don't know.

3.Exemplar seems to be very good. We have only uncovered 1 bug with 
this package so far, and that has been rather obscure.

The user interface could be a improved and there is at least 1 
peculiarity with it when running under Win95.

We are not aware of a schematic front end that can be used with 
exemplar for the PC environment although Aldec will supposedly provide 
this.

4.With regard to state machine support, Aldec does offer a graphical 
entry mechanism for state machines which automatically creates vhdl code. I 
haven't personally used it, but it seems to be very good.

5.The 4 5206 designs varied from 60% to about 85%. The 5210 design was 
about 70%. 

6.All designs were done completely in vhdl.

This approach naturally facillitates porting of designs to different 
products or families. For example, as a cost reduction exercise, we were 
able to port these designs to altera's 10k parts, although both exemplar 
and maxplusII could not handle the internal tri-states.

For future designs however, I would consider using a combination of 
the two simply from a documentation stand point. Also, if aldec is used, 
state machines can be documented very easily and clearly. I still think a 
picture is worth some portion of a thousand words.

The best tool I've seen for this has been Mentor's System Architect. 
This may be a really good way to go if you can afford it.



Lasha Aponso

======================================================================

From: Jean-Paul Smeets
Subject: Re: ViewSynthesis and Xilinx

We have just baught the ViewLogic package, so i'm starting with
ViewSynthesis, but I've use Synplify from Synplicity with the
QuickLogic FPGA's as target.

We used Verilog as Synplify supports both Verilog and VHDL. The speed
of the compiler is very high, most simple designs are a synthesized in
a couple of seconds. The quality of the output is very high. In a
large number of cases the result is the same as you would obtain by
drawing it with library parts. Only for the most timing critical
pieces, I still use hard macro's for which I disable all optimisation.

For these designs I mix schematic with HDL input, but for simple
designs I only use the HDL.

If Viewsynthesis performs as Synplify I would be very happy.

==============================================================================
XUMA is an independent mailing list for users of Xilinx devices and tools.

To subscribe send an email to xuma_request@ecla.com

Would you like to contribute an article?
Share an experience, good or bad, with xilinx parts or tools, 
Describe a bug,
or need some help, have a question or just a comment:

then, post an article, send an email to xuma@ecla.com

The digests are archived at http://www.ecla.com





==============================================================================
XUMA is an independent mailing list for users of Xilinx devices and tools.

To subscribe send an email to xuma_request@ecla.com

Would you like to contribute an article?
Share an experience, good or bad, with xilinx parts or tools, 
Describe a bug,
or need some help, have a question or just a comment:

then, post an article, send an email to xuma@ecla.com

The digests are archived at http://www.ecla.com



Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarApr2017

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

Custom Search