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 20050

Article: 20050
Subject: Re: Easy to program PLD
From: Shawn D'Alimonte <sdalimon@endor.ee.ryerson.ca>
Date: 25 Jan 2000 15:43:14 GMT
Links: << >>  << T >>  << A >>
Klaus Falser <kfalser@durst.it> wrote:

: You could use the smallest Xilinx XC9500 device.
: Programming software is free and downloadable from there site (WebPack)
: and programming is done through JTAG (6 pins ) form the parellel port
: of a PC. The adapter should cost around 50$, but if you want to
: build it itself there should be the schematics somewhere on Xilinx's
: Web-sites.

I will look into it.  Someone else also suggested one of Lattice's
small CPLD's.  These may be better than the GAL.

Thanks for the suggestions.

-- 
Shawn D'Alimonte - sdalimon@ee.ryerson.ca
"Faster processors are nice, but they are not truly revolutionary. And
   neither are colors." - Jim Collas, Amiga Inc.
Article: 20051
Subject: XC9500 0,5u Mask: Errors?
From: +Pablo+ <anon@vapor.net>
Date: Tue, 25 Jan 2000 11:35:02 -0500
Links: << >>  << T >>  << A >>
On Thu, 30 Dec 1999 11:35:10 +0100, Genio Kronauer
<g.kronauer@kronauer-digital.de> wrote:

>Recently Xilinx has changed process for XC900 series from 0.6u to 0.5u,

Die step often means you are getting a much faster part than you are
used to. I suspect you are seeing noise related problems, probably
caused by higher ground bounce of the faster 0.5u device. If this is
so, the problem can be fixed by better power/ground distribution and
decoupling.

The faster part may also cause problems with setup and hold
requirements to external devices from the 9500 device. 

It is a good idea to requalify your designs over your operating
temperature and VCC range when a manufacturer changes the process on a
device.

See http://support.xilinx.com/techdocs/271.htm
and http://www.xilinx.com/xcell/xl30/xl30_36.pdf

I hope this helps you out.

>see also http://www.xilinx.com/partinfo/notify/pcn9803.htm .
>Since that, some 30% of chips of a certain design work, other designs
>work
>on 70% of chips, and some don?t work at all.
>Has anybody the same problem?
>Are those chips suffering from ESD?
>Or do they have some problem with powersupply or do they latch up more
>easily?
>Do failiures affect 9572-15, or also other types?
>Jedec's are not under suspect anymore: Failiures are seen also in
>designs
>without any asynchronous path at all.
>
>Any hints appreciated.
>BR
>Genio
>
>

Article: 20052
Subject: Re: Xilinx Foundation: VHDL to symbol
From: "Holger Kleinert" <Kleinert@ibpmt.com>
Date: Tue, 25 Jan 2000 17:52:55 +0100
Links: << >>  << T >>  << A >>
Hi !

Its easy :
Open the VHDL Windows.
Go to Menu 'Project'
Choose 'Create Macro'

Maybe you have to systhesize the code first.

Regards,

--
Holger Kleinert
Development / Support

IBP Instruments GmbH
Sutelstrasse 7a
D-30659 Hannover, Germany

http://www.ibpmt.com
Fon : +49-511-651647
Fax : +49-511-652283



Steven Sanders <sanders@imec.be> schrieb in im Newsbeitrag:
388DA0C9.83D9D662@imec.be...
> Hello,
>
> I`m a newbie to the Xilinx Foundation Series. I have 5 VHDL files
> (describing 5 components)
> and I want to link these in a top file, but graphical! How do I make
> symbols out of my VHDL code?
>
> Thanx
>
> Steven.
>


Article: 20053
Subject: Re: Atmel config PROMs
From: Peter Alfke <peter@xilinx.com>
Date: Tue, 25 Jan 2000 09:53:54 -0800
Links: << >>  << T >>  << A >>


Peter Fenn wrote:

> Please advise:
> Can Atmel AT17c256 be used in place of Xilinx XC17s20 config PROM?
>

If we assume that all these SPROMs can spit out a serial bitstream,
here are some subtle points:
RESET polarity. You have to program this, and you should go for
active Low RESET, so that you can drive this pin from INIT. Reset
polarity programming is the most confusing aspect of SPROMs.
Power-on delay. If the SPROM takes longer to power up than the FPGA,
you are in trouble, unless you find ways to stall the FPGA.

Peter Alfke, Xilinx Applications


Article: 20054
Subject: Re: Xilinx programming from a Linux PC
From: "rodger" <brownsco@frii.com>
Date: Tue, 25 Jan 2000 14:51:29 -0700
Links: << >>  << T >>  << A >>
While I have not done this, it should be very doable to
prgram an FPGA on a Linux PC.

Check out the PDF at:

http://www.support.xilinx.com/apps/config.htm

Take a look at the App Note "Configuring FPGAs Over a
Processor Bus". It contains diagrams and code to program
an FPGA from a PCs ISA bus.

Another alternative would be to write to either the parallel or
serial ports by modifying the example code.

Another app note is XAPP058. This could be adapted also
with some research.

Have fun.

-Rodger

Phil Endecott <phil_endecott@spamcop.net> wrote in message
news:388C7A42.BA29F505@spamcop.net...
> Hi FPGA Experts,
>
> I have a Virtex prototyping board from VCC (www.vcc.com) which I
> currently program from an NT PC running the Xilinx alliance tools.  The
> problem is that I run all of the other tools (VHDL simulation &
> synthesis, place & route) on a Sun; this machine is locked away in a
> machine room somewhere, and I have a Linux PC on my desk which I use as
> an X-terminal.  I'd like to be able to get rid of the NT PC and plug the
> VCC board directly into my Linux box, but I can't find a way to download
> the bit file.  What I need is either a utility that will download a bit
> file from Linux (I don't need anything like the hardware debugger), or
> alternatively a way to make the Sun version of the alliance tools talk
> to a serial port that's not on the local machine.  Xilinx support tell
> me that neither is possible.
>
> I can't believe that I'm the only person ever to try to do this.  Does
> anyone have any experience to share?  If someone knows the protocol
> needed to talk to an X-checker or MultiLINX cable I'm happy to write my
> own tool.
>
> Thanks in advance,
>
> --Phil.


Article: 20055
Subject: Re: Virtex Fine Pitch BGA pcb layout
From: eml@riverside-machines.com.NOSPAM
Date: Tue, 25 Jan 2000 22:11:34 GMT
Links: << >>  << T >>  << A >>
On Tue, 25 Jan 2000 01:57:53 GMT, bobperl@best_no_spam_thanks.com (Bob
Perlman) wrote:

> The plane-to-plane capacitance is
>nice, being low-inductance, but there isn't much of it, particularly
>if you have lots of stuff switching.  CMOS-based designs can have
>rather astonishing dI/dt requirements; I'd think a long time before
>trying to build a board without high-frequency bypass caps.

I'm not sure that it takes any thinking about. The last time I
checked, a reasonably-sized board with one power and one ground plane
had a total plane capacitance of, IIRC, a few 10's of nF. A given chip
is only going to have access to the stored charge in its own local
region, and so it has a local capacitance of maybe 100pF. Putting in
some ballpark numbers shows that the most you can get out of this cap
is maybe 30 or 40mA, over a 0.5ns interval, for a reasonable supply
voltage drop of 5%. As you say, modern CMOS chips have *very* much
greater transient current requirements than this. 30 or 40mA may be
enough for one switching output, but that's about it.

Evan



Article: 20056
Subject: Anyone changed an NT disk serial number?
From: eml@riverside-machines.com.NOSPAM
Date: Tue, 25 Jan 2000 22:12:45 GMT
Links: << >>  << T >>  << A >>
Anyone managed this? I've tried modding the boot sector with a '95
boot disk and Debug, but Debug can't even see the C drive, let alone
modify the boot sector. And does anyone know where the serial number
is? Presumably the NTFS location is different from the FAT16 and FAT32
locations.

By the way, this is (I think) legal.

Thanks

Evan



Article: 20057
Subject: Re: Xilinx programming from a Linux PC
From: ldoolitt@recycle (Larry Doolittle)
Date: 25 Jan 2000 22:24:06 GMT
Links: << >>  << T >>  << A >>
rodger (brownsco@frii.com) wrote:
: While I have not done this, it should be very doable to
: prgram an FPGA on a Linux PC.

Of course it can be done.

Right now my toolchain for a Xilinx Spartan XCS10 is:
  * Icarus Verilog on Linux    http://www.geda.seul.org/tools/verilog/
  * Xilinx Alliance on Solaris
  * XStools     http://www.xess.com/
      ported to Linux twice, see
     either  ftp.microux.com/pub/pilotscope/xstools.tar.gz
     or      http://recycle.lbl.gov/~ldoolitt/fpga/

One of these days Alliance will run on Linux, and I can stop
mooching off of another group's Sun.  Right now that step is
embedded within a Linux Makefile using ssh.  Downloading 12 Kbytes
from Linux to the XCS10 takes 1.93 seconds.

: Take a look at the App Note "Configuring FPGAs Over a
: Processor Bus". It contains diagrams and code to program
: an FPGA from a PCs ISA bus.

Cute.  But not relevant for an existing development kit.

: Another alternative would be to write to either the parallel or
: serial ports by modifying the example code.

The parallel port works trivially.  Lots of people have made
it work for the XESS prototyping board listed above.

: > I have a Virtex prototyping board from VCC (www.vcc.com) which I
: > currently program from an NT PC running the Xilinx alliance tools.  The
: > problem is that I run all of the other tools (VHDL simulation &
: > synthesis, place & route) on a Sun; this machine is locked away in a
: > machine room somewhere, and I have a Linux PC on my desk which I use as
: > an X-terminal.  I'd like to be able to get rid of the NT PC and plug the
: > VCC board directly into my Linux box, but I can't find a way to download
: > the bit file.  What I need is either a utility that will download a bit
: > file from Linux (I don't need anything like the hardware debugger), or
: > alternatively a way to make the Sun version of the alliance tools talk
: > to a serial port that's not on the local machine.  Xilinx support tell
: > me that neither is possible.

Outrageous.  Either they don't understand the question,
or they are putting short-term stockholder greed ahead
of your desire to get the job done.

: > I can't believe that I'm the only person ever to try to do this.  Does
: > anyone have any experience to share?  If someone knows the protocol
: > needed to talk to an X-checker or MultiLINX cable I'm happy to write my
: > own tool.

If this cable connects the programming pins of the Virtex to
the parallel port of the PC, like the XESS board does, you just
look at the Virtex manual.

     - Larry Doolittle   <LRDoolittle@lbl.gov>
Article: 20058
Subject: xilinx foundation: map exceptions
From: Chuck Carlson <chuck@sanna.com>
Date: Tue, 25 Jan 2000 14:26:10 -0800
Links: << >>  << T >>  << A >>
Hello all,

I've just obtained the 1.3 student edition and am running on NT.
Whenever I
try to implement a design using XACTstep, map always produces an
exception
error.

Thanks for any ideas,

Chuck


Article: 20059
Subject: Re: Xilinx programming from a Linux PC
From: "Steve Casselman" <sc@vcc.com>
Date: Tue, 25 Jan 2000 17:16:02 -0800
Links: << >>  << T >>  << A >>

Larry Doolittle <ldoolitt@recycle> wrote in message
news:86l7q6$if71@overload.lbl.gov...
> rodger (brownsco@frii.com) wrote:
> : While I have not done this, it should be very doable to
> : prgram an FPGA on a Linux PC.
>
> Of course it can be done.

A small point... Jbits will run on Linux. You don't need any other tools.

Steve Casselman, President
Virtual Computer Corp.


Article: 20060
Subject: CFP - Reconfigurable Technology
From: "John Schewel" <jas@vcc.com>
Date: Tue, 25 Jan 2000 17:18:31 -0800
Links: << >>  << T >>  << A >>
CALL FOR PAPERS

****************************************************************************

Reconfigurable Technology: FPGAs & Reconfigurable Processors

for Computing and Applications


To be held as part of:

SPIE's Photonics East Symposium on Voice Video, and Data Communications,

5-8 November 2000, Boston Massachusetts

****************************************************************************



The past decade has seen a dramatic increase in the use of reconfigurable
logic devices in commerical applications. Among the most significant
milestones in this field are the arrival of million plus gate logic devices
and the introduction of a series of new reconfigurable processors. These
devices are ideal for data-intensive, Internet, DSP,and other high
performance embedded telecom and datacom applications.

Many systems engineers are using reconfigurable technologies to overcome
computation and product development bottlenecks. The advent of million plus
gate counts and advanced manufacturing techniques have made these
reconfigurable devices more economical and practical for computing systems.

This conference, in it's fifth year, focuses on three areas of
reconfigurable technology:

1) new devices and systems

2) tools and techniques

3) high-performance applications

The conference will present papers that illustrate applications and
techniques for using reconfigurable technology in both design and production
cycles. Papers relating to the following areas are solicited:

* field programmable devices

* reconfigurable processors

* programming tools and methodologies for reconfigurable devices & systems

* applications and platforms utilizing reconfigurable technology for:

- network & data intensive applications

- hardware/software codesign

- rapid product development

- high-performance computing

- image, signal, and communication processing

- robotics

- evolvable algorithms



Conference Chairs:

John Schewel, Virtual Computer Corp.;

Peter M. Athanas, Virginia Polytechnic Institute & State Univ.;

Chris Dick, Xilinx Inc.;

John T. McHenry, National Security Agency



IMPORTANT DATES

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


On-site Proceedings requires the following paper submission schedule.

* Deadline for Abstract submission: March 27, 2000 *

* Deadline for final Manuscripts: August 14, 2000 *


SUBMISSION INSTRUCTIONS:

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

Accepted papers are required to make an oral presentation at the conference
in Boston,MA.

Internet http://www.spie.org/info/vv

E-Mail pecalls@spie.org

Fax 1(360)647-1445

Mail SPIE - The International Society for Optical Engineering

PO Box 10

Bellingham, WA 98227

If you have any questions regarding the conference contact the Conference
Chairman, John Schewel jas@vcc.com (+1 818-342-8294).


****************************************************************************



Article: 20061
Subject: Re: Virtex Fine Pitch BGA pcb layout
From: John Larkin <jjlarkin@highlandSnipSniptechnology.com>
Date: Tue, 25 Jan 2000 20:32:46 -0800
Links: << >>  << T >>  << A >>
On Tue, 25 Jan 2000 01:57:53 GMT, bobperl@best_no_spam_thanks.com (Bob
Perlman) wrote:

|On 25 Jan 2000 00:47:04 GMT, krw@attglobal.net (Keith R. Williams)
|wrote:
|
|<stuff snipped>
|>On Virtex/VirtexE?  That's not what Xilinx is saying.  That's not
|>my experience either.  Our product is a BGA and we're having all 
|>sorts of problems supplying the power to the part (note that this
|>is on testers with their inherent problems, so far).
|>> 
|>> 460 is a LOT of caps! 
|>
|>Not really.  I have lots more space for them.  ;-) In fact I'm 
|>using *many* 470uF SMT caps in parallel instead of PTH bulk 
|>capacitors. Parallel is an advantage.  Putting them on the 
|>back-side is an advantage.  I gotta buy the things in reels 
|>anyway so it doesn't matter much if I need one or use a hundred. 
|<more stuff snipped>
|
|One of the best how-many-and-what-kind-of-bypass-capacitors-do-I-need
|articles I've seen is by some signal integrity engineers at Sun,
|titled, "Power Distribution System Design Methodology and Capacitor
|Selection for Modern CMOS Technology."  You can find it online at:
|
|http://www.qsl.net/wb6tpu/si_documents/docs.html
|
|One comment about capacitor placement: if you have a 500ps risetime
|(as more than a few modern devices do) and the bypass capacitor is
|more than an inch or so away from the device's power pins, that cap
|isn't helping source current for that edge, except to replenish plane
|charge after the fact.  Placement does matter, at least for the
|smaller, high-frequency bypass caps (the larger, lower-frequency bulk
|capacitors are another matter).  The plane-to-plane capacitance is
|nice, being low-inductance, but there isn't much of it, particularly
|if you have lots of stuff switching.  CMOS-based designs can have
|rather astonishing dI/dt requirements; I'd think a long time before
|trying to build a board without high-frequency bypass caps.
|
|Take care,
|Bob Perlman
|
|-----------------------------------------------------
|Bob Perlman
|Cambrian Design Works
|Digital Design, Signal Integrity
|http://www.best.com/~bobperl/cdw.htm
|Send e-mail replies to best<dot>com, username bobperl
|-----------------------------------------------------


Bob,

thanks for the reference. The papers are interesting, but both ignore
plane capacitance. A power-ground plane pair with a small dielectric
gap (a few mills can be easily had) is in fact a very low impedance
transmission line. Transients looking into the parallel planes (as I
see with my TDR experiments) see low-Z transmission line in all radial
directions.  My testing indicates that the highest impedance a chip
pin sees is the via that reaches down from the pin to the power or
ground plane; the planes themselves (if the dielectric gap is small)
are very stiff for fast transients. Of course, if you need X colombs
of charge to keep your chip happy, you'd better have n*X farads
available somewhere to supply it. 

I think my point is that, for properly designed multilayer boards, the
planes themselves are an enormous asset in supplying low-impedance
power for fast transients. My TDR tests indicate that additional
bypass caps can be scattered about, and improve the apparent
capacitance a pin sees, without having to be really close to every
pin.  I have never observed the multiple-resonances effects predicted
by the Sun paper Spice models for paralleled caps; I suspect they
didn't model the planes.

As I get braver, I am using fewer bypass caps, spread farther apart,
and am still measuring very clean Vcc and ground noise levels. As
parts get smaller, and have *lots* of pins, one eventually will be
faced with brickwalling the backside of the board with thousands of
head-to-butt capacitors. This will eventually get silly, and our
Manufacturing people will stop saying 'hello' in the hallways.

John

-----------------

"If anybody ever marries you, it will be for the pleasure 
 of hearing you talk piffle," said Harriet, severely.
Article: 20062
Subject: Re: How to access standard sdram ?
From: "Joel Kolstad" <Joel.Kolstad@USA.Net>
Date: Tue, 25 Jan 2000 22:30:58 -0800
Links: << >>  << T >>  << A >>
Free Spirit <Burn@D.man> wrote in message
news:388cfa84.417978246@enews.newsguy.com...
> The read and write are pretty easy, but I had to use a lot of states
> to properly initialize the darn things.

Hmm... no CPU interface around?  Tossing a multiplexor in front of the
registered CAS, RAS, WE and address line outputs and handing the SDRAM data
sheet to the device driver guy can do wonderful things... :-)

---Joel Kolstad



Article: 20063
Subject: Pin to pin
From: =?iso-8859-1?Q?J=E9r=E9mie?= WEBER <jwr@digigram.com>
Date: Wed, 26 Jan 2000 09:57:33 +0100
Links: << >>  << T >>  << A >>
Do you know how to link two pins of a virtex fpga like if it is a wire
between this two pins ( in order to bypass a fpga... )

Article: 20064
Subject: Re: Anyone changed an NT disk serial number?
From: Steve Bird <steveb@vizef.demon.co.uk.island>
Date: Wed, 26 Jan 2000 09:04:19 +0000
Links: << >>  << T >>  << A >>
In article <388e1f97.4722839@news.dial.pipex.com>, eml@riverside-
machines.com.NOSPAM writes
>Anyone managed this? I've tried modding the boot sector with a '95
>boot disk and Debug, but Debug can't even see the C drive, let alone
>modify the boot sector. And does anyone know where the serial number
>is? Presumably the NTFS location is different from the FAT16 and FAT32
>locations.
>
>By the way, this is (I think) legal.
>

Depends on why you would want to do this... If you're trying to subvert
a s/w licensing schema (say FLEXlm) then its not.

Article: 20065
Subject: Re: Atmel config PROMs
From: dulik@my-deja.com
Date: Wed, 26 Jan 2000 09:10:58 GMT
Links: << >>  << T >>  << A >>
Yes, you can use it - they behave the same way as XC17S20. And they
bring you the possibility of in system programming, too - you can
switch them to serial programming mode and change the bitstream without
removing the EEPROM from the board. This can be done without any extra
logic, just modifying the Xilinx design.

Tomas Dulik

In article <86k3gn$gmo$1@nntp5.atl.mindspring.net>,
  "Peter Fenn" <PeteFenn@mindspring.com> wrote:
> Please advise:
> Can Atmel AT17c256 be used in place of Xilinx XC17s20 config PROM?
>
> Regards
> Peter Fenn
>
>


Sent via Deja.com http://www.deja.com/
Before you buy.
Article: 20066
Subject: Anyone has Synplify 5.1 eval ?
From: dulik@my-deja.com
Date: Wed, 26 Jan 2000 09:15:34 GMT
Links: << >>  << T >>  << A >>
Hello, does anybody have the Synplify 5.1.4 eval version ? I have the
5.2.2 and 5.3.1, but don't like them - they both stopped working after
20 days :-)

Tom


Sent via Deja.com http://www.deja.com/
Before you buy.
Article: 20067
Subject: Re: Atmel config PROMs
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Wed, 26 Jan 2000 09:43:09 +0000
Links: << >>  << T >>  << A >>


dulik@my-deja.com wrote:

> Yes, you can use it - they behave the same way as XC17S20. And they
> bring you the possibility of in system programming, too - you can
> switch them to serial programming mode and change the bitstream without
> removing the EEPROM from the board. This can be done without any extra
> logic, just modifying the Xilinx design.

I used these parts a couple of years ago & they were fine EXCEPT for one
problem. The first byte of config data was only loaded into the serialiser
on power-up and not every time RESET was asserted. In fact it was much
worse than this since even a power cycle didn't always load the first byte
correctly.

Checking with Atmel I found that this was a known problem since the first
byte is cached at power on, a new version of the device was supposed to fix
it but it never came out while I was using the parts.

You might be better off with the Xilinx XC1800 series parts that are ISP
via JTAG.



Article: 20068
Subject: Re: Anyone changed an NT disk serial number?
From: Jo Depreitere <jdp_nospam@elis.rug.ac.be>
Date: Wed, 26 Jan 2000 11:59:53 +0100
Links: << >>  << T >>  << A >>
Steve Bird wrote:
> 
> In article <388e1f97.4722839@news.dial.pipex.com>, eml@riverside-
> machines.com.NOSPAM writes
> >Anyone managed this? I've tried modding the boot sector with a '95
> >boot disk and Debug, but Debug can't even see the C drive, let alone
> >modify the boot sector. And does anyone know where the serial number
> >is? Presumably the NTFS location is different from the FAT16 and FAT32
> >locations.
> >
> >By the way, this is (I think) legal.
> >
> 
> Depends on why you would want to do this... If you're trying to subvert
> a s/w licensing schema (say FLEXlm) then its not.

But if you buy a new (faster) PC, shouldn't they allow you to use your
old software? And with FLEXlm you have to fiddle with the serial number
(i.e., if you have a student license which has no support). It's not
exactly legal, but they can't really object to this, since I bought
the software and I want to keep on using it(?)

Jo

-- 
=======================< mailto:jdp@elis.rug.ac.be >=======================
"How things look different on the way down",  Cries on the Wind -- Anathema
====================< http://www.elis.rug.ac.be/~jdp >=====================
Article: 20069
Subject: Re: Virtex Fine Pitch BGA pcb layout
From: eml@riverside-machines.com.NOSPAM
Date: Wed, 26 Jan 2000 11:07:01 GMT
Links: << >>  << T >>  << A >>
On Tue, 25 Jan 2000 20:32:46 -0800, John Larkin
<jjlarkin@highlandSnipSniptechnology.com> wrote:

> My testing indicates that the highest impedance a chip
>pin sees is the via that reaches down from the pin to the power or
>ground plane; the planes themselves (if the dielectric gap is small)
>are very stiff for fast transients. Of course, if you need X colombs
>of charge to keep your chip happy, you'd better have n*X farads
>available somewhere to supply it. 
>
>I think my point is that, for properly designed multilayer boards, the
>planes themselves are an enormous asset in supplying low-impedance
>power for fast transients. My TDR tests indicate that additional
>bypass caps can be scattered about, and improve the apparent
>capacitance a pin sees, without having to be really close to every
>pin.

We're obviously talking about two different things here. If you're
discussing transmission lines, then yes, the planes provide great
return paths, and your edge rate and via-crossing results are very
interesting and useful.

What's normally meant by 'decoupling' is the ability to provide lots
of instantaneous current to your chip, quickly, without relying on a
remote power supply, and this is the first-order problem. The critical
word here is quickly - this means that capacitor placements, and
values, are *critical*. There's a vast amount of literature on this.
I've heard, second-hand, of one person who attempted to build a
digital board without decoupling caps; he claimed the planes were
enough. The description I had of him didn't include the word 'brave'. 

>  I have never observed the multiple-resonances effects predicted
>by the Sun paper Spice models for paralleled caps; I suspect they
>didn't model the planes.

I'm never observed it either, but the theory is simple enough. A real
capacitor is a series LCR. Put two next to each other, add any extra
inductance at the pads, and you're in trouble. Specifically, the 2
LCRs in parallel form a resonant tank circuit. I'm not sure how the
planes can make any significant difference to this, apart from adding
another minimum at very high frequencies. I've got a graph in front of
me (from an old Cypress Hotlink appnote) plotting combined reactance
from a parallel 22nF//100nF combination. This assumes that each cap
has an ESR of 30mohm, and an ESL of 5nH, which they believe is
"achievable on real PCBs using 'good' layouts and surface-mounted
capacitors" (IIRC, you can probably reduce this to about 1nH with very
careful cap selection and an impractical layout). The plot shows a
reactance peak of over 100ohm at about 105MHz, which is not good.

Evan

Article: 20070
Subject: Re: Xilinx programming from a Linux PC
From: eml@riverside-machines.com.NOSPAM
Date: Wed, 26 Jan 2000 11:17:54 GMT
Links: << >>  << T >>  << A >>
On 25 Jan 2000 22:24:06 GMT, ldoolitt@recycle (Larry Doolittle) wrote:

>One of these days Alliance will run on Linux, and I can stop
>mooching off of another group's Sun. 

Roll on. ModelSim's on Linux now, and this probably means that
Spectrum will have another go at Linux before long. At that point I
can find out whether the 'Recycle Bin' can cope with \windoze...

Evan
Article: 20071
Subject: Re: Xilinx Foundation: VHDL to symbol
From: "Alex Sherstuk" <sherstuk@iname.com>
Date: Wed, 26 Jan 2000 16:29:22 +0300
Links: << >>  << T >>  << A >>
1) Use XILINX Foundation Accessories -> Foundation Express to compile your
VHDL files to XNF

2) Use Foundation Schematic Editor -> Hierarchy -> Create Macro Symbol from
Netlist to create schematic symbol, based on XNF-file


Steven Sanders wrote in message <388DA0C9.83D9D662@imec.be>...
>Hello,
>
>I`m a newbie to the Xilinx Foundation Series. I have 5 VHDL files
>(describing 5 components)
>and I want to link these in a top file, but graphical! How do I make
>symbols out of my VHDL code?
>
>Thanx
>
>Steven.
>


Article: 20072
Subject: Re: Virtex Fine Pitch BGA pcb layout
From: John Larkin <jjlarkin@highlandSnipSniptechnology.com>
Date: Wed, 26 Jan 2000 07:58:55 -0800
Links: << >>  << T >>  << A >>
On Wed, 26 Jan 2000 11:07:01 GMT, eml@riverside-machines.com.NOSPAM
wrote:

|On Tue, 25 Jan 2000 20:32:46 -0800, John Larkin
|<jjlarkin@highlandSnipSniptechnology.com> wrote:
|
|> My testing indicates that the highest impedance a chip
|>pin sees is the via that reaches down from the pin to the power or
|>ground plane; the planes themselves (if the dielectric gap is small)
|>are very stiff for fast transients. Of course, if you need X colombs
|>of charge to keep your chip happy, you'd better have n*X farads
|>available somewhere to supply it. 
|>
|>I think my point is that, for properly designed multilayer boards, the
|>planes themselves are an enormous asset in supplying low-impedance
|>power for fast transients. My TDR tests indicate that additional
|>bypass caps can be scattered about, and improve the apparent
|>capacitance a pin sees, without having to be really close to every
|>pin.
|
|We're obviously talking about two different things here. If you're
|discussing transmission lines, then yes, the planes provide great
|return paths, and your edge rate and via-crossing results are very
|interesting and useful.
|
|What's normally meant by 'decoupling' is the ability to provide lots
|of instantaneous current to your chip, quickly, without relying on a
|remote power supply, and this is the first-order problem. The critical
|word here is quickly - this means that capacitor placements, and
|values, are *critical*. There's a vast amount of literature on this.
|I've heard, second-hand, of one person who attempted to build a
|digital board without decoupling caps; he claimed the planes were
|enough. The description I had of him didn't include the word 'brave'. 
|
|>  I have never observed the multiple-resonances effects predicted
|>by the Sun paper Spice models for paralleled caps; I suspect they
|>didn't model the planes.
|
|I'm never observed it either, but the theory is simple enough. A real
|capacitor is a series LCR. Put two next to each other, add any extra
|inductance at the pads, and you're in trouble. Specifically, the 2
|LCRs in parallel form a resonant tank circuit. I'm not sure how the
|planes can make any significant difference to this, apart from adding
|another minimum at very high frequencies. I've got a graph in front of
|me (from an old Cypress Hotlink appnote) plotting combined reactance
|from a parallel 22nF//100nF combination. This assumes that each cap
|has an ESR of 30mohm, and an ESL of 5nH, which they believe is
|"achievable on real PCBs using 'good' layouts and surface-mounted
|capacitors" (IIRC, you can probably reduce this to about 1nH with very
|careful cap selection and an impractical layout). The plot shows a
|reactance peak of over 100ohm at about 105MHz, which is not good.
|
|Evan

Evan,

to paraphrase an old saying, Spice is not Life.

John


-----------------

"If anybody ever marries you, it will be for the pleasure 
 of hearing you talk piffle," said Harriet, severely.
Article: 20073
Subject: Re: Anyone changed an NT disk serial number?
From: Jerry English <jenglish@planetc.com>
Date: Wed, 26 Jan 2000 11:55:37 -0500
Links: << >>  << T >>  << A >>
Jo Depreitere wrote:

> Steve Bird wrote:
> >
> > In article <388e1f97.4722839@news.dial.pipex.com>, eml@riverside-
> > machines.com.NOSPAM writes
> > >Anyone managed this? I've tried modding the boot sector with a '95
> > >boot disk and Debug, but Debug can't even see the C drive, let alone
> > >modify the boot sector. And does anyone know where the serial number
> > >is? Presumably the NTFS location is different from the FAT16 and FAT32
> > >locations.
> > >
> > >By the way, this is (I think) legal.
> > >
> >
> > Depends on why you would want to do this... If you're trying to subvert
> > a s/w licensing schema (say FLEXlm) then its not.
>
> But if you buy a new (faster) PC, shouldn't they allow you to use your
> old software? And with FLEXlm you have to fiddle with the serial number
> (i.e., if you have a student license which has no support). It's not
> exactly legal, but they can't really object to this, since I bought
> the software and I want to keep on using it(?)
>
> Jo
>
> --
> =======================< mailto:jdp@elis.rug.ac.be >=======================
> "How things look different on the way down",  Cries on the Wind -- Anathema
> ====================< http://www.elis.rug.ac.be/~jdp >=====================

 Allow me to regurgitate my experience about this matter.

I too wanted to upgrade my PC to a faster machine. A PII at

266 Mhz is a snail against a 700 Mhz Athlon. So I called up

the FPGA vendor whose package I wanted to move. I was informed

that for them to cut a license for a new machine would involve

obtaining another license from the original vendor and the charges

would be the same, in other words, I had to purchase a new package at

full price. SO off I went to the net and search engines and found several

packages that would allow me to change the serial number on the disk.

I have not upgraded to another PC yet and have not fully researched the license

agreement to see if there was some restriction about moving the software

from one machine to another.  What I have done is used the disk editor to

changed the serial number on another NT drive.

I would like to hear from software vendors who license their products

using the host id. What is their position on moving the software to another

machine.

So the short answer is YES, one can change the serial number on disk drive

that is used under the NT operating system.

regards

Jerry English



Article: 20074
Subject: Has anyone created VHDL code to interface to a 68HC11 SPI port yet?
From: "Don Golding" <dgolding@angelusresearch.com>
Date: Wed, 26 Jan 2000 17:05:48 -0000
Links: << >>  << T >>  << A >>
I am rather new to VHDL and FPGA.  I want to hook a XC4005 using Xilinx
Foundation 1.5 tools and talk to the FPGA through a 68HC11.
Currently, I am taking a course in VHLD at UCI.


Master/Slave arangement...

Thanks
Don Golding





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