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
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

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 55400

Article: 55400
Subject: Re: Thermal Data for Logic Devices
From: Ray Andraka <ray@andraka.com>
Date: Tue, 06 May 2003 23:16:37 GMT
Links: << >>  << T >>  << A >>
Better for power consumption?  Probably not.  Better for estimating power,
maybe, but not without a lot of work.

Glen Herrmannsfeldt wrote:

>
>
> Well, I was trying to convince someone that an FPGA based design would be
> better than an ASIC design, before either one was built.
>
> >  "They that give up essential liberty to obtain a little
> >   temporary safety deserve neither liberty nor safety."
> >                                           -Benjamin Franklin, 1759
>
> -- glen

--
--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: 55401
Subject: Re: Xilinx VirtexII Pro Rocket-IO
From: chopra_vikram@excite.com (Vikram)
Date: 6 May 2003 16:20:27 -0700
Links: << >>  << T >>  << A >>
> Andreas Wortmann wrote:
> 
> > hi,
> > 
> > i start using the rocket-io transceiver macro on the xilinx virtex-II pro
> > device. does anybody have
> > - or know of - some simple sample vhdl-code to start out with ? at this
> > stage i do not aim at a specific
> > transmission protocol, but i want to prove the device working. are there
> > other important things i have
> > to account for ?
> > 
> > thanks,
> > andreas

Xilinx has a demo board/design for the RocketIO MGT available at -

http://www.xilinx.com/ipcenter/catalog/search/reference/reference_xapp661_rocketio_transceiver.htm

This is a pretty good reference design for anyone starting on
RocketIO.

-Vikram.

Article: 55402
Subject: Re: PLL chips
From: "John_H" <johnhandwork@mail.com>
Date: Tue, 06 May 2003 23:23:18 GMT
Links: << >>  << T >>  << A >>
Sorry for not posting to the original message - newsgroup troubles at work.
I just saw this part and the newsgroup thread wasn't that old...

The Maxim MAX2150:

"The super-high-resolution sigma-delta fractional-N synthesizer is capable
of better than 50mHz resolution when used with a 10MHz reference. Other
features include fully differential I/Q modulation inputs, an internal LO
buffer, and a 50 wideband output driver amplifier"

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/3430/ln/en


> hmurray@suespammers.org (Hal Murray) wrote in message
news:<vb8rj1h4742fe1@corp.supernews.com>...
> > Is there any obvious PLL chip to use with an FPGA?   I want to do
> > things like generate 100.0003001 MHz from 10 MHz.  (Or generate
> > 100 MHz from 9.999999371 MHz)


begin 666 ohm.gif
M1TE&.#EA# `)`)$#`!(2$HF)B<G)R?___R'Y! $```,`+ `````,``D```(@
BG!498BW QHL&A#8"<!#/TW&4A6E>^!G+Q%X9``L/?!4`.P``
`
end


Article: 55403
Subject: Re: I want a 800 k gates FPGA in 40 pin DIL
From: Ray Andraka <ray@andraka.com>
Date: Tue, 06 May 2003 23:28:38 GMT
Links: << >>  << T >>  << A >>
I was referring to FPGAs, not CPLDs.  THe FPGA boards typically come with an
FPGA, power supply regulators and perhaps a brick, interfaces, PROM etc.  If
you are really bent on getting a 50K device in hobby package, use the 144 pin
tqfp package.  With a few minutes practice, you can solder these onto a board
with a decent soldering iron.  They really are no harder to work with than SOIC
packages.  As I mentioned earlier, you can also mount parts on an adapter such
as those made by Ironwood if you don't mind spending more for the adapter than
you did for the chip.  As for the robot peripherals, I think you'd find
yourself bumping into pin limits pretty quickly in the small pinout devices.
Also, BGAs can work well for low pinout applications on two layer boards.  In
that case, select your I/O to be on the outer ring, to keep the signal routing
on the top layer.  Power goes to inner ring pins, but that can be rotued on the
bottom layer with vias.  You can mount the BGAs using a toaster oven or with a
hot air gun for one up stuff (see the numerous threads that have appeared over
the years here).

Torquemada wrote:

> I have to agree with you. I'm definitely interested in accessible FPGAs
> for hobbyists. Robotics is a pretty popular hobby these days and most
> amateur projects are centred on microcontroller devices. But CPLDs
> and FPGAs are in many ways much more logical. With a microcontroller,
> dealing with multiple tasks with interrupts can be a headache.
> With a programmable logic device the design flows much more naturally
> and dealing with asynchronous events like collisions with a wall becomes
> much easier. But there are few devices that are easy for hobbyists to
> physically work with. The XCR3064XL is OK as you can get 0.1"
> spacing PLCC44 sockets. (Incidentally the lowest cost board for this is
> about $50 but the component itself costs only $3 so that stuff about
> the board being as cheap as the parts is plainly false. I built a working
> programmer for one of these things on a breadboard for about $4).
> But at 1600 gates there's a limit to how much behaviour you can get. I'd
> love to see a 50k gate device, say, in a similar package. There'd be plenty
> of I/O for robot peripherals and enough logic to have it do some fun stuff.
> --
> Torque

--
--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: 55404
Subject: Re: I want a 800 k gates FPGA in 40 pin DIL
From: Ray Andraka <ray@andraka.com>
Date: Tue, 06 May 2003 23:35:48 GMT
Links: << >>  << T >>  << A >>
>

If you really believe that the hobbyist market is a huge untapped market, why
don't you get a BGA or TQFP to dip adapter board made and assembled in some
reasonable quantity and sell them?  There are plenty of fab/assembly houses
that are hungry for work right now, so I'm sure you could wrangle something.  I
would at least put bypass caps on the adapter module.  You might also consider
a socketed serial prom so as to not use up valuable pins for programming.


--
--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: 55405
Subject: Functional simulation model for SelectMAP and/or System ACE MPU port
From: amyks@sgi.com (Amy Mitby)
Date: 6 May 2003 16:48:07 -0700
Links: << >>  << T >>  << A >>
Hi,

I would like to design one fixed FPGA to program (and reprogram) other
FPGAs, directed by the CPU. I asked Xilinx and they said that they
didn't have any currently available functional simulation models for
either the SelectMAP configuration interface or the System ACE MPU
port, two mechanisms I'm considering. Does anyone know where any
non-official but used & tested functional simulation models are
available?

Thanks,
Amy Mitby

Article: 55406
Subject: Re: I want a 800 k gates FPGA in 40 pin DIL
From: "Torquemada" <torquemada1@nospam.sigfpe.com>
Date: Wed, 07 May 2003 00:04:53 GMT
Links: << >>  << T >>  << A >>
> If you really believe that the hobbyist market is a huge untapped market,
why
> don't you get a BGA or TQFP to dip adapter board made and assembled in
some
> reasonable quantity and sell them?  There are plenty of fab/assembly
houses
> that are hungry for work right now, so I'm sure you could wrangle
something.  I
> would at least put bypass caps on the adapter module.  You might also
consider
> a socketed serial prom so as to not use up valuable pins for programming.

If I didn't already have a decently paying job that takes up most of my day
I'd do
exactly this! :-) I'll have to see if I can get some prototypes working...
--
Torque



Article: 55407
Subject: Re: spartan 3 development board
From: christian@navitronic.dk (christian sternell)
Date: 6 May 2003 23:24:18 -0700
Links: << >>  << T >>  << A >>
Well, in case anybody is interested - it seems that memec has a real
nice s3 board (practically) available now.  According to this fellow I
spoke to the other day they'll be delivering them in a couple of
months (in Scandinavia at least).

Article: 55408
Subject: Xilinx XC4000 slave serial programming Q
From: javiercero1@yahoo.com (Frank Martinez)
Date: 6 May 2003 23:49:35 -0700
Links: << >>  << T >>  << A >>
Hi! We have a prototype board with an XC4000 FPGA on int, we are
trying to program it using the serial slave mode, we have the traces
on the board for every pin in the parallel III cable, except the DONE
pin.

Theoretically the DONE pin only lets the programmer know that the
devide was configured correctly right?

The foundation tools' programmer refuses to even download the code
because it claims it can not see a low DONE signal (I assume the
signal transitions from hi to low while the device is being
programmed, and back to low to hi once it is programmed).

Is there a way of telling the programmer in the foundation tools to
ignore the DONE signal (I have not found it), or do we need to "fool"
the program by generating a DONE signal on our own using some discrete
logic. Or does anyone know of a freeware program that downloads the
bitstream files down the parallel cable, and that comes with source so
that it can be modified to ignore DONE?

Please any help will be very very appreciated!!! We do not want to
resubmit the proto board just because a single signal went MIA!!!

Article: 55409
Subject: Re: spartan 3 development board
From: Uwe Bonnes <bon@elektron.ikp.physik.tu-darmstadt.de>
Date: Wed, 7 May 2003 07:52:28 +0000 (UTC)
Links: << >>  << T >>  << A >>
christian sternell <christian@navitronic.dk> wrote:
: Well, in case anybody is interested - it seems that memec has a real
: nice s3 board (practically) available now.  According to this fellow I
: spoke to the other day they'll be delivering them in a couple of
: months (in Scandinavia at least).

"(practically) available now" ... "in a couple of months"

:-(
-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Article: 55410
Subject: Re: PLL chips
From: hmurray@suespammers.org (Hal Murray)
Date: Wed, 07 May 2003 09:08:24 -0000
Links: << >>  << T >>  << A >>
Thanks for all the suggestions/ideas.

Yes, the context is GPS and a very good crystal.

Yes, I'm willing to wait a minute for it to lock up
and I'm willing to consider using a huge cap.

This is a hobby/hack, so the "specs" get adjusted to
fit budget and whatever seems neat at the time.  (Or it
may never get off the drawing board.)

I was sloppy with the number of digits I typed in, but the idea
is that I wanted lots of low order bits.  Should be easy to do
in an FPGA.

The good/fancy crystals I've seen come with an analog input
to tweak the frequency.  With a good crystal, you end up needing
a 16 bit D/A and the crystal aging has to be measured and factored
in when coasting if the GPS runs out of satellites.

I have a HP Z3801A GPS/crystal box.  Ask Google if you
are a geek and like neat toys.  They are surplus from the
cell phone people.

The basic idea is to build a Unix time-of-day clock in a FPGA on
PCI.  That's 32 bits of microseconds-in-this-second, and 32 bits
of seconds since 19xx.  I've got a real-good 10 MHz clock so that
much should be reasonable.

32 bits will hold nanoseconds-in-this-second for more accurate
timimg.  So a faster clock is interesting.  (No real need,
just feels like fun.)  First idea is just a 10x PLL, or something
like that to get a few more low order bits.
(No point in going crazy.  PCI is only 33 MHz.)

Second idea is to put a GPS unit and good crystal on the card too.

OEM GPS units are under $100.  But round up for an antenna.

Good crystals are now available that will fit on a PCI card.
(Maybe not quite within all the rules, but close enough for
this hack.)  I'm not sure about price yet.  But the general
idea should still work with a not super fantastic crystal.

So then I got to thinking about doing the corrections
in the FPGA rather than tweaking the analog knob.

Now I have another batch of ideas to consider...
The board will be mostly empty.  It's free to add a few
more footprints that don't get used.

Thanks again.

-- 
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.


Article: 55411
Subject: Re: PLL chips
From: Jim Granville <jim.granville@designtools.co.nz>
Date: Wed, 07 May 2003 21:42:09 +1200
Links: << >>  << T >>  << A >>
Hal Murray wrote:
> 
> Thanks for all the suggestions/ideas.
> 
> Yes, the context is GPS and a very good crystal.
> 
> Yes, I'm willing to wait a minute for it to lock up
> and I'm willing to consider using a huge cap.
> 
> This is a hobby/hack, so the "specs" get adjusted to
> fit budget and whatever seems neat at the time.  (Or it
> may never get off the drawing board.)
> 
> I was sloppy with the number of digits I typed in, but the idea
> is that I wanted lots of low order bits.  Should be easy to do
> in an FPGA.
> 
> The good/fancy crystals I've seen come with an analog input
> to tweak the frequency.  With a good crystal, you end up needing
> a 16 bit D/A and the crystal aging has to be measured and factored
> in when coasting if the GPS runs out of satellites.
> 
> I have a HP Z3801A GPS/crystal box.  Ask Google if you
> are a geek and like neat toys.  They are surplus from the
> cell phone people.
> 
> The basic idea is to build a Unix time-of-day clock in a FPGA on
> PCI.  That's 32 bits of microseconds-in-this-second, and 32 bits
> of seconds since 19xx.  I've got a real-good 10 MHz clock so that
> much should be reasonable.
> 
> 32 bits will hold nanoseconds-in-this-second for more accurate
> timimg.  So a faster clock is interesting.  (No real need,
> just feels like fun.)  First idea is just a 10x PLL, or something
> like that to get a few more low order bits.
> (No point in going crazy.  PCI is only 33 MHz.)
> 
> Second idea is to put a GPS unit and good crystal on the card too.
> 
> OEM GPS units are under $100.  But round up for an antenna.
> 
> Good crystals are now available that will fit on a PCI card.
> (Maybe not quite within all the rules, but close enough for
> this hack.)  I'm not sure about price yet.  But the general
> idea should still work with a not super fantastic crystal.
> 
> So then I got to thinking about doing the corrections
> in the FPGA rather than tweaking the analog knob.
> 
> Now I have another batch of ideas to consider...
> The board will be mostly empty.  It's free to add a few
> more footprints that don't get used.
> 
> Thanks again.

 Since this is a corrected timebase, that will mostly be 
extrnally locked, look at the digital calib schemes used on
some watch chips. They use a fixup scheme, of "so many LSB's" to 
nudge the timebase. - No caps, or analog, needed.

 eg I asssume you will want seconds to be as precise 
over 12 months as possible, but are perhaps not so worried 
if the counts per millisecond deviates by some few ppm.

 Also look at the Dual Crystal precision oscillators. 
Rather than a thermister and xtal, they use two crystals of 
different cuts, and the delta Freq gives the nett 
effective temperature.
 Measuring two Freqs sounds like a FPGA task :)

-jg

Article: 55412
Subject: Re: LPM_ROM problem with Altera EP1K50 parts
From: IJ.McCrum@ulst.ac.uk (Ian McCrum, MI5AFL)
Date: Wed, 07 May 2003 09:57:54 GMT
Links: << >>  << T >>  << A >>
On Sun, 04 May 2003 18:26:28 GMT, "FPGA user" <nospam@aol.com> wrote:

>Hi,
>
>Has anyone used LPM_ROM megafunctions in an EP1K50?  They do NOT seem to
>work in simulation.  However, an EP1K30 or an EP1K100 DOES work.
>
>I'm using the latest version of MaxPlus II, and created a design with only
>that megafunction, unclocked, with addr in/data out ports, and an
>initialization file.  I organized the part as 64 16 bit words.
>
>Compilling and changing nothing but the device shows up the problem.
>
>
>Thanks!
>
>
I have used LPM_ROM for a large(wide) microcode design, I recall it
fitted/worked better if I split the ROMs into 8 bit chunks, so two
64x8 roms may be better. I think I retrospectively read something in
the altera literature to do this ... YMMV

Email me if still having probs as I probably have old code on my
website www.eej.ulst.ac.uk try IJ.McCrum as my email username and
ulster.ac.uk ad my email domain.

Regards
Ian McCrum, MI5AFL

Article: 55413
Subject: OT: looking for I/Q mixers/modulators for TX and RX
From: "Ken" <aeu96186_MENOWANTSPAM@yahoo.co.uk>
Date: Wed, 7 May 2003 11:23:10 +0100
Links: << >>  << T >>  << A >>
(I posted a similar message before whose subject was "RF
transmitters/receivers with Xilinx Xtreme DSP Kit" - I am reposting with
better details of what I need - this is not strictly do do with FPGAs but I
hope some of you guys may have experience with this stuff - I may well do
the I/Q modulation/demodulation on FPGA anyway so it is somewhat relevant...
;-)   )



Hello,

I am looking for several pieces of off the shelf equipment to perform RF
transmission and reception across a table-top/room.  The gear can therefore
be very low power.  The RF carrier frequency is flexible and a few hundred
MHz would do although GHz would be nice - I guess if I track down a range of
suitable equipment I will be able to trade Hz against cash!

I have seen quite a lot of expensive and complicated pieces of equipment
that will do the simple tasks I require but I cannot justify the cost when
all I really need is some sin/cos mixing and modulation.  I have searched
some online catalogues and spoken to reps from various companies with no
success so far - surely these simple boxes must exist to buy!

Anyway, here is the list:

TX Box 1:
Inputs:  1 BNC input carrying an IF modulated signal
This box just needs to AM modulate the IF input to an RF frequency and
possibly perform some sort of filtering to reject images?
Outputs:  1 - something suitable for feeding to an antenna...? not quite
sure

RX Box 2:
Inputs:  suitable connector coming from an antenna I guess
This box needs to demodulate the RF signal from box 1 and output it at some
IF or baseband.
This box would need to do suitable filtering to demodulate with no problems.
Outputs: 1 BNC output.

TX Box 3:
Inputs:  2 BNC inputs carrying I and Q signals at baseband/IF
This box needs to mix the I/Q signals with cos/sin waves and add them
together to produce an RF signal
Outputs: 1 - something suitable for feeding to an antenna

RX Box 4:
Inputs:  suitable connector coming from an antenna
This box needs to mix the input with a cosine to get the I output and mix
the input with a sine to get the Q output.
This box would also need to do suitable filtering to demodulate with no
problems.
Outputs: 2 BNC outputs carrying I and Q.

Some of the boxes may not need filters as the ADCs/DACs I will be using can
possibly do this job depending on the specs.

Any pointers towards suitable gear/places to buy it would be very much
appreciated!

Thanks in advance for your time,

Ken


--
To reply by email, please remove the _MENOWANTSPAM from my email address.



--
To reply by email, please remove the _MENOWANTSPAM from my email address.



Article: 55414
Subject: Re: looking for I/Q mixers/modulators for TX and RX
From: "Leon Heller" <leon_heller@hotmail.com>
Date: Wed, 7 May 2003 11:15:53 +0000 (UTC)
Links: << >>  << T >>  << A >>



"Ken" <aeu96186_MENOWANTSPAM@yahoo.co.uk> wrote in message
news:b9ams2$bjb$1@dennis.cc.strath.ac.uk...
> (I posted a similar message before whose subject was "RF
> transmitters/receivers with Xilinx Xtreme DSP Kit" - I am reposting with
> better details of what I need - this is not strictly do do with FPGAs but
I
> hope some of you guys may have experience with this stuff - I may well do
> the I/Q modulation/demodulation on FPGA anyway so it is somewhat
relevant...
> ;-)   )
>
>
>
> Hello,
>
> I am looking for several pieces of off the shelf equipment to perform RF
> transmission and reception across a table-top/room.  The gear can
therefore
> be very low power.  The RF carrier frequency is flexible and a few hundred
> MHz would do although GHz would be nice - I guess if I track down a range
of
> suitable equipment I will be able to trade Hz against cash!

I've got a simple receiver design on my web site that will do conversion
from up to 500 MHz down to baseband I and Q, if you use two of them with a
quadrature oscillator. It's quite easy to make with my PCB design.

Leon
-- 
Leon Heller, G1HSM
leon_heller@hotmail.com
http://www.geocities.com/leon_heller



Article: 55415
Subject: Problem erasing Coolrunner
From: "Ondrej Zoubek" <zuban@atlas.cz>
Date: Wed, 7 May 2003 14:56:00 +0200
Links: << >>  << T >>  << A >>
Hi all,
I have problem with erasing XCR3064XL.

I'm programming it from .svf file created with iMPACT. I have read, there
was a problem with it, but in 5.1 SP3 it should be fixed. I have that
version.
I'm using .xsvf file, conversion utility svf2xsvf.exe, attached with
xapp058.

In the .svf file, there is one line (in the erase procedure)
STATE DRCAPTURE DRPAUSE;
which the utility svf2xsvf don't understand. I don't know why, because it
knows STATE commands with multiple parameters.
It must understand it badly, because it says error, 'DRCAPTURE invalid
stable state'.

I rewrite it to STATE DRPAUSE; and see how the states changes.
It's going in that way: PauseIR -> Exit2IR -> UpdateIR -> SelectDR ->
CaptureDR -> Exit1DR -> PauseDR.

Does anybody know what's going wrong?
Can be somewhere found newer version of the utility svf2xsvf? (I have 5.00)

I'm programming it using playxsvf, with rewritten ports.c


Thanks for answers




Article: 55416
Subject: Re: flash-disk
From: "Mikhail" <kostkin@asicdesign.ru>
Date: Wed, 7 May 2003 17:52:15 +0400
Links: << >>  << T >>  << A >>
I mean that disk have to work in hard conditions (vibration, hits etc.)

"Michael Condon" <michael.condon@nrl.navy.mil> сообщил/сообщила в новостях
следующее: news:b98e94$662$1@ra.nrl.navy.mil...
> did you mean external ?!?
> "Mikhail" <kostkin@asicdesign.ru> wrote in message
> news:b983v1$al2$1@news.wplus.spb.ru...
> > Could anyone recommend me flash-disk with IDE/ATA for "extremal"
> > application?
> >
> > I found BiTMICRO (www.bitmicro.com) but few of thier technical
> > characteristic seems strange ...
> >
> >
> >
>
>



Article: 55417
Subject: Re: How failures happen, and how they don't
From: Weifeng Xu <wxu@ecs.umass.edu>
Date: Wed, 07 May 2003 14:26:28 GMT
Links: << >>  << T >>  << A >>

Austin Lesea wrote:

> Michael,
>
> Faults are things like oxide break down, or junction breakdown.
>
> They are most likely from weak or originally defective oxides or junctions
> that eventually broke from eventual overstress (due to their original weak
> nature).
>
> These kinds of defects can not be totally avoided.  One can burn in devices,
> one can overstress devices, but lately there is a lot of data that shows
> that doing so actually increases field failure rates (if you stress too
> much, if you stress not much at all, it does very little for the long term
> failure rate).
>

   I am wondering whether the soft errors (if it does occur) in configuration
memeory cell will
destroy the device and bring in some permanent faults. From above discussion, it
seems even
even contension won't break the device easily.
   For example, what if a bidirectional hex line is driven by two drivers
because one tri-buffer
was opened due to SEU effect. Can the virtex device still tolerate that, if so,
for how long?
   Thanks!

>
> This whole subject has nothing to do with FPGAs, so you should research
> semiconductor reliability.
>
> Soft fails are another issue, and the following link covers that:
>
>  http://www.xilinx.com/support/techxclusives/1000-techX35.htm
>
> Austin
>
> Michael Garvie wrote:
>
> > Given your FIT rates for high temperature, high moisture, electrical
> > overstress, or purely random; what is the fault mode induced?  Are these
> > localized and permanent?  Or do they go away with scrubbing?  Or are
> > they at the FPGA subsystem level?
> > Regards,
> > Miguel
> >
> > Austin Lesea wrote:
> >
> > > I suggest that if you have suddenly experienced a failure, that is the
> > > kind of failure known as "random" and fits well within our FIT rate
> > > predictions.
> > >
> > > As everyone knows, failures do happen, and if the designers, industry
> > > and foundries could prevent that from happening, we would all be
> > > happier when we have to design fail-safe systems.
> > >
> > > So, until then, fail safe systems fail, by failing to fail safely.
> > >
> > > Austin




Article: 55418
Subject: Re: Using Cyclone's PLL
From: Austin Lesea <Austin.Lesea@xilinx.com>
Date: Wed, 07 May 2003 07:42:54 -0700
Links: << >>  << T >>  << A >>
Greg,

You forgot to mention that you have the jitter transfer function in an
app note (I'll let you reference it), and the PLL is well behaved with
no attenuation out to ~ 6 MHz, and then a small peak (typical of a
critical damped stable loop) of a little jitter gain at around 6 MHz,
and then the 20dB/decade classic slope attenuating jitter after that.

The same plot also shows the Virtex II DCM jitter transfer function,
which is identical from DC to 6 MHz, with just slightly higher jitter
due to the 60 ps tap changes, and then it remains flat beyond 6 MHz (no
attenuation at all).

In comparison, a ICS PLL (8745) has a cutoff pole at 300 KHz, and thus
would attenuate far more jitter (what we recommend if one needs to
"clean up" the jitter).

PLLs in FPGAs can not afford to have the pole set that low, as the
capacitor size becomes huge, and that makes it a cost prohibitive
feature. Thus the PLLs have the smallest possible capacitor to meet the
intended application, and remain stable over process, voltage, and
temperature.

As well, these plots do little to show how they behave in a real design
with IOs switching, and logic in the core toggling.

"We have not come to bury PLLs, but to praise them...." (apologies for
the miss-qoute of Brutus to Ceasar....).

(We use PLLs in the Multi-gigabit transceivers in Virtex II Pro to
multiply the reference clock by a factor of 20 from 156.25 MHz to 3.125
GHz, so we know all about PLLs, too.  Thanks to Altera for the jitter
transfer curves.  By the way, how about publishing the jitter tolerance
curves, and residual jitter output curves, too?  One of the classic set
of three is insufficient to tell the whole story about a PLL.  It may be
good at attenuating, but will it remain locked in the presence of input
jitter?)

Austin

PS: Phil, I have found the button, and I apologize for being "newgroup
archive hostile."


Greg Steinke wrote:
> 
> Joona,
> On Cyclone, the way to change the PLL configuration is to reconfigure
> the entire part. So for your particular application, what you could do
> is store 8 possible configurations, and choose which one configures
> the part depending on the user input. Since the user is going to
> select the multiplication factor, it is probably not changing very
> often. A way to do this would be to store the configurations in a
> parallel Flash memory and use a MAX device (CPLD) to select which
> configuration is used.
> 
> The Enhanced PLLs on the Stratix devices do support real-time
> reconfiguration without the need to reconfigure the entire device. For
> example, logic within the device can change the multiplication and
> division factors to meet your needs. This process is documented in
> AN282:
> http://www.altera.com/literature/an/an282.pdf
> 
> For your final question on the jitter:
> The relationship of input jitter to output jitter on a PLL is somewhat
> complicated. It is not the case that multiplying the frequency by 5
> (as in your example) will multiply the jitter by 5. In general, a PLL
> will pass through low-frequency jitter, but reduce high-frequency
> jitter. A high-level explanation is that the PLL actually regenerates
> the clock and does not simply pass it through with some delay, so
> depending on its characteristics it will not pass through all
> high-frequency variation of the clock period. Therefore high-frequency
> jitter is attenuated. On the other hand, a slow variation of the clock
> period (low-frequency jitter) will be tracked by the PLL oscillator
> and so will be passed through to the output.
> 
> The documentation for the various devices shows what the output jitter
> will be, and this is specified for any multiplication factor. In some
> cases there is a different spec depending on what multiplication
> factor is used.
> 
> For more details on the Stratix PLL and how it implements frequency
> division/multiplication please consult:
> http://www.altera.com/literature/hb/stx/ch_1_vol_2.pdf
> For more details on the Cyclone PLL please consult:
> http://www.altera.com/literature/an/an251.pdf
> 
> Sincerely,
> Greg Steinke
> gregs@altera.com
> 
> jonesky1@hotmail.com (Joona R) wrote in message news:<2f3990c3.0304261139.3be78e8c@posting.google.com>...
> > Hello!
> >
> > Can I change Cyclone PLL's divisor and multiply counter values on the
> > fly? I will use Quartus, after license has arrived.
> > I want that user can change clock frequency of device by pressing a
> > button. Needed frequencys are between 30MHz and 100MHz. Frequencys may
> > be for example 10MHz distances (30, 40, 50, 60, 70, 80, 90 and
> > 100MHz).
> >
> > I was thinking, if input frequency will be 20MHz. Then it is divided
> > by two and then multiplyed with user's wanted value (3 to 10). Is this
> > reasonable way? How is input frequency's jitter effecting to output
> > freq? If fIn is 20MHz with 50ppm, is fOut then 100MHz with 250ppm?
> > (M=10, N=2)
> >
> > Is this possible or is there some better way to do this? Good ideas
> > are welcome! =)
> >
> > Best Regards,
> >  Joona

Article: 55419
Subject: Re: How failures happen, and how they don't
From: Austin Lesea <Austin.Lesea@xilinx.com>
Date: Wed, 07 May 2003 07:53:41 -0700
Links: << >>  << T >>  << A >>
Weifeng,

We took a device, (actually a number of them) and placed them in
contention purposely. Current densities of 10-15 million amperes per cm
squared.  Far over any electromigration limits.

We then took readings every month.

That started over four years ago.

No problems whatsoever.

When we say these parts are tough, we mean it.  We have no idea what the
customer will do, and sometimes the static stress of contention is less
than the operating stress a typical customer may require!

Some anit-fuse FPGA vendor(s) have a lot of foolish FUD (fear,
uncertainty, doubt) out there which is totally wrong, false, and
intentionally mis-leading.  We have countered that with the "1000 years
between single event upsets" tech exclusive on the public website.

This tech exclusive by Peter and myself has been a best seller, as
evidenced by doing a search in Google.  It is #1 on the "hit" list for a
few months now.  Never been #1 on Google before.......

(search for:  1000 years between single event upsets)

Austin

Weifeng Xu wrote:
> 
> 
> 
> Austin Lesea wrote:
> 
> > Michael,
> >
> > Faults are things like oxide break down, or junction breakdown.
> >
> > They are most likely from weak or originally defective oxides or
> > junctions
> > that eventually broke from eventual overstress (due to their
> > original weak
> > nature).
> >
> > These kinds of defects can not be totally avoided.  One can burn in
> > devices,
> > one can overstress devices, but lately there is a lot of data that
> > shows
> > that doing so actually increases field failure rates (if you stress
> > too
> > much, if you stress not much at all, it does very little for the
> > long term
> > failure rate).
> >
> 
>    I am wondering whether the soft errors (if it does occur) in
> configuration memeory cell will
> destroy the device and bring in some permanent faults. From above
> discussion, it seems even
> even contension won't break the device easily.
>    For example, what if a bidirectional hex line is driven by two
> drivers because one tri-buffer
> was opened due to SEU effect. Can the virtex device still tolerate
> that, if so, for how long?
>    Thanks!
> 
> >
> > This whole subject has nothing to do with FPGAs, so you should
> > research
> > semiconductor reliability.
> >
> > Soft fails are another issue, and the following link covers that:
> >
> >  http://www.xilinx.com/support/techxclusives/1000-techX35.htm
> >
> > Austin
> >
> > Michael Garvie wrote:
> >
> > > Given your FIT rates for high temperature, high moisture,
> > electrical
> > > overstress, or purely random; what is the fault mode induced?  Are
> > these
> > > localized and permanent?  Or do they go away with scrubbing?  Or
> > are
> > > they at the FPGA subsystem level?
> > > Regards,
> > > Miguel
> > >
> > > Austin Lesea wrote:
> > >
> > > > I suggest that if you have suddenly experienced a failure, that
> > is the
> > > > kind of failure known as "random" and fits well within our FIT
> > rate
> > > > predictions.
> > > >
> > > > As everyone knows, failures do happen, and if the designers,
> > industry
> > > > and foundries could prevent that from happening, we would all be
> >
> > > > happier when we have to design fail-safe systems.
> > > >
> > > > So, until then, fail safe systems fail, by failing to fail
> > safely.
> > > >
> > > > Austin

Article: 55420
Subject: Re: I want a 800 k gates FPGA in 40 pin DIL
From: Ray Andraka <ray@andraka.com>
Date: Wed, 07 May 2003 15:00:09 GMT
Links: << >>  << T >>  << A >>
It shouldn't take you more than a week or so worth of nights and weekends to
get a netlist and layout package together for a board house to produce it.  You
might talk to a few board houses first to get pricing as well as options for
input format.  Some only work with specific board layout packages, others are
far more flexible.  Some will even do the netlist capture for an additional
fee.

Torquemada wrote:

> ? If you really believe that the hobbyist market is a huge untapped market,
> why
> ? don't you get a BGA or TQFP to dip adapter board made and assembled in
> some
> ? reasonable quantity and sell them?  There are plenty of fab/assembly
> houses
> ? that are hungry for work right now, so I'm sure you could wrangle
> something.  I
> ? would at least put bypass caps on the adapter module.  You might also
> consider
> ? a socketed serial prom so as to not use up valuable pins for programming.
>
> If I didn't already have a decently paying job that takes up most of my day
> I'd do
> exactly this! :-) I'll have to see if I can get some prototypes working...
> --
> Torque

--
--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: 55421
Subject: Re: OT: looking for I/Q mixers/modulators for TX and RX
From: Ray Andraka <ray@andraka.com>
Date: Wed, 07 May 2003 15:06:56 GMT
Links: << >>  << T >>  << A >>
Ken,

Do yourself a favor and buy, borrow or steal a copy of the ARRL Handbook for
Radio Amateurs.  It has simple circuits in it to do exactly what you want at a
cost far lower than any commercially available solutions.

Ken wrote:

> (I posted a similar message before whose subject was "RF
> transmitters/receivers with Xilinx Xtreme DSP Kit" - I am reposting with
> better details of what I need - this is not strictly do do with FPGAs but I
> hope some of you guys may have experience with this stuff - I may well do
> the I/Q modulation/demodulation on FPGA anyway so it is somewhat relevant...
> ;-)   )
>
> Hello,
>
> I am looking for several pieces of off the shelf equipment to perform RF
> transmission and reception across a table-top/room.  The gear can therefore
> be very low power.  The RF carrier frequency is flexible and a few hundred
> MHz would do although GHz would be nice - I guess if I track down a range of
> suitable equipment I will be able to trade Hz against cash!
>
> I have seen quite a lot of expensive and complicated pieces of equipment
> that will do the simple tasks I require but I cannot justify the cost when
> all I really need is some sin/cos mixing and modulation.  I have searched
> some online catalogues and spoken to reps from various companies with no
> success so far - surely these simple boxes must exist to buy!
>
> Anyway, here is the list:
>
> TX Box 1:
> Inputs:  1 BNC input carrying an IF modulated signal
> This box just needs to AM modulate the IF input to an RF frequency and
> possibly perform some sort of filtering to reject images?
> Outputs:  1 - something suitable for feeding to an antenna...? not quite
> sure
>
> RX Box 2:
> Inputs:  suitable connector coming from an antenna I guess
> This box needs to demodulate the RF signal from box 1 and output it at some
> IF or baseband.
> This box would need to do suitable filtering to demodulate with no problems.
> Outputs: 1 BNC output.
>
> TX Box 3:
> Inputs:  2 BNC inputs carrying I and Q signals at baseband/IF
> This box needs to mix the I/Q signals with cos/sin waves and add them
> together to produce an RF signal
> Outputs: 1 - something suitable for feeding to an antenna
>
> RX Box 4:
> Inputs:  suitable connector coming from an antenna
> This box needs to mix the input with a cosine to get the I output and mix
> the input with a sine to get the Q output.
> This box would also need to do suitable filtering to demodulate with no
> problems.
> Outputs: 2 BNC outputs carrying I and Q.
>
> Some of the boxes may not need filters as the ADCs/DACs I will be using can
> possibly do this job depending on the specs.
>
> Any pointers towards suitable gear/places to buy it would be very much
> appreciated!
>
> Thanks in advance for your time,
>
> Ken
>
> --
> To reply by email, please remove the _MENOWANTSPAM from my email address.
>
> --
> To reply by email, please remove the _MENOWANTSPAM from my email address.

--
--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: 55422
Subject: Re: OT: looking for I/Q mixers/modulators for TX and RX
From: "Ken" <aeu96186_MENOWANTSPAM@yahoo.co.uk>
Date: Wed, 7 May 2003 16:11:49 +0100
Links: << >>  << T >>  << A >>

OK Ray - you've convinced me - if it has circuit diagrams then I can get our
workshop to build exactly what I want.

Cheers,

Ken


"Ray Andraka" <ray@andraka.com> wrote in message
news:3EB921E5.4783FC52@andraka.com...
> Ken,
>
> Do yourself a favor and buy, borrow or steal a copy of the ARRL Handbook
for
> Radio Amateurs.  It has simple circuits in it to do exactly what you want
at a
> cost far lower than any commercially available solutions.
>
> Ken wrote:
>
> > (I posted a similar message before whose subject was "RF
> > transmitters/receivers with Xilinx Xtreme DSP Kit" - I am reposting with
> > better details of what I need - this is not strictly do do with FPGAs
but I
> > hope some of you guys may have experience with this stuff - I may well
do
> > the I/Q modulation/demodulation on FPGA anyway so it is somewhat
relevant...
> > ;-)   )
> >
> > Hello,
> >
> > I am looking for several pieces of off the shelf equipment to perform RF
> > transmission and reception across a table-top/room.  The gear can
therefore
> > be very low power.  The RF carrier frequency is flexible and a few
hundred
> > MHz would do although GHz would be nice - I guess if I track down a
range of
> > suitable equipment I will be able to trade Hz against cash!
> >
> > I have seen quite a lot of expensive and complicated pieces of equipment
> > that will do the simple tasks I require but I cannot justify the cost
when
> > all I really need is some sin/cos mixing and modulation.  I have
searched
> > some online catalogues and spoken to reps from various companies with no
> > success so far - surely these simple boxes must exist to buy!
> >
> > Anyway, here is the list:
> >
> > TX Box 1:
> > Inputs:  1 BNC input carrying an IF modulated signal
> > This box just needs to AM modulate the IF input to an RF frequency and
> > possibly perform some sort of filtering to reject images?
> > Outputs:  1 - something suitable for feeding to an antenna...? not quite
> > sure
> >
> > RX Box 2:
> > Inputs:  suitable connector coming from an antenna I guess
> > This box needs to demodulate the RF signal from box 1 and output it at
some
> > IF or baseband.
> > This box would need to do suitable filtering to demodulate with no
problems.
> > Outputs: 1 BNC output.
> >
> > TX Box 3:
> > Inputs:  2 BNC inputs carrying I and Q signals at baseband/IF
> > This box needs to mix the I/Q signals with cos/sin waves and add them
> > together to produce an RF signal
> > Outputs: 1 - something suitable for feeding to an antenna
> >
> > RX Box 4:
> > Inputs:  suitable connector coming from an antenna
> > This box needs to mix the input with a cosine to get the I output and
mix
> > the input with a sine to get the Q output.
> > This box would also need to do suitable filtering to demodulate with no
> > problems.
> > Outputs: 2 BNC outputs carrying I and Q.
> >
> > Some of the boxes may not need filters as the ADCs/DACs I will be using
can
> > possibly do this job depending on the specs.
> >
> > Any pointers towards suitable gear/places to buy it would be very much
> > appreciated!
> >
> > Thanks in advance for your time,
> >
> > Ken
> >
> > --
> > To reply by email, please remove the _MENOWANTSPAM from my email
address.
> >
> > --
> > To reply by email, please remove the _MENOWANTSPAM from my email
address.
>
> --
> --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: 55423
Subject: Re: OT: looking for I/Q mixers/modulators for TX and RX
From: brad@tinyboot.com (Brad Eckert)
Date: 7 May 2003 09:21:19 -0700
Links: << >>  << T >>  << A >>
"Ken" <aeu96186_MENOWANTSPAM@yahoo.co.uk> wrote in message news:<b9ams2$bjb$1@dennis.cc.strath.ac.uk>...
> 
> Any pointers towards suitable gear/places to buy it would be very much
> appreciated!
> 
Here's a web page with pointers to newsgroups that can help:
http://www.ve2dm.net/ve/hamnews.shtml

Article: 55424
Subject: Re: PLL chips
From: Larry Doolittle <ldoolitt@recycle.lbl.gov>
Date: Wed, 7 May 2003 16:29:42 +0000 (UTC)
Links: << >>  << T >>  << A >>
In article <3EB8D4F1.427@designtools.co.nz>, Jim Granville wrote:
>> 
>> Yes, the context is GPS and a very good crystal.
>> I was sloppy with the number of digits I typed in, but the idea
>> is that I wanted lots of low order bits.  Should be easy to do
>> in an FPGA.
>> The good/fancy crystals I've seen come with an analog input
>> to tweak the frequency.  With a good crystal, you end up needing
>> a 16 bit D/A and the crystal aging has to be measured and factored
>> in when coasting if the GPS runs out of satellites.
>> 
>> So then I got to thinking about doing the corrections
>> in the FPGA rather than tweaking the analog knob.
> 
>  Since this is a corrected timebase, that will mostly be 
> extrnally locked, look at the digital calib schemes used on
> some watch chips. They use a fixup scheme, of "so many LSB's" to 
> nudge the timebase. - No caps, or analog, needed.

I agree.  In general, if you can replace hardware with firmware you
come out ahead.  In this case, you want to build the inverse of a
DDS.  Calibrate the number of femptoseconds per tick of the crystal,
and just do a very wide accumulate on every tick.  The high order
bits will give you the time (you probably don't want to expose the
femptoseconds part to the software).

>  eg I asssume you will want seconds to be as precise 
> over 12 months as possible, but are perhaps not so worried 
> if the counts per millisecond deviates by some few ppm.

Direct phase accumulation is never off by more than one lsb.

>  Measuring two Freqs sounds like a FPGA task :)

_Everything_ sounds like an FPGA task these days.  :-)

    - Larry



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
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

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