[SI-LIST] : RE: [SI-LIST]: Spectraquest Vs. XTK/XNS

About this list Date view Thread view Subject view Author view

From: Todd Westerhoff (twester@hhnetwk.com)
Date: Fri Nov 10 2000 - 06:11:03 PST


SQ has a "timing spreadsheet" built in that can calculate your timing
margins, given some critical assumptions. Basically, you:

a) Load component timing value (Tco min/max & setup/hold) data into the
Allegro database
b) Specify clock speeds, and set budgets for clow skew/jitter
c) Run signal integrity analysis on the selected signals
d) The system calculates the margins

The "critical assumption" part is that the timing model assumes you're
talking about buses with clocked-register to clocked-register interfaces,
without intermediate asynchronous logic. Given that, you can use the
equations

Clockperiod < Tcomax + Setup + Clockskew + Jitter + Flighttimemax
Flightimemin > Hold - Tcomin - Jitter - Clockskew

There is actually an on-line seminar (slides, audio) that explains this
stuff, at:

http://www.netseminar.com/index.cgi?sem_num=308

The seminar explains the timing model in more detail, and goes through cases
where the timing model applies, and where an external static timing analyzer
would be used to supplant the analysis done by the timing spreadsheet.

Hope this helps!

Todd.
---------------------------------------------

Todd Westerhoff
Signal Integrity Engineer
Hammerhead Networks
5 Federal Way
Billerica, MA 01821
ph: 978-671-5084

-----Original Message-----
From: owner-si-list@silab.eng.sun.com
[mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Ken Cantrell
Sent: Thursday, November 09, 2000 8:53 PM
To: Ken Willis; si-list@silab.eng.sun.com
Subject: RE: [SI-LIST] : RE: [SI-LIST]: Spectraquest Vs. XTK/XNS

Ken,
You're a born politician. That was a very diplomatic way of saying that he
CAD guys don't like it when you auto-override them. Same here. Slow but
slow. How about the timing analysis engine, any comments on that?
Ken

-----Original Message-----
From: owner-si-list@silab.eng.sun.com
[mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Ken Willis
Sent: Thursday, November 09, 2000 4:40 PM
To: si-list@silab.eng.sun.com
Subject: [SI-LIST] : RE: [SI-LIST]: Spectraquest Vs. XTK/XNS

Hi everyone,

The interconnect modeling capability of SQ is actually very
strong, with its frequency dependent lossy coupled TLine
functionality, a la W element. If you put in the right
loss tangent values for the materials you are using, you
can get very good agreement. I have gotten very good
eye pattern correlation up to data rates of 2.5 GHz, through
backplanes, connectors, etc. You have to do some extra effort
on via modeling (simple lumped via models don't work well at
those frequencies). But since SQ is spice-based, you can write
your own spice subcircuits for vias when you get to those
frequencies. The spice-based nature allows you to model all
kinds of detailed stuff very well.

But SQ is definitely a toolset for people using Allegro as the
back end. There is some loose integration to import PADS
and Mentor databases, but I don't see this as the strength at all.
The integration with Allegro is a huge plus, and the ability to
quickly extract a graphical electrical topology is something
I use daily, and find invaluable for troubleshooting. It is
also pretty easy to hook multiple (Allegro) databases together
and do massive batch simulations. I have now simulated well over
a dozen complete (big) designs and we have had very good lab
correlation, and caught lots of stuff before going to fab.

On the negative side, documentation is weak compared
to HL or XTK, and the support/training can vary a lot. But if your
company uses an Allegro back end and you know what you are doing,
SQ can be a lifesaver if you need to do industrial-strength levels
of analysis. There is also huge potential to enable
formal constraint-driven design downstream in Allegro, which I
think is largely untapped at this point.

Ken Willis
Sycamore Networks

-----Original Message-----
From: Ken Cantrell [mailto:Ken.Cantrell@srccomp.com]
Sent: Thursday, November 09, 2000 4:05 PM
To: Todd Westerhoff; si-list
Subject: RE: [SI-LIST] : RE: [SI-LIST]: Spectraquest Vs. XTK/XNS

Re: Todd's 2 cents,
I would add that if you are in a large product line environment (on stop
shopping), SpectraQuest is great with it's global auto-update features,
saves you significant time. I thought their interconnect modeling was
weak
though, and their timing analysis depended on your already knowing your
timing margins. Have they updated those features?
Ken

-----Original Message-----
From: owner-si-list@silab.eng.sun.com
[mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Todd Westerhoff
Sent: Thursday, November 09, 2000 12:30 PM
To: si-list
Subject: [SI-LIST] : RE: [SI-LIST]: Spectraquest Vs. XTK/XNS

Okay, there's been enough commentary, so I'll throw my two cents in ...

Hyperlynx can't be beat for ease of use. Michael's original comments
reflect that, and I agree. Hyperlynx is the sort of tool you can
install
and have running in a morning. On the other hand, the waveform display
capabilities of Hyperlynx are limited, so when you get to doing
precision
work, or when you want to enter complex topologies (LineSim), things
tend to
come up short. And .... since Hyperlynx is intended to be a
*simulator*,
you can create and edit the PCB database only through an external CAD
tool.
BoardSim will let you view, but not edit, the PCB database.

XTK is the most established game in town. It's a simulator, too, so you
have to create and edit your PCB database in a separate CAD tool as
well.
Innoveda has a transmission line editor (scratchpad) and a few different
options for viewing the PCB database (PreView, ePlanner) ... but you're
still not really able to edit the PCB database directly. XTK's
algorithms
make the simulation fast, and, because the XTK engine is an ASCII,
netlist-based tool, you can script it to do just about anything.
Because
it's been around a while, chances are there's a switch somewhere to do
just
about anything you can think of.

SPECCTRAQuest is the (relatively) new kid on the block, although the
simulation technology (SigNoise) has been around for a while and is
quite
solid. The simulator is SPICE-based, so don't expect it to outrun XTK.
On
the other hand, with the abundance of fast processors and memory, PCB
simulation speed isn't quite the battleground it used to be.
SPECCTRAQuest
reads and writes the Allegro database format directly, without
translations.
You set the electrical properties (stackup, model assignments, power
nets)
directly in Allegro database and that's it. Once the database is set
up,
you can run signal integrity & crosstalk simulations by selecting the
nets
whose behavior you want to see. You can also "extract" an electrical
model
of a pre- or post-route net directly into the transmission line editor,
and
view/edit/simulate it. Plus, you can move/edit parts or nets in the
Allegro
database, and, you can apply/remove/verify design rules that drive the
placement and routing processes with Allegro. If you've got a complex
design and use Allegro for layout, SPECCTRAQuest is hard to beat. It
costs
more than Hyperlynx, and you won't learn it in a morning. For my money,
simulation results between XTK and SPECCTRAQuest are basically
equivalent
(the finer points of each tool's approach notwithstanding).

And the end of the day - simulators are fine, but it's the ability to
create
designs, run analyses, make decisions and verify results that brings
products to market. Keeping the create-analyze-decide-verify loop time
down
is critical - and that means integration plays a key part. Our designs
are
not simple by any measure, and SPECCTRAQuest is working just fine for
us.

That's my $0.02.

Todd.

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:03 PDT