[SI-LIST] Re: SI Engineer Qualification
- From: "Hassan O. Ali" <hassan@xxxxxxxx>
- Date: Sat, 29 Dec 2001 15:23:53 -0500
Sainath/Ken,
Computational electromagnetics (CE) gives you an insight on
basic/advanced electromagnetic theory and the computational methods used
by various SI related tools. But it doesn't necessarily give you much
applied knowledge on high-speed board-level and chip-level SI issues. A
person trained on CE gets much of SI knowledge through working with
high-speed board and chip designers - after his graduation.
There is essentially two things: "SI Theory" and "SI Practice":
"SI Theory" (IMHO) doesn't have to be the knowledge of Maxwell's
equations and such hardcore EM theory. It just entails the knowledge of
what causes "SI Issues" such as reflection/ringing/overshoot/undershoot
(impedance mismatch, topology effects), crosstalk (coupling),
simultaneous switching noise (effects of package and plane parasitics),
media loss (effects of material properties, link length etc), jitter
(device, pattern-dependent, etc.), etc.
"SI Practice" ventures to *prevent* and (in after-thought) *solve* "SI
Issues". It entails "SI Design", "SI Modeling", and "SI Simulation". "SI
Design" deals with the tactics of *preventing* SI issues. It includes
(among other things) the knowledge of choosing:
- appropriate device (parallel/serial) interface technologies, (HSTL,
SSTL, LVDS, CML, etc)
- appropriate interface topologies (point-to-point, point-to-multipoint
- star, daisy-chain, uni-/bi-directional, etc)
- appropriate board/chip layup (stackup) and transmission line types -
stripline (single/diff-pair), microstrip lines, coplanar waveguide, etc.
- appropriate board/chip materials - FR4, Getek, N4000-13, etc.
- appropriate link lengths (min/max length rules)
- appropriate component placement
- appropriate crosstalk (parallelism/spacing) and via count rules
- appropriate termination schemes
- etc.
In much of today's high-speed design we deal with marginal SI
performance, therefore any scrupulous SI designer would not rely only on
her "gut feelings" but would normally make sure that she has covered all
the bases. There are variations everywhere:
- device corners: fast, typical, slow
- tolerances of discretes (resistors, capacitors, used in termination,
decoupling, etc.)
- package parasitic woes
- board/chip layout tolerances (which effect timing design, diff-pair
skews, etc.)
- media variations: manufacturing tolerances for the chip/board which
affects transmission line properties
- link length variations
- variation in via and other interconnect schemes for the same class of
signals
- etc
This is where "SI Simulation" plays its role. An SI designer can
investigate thousands and thousands of cases in her "solution space" and
come up with various "SI Rules" to *prevent* SI issues. SI simulation is
carried out using various simulation tools such as HSPICE,
SPECCTRAQuest, XTK, Hyperlynx (LineSim/BoardSim), etc. The only
essential knowledge is *how to use* those tools. Knowing how the engines
of those tools work is just a bonus which a CE trained person will have
an edge on.
But "SI Simulation" requires a great deal of models. Models for the
device I/Os (HSPICE, IBIS, etc.) and their packages; for the
interconnects - transmission lines, vias, junctions, etc. In most cases
one deals with generic interconnects that are well modelled by the
available SI tools, but in some few cases a weird interconnect can
arrise that may need more accurate characterization using appropriate
interconnect modeling tools (aka field solvers). This is where a CE
trained person will show her appropriate strength. Appropriate use of
field solvers in interconnect characterization requires both aspects of
CE: electromagnetic theory and computational techniques. This is what
Sainath eluded to in his message, but as you can see, it is just a part
(and I may venture to say, a small part) of the whole "SI Design" field.
That's how I understand it.
Regards.
Hassan.
Sainath Nimmagadda wrote:
>
> Ken,
>
> That's a splendid reply. As one who fits your description (except writing
> own ticket :-)), I too believe that is how the modern outlook of SI ought to
> be. To add to your already great reply, please add couple lines on the
> importance of electromagnetics and, in particular, computational
> electromagnetics (interplay of E and H fields, time-marching of fields,
> engineering the material interfaces, discontinuities and boundary conditions
> etc.) in understanding SI. Then, one better appreciates your reply and what
> is at the heart of SI.
--
Hassan O. Ali Email: hassan@xxxxxxxx
12-2159 Elmira Drive Tel/Fax: (613) 721-9047
Ottawa, Ontario
CANADA K2C 1H3 WWW: http://www.glcom.com/hassan
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
