[SI-LIST] Re: Right Angle Bends
- From: Julian Ferry <julian.ferry@xxxxxxxxxx>
- To: Brad Brim <bradb@xxxxxxxxxxx>, 'Jeff Walden' <jwalden@xxxxxxxx>, "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
- Date: Fri, 15 Jul 2011 19:48:09 +0000
I think one of the problems here is that like with most myths or urban legends,
there is a kernel of truth to it.
It is pure, unarguable physics that the capacitance will go down and the
inductance will go up in a right angle bend. The question is whether that
change matters in a particular application.
For some applications, like in high power microwave or RF applications, this
effect can be significant enough relative to other factors that it is
definitely worth worrying about.
But in the grand scheme of things in the current SI world, this change is way
down on the list of potential problems. It is effectively swamped out by many
other effects (like our relatively crappy connectors, for one example...)
I think these guys wrote a pretty decent little paper with some math that can
help determine whether you might need to worry about bends.
http://www.millertechinc.com/pdf_files/mti_tn063_microstrip_right_angle_bends.pdf
Also keep in mind that we in the SI world are mostly dealing with pulsed
signals, where a large portion of the energy is contained in the lower
frequency components. The excess capacitance will only affect the very high
frequency components. But in the microwave environment, it's all about the high
frequency signal, so a bend can be a much greater concern.
Julian Ferry
High Speed Engineering Manager
Samtec, Inc
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Brad Brim
Sent: Friday, July 15, 2011 2:57 PM
To: 'Jeff Walden'; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Right Angle Bends
exactly, Jeff ... and the traces that connect to the bends are also MUCH
longer than typical for the RF/microwave case.
In RF/mw ckt sim libraries/layouts a bend is a separate "component". The
reference planes (i.e. where the traces connect to the bend) are at the edge
of this component. For example, the 90-degree bend reference planes are at
the edge of the square representing the area of the bend. For most SI
applications the bend is NOT a separate component and the two traces simply
meet at a single node. Having worked way too many hours to implement and
test RF/mw ckt sim bend models over a dozen years ago I observed the
parasitics are a delay of length on the order of the node-to-node distance
with additional capacitive parasitics for sharp bends and inductive
parasitics for aggressively chamfered bends. For a 90 degree bend the
different definitions of reference plane imply 2*(W/2) additional length
trace for the SI case. Given approximate parasitic delay of sqrt(2)*W/2, all
implies doing nothing for SI applications is still on the order of only 30%
phase delay error versus a much more precise parasitic model (for an already
small parasitic). The phase delay dominates because bend capacitive
parasitics are small for SI apps relative to other capacitive parasitics not
modeled throughout the system.
Therefore, if the trace are not wide (low impedance) and their lengths
coming in/out of the bend are long relative to the linewidth, then ignoring
the bend is obviously the correct choice.
Where SI apps might not always want to ignore bends is for tight meander
structures used to accumulate phase delay and balance skew. These geometries
sometimes have short distance between bends and could therefore lose some
accuracy from ignoring bend parasitics. In these cases it is probably more
important to include coupling amongst the parallel traces. As we all might
guess, if you need to know a meander behavior accurately you may wish to
model it as a single component with more detailed simulation rather than
treat it as a collection of traces (with or without bend parasitics).
cheers,
-Brad
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx
> [mailto:si-list-bounce@xxxxxxxxxxxxx] On Behalf Of Jeff Walden
> Sent: Friday, July 15, 2011 10:55 AM
> To: si-list@xxxxxxxxxxxxx
> Subject: [SI-LIST] Re: Right Angle Bends
>
> The difference is that today's SI traces are significantly
> narrower than the typical "RF" microstrip of 30 years ago.
> -Jeff
>
>
------------------------------------------------------------------
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 technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
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 technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
