For my application which is SAS-4, the COM metric is not enforced as it is
in other sophisticated ethernet based specs, hence the predicament. Any
links to the samtec paper or is it not out yet?
Thanks
Robert
On Thu, Dec 19, 2019, 3:39 PM Ryan Lott <ryan.lott.ee@xxxxxxxxx> wrote:
I can't and won't answer your question for you, but I will say if you know
the systems you're working with and have fantastic values with better
figures of merit such as COM, then I think you have all the potential to
answer your predicament with confidence. There's a reason the industry has
been doing away (at least partly) with frequency masks for quite some time
now. In fact, Samtec wrote a great whitepaper for designcon 2019 talking
about this. Definitely recommend checking it out...
-Ryan
On Thu, Dec 19, 2019, 2:33 PM <talented.libran@xxxxxxxxx> wrote:
I feel you answered my second question based on the wrapping techniques
or technology which is causing higher bandwidth cables to have MC issues
maybe because of cig wrap which tends to propagate common mode. It looks
like short cables < 0.5m have severe reflection issues which includes RL
and MC as there is not much dissipation of energy even if loss is an
advantage with them. So coming back to my question, should I ignore SCD21
spec violations and pass <1m cables?
Thanks
Robert
On Thu, Dec 19, 2019, 2:21 PM Ryan Lott <ryan.lott.ee@xxxxxxxxx> wrote:
SCD21-SDD21 is first order approximation with some strong assumptions.
It's still really useful. However, a better approximation requires taking
into account more terms e.g. re-reflections. It's very analogous to VNA
calibration theory. Specific inputs to a better approximation of SCD21
output would be common-mode reflection properties, common-mode transmission
properties, differential reflection properties, differential transmission
properties, and the asymmetries of the channel. You can mathematically
model this with certain assumptions. I've derived an approximation before
and can pass on my notes if I can find them. After that, it just becomes
unpragmatic and more.
So think about reflections. Most cable channels have a very high
SCC11/22. Also, shorter and less lossy channels generally have more
reflection energy whether diff or common.
Regarding your question on old vs newer cables... One hypothesis would
be to think about how different wrap styles affect common signals e.g.
helical/spiral vs cig/longitudinal. Helical has far greater common-mode
attenuation than cig wrap (at least this has always been my experience).
Might also be insightful to plot time domain and do some gating
techniques to gain some confidence/intuition.
In the end, MC is complicated and even the best approximations can't
always explain the underlying cause of SCD21 levels.
-Ryan
On Thu, Dec 19, 2019, 9:44 AM <talented.libran@xxxxxxxxx> wrote:
Hello Si-List'ers,
I see from cable vendors, the shorter cables 0.5m and 1m exhibit higher
SCD21 and SCD21-SDD21 data than 2m and 3m high speed cables. I know the
length dependency would affect mode conversion insertion loss, longer
cables would show lower SCD21 and the relevant spec is loss independent
mode conversion.
Can I ignore the SCD21 failures for shorter cables 0.5m and 1m cases and
consider pass? I wonder why do I see high bandwidth 24Gbps cables
exhibit
mode conversion issues than the lower bandwidth for example SAS 2.1
cables.
Any insight would be appreciated.
Thanks
Robert
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
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
