My comments had to do with ripple generated on the Vdd rails by the two types
of signaling. This ripple is the most common source of EMI. Differential
signaling generates far less ripple than does single ended.
True, board to board or box to box, differential is far quieter that parallel.
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On ;
Behalf Of Istvan Novak
Sent: Thursday, July 7, 2016 5:53 PM
To: Jeff Loyer <jeff.loyer@xxxxxxxxxx>
Cc: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Via fence along board edge, stripline vs. microstrip
Jeff,
OK, I now understand what you meant. I think the benefits of differential
signaling for EMI
is not that important if you stay on the same board; the EMI benefits become
more clear
for box-to-box connections. This also may contribute to the different
experiences.
Regards,
Istvan Novak
Oracle
On 7/7/2016 10:12 AM, Jeff Loyer wrote:
Hi Istvan,
What I was alluding to was the fact that "high speed routing" seemed
not to be a factor in EMI, else the transition to differential
signaling would have impacted the EMI signature of Intel systems. I
never heard any mention of that. On the other hand, Lee says that his
experience was otherwise, so single-ended signaling seems to
contribute. I'm guessing at this point that I only had visibility to
well-designed systems where high speed routing was carefully
controlled, over continuous return paths, for instance. Lee might
have been exposed to less carefully designed systems and observed the
benefits of differential signaling. Interesting information, perhaps
explaining something that has bothered me for years (why we, Intel, didn't
see an improvement in EMI when we went to differential signaling).
Thanks all for sharing your insights - always nice to learn more.
Jeff Loyer
-----Original Message-----
From: Istvan Novak [mailto:istvan.novak@xxxxxxxxxxx]
Sent: Wednesday, July 6, 2016 8:12 PM
To: jeff.loyer@xxxxxxxxxx; Sanjeev Gupta
Cc: si-list@xxxxxxxxxxxxx
Subject: Re: [SI-LIST] Re: Via fence along board edge, stripline vs.
microstrip
Jeff,
I am not sure I follow the logic: in case there was no EMI problem
with the single-ended front-side bus, if you make it better by turning
it to differential, it still wont have EMI problem.
So from the perspective of complaints, there is no change (zero
before, zero after). Or are you saying that there were EMI issues
with the single-ende front-side bus and with the differential
front-side bus the EMI problems remained unchanged?
Regards,
Istvan Novak
Oracle
On 7/6/2016 11:18 AM, Jeff Loyer wrote:
Something that I’ve wondered about is: If microstrip (or stripline)
might cause EMI, how come Intel never saw any significant difference
when they went to differential busses between processors? Intel
replaced the single-ended front-side bus (high EMI) with differential
signaling (low
EMI) for communication between the processors and memory controller
and yet I’ve never heard anyone write (or remark) that this made an
impact on the EMI signature of their servers or desktops.
*Jeff Loyer*
*From:* Sanjeev Gupta [mailto:sanjeev@xxxxxxxxxxxxxxxxxxxxxxxxxx]
*Sent:* Wednesday, July 6, 2016 6:06 AM
*To:* istvan.novak@xxxxxxxxxxx
*Cc:* doug@xxxxxxxxxx; si-list@xxxxxxxxxxxxx; Jeff Loyer; Lee Ritchey
*Subject:* Re: [SI-LIST] Re: Via fence along board edge
I have another question to EMI experts on the same lines and add few
more points in the discussion..
Which is good from EMI point of view.. stripline or microstrip routing?
I conducted few EMI simulation experiments long time back and found
something interesting.
experiment was related with microstrip vs stripline radiation using
3D EM tool in real PCB condition and measured far-field@3m .
I found stripline radiation much higher than microstrip. This is
definitely not consistent with established myths..Normally stripline
is supposed to be better than microstrip routing for EMI. But, any
stripline routing in real PCB will come with at least 2 via
transition to make connection with components.
The conclusion was pointing towards that any via-addition in the
layout will create good amount of radiation. Therefore, stripline
radiation was worse than microstrip.
I am not sure how does this conclusion will apply to GND via
stitching fencing and ongoing discussion. But, any via addition in
the layout does not seem to be free insurance.
Another simulation experiment was related with radiation through
bends (90degree, 45 degree,45 degree mitred). I could not find any
difference in radiation till 10GHz. difference was marginal beyond
10Ghz. So 90 degree bend radiation is also a myth. Many earlier
discussion on si-list also confirms the same conclusion.
it is not possible to simulate the absolute EMI radiation from given PCB.
But relative EMI assessment for different layout condition is
definitely possible.
Thanks
Sanjeev
SigIntegrity Solutions
www.sigintegrity-solutions.com
------------------------------------------------------------------
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