[SI-LIST] Re: AFR
- From: Scott McMorrow <scott@xxxxxxxxxxxxx>
- To: Jim Nadolny <jim.nadolny@xxxxxxxxxx>
- Date: Wed, 31 Dec 2014 05:42:33 -0500
Jim and all,
Actually, a significant amount of the problem with calibration structures
is non-identicality of the launch structures. This is especially a problem
for non-isotropic epoxy-fiberglass composites, since the launch entry point
Er varies randomly with placement on the fiberglass, and has a
manufacturing slop w.r.t. alignment on the pads.. I found that most of the
problems went away with isotropic materials (like ABT films for package
buildup layers), but I still had to deal with launch identicality issues
from manufacturing variation of the connectors, and misalignment of
placement. Ultimately I had to screen 2.92 mm Southwest Microwave edge
launch connectors for similar TDR profiles, and align them on the board
under the microscope, to get best TRL and GMS-parameter performance. Since
the GMS approach for material deembedding is similar to Aglient AFR, I
expect the same issues to occur.
I found that deembedding was less sensitive to trace impedance variations
with isotropic materials, since the signal propagation velocity was still
constant. The big hit I took was in launch non-identicality, as this
greatly impacts the return loss profile of the cal structures. Variations
of return loss between the cal structures messed up the deembedding
process, especially at the TRL stitch points. Since AFR has fewer cal
structures, and therefore fewer potential launch mismatches, it should
generally be more robust than TRL.
The other issue to mention w.r.t. causality is incorrect placement of the
reference plane. A mistake many make in choosing a reference plane
location is to place it too close to a non-uniform structure, such as
bends, pads, vias, Y-transitions, etc. In order to intelligently choose a
reference plane, one needs to know the potential worst case variation of Er
across the de-embedded trace segments. That is, the traces removed from the
DUT in the calibration process. If the reference plane is placed too
close, then causality problems are guaranteed to occur.
Scott
Scott McMorrow
Consultant - R&D
16 Stormy Brook Rd
Falmouth, ME 04105
(401) 284-1827 Business
http://www.teraspeed.com
On Tue, Dec 30, 2014 at 2:48 PM, Jim Nadolny <jim.nadolny@xxxxxxxxxx> wrote:
> Resending - 1st transmission was garbled...
> Hey Jeremy -
>
> Excellent point on causality. Al Neves points to it as well - you need
> "perfect" cal standards and test fixtures to get "acceptable" causality.
> And causality correction is sort of a misnomer IMO. You can create more
> problems correcting for causality than you fix. My colleague Stefaan is
> going to cover causality correction at DesignCon this year (Paper 14-TH1,
> "Causality Demystified").
>
> ISD and iTRL are what I refer to as custom Matlab code. As I understand
> it, these methods synthesize the exact test fixture response from
> measurement and that leads to significant improvements in the resultant
> S-parameter quality. TRL/LRM/AFR rely on cal structures to mimic the test
> fixture response. Given the tolerances of PCB fabrication (+/-5% or so)
> the cal structure is never going to be a perfect match to the test fixture
> and therein lies 90% of the problem.
>
> ISD and iTRL are one way to get better data. Al Neves/WRT has a second
> way that he is going to cover at his DesignCon tutorial and alludes to
> other methods that are forthcoming at the conference. Should be
> interesting as always.
>
>
>
> From: Jeremy Buan [mailto:jeremybuan10@xxxxxxxxx]
> Sent: Tuesday, December 30, 2014 11:29 AM
> To: Jim Nadolny
> Cc: han.guobing@xxxxxxxxx; si-list
> Subject: Re: [SI-LIST] Re: AFR
>
> Hi Jim,
>
> I've never used AFR before, since I use ISD from Ataitec. You mentioned
> that AFR has been validated compared to TRL calibration. In my experience,
> TRL calibration results in non-causal s-parameters, which can be seen
> clearly when converted to time-domain. I'm curious, does AFR also yield
> this kind of causality error?
>
> Regards,
> Jeremy
>
> On Tue, Dec 30, 2014 at 6:10 AM, Jim Nadolny <jim.nadolny@xxxxxxxxxx>
> wrote:
> Greetings Robin -
>
> Jim Nadolny here and our team at Samtec uses AFR on a regular basis. We
> use custom designed test boards with the required cal structures for AFR
> de-embedding. We typically test connectors, cable assemblies and PCB
> structures (via fields, coupled traces, etc.) in a controlled lab
> environment.
>
> The power of AFR is in its simplicity. You don't need to be an expert in
> microwave network theory to get reliable data. That said, it is easy to
> fall into the trap of simply trusting the results without checking a few
> simple parameters of the cal structure. I would argue that using a 3rd
> party tool has advantages over custom matlab code, particularly when
> supporting customers. No one can reproduce your results if you are using
> home brew code and this can be problematic as it requires
> peers/clients/customers to trust you, your code, your test methodology,
> etc. In contrast, numerous papers/studies have been written validating AFR
> compared to TRL/LRM de-embedding methods.
>
> The weakness of AFR is that it does not de-embed crosstalk in the test
> fixture. Most de-embedding algorithms do not, so this is not unique to
> AFR. This crosstalk limitation that I refer to applies to inter-pair
> coupling, intra-pair coupling effects are accounted for in AFR. So if you
> are testing a high port count coupled structure (like a high density
> connector) you need to design the test vehicle accordingly.
>
> We don't have a lot of experience using AFR on prototype/production
> electronic hardware and I would welcome comments from users who do. I
> understand that you can use AFR in a "1 port mode" to probe a populated PCB
> for S-parameters, but I have not done this myself.
>
> Regards,
> jn
>
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
> On Behalf Of Han, Guobing
> Sent: Tuesday, December 30, 2014 3:06 AM
> To: si-list
> Subject: [SI-LIST] AFR
>
> Hello Experts,
> Does anyone have the experience of agilent AFR?
> Who can tell me what's the weakness of AFR?
>
> --
> Thanks,
> - Robin (Han, Guobing)
> TEL: 86-21-61094805
> MSN: han_guobing@xxxxxxxxxxx
>
>
> ------------------------------------------------------------------
> 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
>
>
> ------------------------------------------------------------------
> 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
>
>
> ------------------------------------------------------------------
> 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
>
>
>
------------------------------------------------------------------
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
Other related posts:
