Post subject: Signal Integrity Effect of Magnets glued
to 400-800 MHZ PCB Posted: Fri Oct 19, 2007 2:11 pm
Joined: Thu Oct 18, 2007 8:47 pm
a Process/Fabrication guy working with assemblies that contain a Ceramic(LTCC)
PCB operating in the 400-800 MHz (Digital) Clock range. No Analog(Yet.)
Signal lines may be 4inches total length, multilayer. But there are
no ICs, its just a big fanout. Maybe a few 10050 or 0102 caps. Now I
want to stick another ceramic circuit card onto the surface using small
embedded micro magnets about the size of a grain of salt but there will
be hundreds. Some of them will be glued directly over signal lines but
all will be electrically isolated by dielectric layers. Their positions
will also be fixed so there will be no possibility of moving magnets
relative to conductors like in motors. Now my question for you Signal
Integrity Experts here at RFCafe, because nobody I talk to yet can give
me a straight answer on this: Will these isolated fixed magnets influence
the Integrity of typical 400-800 MHz digital signals nearby? If so,
HOW? For example, You hook this assembly without magnets up to your
TDR and to other expensive analyzers and give me a $30,000 SI report.
Now I place the magnets and you re-run the exact same test. Can you
see any difference? Can you derive the presence of fixed magnets inside
the assembly from this data? HOW? I only seem to get nebulous answers
to this, and I ask "Experts." I'm a CHEM Engr, not an EE, so don't go
PhD on me. But if someone can educate me on this topic I would be most
Posted: Mon Oct 22, 2007 8:15 am
Fri Feb 17, 2006 12:07 pm
Location: London UK
Basically, any effectsd will be due to a 3 part mixture of
dielectric material into the RF electro-magnetic field around the clock
line ceramic lines.
Both permittivity and permeability of the magnetic
material comprising the magnets will be frequency dependent. Not knowing
the magnetic material you are using makes it hard to give a definite
answer, but I suspect the molecules of material will not be able to
follow the changing EM field, whereas the dielectric material of the
adhesive possibly will. Hence I predict the major effect will be due
to the permeability of the mixture of magnetic material and adhesive.
If the magnetic material is mainly a Barium Titanate type this has a
high permittivity and loss factor. It will be mixed with a low permittivity
relatively low loss adhesive, of which again we do not know the properties.
When one knows the permittivity of both and the relative concentrations
of each, one can compute the loss and permittivity of the mixture using
Lichteneker's Mixing Formula (basically a logarithmic relationship).
To predict qualitatively what effect the mixture would have, we
would need to have detailed info on the various materials comprising
the magnets and adhesive, the percentage of each in the adhesive mixture,and
how these behave with increasing frequency through the UHF spectrum.
If the k value remains significant compared to the substrate material
in the UHF range, it will show up as high refelction coefficient. If
the magnetic mixture has high loss at UHF, there will be a greater loss
through the power divider,
Post subject: Posted: Mon Oct 22, 2007 2:29 pm
Joined: Thu Oct 18, 2007 8:47 pm
Thanks so much.
If I understand you, the magnet is no different than any other metallic
object despite its magnetic field and particular permittivity. Its effect
is simply capacitative... So to abate this effect am I correct in concluding
that I could do 2 things: 1. shield the signal lines. 2. Route the lines
a bit out of the way and open the dielectric above a ground plane and
maybe use a conductive epoxy and glue the magnets down onto ground plane
to render them continuous with GND? Seems this would kill the cap effect.
But am I forgetting anything?
subject: Posted: Tue Oct 23, 2007 6:22 am
Joined: Fri Feb 17, 2006 12:07 pm
Yes, those precautions would be appropriate.
I had (probably wrongly) assumed the magnets were there to interact
with an RF device, but seemingly that is not the case. Therefore keeping
the magnets away from or shielded from the trace line on the substrate
will ensure no interaction.
If they must be within a close distance
of the trace, say a few millimeters, then using a low density of magnet
grains mixed in an adhesive using polystyrene instead of epoxy should
also have minimum effect.
Epoxy is quite lossy at UHF, but polystyrene
has low loss characteristics by comparison. Melted polythene is also
OK so long as it is not so over-heated that it chars.
with the project