copper traces to nowhere on MSI motherboard

[larrymoencurly]

I have an MSI Z68A-G43(G3) desktop motherboard with an Asmedia ASM1083 PCI-E to PCI bridge chip. That chip seems to be designed to connect to 2 thin serpentine copper traces, each about 2" long, that go nowhere. On this particular motherboard they're not used at all because the jumpers that normally connect to the chip (labeled SIM1 and SIM2 in the picture below) aren't installed, and neither are the header pins for the jumpers. Notice that each serpentine trace has 2 traces adjacent to it that follow it, only they're designed to not be connected to anything at either end.

Asmedia hasn't replied about their chip, and I can't find a datasheet or pinout, but does anybody know the purpose of unconnected serpentine traces like these?

 

[larrymoencurly]

I forgot the link to the photos of the traces to nowhere:


http://imageshack.us/photo/my-images/22/signalstonowhere.jpg/

 

[Loren Pechtel]

I forgot the link to the photos of the traces to nowhere:


http://imageshack.us/photo/my-images/22/signalstonowhere.jpg/

I think that's probably to support something that's not installed on
that board.

 

[DevilsPGD]

In the last episode of

I have an MSI Z68A-G43(G3) desktop motherboard with an Asmedia ASM1083
PCI-E to PCI bridge chip. That chip seems to be designed to connect
to 2 thin serpentine copper traces, each about 2" long, that go nowhere.
On this particular motherboard they're not used at all because the
jumpers that normally connect to the chip (labeled SIM1 and SIM2 in the
picture below) aren't installed, and neither are the header pins for
the jumpers. Notice that each serpentine trace has 2 traces adjacent
to it that follow it, only they're designed to not be connected to
anything at either end.

Asmedia hasn't replied about their chip, and I can't find a datasheet
or pinout, but does anybody know the purpose of unconnected serpentine
traces like these?

If I were to take a guess, perhaps there's another model that supports
SIM cards (either for encryption, or maybe an optional cellular model?)
and these traces would connect up to the pins on that model.

Could be for any sort of component, but it's likely something that
exists on other versions of your board, but not on yours.

 

[Paul]

I have an MSI Z68A-G43(G3) desktop motherboard with an Asmedia ASM1083 PCI-E to PCI bridge chip. That chip seems to be designed to connect to 2 thin serpentine copper traces, each about 2" long, that go nowhere. On this particular motherboard they're not used at all because the jumpers that normally connect to the chip (labeled SIM1 and SIM2 in the picture below) aren't installed, and neither are the header pins for the jumpers. Notice that each serpentine trace has 2 traces adjacent to it that follow it, only they're designed to not be connected to anything at either end.

Asmedia hasn't replied about their chip, and I can't find a datasheet or pinout, but does anybody know the purpose of unconnected serpentine traces like these?

At first I thought differential TDR and test coupon.

(If you're bored...)

http://www.ewh.ieee.org/r5/denver/rockymountainemc/archive/2000/diffimp.pdf

But the inner layers of that four layer stackup, should be
copper planes, and the layup would likely be microstrip.
Meaning the top and bottom traces can't "see" each other.

-
----- VCC
----- GND
-

And that tells me, it's more likely to be two single-ended
TDR measurements, to verify single ended impedance on the
top and bottom layer microstrip.

If they'd put more tracks on the top layer, then it might be
a test coupon for differential impedance check.

The outer tracks could have been intended as guard traces,
but with no pads to make connections to them at all, I'm
not at all sure what their purpose would be.

Could you do diff TDR between the outer traces ? Would
the center copper track invalidate the results ? With
no pads on the ends of the tracks, it's hard to see
how they'd connect the test head.

Diff TDR? Clamp right to the track ?
| |
v v
- - -
----- VCC
----- GND
- - -

Just a guess,
Paul

 

[David]

I forgot the link to the photos of the traces to nowhere:


http://imageshack.us/photo/my-images/22/signalstonowhere.jpg/

Impedance matching stub?

 

[Paul]

Why wibbly wobbly?

Something tells me that this forms an inductance.... but going to
nowhere! Ah, but wait! Notice the outer tracks... this would form a
capacitance to the opposite layer, if UHF circuitry this is what they do
to form 'tuned circuits' in UHF circuits so we'd end up with a self
resonant series tuned circuit - these are used in traps etc, to null a
specific frequency.

It could of course form a capacitance which would only be a pF (pico
Farad) or so.

Mobo's use multilayer boards, if there was anything connected to this
there *certainly* have been a via - a plated through hole connection, it
could also add coupling to whats on the other side of the board, the
board forming the dielectric of a capacitor.

If it was an antenna then it certainly would not have had a guard track
on the outside.

The length, the shape and the outer track are telling something, let you
decide.

davy

It's not an antenna. Of that, I'm sure. You could
design a better antenna than that.

I worked with an engineer that did 10GHz circuitry, and
he used to draw structures in copper (like, filters), and
his handy work looks nothing like that. That track was
drawn by a digital designer, not an RF designer. An RF
designer could not resist the temptation to augment
that simple design. That's the kind of track I'd
draw in a PCB (as a digital designer), as I know next
to nothing about good microwave RF practice.

The serpentine layout is typically used to equalize track length on
multi-lane busses. The designer of that PCB, is using serpentine,
in order to get a decent length of track for the TDR. The TDR needs
enough track to work with, so you can see the "impedance bump" in the
trace, and make sense of it. The serpentine uses "soft curves"
so the width of the copper track will not vary at the corners.
If you did a "square" layout for the serpentine, the
track impedance would be disturbed above 1GHz (corners
start to become an issue).

I recognized immediately, the lack of vias. If there were
vias, I would have dismissed the idea of TDR and test coupon.
If there were vias present, then the track could be connected
to some circuit. As near as I can tell, and especially with
the "SIM1" thing printed on the silk screen, I don't
get the impression it's intended to be part of the design.
It's not a functional part.

With no vias, it's a relatively clean setup to test the
impedance of a track. If the engineer specifies "controlled impedance"
when ordering the PCB, then both parties (PCB house, engineer)
need a means to verify they got what was expected in the order.
Normally, the test coupon appears outside the perimeter
of the final board shipped to customers, for test coupons
intended to verify the PCB shop fulfilled specs.

I've had lots of PCBs that failed the test coupon (this was
caught in production, and the check is done during incoming
inspection of materials). The impedance might be controlled
to around the 5% level. The shop making the PCBs, the operator
there is very experienced in selecting materials for the PCB
and ensuring they give the desired result. For example, I
might specify a certain thickness for a layer, and the operator
at the shop will use a $5000 software package, phone me up and
tell my I'm full of crap. They also see lot to lot variation
in materials, and the operator will adjust the stackup as he
sees fit. But in the end, at the engineering end, the impedance
can't vary too much from the target, or it could affect signal
integrity, as high speed signals leave the PCB and go
to add-in cards (like, PCI Express tracks).

As long as those tracks show no signs of being connected
to some other part of the board, then, the structure
is for testing. Not functionality. The structure exists
on both sides of the board, as if the designer wants
to check layer 1 and layer 4 impedance.

If the solder mask had been rolled back a bit, so
the outer two tracks of the three track set was
tinned on the end, I'd assume the intention was to
connect to those tracks. But it just doesn't look
like the designer intended for test equipment
to be connected to those tracks. So I don't see
what was intended for the outside tracks. You wouldn't
want to have to scrape the solder mask off, to make
connections to those outside tracks. Even tinning
the ends of the tracks would help in that case.

Paul

 

[hp]

I think that's probably to support something that's not installed on
that board.

How about thinking that this an Antenna???

 

[Paul]

How about thinking that this an Antenna???

To see if this is the case, why not do an "image search" on
your favorite search engine, and see how patch antennas
are constructed. Then compare to the pattern in the above
image.

It's not one of these.

http://ts2.mm.bing.net/th?id=I.4983532932564697&pid=15.1

This one has copper plates on the end of each squiggle.

http://www.digdice.com/wp-content/uploads/2010/04/15-dbi-cpe-panel-antenna.jpg

More plates on this one. Note the dimensions of the conductors
are constantly changing. A microwave RF designer designed this.
PCB CAD tools don't like to draw apertures like this, and this
was "hand forced" into the CAD tool.

http://ts3.mm.bing.net/th?id=H.4845359590998262&pid=15.1

The antennas on things like USB Wifi, are rather timid looking.
I would say a non-microwave designer drew this. No finesse.

http://www.wireless-home-network-made-easy.com/image-files/usb-pcb-wifi-antenna.png

Anyway, look for images of PCBs, with a built-in antenna,
and see if you can match Larry's structure.

One other thing. An antenna designer, would probably
not use vias in the antenna path, and would draw the
antenna on the same PCB layer, as the semiconductor
feeding it. It generally isn't a good idea to have
header pins on an RF trace.

Paul

 

[hp]

Just to clarify.

Sorry Paul, I didn't say it was an antenna, funny place to put one I
must admit..... I did say, "if it was and antenna", just to clarify.

The shape has gotta to be adding either coupling, inductance or
capacitance... they wouldn't do it to make the board look pretty, no
vias means no connections.

davy

Small brain fart time,

I remember some folks that chased 'easter eggs' inside integrated
circuit chips. Art in the Art. maybe this is a case of someone
designing a PC trace as an 'easter egg' ? Just because they could?

 

[Paul]

Small brain fart time,

I remember some folks that chased 'easter eggs' inside integrated
circuit chips. Art in the Art. maybe this is a case of someone
designing a PC trace as an 'easter egg' ? Just because they could?

You'd be fired pretty fast, if you did that today.

The thing is, too many people get to inspect your work
and comment on it, before it gets into production. Whether it's
a PCB design, or a chip design, it's pretty hard to hide something
like that and get away with it.

In our chip fab, we had an inspection process, that included
a "public display" of your handiwork. It allowed virtually
anyone in the fab, to notice visual defects in designs. And
they did find things. That's what used to amaze me about the
whole process, is the amount of errors that were found by
other designers casually glancing at the finished design. So
the public display idea, was an approved technique for
improving chip quality. (This was back in an era, when the
geometry wasn't quite as tiny as it is today. The plot would be
too big, to display a 3 billion transistor chip. It would cover
the entire employee parking lot, and waste way too much paper.)

*******

That structure was put in that PCB for a purpose. If I were to
do that at work today, I'd have to get two other people
(Library guy and layout guy) in on it, so right there,
two other people know I'm screwing around. So five minutes
from now, my boss knows. And later, we'd get a call from
the PCB shop, asking us if we were serious about that squiggle,
and could they remove it. So lots of busy-bodies would find out.
That squiggle was "signed off, up and down the line".

I had a guy at the PCB shop, actually *change* one of my
designs, without telling me. So sometimes, these guys get
a little carried away. You wouldn't get your Sparkle Pony
drawing past *that* guy at the PCB shop. Five minutes later,
he's phoning my boss.

Paul

 

[Paul]

One thing though.
Covering as much copper area as possible saves on the etchant... makes
it last longer, sure wouldn't get as 'slushy', I had some Ferric
Chloride that lasted well over a year it became like soggy porridge, I
don't know about the other stuff, Sodium Persulphate.

Merry Christmas all and have a Happy New Year!

davy

Yes, but there are fill patterns (crosshatch or lattice) you
can use for that. The layout guy is supposed to balance the fill of
copper on the layers, so that there is less warping in
processing. If you put "air" on one side of the board, and
a solid copper (unetched) plane on the other side of the PCB,
it'll warp on you, and be bowed in the center. And the stackup
is also supposed to be symmetric. There's lots of symmetry in
PCB design requirements.

http://www.multi-circuit-boards.eu/en/pcb-design-aid/copper-balance.html

We actually had a guy at a local PCB shop, who'd come in and give
a lecture on PCB design. Lasted about two hours and an excellent talk.
PCB shops give a fair amount of feedback, which helps ensure less
loss of productivity because someone isn't familiar with the details.

*******

I used to use the ferric chloride at home. Never had a problem with it.
Just don't dump it down your copper plumbing pipes I used to
etch in styrofoam meat trays. So the trays that hamburger comes in,
you save those and they can be used for etching PCBs.

The largest PCB I ever did at home, was for a keyboard. I built
my own keyboard, for my first computer. The PCB was double sided,
but being a "home manufacturing" process, there were no plated holes.
So if you did a "via", it meant inserting a wire, and soldering it top
and bottom. When soldering, you'd do the vias first, then populate
the components.

The hardest part of the keyboard project, was getting the key assembly
(unit with a couple hundred "legs" on the bottom), to line up with the
holes in the PCB. What a nightmare. But, I eventually got it aligned
and plopped into place. The circuit on the PCB was a keyboard matrix
scanner (so one big chip does the circuitry).

The weakest link, in my home PCBs, was protecting the copper. I've
had corrosion on some of those PCBs so bad, it ate right though
the tracks (takes about ten years to happen). In school, we
had a plating solution (AgNO3 based), that you'd apply after etching.
That would put silver plate (electroless) on the two sides of the PCB,
and provide some protection. But that solution, isn't something you'd
use at home (poisonous). The brew was home-made by the techs in the shop,
and as well as silver nitrate, it also had a hint of, I think it was
sodium cyanide. There were a couple other compounds added for stability.
The solution did a pretty good job, considering the crudeness
of the process. Those PCBs would likely last longer than the
ones I made at home. Silver nitrate by itself wouldn't give you
a good plating job, but the additional chemicals gave a "fine"
deposition. If you used silver nitrate by itself, you'd likely
get flakes deposited.

Paul

 

[David]

The weakest link, in my home PCBs, was protecting the copper. I've had
corrosion on some of those PCBs so bad, it ate right though the tracks
(takes about ten years to happen).

The easiest way to do that, of course, is to solder plate the copper.
Most of the fast turn PCB houses will do that for you. A high
reliability commercial board will have something more elaborate such as
ENIG (electroless nickel under immersion gold). A few years ago, some
vendors were having trouble with the ENIG process where the nickel got
contaminated during the process and you ended up with what was known as
'black pad', where the nickel-tin bond would fracture; this was a
particular problem on BGA balls where you could not inspect for it. We
switched to immersion silver for a short time, but that was a case where
the solution was worse than the problem.. ANY improper handling of the
board would cause severe solderability issues. We quickly went back to
ENIG with a trusted vendor, and have not had any issues since.

Of course, for years, almost all boards were treated with HASL (hot air
solder leveling). Some fabricators abandoned HASL due to RoHS concerns,
but there is a lot of focus on lead-free HASL processes today for
commodity circuit boards. We don't use it, because what we make is not a
low cost commodity.

 

[Paul]

The easiest way to do that, of course, is to solder plate the copper.
Most of the fast turn PCB houses will do that for you. A high
reliability commercial board will have something more elaborate such as
ENIG (electroless nickel under immersion gold). A few years ago, some
vendors were having trouble with the ENIG process where the nickel got
contaminated during the process and you ended up with what was known as
'black pad', where the nickel-tin bond would fracture; this was a
particular problem on BGA balls where you could not inspect for it. We
switched to immersion silver for a short time, but that was a case where
the solution was worse than the problem.. ANY improper handling of the
board would cause severe solderability issues. We quickly went back to
ENIG with a trusted vendor, and have not had any issues since.

Of course, for years, almost all boards were treated with HASL (hot air
solder leveling). Some fabricators abandoned HASL due to RoHS concerns,
but there is a lot of focus on lead-free HASL processes today for
commodity circuit boards. We don't use it, because what we make is not a
low cost commodity.

Yeah, we used HASL too.

And the boards we made, were "armor plated" in terms of quality.
Very hard to damage them with clumsy soldering iron work. Not
like the motherboard I tried to repair here at home.

I realized when I was making home made boards, that they
weren't the best. But what came as a surprise, is how
much they suffered over the years. I never expected the
degree of corrosion I was seeing, with the bare copper.

Paul