First-Pass Success for 10 GPBS Backplanes
Dr. Mike Heimlich, Microwave &
RF Marketing, AWR
I was recently talking to a teacher friend of mine--internationally recognized in her field. Her stature is a source of some sort of pride on my part since I’ll never have a beer with a Nobel laureate, so this is the closest I’ll come. The conversation meandered (no pun intended as you’ll see in a minute) to what was currently interesting with my job and than to what was interesting with hers. We noticed that both of us were involved in teaching and technology despite the fact that we didn’t have the same jobs or work in the same fields. The part I remember most about the conversation is two things she said about how people learn—you can’t teach someone anything (they must want to learn it) and people learn the most when they make mistakes. In other words, we teach ourselves the most when we learn from mistakes. I was blown away. With all due respect to everyone from my Ph.D. advisor down to Mrs. Cohen, my Kindergarten teacher (wherever she may be), I want my money back! But after thinking about it for a while, I could see the logic and how it maps into what we do as circuit designers. If you were with me right out of college when I was designing MMICs, you would know that a “Heimlich Maneuver” is not related to the first-aid procedure for saving a choking victim per se—it’s when, in the days before LVS, a young engineer accidentally tied Vdd to ground through an airbridge. I learned a lot that day. I have never made that mistake again in 20+ years, and I found out how to pop an airbridge to get an IC working for free in 5 minutes instead of for $50K and waiting 3 months.
I recently had the chance to design an SI
demo board with an AWR colleague and a couple of engineers from Anritsu. The idea was to try and drive 10Gbps across
at least 12” of FR4 to compare what AWR’s design software would predict to what
could be measured by the Anritsu VectorStar VNA. We used Microwave Office® software to get a
good 50-ohm differential microstrip line channel and then laid down two coupled
lines to take advantage of the VectoStar’s four-port option. AXIEM™ screamed through some transitions for
SMA to microstrip on the top and then for some microstrip-to-stripline via
transitions for a few channels in stripline. We finished the whole thing up by comparing the circuit models to AXIEM
simulations (sims for short) across the whole length of each channel by taking
advantage of 64-bit AXIEM on an 8-processor PC. Everything looked beautiful....and then I remembered my teacher friend’s
advice… “we learn from our mistakes!”
So to “learn” something, I broke a couple
of things. On the microstrip layer, I
added a channel with an improper fill—too close to the microstrip—that we could
further “break” by dropping in some SMT shunt caps since the solder mask was
left off. On the stripline, we meandered
the line to add some skew by jogging the line to one side only. I thought to myself, “ahhh, yes, this will be
a great experience for those attending the demo—the board will be ‘broken’ and
they can learn something by suggesting how to fix it!” The designs were shipped off to the PCB
layout engineer with instructions to add the usual fill and ground vias to the
top to properly ground the microstrip. And that’s where the joke was on me, as you’ll see in a minute. The board came back and the Anritsu
engineers started testing it. They
couldn’t understand. Based on the AWR sims,
why would the board only support a couple of Gbps instead of up to 10 Gbps? Finally one of them looked at the transitions
and saw that there were no ground vias! My instructions to the layout engineer had been specific about the
microstrip lines...but not the transitions!! I learned a lot from that mistake, let me tell you!
But thanks to my now good friends at
Anritsu, some vias were drilled and filled and we got the board to work despite
me wanting to make sure that the board was broken. So, if you want to teach yourself something
new by fixing some of my “mistakes,” stop by and see the demo board at IMS in
the Anritsu booth or view the demo from DesignCon 2010 over on AWR.TV!

Comments