Archive for August, 2011

Published by Eric Bogatin on 29 Aug 2011

Measure and Simulate Twice, Build Once, is Their Motto

clip_image002
Example of rule check violation: critical net too close to I/O net, as revealed by EMSAT

“Measure and simulate twice, build once,” is the motto of Moss Bay EDA, the distribution arm for software developed at IBM and with other engineers, for engineers. This is also the motto of most high-end OEM companies that have come to recognize the value of simulation and measurement in getting to the correct answer faster.

Moss Bay EDA’s flagship product, EMSAT, is a rule checker that evaluates a board layout file against 16 customizable rules developed exclusively by IBM. These rules check for violations of common EMC guidelines any one of which could contribute to a compliance test failure. “It’s like having an EMC Expert sitting on your shoulder reviewing your board,” Gene Garat, founder of Moss Bay EDA says.

While many EDA companies are talking about providing a cloud computing platform for their tools, Moss Bay EDA is trying a different approach. For those interested in “kicking the tires” before buying their own copy of EMSAT, Moss Bay EDA offers a pay-as-you-go-approach. The software is downloaded and runs on your local computer, but the license resides in the cloud. Periodically the local software checks the license on the cloud and tracks usage.

A fixed fee covers all the automated PCB EMC design review necessary for a single board file. This is a low cost entry path for a small company wanting to move farther up the experience curve and add an EMC design review in their design flow. With their extensive contacts in the EMC industry, Moss Bay EDA also offers recommendations for expert EMC consultants who can not only fix a current board, but also customize EMSAT design rules for all future boards.

Published by Eric Bogatin on 22 Aug 2011

ExpressoEngineering.tv Covers the 2011 IEEE EMC Conference with Video Interviews

image

A website begun by Mike Violette of Washington Laboratories, Ltd offers some relief for those suffering from information overload. He and his team of “Larry King” like-interviewers, including Coco Bean, talk with the top players in the signal integrity and EMC world. They also cover trade shows and other live events relevant to our industry.

They were at the recent 2011 IEEE EMC Conf in Long Beach in full force, creating 30 short video interviews of many of the key companies showing off new products at the trade show.

If you have time for only one video, you have to watch Coco Beam rapping “Interference on My Mind”.

image

Published by Eric Bogatin on 19 Aug 2011

Guidelines for Technical Writing Learned the Hard Way From Many Great Editors

hands

I’ve been writing technical books, feature articles and columns for more than 40 years. Based on the old guideline, “An expert is someone who’s made all the mistakes possible,” I am well along my way to being an expert.

Most of my writing has been for a general engineering audience, written in readable prose, rather than for a peer reviewed technical publication, written in “high technish”. While I always enjoy getting feedback from readers, the most valuable feedback I have gotten has been from my editors, especially those at Prentice Hall. These folks see a lot of pieces and can clearly articulate what works and doesn’t work.

Over the years, I’ve collected the guidelines and advice passed along to me from my editors and dedicated readers which I think about constantly when I write. Writing is a process. The more you practice applying these guidelines, the faster you can accelerate up the learning curve. 

In the interests of passing along to others the great help I’ve received from experts in my career, here in one place are the top ten most important guidelines.

  1. You can never be too unambiguous- do not assume the reader understands what you are saying.

  2. Use every opportunity to teach the reader your specific nuances and specialized vocabulary that you want them to learn

  3. At the point you lose someone, you have lost them for the rest of the article. Start at ground floor and build up complexity as you go. Lead up to the most complex topics and use visuals and imagery where possible to make clear the points.

  4. Always start at the end. Ask, what is the most important take away(s) for the piece. Focus on these. After the paper is written, ask, for each word, does it support the take away you want. If it doesn’t, take it out. It’s how a sculptor carves an elephant from a block of stone. He imagines the elephant and cuts away everything that is not elephant.

  5. A key message for all successful products is that they help the user get to the correct answer faster. This is one of the elephants. Use every opportunity to illustrate and emphasize how a product or tool helps the user get to the correct answer faster.

  6. First drafts can be backstories to help you articulate the details and the story. After you write the first draft, “re-write” (not edit) to emphasize the key points (the elephants).

  7. Avoid the “novel” version of how you got there, unless that is the key take away.

  8. The more technical the content, the shorter should be the paragraphs

  9. You should be able to tell a parallel story in pictures and figure captions that a reader can use to skim the article

  10. While you tend to think of every word as a cherished child, never hesitate to kill off your children when they don’t add to the elephant.

Good luck in your writing adventures!

Published by Eric Bogatin on 17 Aug 2011

EMSCAN Near Field Magnetic Field Sensing Shows Return Current Distribution in Circuit Boards

NP Spatial close-up

I attended the 2011 IEEE EMC Symposium in Long Beach and encountered a number of exciting companies, people and products. The Canadian company EMSCAN is one of these.

Their EMxpert can measure, plot and analyze the current distribution in circuit boards at specific frequencies. It uses an array of high bandwidth magnetic field sensors on a flatbed which picks up the near field magnetic field from a circuit board.  When the board is turned on, return currents in the planes generate local magnetic fields with frequency components based on the signal’s spectrum and a spatial distribution and magnitude based on the current distribution.

When you superimpose on the map of the current distribution, the component outlines on the board you get an immediate visual clue about where currents are actually flowing on the board. Usually the currents on the bottom of the board that can be sensed outside the bottom are due to the power and ground distribution, but may also relate to the aggregate of signal returns. You might use this information to consider adding more capacitors to reduce the local currents and reduce the near field radiated emissions.

The near field external magnetic fields are not always a direct indication of far field radiated emissions, but it offers a new window into what’s going on with a functional board. 

NP Spectral Spatial ScriptingEach magnetic field pick-up-loop sensor can measure the spectral response of the current.  Depending on the specific signal, you will see frequency components at harmonics of the signal.

You can then sit at one frequency and scan across the array to measure the peak amplitude of the magnetic field, at that frequency. This maps the current distribution at that frequency in the board that leaks through to the bottom. 

On my visit to the EMSCAN booth, I recorded this short video demonstration performed by Cedric Caudron, an applications engineer with EMSCAN in which he walks through the process of looking at the spectrum, the current distribution, then adding a capacitor and showing that the current distribution in the planes is reduced by adding the capacitor. It’s pretty darn cool.

Demo of the EMxpert at the 2011 EMC Symposium

Published by Eric Bogatin on 14 Aug 2011

How Important is the Dielectric Constant in the Power and Ground Distribution Network?

image
Simulated impedance profile of 5 x 5 inch power and ground planes with different Dk and dielectric thickness, simulated with HyperLynx PI

Which is better, a high Dk or a low Dk between the power and ground planes in a board? This is the question Istvan Novak addresses in his recent column, Quiet Power at the Iconnect007 web portal. The answer, he points out, like so many other signal integrity questions is, “…it depends”.

The only way to evaluate “it depends” questions is by analysis. If you haven’t read his analysis, it is worth your time. He shows that if all you do is look at the total capacitance between the planes, yes, a higher Dk laminate between the power and ground planes enable higher capacitance. However, that is often not the most important issue.

As Istvan says, the capacitance contribution of the planes is not so important, as there is far more capacitance in the discrete capacitors added to the board.

If you worry about capacitance, he goes on, maybe a better term might be the “capacitance of the service area.” This is the amount of capacitance that is accessible to a single point on the plane during some short period of time.

The higher the Dk the slower a signal will travel between the planes. This means that in a short discharge time, there is a smaller area that contributes to the discharging capacitance. This is the service area. While the service area decreases with higher Dk, the capacitance per area increases with Dk.

Istvan so clearly shows that these two contributions cancel and the capacitance of the service area is independent of the Dk.

At higher frequency, the impedance in the planes is dominated by the spreading inductance. This is all about the magnetic fields, which do not interact with the dielectric materials. To first order, he points out, the spreading inductance is just about independent of the dielectric constant of the laminate between of the planes.

When the impedance from a central point is simulated, these two regimes are clear. Istvan used his own home-grown tool to simulate the impedance of the planes, I used HyperLynx PI. The plot above compares the simulated impedance looking from the center of two planes 5 inches x 5 inches, for the combinations of h = 2 and 20 mils and Dk = 4 and 16.

Above the plane self resonant frequency, the impedance to first order depends only on the dielectric thickness. There is a second order effect, related to the specifics of the self resonant interaction.

At the highest frequency, the slope up in the plane impedance is related to the spreading inductance. This is smaller for thinner dielectric thickness and independent of the Dk.

There is a second order factor that affects the peak heights of the resonances. The lower the characteristic impedance, the more the attenuation and the smaller the peak heights. For the same h, a higher Dk means lower impedance and more damping.

Which is better, a higher or lower Dk? The most important feature is the dielectric thickness. Which ever gives the lower h is the laminate to use. However, if the same dielectric thickness is possible then higher Dk will offer slightly lower peak plane resonances.

Published by Eric Bogatin on 05 Aug 2011

Printed Circuit University, the one-stop-shop for Educating the Printed Circuit Board Industry

image

Starting in February 2011, we moved all of the online signal integrity training on the beTheSignal.com web site to a new web site started by Pete Waddell, the publisher of Printed Circuit Design and Fabrication Magazine, and the president of UP Media Group.

Pete’s vision is to be a one-stop-shop for everything related to successful printed circuit board based products, including fabrication technology, assembly, inspection, measurement, processing, design, and performance. While most of the current content focuses on signal integrity, Pete has plans to add many other important topics for designers and engineers form many industry experts. If you want to join the “faculty”, be sure to drop him a note at pwaddell@upmediagroup.com.

Just as we had on my old site, PCU has three levels of users, visitors, members and subscribers. Anyone can visit the site and check out a limited amount of content just by browsing around. If you create a free account, you become a member and can access quite a bit more content, mostly pdf papers for download and some of the online lectures. As a member, you can download more than 100 papers and view more than a dozen streamed videos.

For a nominal fee, you can sign up for and access specific curriculums. For example, I created two different professional development certification signal integrity classes, Essential Principles of SI, a foundation building class, and Advanced SI Design, covering additional topics at a more technical level. These are each about 40 hours of structured content with a final exam and a certificate of completion.

There is also an all-you-can-eat option. As a subscriber, you can access all the content on the web site, including all the curriculums, all the webinars, online lectures, hands on labs, application notes, book chapters and pdf articles.  Options are available for individuals and for entire companies.

While online training will never be the same as live training, it has an important role to play, especially as travel becomes more expensive and inconvenient. It gives you the option of the specific training you want when you want it, right at your desk.

If you deal with printed circuit board technology, either in the design, fabrication or assembly, PCU should be on your destination list.

Let me know what you think!