Home    Bloggers    Messages    Resources    Webinars    DesignCon Event   
Tw  |  Fb  |  In  |  Rss
Alfred P. Neves

Become a Measurement Master

Alfred P. Neves

About 22 years ago, I decided to get my fly fishing mojo together. Understanding that casting was fundamental to my goal, I took a class with an instructor donned with a colorful name by the local fly-fishing club. I practiced my casting, dealt with the instructor's salty personality, and of course, fished.

For the first couple of years, I was a bit unbalanced. Then I went on a salt water fly-fishing trip for redfish in a guided flats boat in Florida. The wind was howling over 25mph, the boat was moving, the fish were moving, and the guide was shouting over the wind in frustration trying to help me place the fly. I felt so defeated, seeing how long the journey would be before I would become a fly fishing "Master." I wasn’t even close.

Redfish went back into the water.
Redfish went back into the water.

But eventually, I got better.

You can apply the same persistence to making better measurements.

Assuming you’re past the basic issues with making a good measurement, what does it take to be a master at making VNA measurements from a signal-integrity practitioners' perspective? I am talking 40GHz, hybrid or multi-tier calibrations, launch design -- the whole picture.

A good fraction of the S-parameters I encounter have serious issues. They're plagued with passivity violations for any number of reasons -- the VNA wasn’t set up properly, there are fixturing issues, the calibration was not validated or even checked, the cables are loose, etc. Around 2002, I proposed a Test-and-Measurement track in DesignCon, and after a bit of arm wrestling, it was adopted and has turned out to be a valuable track. I now co-chair that track with a few other engineers, so I see a lot of energy focused on getting good S-parameter models.

My technical background includes making a lot of measurements to 50GHz, including fast TDR, designing platforms to validate S-parameters, and corresponding them to 3D-EM analysis. My company has conducted quite a bit of VNA training and we market products for the SI type engineer. I want to share my observations in this blog.

Although I consider myself good at using a VNA, I'm still no master, but I keep trying. My next blog will show how to work towards becoming a VNA measurement master.

Al Neves
Al Neves
11/14/2012 4:12:01 PM
User Rank Blogger
Re: SPECTRUM ANALYZERS
Good post!  This method is rooted in some good Stochastic process theory.   The Power Spectral density of the 2T TX output is very useful because it seperates the data dependent jitter issues, from the random jitter issues due to VCO, peaking of the loop response of the PLL (often 10X).

You can tie the PSD, frequency domain, into the FFT of the autocorrleation function, which is time domain.   Interestingly, you can measure the autocorrelation or variance record with either a TIA (recall Gigamax or Wavecrest) or a sampling oscilloscope with the delay (DL-11 I recall)?

LeCroy has a good white paper:http://cdn.teledynelecroy.com/files/whitepapers/wp_techbrief_var_of_time.pdf

 

50%
50%
Al Neves
Al Neves
11/14/2012 4:03:31 PM
User Rank Blogger
Re: Arm twisting
DesignCon is THE conference for our Company in terms of integrating a marketing message, product development cycles, our network, our connection to the industry, and it is frankly enjoyable as well. Thanks, but the Test/Measurement co-chair folks (especially Heidi Barnes) deserve a lot of credit for developing this track as well.

50%
50%
Barry J. Sullivan
Barry J. Sullivan
11/14/2012 2:27:19 PM
User Rank Blogger
Arm twisting
I remember your passionate argument for a DesignCon test and measurement, Al. The arm twisting I recall was getting you to agree to organize it during the early years.

Others have helped, but we owe thanks to you for getting it started, and for remaining a strong advocate for this area at DesignCon.

50%
50%
RHaller
RHaller
11/12/2012 1:40:35 PM
User Rank Apprentice
SPECTRUM ANALYZERS
I have also found myself in the Lab often trying to refine measurment techniques, and looking for new ways to look at and solve problems on High Speed SERDES. One of the more recent techniques I have found invaluable is the addition of a Spectrum Analyzer. When I have a single misbehaving SERDES link in a chassis full of links, after examining the usual culprits (clocks, Power), I find hooking up a spectrum analyzer and inserting a 101010 patterm enables one to examine the spectral Content of the channel for abnormalities.This is a view in the frequncy domain but takes into respect that you have a fully running system. ...

50%
50%
Al Neves
Al Neves
11/5/2012 4:46:45 PM
User Rank Blogger
Re: Passivity violations
We see a lot of passivity issues with both low loss devices and resonant devices, where the calibration had significant systematic error.   

Interestingly, we run into a lot of passivity issues with de-embedding approaches using T-matrix, and Automatic Fixture Removal, especially if the de-embedding structures are designed improperly and/or with poor signal integrity (the launches are crappy return loss for example).   

Anritsu and WRT presented a 3.5 Hour tutorial last DesignCon that provided some examples of passivity issues.   See our website for a copy.   

Many of the passivity issues are at lowest VNA swept frequency, and my guess is the directivity of the VNA coupler rolls off a bit and that is what is causing some minor transmission tracking error issues, especially for older VNA architectures.

 

50%
50%
Martin Rowe
Martin Rowe
11/5/2012 9:24:30 AM
User Rank Blogger
Passivity violations
"They're plagued with passivity violations for any number of reasons."

What sort of violations and what kind of reasons?

50%
50%
More Blogs from Alfred P. Neves
Conferences are a great way to build your personal engineering brand.
What does How to Solve It author George Polya have to do with signal integrity?
Understanding how to use complex test equipment is different from knowing how to apply measurement results to complex engineering problems.
To be a Measurement Master in the frequency domain, start with this list and modify as needed.
information resources
sponsored content by Agilent
All resources
flash poll
follow us on twitter
like us on facebook
webinars
upcoming
7/17/2013 12:00:00 PM

Navigating the compliance standards for high-speed digital standards can be a difficult process. This panel comprised of the standards compliance gurus will help attendees understand how to successfully negotiate the standards process by understanding how standards organizations operate, and define compliance. Each of the 3 presenters will give a key insights to standards compliance such as: What happens when you don't pass compliance, interoperability issues, compliance workshops, and what to do when there isn't a formal compliance program. Lastly, we'll open the floor for questions from the audience.
8/13/2013 12:00:00 PM

The Mobile Industry Processor Interface (MIPI) Alliance standards body standardizes interfaces for mobile designs. The MIPI D-PHY and M-PHY are introduced as high-speed physical layers serving as base for multiple applications such as camera, display, memory and baseband running protocols. Other, originally non-mobile, protocols are being adapted to work with the M-PHY physical layer. Being optimized for both performance (high data rates) and (low) power consumption introduces many new validation (test) challenges outlined in the MIPI Conformance Test Suites (CTS), which are key to system interoperability. The new M-PHY specification rev.3 will be adding Gear 3 support (~6 Gbps data rate).
DesignCon Community
About Us     Contact Us     Help     Register     Twitter     Facebook     RSS