An engineer battles time zones and a customer who isn't talking to investigate a problem with an application involving a high-speed D/A converter
I had just wrapped up some work on high-speed A/D converters evaluating the use of our A/D in Zero-IF (a.k.a. Software Defined Radio) applications, when my manager asked me to help out a junior engineer with an issue that a BIG customer in Japan was having with our D/A products. The problem had dragged on for more than two months with no clear progress being made up until this point.
As I worked with the junior engineer he was clearly frustrated as he would always say, "I already looked at this/that...." Finally, I realized I should take it over, thanked him for his time in helping me to come up to speed on the problem, and said that I should be good the rest of the way by myself. After being put into direct contact with the customer, I would ask questions about the system’s design. The process was tedious, as the response would take 24-hours each time, due to time zone differences. Moreover, every answer came back as “NO...TRADE SECRET.”
However, they did provide me with an image of the "issue" they were seeing on the oscilloscope. I only had an image of the "problem" to work from, which was simply an image of a step function for the brightness (technically the luminosity) levels for a standard video color bar test pattern. There was however unusual ringing on the upper steps, but oddly the ringing was not present on the lower steps.
From their image, I could clearly see that they were trying to effectively do a stair step functional pattern with our D/A, and so I drove our D/A to produce the same waveform; with the same timing. To my surprise, I was able to reproduce the exact problem in our lab!
Clearly there was a Gremlin in the "system." Given that the D/A was the high-speed type, it was a current steering D/A architecture and required a 50-ohm resistor to ground to develop the signal. Our evaluation board had the 50-ohm resistor on it as required, but when I removed the resistor from the board and simply used the 50-ohm resistor built into the scope, the Gremlin was gone! Effectively there was an antenna effect at play here...but a rather odd one.
Two weeks later, I shared my findings with the customer by showing them two images: One of how I reproduced their waveform with their "issue" and another with the "issue" gone when the signal was properly terminated. Like many other application engineers who solve problems, I was given no thanks or told that “yes, it worked.” There were simply no more complaints from the customers or managers.
More often than not, that is what an application engineer HAS to accept. And come to think of it, no more haranguing customers or managers is a pretty good payoff in itself.
Manny Muro is an electrical engineer alumnus from N.C. State University, who is taking his more than 15 years of industry experience to found his first high-tech company, Digital Data Innovations - DDI, after being involved in more than six tech start-ups since high school. He currently resides in Portland, Oregon and can be reached via e-mail at email@example.com..
The images with the distortion was with the MegaOhm scope probe. Without geeking out too much, the issue lies in that the high signal levels involved the switching of larger current sources to drive out the desired level. Where at lower levels the amount of current sources being switched in, or out, were handled more smoothly with thermometer decoding. So just as there was an impedance mismatch with the MegaOhm scope probe there was a mismatch with the board trace to the target CMOS I.C. The DAC supported up to 250Msps rates, which requires fast settling times, so the edges where rather sharp; hence the antenna effects occurred that we were seeing.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.