27 March 2014

The Applicos Advantage

Written by 
The Applicos Advantage

Several years ago I was using Applicos instruments to test an 8 bit microcontroller with an integral A/D and D/A for the purpose of selling the Nextest Lightning test platform. The microcontroller was made by a very large semiconductor manufacturer and I chose this device partly because I thought the choice might curry favor with the maker and sell them some testers, perhaps a whole fleet if we were lucky. However, I came upon a stumbling point on the DAC test. My test program showed a rather severe DNL failure right at the 0x3F to 0x40 transition. You can see the DNL error in this plot.

Thinking that there must be some error with my test program I worked for days to solve the problem. This behavior could not be seen on the scope and I began to suspect that the Applicos instrument was wrong. But then I setup the Applicos capture strobe timing to capture the event at a very high rate using equivalent-time sampling techniques and the problem came into focus.

dnl error

What was happening was that the DAC in the microcontroller had a condition where the lower 6 bits were faster than the upper bits causing the transition from 0x3F to 0x40 to go 0011_1111 to 0000_0000 to 0100_0000 which caused this large negative spike in the DAC response. I broke out the scope again and finally saw it, once I knew what I was looking for.

The Applicos capture instrument had captured the glitch during the INL/DNL test and reported a large DNL failure. The placement of the capture strobe was fortuitous because had I captured it a little later it would have missed the glitch and reported everything as being fine. Instead it showed me a glaring error in the part that someone at this highly respected company must have found.

I was ecstatic that I would be able to show this to the maker of the microcontroller and confirm what they had seen.

So with these wonderful pictures in hand, I flew around the world touting the tester and the good fortune I had using Applicos instruments. I sold several million dollars worth of testers with this demo and then one day I finally got the chance to present to the maker of the microcontroller. I gave my presentation and proudly showed him the graphs above knowing that he would be thrilled that I had found the same thing that his test engineers had found.

I noticed a puzzled look spread over his face. He had never seen anything like this. His company, a top ten player in the world of semiconductors had completely missed this aberration in one of their top selling products. After several minutes of discussion he asked me a rather embarrassing question. “Have you shown these plots to anyone else?” I could not lie, I said, “Of course, I’ve shown these pictures all around the world!”.

Not the best way to make a sale I realized later. He asked that I not mention his company name when showing these plots, and so his secret remains with me to this day. However, as an epilogue, a year later the very same gentleman came in for a demo, I showed him the same plots and he was once again surprised that no one in his company had never noticed this glaring problem. It was easy to tell him the truth, “You are clearly using the wrong instruments to test your products, perhaps you should give Applicos a try.”

Engineer at Applicos