# Some data sheet mysteries examined

This is actually based on a collection of calls that have a common theme, and is a salute to those hard working people that write our data sheets. My first caller sounded like a young engineer that was still trying to find the limits of this new world of industry where he had just landed. The conversation started out something like this.

“About the power supply voltages for this part, the data sheet says the specification voltage is 15V. Well, I have the circuit running on 20V and everything looks OK. What's the deal?”

At this point I introduced my caller to the Absolute Maximum Table in the data sheet with the warning that he was operating above the stated maximum of 18V. Not only was there a good chance that one or more of the specifications would not be met but that he was compromising the reliability of the part. I then continued to remind him that the semiconductor fabrication process resulted in a distribution of performance parameters. Statistics played a major role in describing the part. We determine the absolute maximum specifications on the safe side so that he can count on all parts surviving all the time. He was on thin ice and when parts started failing I could not offer any support.

At the other end of the spectrum is the old hand that has been bitten too many times by a “typical” spec. The phone call from him can start with the question, “OK, what are you guys trying to hide? If I can't get an absolute minimum or maximum on that specification I will not use the part. You guys are trying to hide a flaw in the performance.”

My reply to this caller is an explanation of economics. Either rigid design analysis and characterization, or actual 100 percent testing before the part is shipped must substantiate every spec in the data sheet that is a minimum or maximum. Applying that standard to the every specification in the data sheet would drive the cost to a point where he could not afford the part. As a solution we include typical data with the goal that this number is valid for well over half of the parts shipped (we like to see at least 1phi distribution meeting the spec).

In this last case the chore of describing the operation of the part goes to the writer of the data sheet. Within the Precision Linear section at TI-Tucson this task falls to two Technical Marketing Engineers. Their task is to pick the brains of the design engineers and test engineers to determine the performance of each part and then translate that knowledge into text and tables so the engineer customer that is designing the part into his system can understand exactly how the part functions. This is no small task.