Making 4 Hours of Work Take Only 3 Mins
I have to start by saying that I don't in anyway take full credit for this program. I did the lions share of the programming and interface design, but the initial idea came from my friend Rashmi Jha when we were students together (she's a professor now, and I'm still a student). The CVC program itself was written by Dr. Hauser, and Dr. Ahmed.
In a nutshell, the semiconductor industry learns a lot from making and measuring silicon MOS capacitors. But to get accurate information out of modern devices, the measurements have to be fit to models. Dr. Hauser and Dr. Ahmed wrote the CVC program to do this, but it's a command line program that processes 1 file at a time. On top of that, each file to be processed has to have only two columns of data and they need to be in the proper order.
One single experiment can yield 20 or more samples, of which we have to measure a minimum of 5 capacitors each for statistical reasons. For my own work, I typically measure these five different capacitors under 5 or so different conditions. So we are talking about hundreds of files which need to be processed and examined one at a time. This took hours, or days.
So Rashmi and I made AutoCVC. It is first a GUI that can access all the features of CVC. Second a batch processor for the CVC program (meaning it will run CVC on ever file in the given directory). Third it will isolate any two data columns and remove any arbitrary amount of header data. Fourth it will run simple statistical analysis of the extracted data for an arbitrary combination of the output files, including setting a fitting error threshold and scanning the files to see if they fall below it. And last, it will extract data from the output files and put them into a new file which can easily be imported into a spreadsheet and further manipulated or graphed.
One single experiment can yield 20 or more samples, of which we have to measure a minimum of 5 capacitors each for statistical reasons. For my own work, I typically measure these five different capacitors under 5 or so different conditions. So we are talking about hundreds of files which need to be processed and examined one at a time. This took hours, or days.
So Rashmi and I made AutoCVC. It is first a GUI that can access all the features of CVC. Second a batch processor for the CVC program (meaning it will run CVC on ever file in the given directory). Third it will isolate any two data columns and remove any arbitrary amount of header data. Fourth it will run simple statistical analysis of the extracted data for an arbitrary combination of the output files, including setting a fitting error threshold and scanning the files to see if they fall below it. And last, it will extract data from the output files and put them into a new file which can easily be imported into a spreadsheet and further manipulated or graphed.
Well it all sounds pretty unimpressive, but the bottom line is that what used to take hours to do, now takes seconds. We can process our data much quicker, and I'm happy. The downside: now our boss expects us to do a lot more work and to get the results to her quicker. Oh well....
No comments:
Post a Comment