Requirements, or “Where is the pie [chart]?”


Requirements engineering is an important part of software engineering, but how does it actually work in a company that provides a yield analysis software?

Requirements engineering is defining, analyzing and managing customers’ software requirements, resulting in software working successfully for years to come. It is about solving real-world problems by developing digital world solutions.

During requirement solicitation, software engineers collect the requirements from the customer. In the semiconductor industry, the customers’ needs vary greatly depending on, for example, whether they are IDMs, fabless, foundries or OSATs, so configuring their solution is always part of the process.

Communication between the customer and the software company is key in the solicitation stage. The problems and requirements have to be stated clearly and completely so that the solution developed is correct and effective.

During the initial conversations with a new customer, the focus of the customer’s questions is often about what we can offer with our yield software. However, in order to find the optimal solution for each customer, defining the customer’s needs, or requirements, is much more important.

To get the complete requirements, we ask our customers some relevant questions:

How much data do you produce per day/week/month?

How long do you want to store the data?

What data sources are you considering now, and in the future, e.g., product test, wafer sort, PCM/WAT, AOI, defect inspection…?

It is important for us to know things like:

How many devices do you have on a wafer (mean and max values)?

That number can vary from a few hundred ICs to hundreds of thousands or even millions of discrete devices. We have an optimized wafer-map display for any amount of data, but we need to plan the storage requirements for each customer. 

How many parameters do you typically test on a device?

Again, this can range from zero (“binning only”) to several ten-thousands of tested parameters, and proper planning will be the key to a successful roll-out with optimal user experience.

What is the quality of your raw data?

It makes a big difference for the project scope if there’s already a central data warehouse with clean data or whether files from various test houses (“with varying naming conventions…”) are to be consolidated into a YieldWatchDog database.

And finally:

What analyses do you need?

We have dozens of visualizations, reports, and charts in YieldWatchDog, and we probably already cover anything you’ll ever need. But you might have a special requirement that is not currently available.

For example, there is no pie chart in YieldWatchDog. Which is somewhat surprising, given that we like pie so much… The simple reason for that is that for the past 16 years, no customer has asked for a pie chart. But if you are the first customer who wants one, we will simply add it. That’s what our software developers do all the time: implement solutions for our customers’ requirements.

So, the better we know your requirements before the roll-out, the better we can address your needs, create custom features and ensure long-term success. We love taking care of our customers, helping them achieve their goals and expanding our very satisfied customer base, so be prepared to answer some key questions during our first meeting… and perhaps can also enjoy some pie together.

Want to know more about YieldWatchDog?
About DR YIELD software & solutions GmbH

DR YIELD provides the leading-edge advanced analytics software YieldWatchDog for analysis and control of semiconductor manufacturing and test data. This enables semiconductor manufacturers as well as Fabless companies to improve important manufacturing dimensions such as equipment availability, throughput, operating costs and yield. Once installed YieldWatchDog allows you to get actionable insights into your data. For detailed information click here 

Comments are closed.