Previous blog posts have discussed the merits of choosing a commercial software platform for implementing the equipment side of EDA (Equipment Data Acquisition) and how you would use that package to differentiate your equipment data collection capabilities from your competitors.
In this post, we discuss how to design the equipment model to contain enough information to make it useful without publishing so much data that it becomes cumbersome for your factory customers to find the data that is most important to them.
Data to Publish
The automation requirements for the most advanced fabs call for the latest versions (Freeze II) of all the standards in the EDA suite, including the EDA Common Metadata (SEMI E164) standard. In addition to providing an excellent foundation for a new equipment model, E164 enables consistent implementation of GEM300, commonality across equipment types, automation of many data collection processes, less work to interpret collected data, and true plug-and-play client applications—all of which contribute to major increases in engineering efficiency. These capabilities benefit both the equipment suppliers and their factory customers alike. Therefore, equipment models should make all E164-compliant data available.
To summarize, those who remember the complexity of implementing SECS-II before GEM came along (pre-1992) will understand this analogy: E164 is to EDA what GEM was to SECS-II.
- Fab-specified Data
The second blog post made the following statement:
“This is why the most advanced fabs have been far more explicit in their automation purchase specifications with respect to equipment model content, going so far as to specify the level of detailed information they want to collect about process performance, equipment behavior, internal control parameters, setpoints and real-time response of common mechanisms.”
You only have to read the latest requirements specs for these fabs to get more specifics. Pick the one from your customer base that sets the bar highest and let that be your target.
Data to Avoid in the Model
It is easy to fall into the mindset that if publishing some data through the EDA interface is desirable, the more data we can publish, the better. This is not always the case. In his fascinating book, The Paradox of Choice, Barry Schwartz makes the case that freedom is defined by one’s ability to choose, but more choice doesn’t mean more freedom. In fact, too many choices actually cripple one’s ability to choose. The same can be said of data published in an EDA interface. Making too much data available actually hinders the creation of EDA client applications.
We were recently working with a fab to perform a proof-of-concept where we connected an EDA client to a piece of equipment with an EDA interface. We were able to connect to the equipment in a matter of minutes, but finding suitable data to collect for our proof-of-concept took almost an hour because there was so much superfluous data published from the equipment.
Publishing everything including the kitchen sink reduces the ability to create an efficient EDA client application.
Some examples of data to avoid publishing in the model include:
- Parameters that have no value – If a parameter is available in the model, but the value is not published by the equipment control application, that parameter is just extra noise in the interface. Consider not adding it to the model.
- Parameters with values that do not change – If a parameter value does not change during the life of the application, it does not make sense to collect that parameter’s data. For example, if an application uses an equipment constant, it may not be necessary to publish that constant through the EDA model.
- Irrelevant data – If a parameter contains data that is irrelevant to data publication, it should not be added to the model. For example, having parameters in the model that contain the IP address or port number for connection are not very useful in the equipment model. This information is necessary in connecting with an EDA client, but is not relevant for data collection in the model.
The takeaway: Publish data required by E164 and additional fab-specified data, but carefully evaluate other data to be published to make sure it is relevant and useful for data collection.
If you have questions about Equipment Data Acquisition or would like a demo of the functionality described above, please contact Cimetrix to schedule a discussion.
You can download an introduction to EDA White Paper any time.