Over the years, I have had one foot in the metrology/test and measurement world and the other foot in software and software architecture. With two different perspectives, I tend to look at things from multiple angles, which causes me to question everything – even those things we have been doing the same way for hundreds of years.
In the software world, we have the term “waterfall development.” Typically, a waterfall development process is very long because it does everything in one giant step. It starts with gathering requirements and then documenting the software specifications. Those specifications are passed on to the development team, who write the application. When it meets the specifications, they call the software done.
Most of the calibration labs I have seen use a waterfall approach to calibration. When an instrument comes into the calibration lab, we assign the work to one technician. The technician looks up the procedure and follows the steps until the unit either passes calibration or is returned out-of-tolerance.
What makes this a waterfall method is the singular approach. One technician works on one instrument until it is finished, then moves on to the next item in the queue. In larger labs, the next item in the technician’s queue is usually similar to the previous instrument.
One major exception to this process is in the temperature lab, where the technician typically performs several calibrations simultaneously. It can seem a little crazy at first, but temperature techs have several timers running in their heads as they track all the probes they have in different baths.
This got me thinking, “We need to move away from waterfall metrology, ” just like the software industry moved from waterfall to an iterative development model. In this move, we can change some additional things along the way.
First, is the idea that a single technician needs to calibrate 100% of the instrument. In the past, when calibrations were shorter, we could have one technician calibrate everything. Today, many calibrations are hours upon hours of work, and we have computers and databases. We can track who performed what test and when; we can pass the calibration task from one technician to the next in the chain.
We can also include the Henry Ford factory approach to calibration, where a technician performs the same part(s) of a calibration then passes the instrument to the next technician/test station.
Next, is test station configuration. Today, most labs have a rack of equipment meant to do the whole calibration. But if you watch the technician work, he will use 1-3 of the 15 instruments in the rack. With automation and better test station management, the same 15 instruments could be testing 10 or more instruments. This is how the temperature technicians can be so productive. They are running multiple calibrations in parallel.
Finally, automation and why this is part of the automation corner. In the calibration lab, we still think about automation from the waterfall perspective, the all-or-nothing point of view. This is our biggest obstacle and what is preventing metrology production from moving forward at a faster pace.
It is a bold idea for a developer to think, “I don’t have to automate all of it!” I can automate just some of it. Focus on automating the longer, time-consuming tasks. If you take a 20-hour calibration and automate just a 4-hour section, decreasing the time from 4 hours down to just 1 hour is still a 75% improvement.
Think about it for a second, using the 80/20 rule. If 20% of test points in a procedure will take 80% of the time to automate in a waterfall environment, it’s not very efficient.
By unlocking the automation requirements tied to automating the entire procedure, to a more modular approach of automating what takes the most time, flips the 80/20 rule. Now, those test points that take 80% of the manual calibration time become the focus for automation. They have the biggest cost savings, not just in the calibration time but also in the time it takes to create and test the automation. Plus, those longer tests are usually long on similar instruments, saving even more time as the technique is applied to more and more similar instruments.
In conclusion, by breaking down this monolith of calibration into smaller parts, we can increase efficiency, as we put the pieces back together. We don’t always have to think of calibration as a waterfall process. By mixing and matching technicians, lab standards, and automation, process efficiencies can increase exponentially.
