It is common that our projects are to replace legacy measurement systems. For example, perhaps going from an analogue XY plotter to a digital logging system. So what are the steps you need to go through in these projects that are different to a standard new development?
Before you think about doing any work, you need to understand why you are performing the update. If you don’t do this, you can’t tell whether the project was worth doing! It also helps a lot for other people involved in the project to understand why so they can help point you in the right direction.
There tend to be two significant categories here:
The next step is understanding how the existing system works. If it has been around for a while, then there are probably questions to ask that haven’t been considered since it was installed!
One example is what measurement technology does it use? Are the same sensor types available or do we need to change this? What is going to be the change in performance if we change?
Another common question you should ask is what the configurations you use actually mean. Configuration terminology can often vary between vendors and processors. As one example, an upgrade we completed used a gain value to specify some measurement scaling. It turned out this was specific to the instrument sensitivity and excitation, so we had to convert this back to engineering units to translate it into the new technology.
Now you know what you have and what you need, you can identify appropriate replacements.
The exact steps here will vary from system to system, but the important thing is to feed in the information from previous stages. Use these details to ensure the new system fulfils your goals and opportunities while being a consistent replacement for the existing system.
One thing to start considering here is what impact you expect the new technology to have. For example, if we are running faster or with a higher resolution, we might pick up noise that wasn’t detectable before. Or if you are expecting a significant improvement in accuracy, how do you account for that? What is acceptable to say it is equivalent to the old system, or do we need to plan in characterisation so we can understand the relationship between old and new measurements?
The development stage is the one stage that is typically the same regardless of a legacy system or a greenfield development project. Once specified the new system needs building.
The important thing here is to consider the users and the upgrade process during the build. Can we use the same configuration terms and formats? Can we validate subsystems against the existing system as we go?
The transitioning is an interesting aspect. For example, can we phase in parts of the system at a time to catch errors earlier – at least in a test scenario.
So once you complete the system, you plug in and start using it right?
Wrong in a legacy cutover. We need to prove that the new system produces results that are consistent with the old system, accounting for any variations expected in the new design.
When we work on these systems, we plan in a system test before cutover to verify that the new system is good to go. What this looks like depends on the system. Assuming you are doing something non-destructive you can run the old system side by side, or on the same test parts, as the new system and compare the values.
At this point make sure you review back to your original ambitions as well. Have you seen the benefits you wanted? If not can we make final modifications to make sure we hit all of those goals?
If you are upgrading a legacy measurement system, make sure you consider these questions and steps. If you have done this before and think we have missed something, let us know in the comments.