Ventilator Test System
How S5 Solutions helped a client speed up their test system by 30% to keep up with production to support ventilators for COVID-19
By Tony Kuiper and Stuart Wilson, S5 Solutions
By Tony Kuiper and Stuart Wilson, S5 Solutions
With increased demand for ventilators brought on by the COVID-19 pandemic, our client had to significantly increase their production to match. Manufacturing was able to increase production capacity by adding shifts.
The testing is in-depth, involving precise measurements over long test cycle times. A single station tests multiple UUTs to help optimize the time and use of the instruments. The test department was able to see some increased capacity by running the tester full time, through all shifts. However, the end-of-line test stations quickly became a bottleneck. |
The client decided to build an additional test station. This offered us a short window of opportunity to make some software changes. S5 Solutions had one week to make software changes to decrease the test time.
The test station software was written in LabVIEW and TestStand and has provided the client a solid test system for years. We needed to develop a strategy to improve testing speed without sacrificing accuracy in testing.
The test station software was written in LabVIEW and TestStand and has provided the client a solid test system for years. We needed to develop a strategy to improve testing speed without sacrificing accuracy in testing.
S5 engineers first examined and profiled the code to find time-consuming operations. We were able to make improvements in some of the handshaking timing with one of the major third-party instruments. The overall communication was able to be sped up without compromising the connection. Since this instrument is used extensively during test, any slight time saving is multiplied by the thousands.
The next phase of speed increase was more complex. We improved the method of calibration to reach the same target point faster. The new algorithm is able to get much closer with a coarse approximation before the slower fine tuning of the final set point. The overall result is the same control accuracy with much less time taken in arriving at the calibration point.
Finally, the original architecture of the test system caused an inadvertent dependency between the three UUTs in the batch. We improved the logic used to control the way the three units interact, making them less interdependent.
These software efforts had a dramatic effect, resulting in a net time savings of 30% of the overall test time.
The next phase of speed increase was more complex. We improved the method of calibration to reach the same target point faster. The new algorithm is able to get much closer with a coarse approximation before the slower fine tuning of the final set point. The overall result is the same control accuracy with much less time taken in arriving at the calibration point.
Finally, the original architecture of the test system caused an inadvertent dependency between the three UUTs in the batch. We improved the logic used to control the way the three units interact, making them less interdependent.
These software efforts had a dramatic effect, resulting in a net time savings of 30% of the overall test time.