Leo Photo

The Role of Leak Testing in Building a Reliable IoT/IIoT Sensor

July 8, 2020
Which test methods work, which ones don’t—and why.

The growth of IoT/IIoT is driving the development of sensors for a multitude of environmentally harsh applications. These range from volcanology, to securing remote job sites, monitoring oceanic conditions at sea, and even helping you to avoid that fender bender.

Regardless of the application and the required functionality, the sensors in question all have the same thing in common–they must have a sealed enclosure capable of protecting them from penetration by water, dust or even gas, or from any loss of internal equilibrium.

This requires a leak (or penetration) test on the production line, but not just any such test will do.

Why a dunk test isn’t the answer

In the context of this article, when I reference a “sealed sensor,” I am referring to any such device that includes over-molded parts, has potting for sealant, is welded closed and that may also have electrical connections with trailing wires. I am also covering specifications ranging from IP67 to hermetically sealed, impermeable or simply waterproof.

Before we go further, let’s discuss dunk testing—a common, and problematic approach that’s often been relied on for sealed devices.

An enduring misconception about dunk testing is this: If you put a test part underwater and it has a leak, that leak is going to form a visible bubble. This is not always the case. When you do have a bubble, it’s because water is displacing air. The leak has to be relatively large for the water pressure to displace enough air to form a bubble. You may have an undesired leak, but not so large that it will form a visible bubble. The only way you can be sure that water has penetrated the sealed part is to pull it apart and inspect for signs of moisture.

The other consideration is that when you put a part underwater, that water pressure can also have uncontrolled variable effects—if the seal is designed to squeeze together under pressure, it may mask a leak.

So in closing, dunk testing is unreliable and ultimately destructive—particularly if there are electronic components inside that are easily damaged by water. Also, if the cause of a leak is easily addressed, such as screws that are not being tightened down correctly at an upstream production station, and the value the part warrants making the effort to repair it, why risk destroying it?

Test with an air-tight chamber instead

Testing with air is a much better alternative. It’s non-destructive, measurable, and quicker to keep pace with production. But conventional pressure or vacuum decay testing methods that rely on testing the part by filling it with pressurized air won’t work with sealed sensors. That’s because a sealed sensor lacks any ports, valves, or other means of creating pressure inside of it in a controlled manner for testing purposes.

Instead, you have to place the sensor in a sealed chamber. Because the sensor is likely to have very little internal free air space, it must be tested in a leak-tight chamber with a minimal volume of free air around it. A best practice is to have no more free air space around the sensor then there is within the sensor. This optimizes both the cycle time and the sensitivity of the test, to ensure its repeatability and reliability.

This approach also addresses challenges when you have a sensor that may be trailing a length of wire. One example I have encountered is those floating sensors I referenced in my introduction. In one case, the sensors in question trailed dozens of meters of wire—a custom-fitted test chamber is the only way to accommodate this and ensure the test cycle remains fast enough to keep pace with production.

Then pick the right test for the job

Depending on the acceptable leak rate and the usage scenario of the device in question, you have three options:

Volumetric fill testing with pressure: This is a common approach that is suitable for sealed parts that must meet the IP67 standard for resistance to penetration. To meet this standard, the device in question must be sufficiently dust and watertight to withstand immersion in one meter of water for 30 minutes.

We then test using first a gross leak test and then a fine leak test. This two-step approach is necessary because a sealed part with a gross leak (large hole) will pass a fine leak test when tested without a controlled fill. This occurs because test pressure will fill the chamber and the part at the same time. As a result, no pressure change is measured during the actual test. We must first flag those parts that have a gross leak before we can carry out the crucial fine leak test that will determine pass/fail. The fine leak test measures pressure loss in the test chamber over an extended period of time. Any change in pressure indicates that air from the chamber is infiltrating the sensor.

Volumetric fill testing with a vacuum: In this case, we evacuate air from the test chamber to create a vacuum instead of pressurizing the chamber with air. Again, we must first perform a gross leak test before a fine leak test, for the same reasons noted above. With a vacuum test, any pressure change during the fine leak test indicates that air is leaking into the chamber from the sensor being tested.

We may use a vacuum in those instances where the sensor will undergo atmospheric pressure changes during normal use, like electronic components in a vehicle. The vacuum can simulate those lighter atmospheric pressures experienced when, for example, you drive up and through the mountains.

A tracer gas method: In this case, the sensor is placed in its custom, leak-tight chamber, and bombarded/saturated with helium. If there are leak paths present, the helium will be forced through them and into the sensor. A mass spectrometer is then used to “sniff” for any evidence that this helium is then leaking from the sensor. (Another approach is to assemble the sensor in a helium-rich environment so that it arrives at the test station containing inside a certain percentage of helium.)

We use this approach with sealed sensors that must meet the most stringent acceptable leak rates. Helium enables us to detect any movement of material at the molecular level.  We may be testing for either penetration of the sensor, or loss of a desired gaseous state of equilibrium from within the sensor.

The right leak test is critical to performance

We are relying more and more on deployed sensors to collect and relay critical data, free humans from laborious or dangerous tasks, and equip our everyday living and working environments with more ease and convenience. The challenge is to ensure these devices are capable of withstanding the environmental rigors they face so that they function reliably throughout their expected operational lifespans. Employing the right leak test on the production line is a critical part of this quality assurance process.