Does the Internet of Things force us to rethink our test strategies?
download the full paper here
The Internet of Things (IoT) seems to be on a good growth trajectory. Novel sensor devices for the end nodes, combinations thereof are launched every month.
Many sensors – many of them being MEMS based – require a thorough calibration and test prior to being assembled into the end node device. Sensor test typically includes providing a defined stimulus signal – e.g. acceleration, turn, magnetic field, sound, moisture, gas – while the device is being tested not only to validate the device function but also to calibrate its output signals. Providing such a stimulus during test is no minor challenge: consider what can happen to a contacted device under test do when subjected to an acceleration as high as 100g – 10 times that what electronics in a fighter jet are typically exposed to.
Over the past two decades handler manufacturers have provided test handlers with a stimulus by merging a standard test handler – gravity or pick & place or turret – with a stimulus module. Some sensor handling equipment has also been built for specific stimuli and using traditional pick & place concepts.
It is clear though that standard gravity, pick & place and turret handlers were designed for different applications – other package types, test times, tester constraints, temperature requirements etc..
Sensor devices have very unique features much different from the “average” IC:
- low number of IOs (6-14)
- small package outline – 4mmx4mm until a few years ago, now 2mmx2mm or smaller, the first wafer level sensors are emerging
- “active areas” on the package – pressure or uPhone or optical ports
- test times can be long, e.g. >30secs for pressure sensors, or short, e.g. <5secs for low g accelerometers
- tester resources are hardly a limit and test parallelism of 256x can be achieved
Particularly consumer applications sensors have other important attributes:
- they need to be low cost to enable a high volume market penetration
- if they ramp, they can ramp steeply – but many devices do not make it to volume
- product life cycles are short (2-3 years)
- the next product generation often requires an altered or totally different stimulus
While testers can be reconfigured and reused, it is not uncommon that test handlers can not be upgraded to meet the next generation requirements and are essentially obsoleted after a period much shorter than the depreciation time. This trend is often exacerbated by device package outlines shrinking beyond the practical capability of the handling concept initially chosen for the first generation sensor device.
In the presentation we will discuss application areas and limits of “classical” gravity, pick&place or turret sensor handling equipment and their economics when it comes to re-tooling for new stimuli.
We will also discuss alternatives to the “classical” gravity, pick&place or turrent sensor handling with a stimulus box and their economics. They alternatives include sensor calibration by electrical stimulus, wafer level stimulus, strip testing and carrier based testing. Pros and cons will be reviewed in detail.
At the end we will review challenges of upcoming sensor fusion into single packages (“combos”) and the integration with radios and uControllers as driven by the IoT.
Presented at: Test Vision 2020
Presented by: Reinhart Richter
Date: July 2014
download the full paper here