The IAC team has of course been watching the situation in Japan and one early prediction we made in a recent Level Measurement web-cast is that there would be some shortages of sensor modules. It has been surprising that the impact is as small as it is up to this point. This has been a testament to good supply line management practices of keeping some buffer in the pipeline and using multiple sources.
As of this week, we are seeing that one particularly specific automotive sensor has become critically short in supply. This is causing many auto manufacturers to slow down or halt production. The sensor in question is a Mass Air Flow sensor module made by Hitachi. This is a fairly precision device that measures the characteristics of the incoming air supplied to the engine. The engine's computer then uses data from this sensor to precisely control the fuel/air mixture for maximum economy and minimum pollution. From what we see, the problem is actually a shortage of a particular semiconductor at the heart of the assembly. Hitachi has approximately 60% of this sensor module market and, key parts were made/tested in at a plant strongly affected by by the earthquake and subsequent losses of electricity.
I have significant experience in the semiconductor Automatic Test Equipment (ATE) industry so I am going to speculate a little about why we are seeing problems for this type of part. A vast majority of semiconductors are now produced and tested at companies that specialize in semiconductor fabrication and/or testing. Therefore having redundant regionally diverse supply channels is fairly easy. In the case of a few types of semiconductors like accelerometer type (MEMS) chips used in airbag and other automotive applications and the aforementioned air flow chip, specialized testing is needed. In other words, you have to apply carefully controlled stimuli while testing/calibrating the chip. In the case of the MEMS chip, a g-force. In the case of the air flow sensor, air mixtures of differing force and temperature. Once you set up a test cell for this type of work, it's pretty specialized and not really flexible to test anything else. Therefore, there is little economic incentive for a wafer fab/test house to do this kind of work unless you are going to pay them for maximum utilization.
One last point before we move on. You ideally want to do most of the testing at the wafer level or at least the packaged device. You never want to be finding bad devices at the finished air flow sensor module product level.
In conclusion, VDC estimates that as a result of this event, in order to preserve its market lead, Hitachi will likely have to address the single source problem by diversifying its production test and final assembly sites and/or adding inventory buffer.