Well ... that is very interesting. I consider one thing not mentioned also a potential road block - conductivity. I don't pretend to know the details of this process, or for that matter the ball-and-pad method.
I do know that it is difficult to make a mechanical connection conduct as well as a soldered connection. If speeds continue to increase at the rate they have, and geompety continues to shrink as it has, then I would have to be convinced there is NO performance, and long-term RELIABILITY issues associated with the process.
Almost all high-reliability work I have been involved with has required soldered vs. mechanical connections. A good portion of the need for these requirements was a result of the demands of environmental conditions (thermal shock, vibration, applied forces, humidity, etc.) and a desire to maintain reliability over a period of time.
Loose connections can conduct poorly, which is pretty bad news for computing. A proper solder connection is equivalent to a continuation of the base metal, whereas a mechanical connection has an associated loss. Much better results in mechanical connections are created when a "cold weld" is created under pressure (for example, resulting from crimping), but I don't think this process involves that.
Decreasing geometry may favor this process, as anyone who has had to deal with solder balls, pitting, voids, and contamination problems can testify. But, it will be interesting to see how these issues are affected as the move to copper continues. Copper is a pretty well behaved material, and it may by itself take care of a lot of current problems. Then again, it will most likely introduce its own unique ones.
Testing the whole wafer DOES sound like a good idea, but I am skeptical that it will be incorporated industry wide. The article made many good points to support this argument. The cast of charachters is impressive, but NEEDS, much more so than DESIRES, drive this industry. RAM makers may need this technology, RAM testing equipment makers may need this technology, and maybe a company like SMTC may need this technology in the manufacture of analog components. But, there is a trend towards out-sourcing fab work, so it may simply be adopted by only a few specialty contractors.
It seems the story did infer one less obvious point, Edge Semiconductor will at some point be recognized as being a strategically important part of SMTC (if not already). These ever increasing demands on back-end processes are making proper testing more difficult and costly to implement. I thought I remember reading some where that Edge had gross margins over 50%, which would not suprise me, but I am not certain of this.
Go long ... Go deep ... Go SMTC ... I'm going to sleep :o)
Regards, JB
P.S. Institutions are mugging up the chip-equipment makers, as they begin to report improving book to bill ratios. What are the chances that Edge Semiconductor may at some point soon provide a pleasant earnings suprise for SMTC? I'd say this possibility is improving daily, based on this article and the chip-equipment maker industry. I have not looked at ATE makers yet, but there is a certain progression that must follow for the whole process to work.
Platinum management, hiring the best engineers, proprietary technologies, expanding margins ... duh George, which way did it go, which way did it go?
P.S.S. OFF TOPIC - Concerning ACEC, there is somewhat of a void of information available to digest, and some I have seen appears to be inaccurate. For example, ACEC reported positive earnings last quarter, but I have seen some reports showing it still has negative earnings. I just think it is a right place, right time, right technology case, since there really is nothing spectacular about what they do. However, ACEC IS a perfect example of a company with a NEEDED technology, which really doesn't bother me, as you can probably imagine :o) |
