Occam Files No 8
Package-Under-Package Occam Constructions
Those who read Occam Files No. 7 will recall my admission that many of the ideas presented in this series will likely be considered audacious suggestions by many readers. These suggestions are not in the same class as anti-gravity or perpetual motion machines, which to the best of anyone’s present knowledge are patently impossible (in that regard, the US patent Office will refuse any patent that make’s claim to providing perpetual motion). Rather, these ideas are more akin to the notion that heavier-than-air [LL1] travel was impossible, as was the common belief. Indeed those early visionaries (arguably beginning with Homer’s apocryphal tale of Daedalus and Icarus) who knew that was possible needed only look up and see birds in flight. Birds they realized were clearly heavier than air, yet they flew, not in competition with gravity but in cooperation with it and evolution. This is a key factor in all technological progress.
Technological invention requires the inventor to “paint inside the lines” as circumscribed by the laws of physics and so long as the inventor does, the real limits are those of the inventor's imagination alone.
When the Occam Process was first announced, it was met with both praise and ridicule. It was rejected out of hand by those who were rooted to the ground and praised by those who sensed that there might just be something to the ideas being proposed. As one friend put it, when certain segments of the industry railed against the concept, “Getting ridicule and resistance is evidence of a fear response from those worried that the idea might take hold.” This buoyed my spirits and has kept me afloat to this day. For that I thank you, my friend.
That aside, this article delves into the component issue again to examine another “impossible” or at least improbable component layout solution, package-under-package or “PuP”, in keeping with the other acronyms which have been in play over the last decade or so (see example using legacy components in Figure 1). Specifically, those are package-on-package or “PoP “and package in package or “PiP”. The former package-on-package concept has been around for much longer than a decade. Stacking of dual in-line packages or “DIP” devices was implemented perhaps as early as the 1980s, primarily to increase memory on memory cards and later with other types of packages including BGAs and CSPs.
Package-in-package has also a fairly long history with experiments beginning in the 1990s. These are typically packaged devices which are mounted onto small PCBs which are populated with other devices including discrete devices such as resistor, capacitors and inductors, in addition to other unpackaged semiconductor devices often wire bonded or flip chip mounted to the substrate. The completed assembly is then molded with plastic and assembled onto a PCB using solder. It is a solution that stops just short of the final objective of the Occam Process concept. This example is provided to help the reader appreciate and consider the possibilities for their future designs. Not limited to components alone, a thermal spreader could replace the center package to provide two-side thermal relief for high-power components.
An earlier article with a graphical representation of the structure can be read at the following link. https://www.3dincites.com/2012/12/the-other-3ds-occam-process/
With that background, it is possible to differentiate PuP from Pop. PiP is excluded from this discussion because it is a unique integrated packaged device, typically serving a dedicated purpose. (Note: While the following description is focused on solderless PuP structures and assemblies, it is also possible to surface mount a component or components to the bottom of a package using solder. Such structures have been seen on microprocessors’ assemblies; drawings can be viewed in the patent link provided at the end of this article)
Both PoP and PuP have structures where at least one package resides on another, so what’s the difference? The difference lies in the interconnection method. With PoP, the devices are electrically interconnected to one another by means of circuitry, vias and an electrical joining method, typically solder. With Occam solderless assembly of PuP structures, the interconnections are made primarily by plating microvias which connect to the terminations on the packages. The devices reside in the same space but on different levels. This means that the vias drilled to access the devices will be to different depths. It is not that dissimilar to drilling vias in a multilayer PCB to different depths. So while the differences may at first seem semantical, there are structural differences in approach to interconnection that make them arguably quite different. The advantages of PuP constructions line up nicely with those offered by PiP and PoP. All conserve base substrate area and all can integrate different functions in a reduced space. However, there are some unique, interesting, and even compelling prospects for PuP constructions.
Consider the following. Given that thermal energy generated by IC devices is desirably removed and given that the substrate perhaps best suited to Occam constructions is aluminum as argued in Occam Files No.6, having the device with the greater thermal energy generation on the bottom places it directly on the integral thermal spreader, which is the substrate itself. Other components can be placed on the surface of the device if space is available (and it often is). Component advantageously placed on the surface include termination resistors and capacitors. Others could easily be used as well including QFN devices as the design (and/or the designer) dictates. It is anticipated that assembly of such Occam structures will require preassembly but that is not necessarily an imperative. As previously discussed in earlier Occam Files, it is best that the terminations of all of the components used in an Occam PuP assembly reside on a fundamental grid pitch, which is preferably 0.5mm for reasons discussed earlier.
Turning back to the use of the substrate as a thermal spreader, note that there is greater economy in such a case as thermal management would otherwise require a separate heat spreader which would require attachment to the device. At this point it is probably worth reminding the reader that there is an inverse relationship between thermal excursions and duration and semiconductor reliability. When all active components are thermally relieved, the prospective reliability of the entire assembly is improved.
I’ll leave it here for the reader to ponder for themselves the prospects and to rethink their perceptions of the possible. As has been shown throughout history, the limits of technology are likely more circumscribed by imagination than by physics. Dare to think outside the conventional.
The reader is invited to review a patent discussing the concept in more detail at the following link if interested.
Next time: Thermal management options with Occam