The Occam Files - No.2

A Critique of Current Electronic Design and Manufacturing: Symphony or Cacophony?

The early electronics industry was a bit like the Old West in that there was little in the way of law and order and folks took design and production decisions on their own. This was evident in the first published booklet on the early printed circuit industry titles: “Printed Circuit Techniques” which was released by the US Government Printing Office in 1946 just following WWII. In the publication all of the classified R&D that was being done during the war was released openly for public benefit. In the book, there were shown and discussed numerous different methods for making electronic circuits.

The DNA of many of today’s manufacturing processes can be traced back to the methods described in that early publication. Ideas explored included lithographic and stencil printing of both conductive and resistive inks (printed electronics anyone?), metal deposition by flame spraying (a method akin to sputtering) and others. To say that the early researchers were highly creative,seems an understatement today. However and unfortunately developers largely worked in silos as it appeared both then and now, as there was little evidence of shared knowledge as to how to make printed circuits before the publication that broke down the wall to expose the plethora of novel solutions being showcased and promote cross fertilization of ideas. In short, the shared knowledge provided opportunity to bring some coherence to what was mix of not always compatible solutions for making electronic circuits.

Eventually, the printed circuit industry rallied around the idea of making PCBs using print and etch technology, a method credited to innovator, Paul Eisler.  Print and etch is still the cornerstone technology in use today for PCB fabrication, it is however carried out with much superior processing technologies, materials and equipment. Important improvements to PCB technology followed as the industry grew, including the plated through hole and multilayer circuit constructions which allowed for increased density and performance. Flexible and later rigid-flex circuits increased the offerings and increasingly the printed circuit has taken on other roles in managing signals in the PCB, especially in the area of controlling impedance and managing signal integrity.

Arguably the PCB was the first methods of integrating circuits. The IC and the PCB have mimicked each other’s processing methods over time. Before the integrated circuit, the PCB integrated transistor functions on the board.

The development of components for use in electronics was also interesting. Prior to the development of the IC, transistors were packaged individually in transistor outline (TO) “cans” or small round can-like packages. While TO cans still exist and are in use in some products today, they were eventually largely replaced by IC which economically packed hundreds (today billions) of transistors into small form packages which were mounted on PCBs for the most part using tin-lead solder. The first IC packages were surface mount flat packs, however, the dual in line (DIP) package which employed through hole interconnection became dominant because it was more robust and easier to assemble, first by hand and later by automation.  

The termination pitch chosen for the leads on DIP packages was 0.1 inch or 100 mil centers. This was an arbitrary number but it became a standard and PCBs were designed using the same basic pitch. One need look no further than the developers breadboard which has thousands of holes all on 100 mil centers. The major outcome was that PCB layout and later computer aided design tools were created using a fundamental grid pitch of 100 mils and as a result, designing PCBs was easy and efficient. However as integration increased and performance demands grew, these leaded packages could not meet requirements and the industry turned back to surface mount with improved packaging solutions (SOTs, TSOPS and QFPs) and better production tools.

The problem here was that the cost of owning, operating and managing SMT tools and the materials required became prohibitively expensive and even the larger OEMs who were industry thought leaders abandoned their vertical integration roots where they controlled everything in house from design and manufacture of ICs, packages and PCB substrates through PCB assembly. The assembly process was thus outsourced and the EMS (electronic manufacturing services) industry was born.

Everything progressed well until it was finally observed that area array interconnection would provide superior interconnection density and performance. First was the PGA (pin grid array) which proved an ideal format for microprocessors which were required for the early personal computer industry as it could be plugged into a socket and allowed for users to easily upgrade their microprocessor chips. (It also provided some tax advantages but that is another story.)

The PGA was followed by the ball grid array (BGA) and these did not require the added expense of a socket (though sockets were and are available) This is also where the industry missed an important opportunity.

When the industry shifted to surface mount packaging they wanted to have a roadmap to follow for future smaller and denser packages. One very important thing the American industry also did was to agree to convert to metric units. 100 mils is actually 2.54 mm but metric is 2.5mm. As long a pin counts are low, it does not matter but as pin counts rise it becomes a problem due to run out. The industry also adopted what was called an “80% Rule” which mandated that every future lead pitch should be 80% of the previous generation thus the lead pitch for terminations went from 1.25mm to 1.0 mm to 0.8 mm to 0.65mm to 0.5mm to 0.4mm and then 0.3mm and lower. The 80% rule was again and arbitrary value and arguably not a problem for peripherally leaded packages but for area array it is both discordant and nonsensical when viewed logically. It creates significant problems for routing circuits when multiple grid pitches are used in design and it increases the layer count requirements while increasing potential for noisy designs and reduced signal integrity.

The Occam solution to the problem is to use a fundamental grid pitch for both substrate design and component terminations. The suggestion is to use a base pitch of 0.5mm but the number is not all imortant. The trick is to have all components with their terminations on a base grid pitch which can be depopulated as needed. This is a very harmonious, even symphonic-like solution where all of the elements of electronic design come together to create a beautifully simple and more elegant design with less interferences and fewer layers. Demonstrating this was and is a fundamental objective of the Occam Prize and that is one of the purposes of this series if the reader will allow their minds to open to the potential.

To help accomplish that goal and provide a foundation for understanding, a slide presentation providing some visual representations of some of the key ideas in the material presented here, please cut and paste the link below into your browser.

https://meptec.org/Resources/6%20-%20KEYNOTE%20Fjelstad.pdf

Keep it simple...

Next Installment – A Brief History and Critique of Solder in Electronics