US Electronics: A Metric Peg in an Imperial Hole

By The Metric Maven

Bulldog Edition

The Metric Maven has read countless articles about the decline of American electronics manufacturing. Pundit after Pundit who almost certainly have no idea which end of a soldering iron to pick-up, feign deep knowlege of what makes us “uncompetitive.”  There is always a laundry list of problems which are trotted out by these professional opinion manufacturers. The one item which is never on their list, is the lack of the metric system in the US. This omission is proof by proxy they are armchair commentators who know a thousand ways to make love to a woman, but have never had a girlfriend.

Central to the miniaturization of electronics in the 21st century is the surface mount device (SMD). They are packed into smaller and smaller sized printed circuit boards. This makes “smart” phones, and other devices smaller, and much easier to lose–I mean use. The world standards groups met long ago and defined SMDs in terms of metric. Industry PCB consultant Tom Hasherr sees it this way:

The United States is now the only industrialized country in the world that does not use the metric system as its predominant system of measurement.

All the World Standard Groups involved in the electronics industry (IPC, IEC, NIST, JEDEC, EIA & JEITA) have made the transition to the metric measurement system. They formed an alliance to stop using English units and all the data they publish is in metric units.

Buried within these diplomatic statements is a larger story of America looking down, taking direct aim, and shooting itself in the foot—with inches.

The story goes like this: Once upon a time in the 1980s the world standards organizations banded together and produced worldwide metric standards. The Electronics Industries Alliance (EIA) in the US was given the responsibility to articulate the size of surface mount devices. The world had created all the standards in metric, and the EIA was to publish the new dimensions for all to use. A book was printed with all the component names, dimensions and other pertinent engineering data. It was then released to US manufacturers for implementation.

American component manufacturers refused to make components to metric dimensions. The PCB assembly and etching houses rejected metric dimensioned drawings, and spurned any thought of using them. They repeatedly demanded the EIA publish a version of the standard with Imperial (English) units. The EIA finally did this and unilaterally changed the names of the components.

The metric SMD components were now renamed using inches. Originally, the first two numbers of the chip component names are the SMD length in mm, and the second two are the width in mm. There is an assumed decimal point between each set of paired numbers. For example 3216 is 3.2 mm x 1.6 mm. Here is a short list of the renaming:

Now the 3216 is renamed 1206 which is 0.12″ x 0.06″ with an assumed decimal point at the front, and whatever conversion factor error is introduced. We can see that this US “improvement” introduced the same name designation for different sized electronic SMD components. After this re-naming, should you be interested in an 0402 or 0603 device, one now has the opportunity for a metric/imperial nomenclature mistake, which could precipitate lost time and money.

When the world standards committees discovered what the EIA had done, they released an order for the EIA to cease publication of this non-metric document. The EIA was reminded they were in violation of the international agreement they signed with all the world standards bodies agreeing NEVER to publish ANY standards using Imperial (English) units. In 1991 the EIA stopped publishing the feral document, and it is my understanding, that because of this, there is no longer any official standard  followed in the US.

The US component manufacturers and PCB etching houses returned to the days of the perch, furlong, and barleycorn. With no standard to follow, SMD manufacturers began to game the situation. There were no longer standards for capacitors or inductors, or common three leg transistors known as SOT23.  So 20 different sizes of these small outline transistors (SOT23) appeared. Chaos ensued. Rather than impose order by standards regulation, or metric adoption, the US industry just tried to figure out a way to name the multitude of these ad hoc non-interchangable “standard” parts.

The rest of the world embraced metric measurements and metric standard electronic parts. If you are in Germany and order parts from Japan, or Korea, or Timbuktu, you know they will fit on your printed circuit board. These are all metric countries.  You have no guarantee if you order American electronic surface mount parts, that they will fit. If you were a German, would you take the chance?

The situation is actually far worse than I have explained thus far. In the United States our PCB software puts down grids in mils (a feral unit of the inch) or in inches. The world standard for parts is metric. The standard grid size for which these parts are designed is 0.05mm, so there is no reason to expect the metric parts to fit nicely on an inch based software grid. They are two different measurement units! The software used to connect up parts makes many mistakes in a mixed imperial/metric environment. We often have to fix these “by hand.” There is no guarantee of compliance to layout standards when metric and imperial are mixed. Metaphorically, we are trying to fit square pegs into a set of round holes. Metric parts on a metric grid are interconnected by software to international standards.

Vacuum Tube SchematicOver the last decade, I’ve watched as one small US based printed circuit board house after another have gone out of business. The company where I last had full-time employment, picked up and moved to China. How much of this PCB work might have remained in the US if we had embraced the metric system years ago? It’s hard to say, but it should be clear that metric conversion is of paramount importance—period. Unfortunately, the idea of industrial policy disappeared as a government concept in 1980, along with metric conversion, and has not been contemplated since.

When construction, medical, electronics and other industries are all taken together, one has to wonder just how much it has cost us not to have mandated the metric system in the United States. There were legislative efforts to make metric the official measurement system of the US in 1866, 1902, 1921, and 1975, and they were all either rejected by Congress, or were voluntary and therefore impotent.  We may never know how much it has cost the country, because the US may never become metric.

The technically ignorant pundits who populate our media space, will continue to point to reasons that are divorced from any understanding of Engineering, technology, manufacturing, or the metric system as a problem with US manufacturing “competitiveness,” and the rest of the world will  continue to reap the benefits of their choice to become metric—decades ago.

9 thoughts on “US Electronics: A Metric Peg in an Imperial Hole

  1. You have awoken one of my issues – most of the case sizes are obviously in English or metric, but there are two that are used in both. 0603 and 1005. Because ordering information is still often in English units, I started putting an ‘M’ on the end to keep them separate:

    http://wiki.xtronics.com/index.php/SMT_Case_Size_Codes#Imperial_and_metric_case_size_codes

    There are still enough English unit parts around that CAD systems have to support both – and that produced a long thread on the mailing list about using nanometers as the raw data – the use of this extra precision reduces the maximum size of circuit board – thus a long debate..

    For me, the hard part of transitioning was having the ‘feel’ for feature sizes in mils(metric Inch in this case) and converting my head to thinking in metric. It is not a trivial change, but it is happening.

    It is also common to use the insane standard of numbered drill bits for circuit board hole sizes. I now only work in Metric unless I’m creating a part ‘foot print’ from an English unit drawing.

  2. As you quite rightly state, none of the “economists” trying to get the US economy back on its feet takes metrication seriously. They still live together with their medieval measurements in the past and piously hope America will once more dictate what happens in this world.

  3. I worked as an electronics assembler in the 1970 when attending High School. I remember P C B mounted with number six screws in a chassis. The wire spools in a closet according to A W Standard. Capacitors, transistors, and resistors were measured in SI units. I thought wires, chassis, and fasteners were all going to change to SI as well. It is tragic that manufactures in the United States tried to force the world to inches, foot and pounds as the world was at the end of World War 2. I never thought this current situation was this bad.

  4. One can only wonder what it is that makes people hang on to a Hodge podge of medieval feet and thumbs with a much simpler and universal system on offer for free? Three culprits come to mind, misplaced megalomania nurturing a misplaced patriotism and the worst of all, habit. No country ever managed to convert voluntarily to the metric system. Well if intelligence does not triumph maybe unemployment and the ensuing misery will do the trick in America. Pity, that such a superb idea has to be forced on presumably intelligent people?

  5. Question: Is the P C B grid need to be 1 mm by 1 mm and the components: capacitors, transistors, and resistors 0.05 mm for good fit?

    • Please don’t take this amiss, but you are actually asking for a complete crash course in surface mount PC fabrication. There is simply no way to answer without knowing a great deal more about what you are trying to do: a one-off prototype? a limited production run? or a million-unit run of the next iPhone? The general rule is that the larger the run, the tighter the component spacing engineers are willing to risk. But knowing manufacturing processes is also critical. Will you be wave soldering (almost never used for surface mount these days), or reflow soldering (this term covers a variety of related processes, and you need to know exactly which, and the tolerances of the reflow: “Of course, your reflow may vary.”).

      Or, will you be doing what I personally know a multi-billion dollar aerospace company was doing only a couple of years ago, viz: having superbly talented female technicians, with the patience of saints, spend weeks in front of stereo microscopes, hand placing and soldering 1250+ components as small as 0402 (imperial, dammit!) on a run of twenty-four boards I designed. They wore surgical masks, because soldering an 0402 is about like trying to solder a grain of coarse-ground black pepper, and the slightest breath would blow it away. They also had their own personal portable DVD players set up next to their microscopes, and were playing what I took to be the Chinese-language equivalent of chick flicks, as a sanity-preserving strategem: to occupy those parts of their brains not directly concerned with component placement.

      All this because the American gigabuck company in question did not own a pick-and-place machine, and its board-layout engineers would have been terrified of it, even if they had. Pick-and-place: this is another technology you have to understand in excruciating detail, because it also affects the spacing of components.

      The best I can do is to suggest that a 1 mm grid would be considered incredibly coarse and obsolete by today’s manufacturing standards, even in the U.S. Hope this helps.

  6. I was asking in relation to the article about the grid size alternative to U S C inches.

    • I clicked the hyperlinks to background articles. What grid size is needed to avoid the hand placing?

      • Well, first of all, grid size is not component spacing. And second, you seem to suggest you need greater spacing if you are to avoid hand placement and soldering. Actually, it’s the other way around: machine placement of parts, and reflow soldering, permit much smaller parts, and allow you to pack them far more tightly on circuit boards than any technician could possibly manage. Humans need room for their fat fingers. This may help:

        There’s a program you can download here. Actually, there are three: the full program you have to pay for up front, the slightly sawed-off version is free for ten days, and the Super Galactic Class version is just free. Download the freebie: it’s all you need. You have to register, but it isn’t as creepy as Facebook. What the program will show you, are the outlines of several thousand common electronic parts, along with the “land patterns,” which they are mounted upon, and also rectangles showing how much “real estate” should be left clear around each part. I take it this is really what you are interested in.

        The thing to remember is that these dimensions are really only suggestions, and they vary depending on which library of parts you use. The program comes with three libraries, which I tend to call Papa Bear, Mama Bear, and Baby Bear. The Papa Bear library shows the largest lands and widest clearances. This is the library to use if you are soldering by hand. The Mama Bear library shows tighter spacing, suitable for most pick-and-place and reflow manufacturing operations. The Baby Bear library packs parts together extremely tightly, and should be used only if you are Apple Computer. (And if you were Apple Computer, you wouldn’t be using this program anyway, because your manufacturing processes would already be far better controlled, and completely, natively, metric to begin with.)

        There is no single number for component spacing that applies to all parts, in all situations.