I went down a few rabbit holes while researching the Advent Calendar last year, that didn't make it into the queue because they got too long or went too far afield. Here's one of them!
One thing you notice when you watch a bazillion videos about old games consoles is how the design of circuitry has evolved. If someone says 'circuit board' today, you think a light piece of leafy green board, filled with parallel lines of copper at 45° and 90° angles, dotted with lots of tiny inscrutable plastic and metal doodads. But it took a long, long time for them to get that way.
If you look at really old circuit boards -- and I mean really, really old circuit boards, like from the beginning of the transistor era, they look completely different. They're brownish, for one thing. And kind of... wiggly?
Apologies for the transfer quality. It's not your connection, it just sucks. This piece appears to be some sort of promo-tainment thing from Tektronix themselves, from 1969. The rounded corners and bluish fuzz at the edges is an effect called 'vignetting', and it means this is originally from a 16mm film reel. There's no earthly reason for film to look this terrible. The uncentered picture means someone copied it by pointing a camera at a projection screen instead of bothering to get a proper kinescope setup, and the fact that it only goes up to 240p makes me feel like it was originally transferred over two decades ago for RealPlayer and nobody bothered to fix it for YouTube. VHS is about 240 lines, but if this were a crap transfer from a VHS tape you'd also see scanlines. It's possible there's a better copy at VintageTek, a museum dedicated to the history of Tektronix; they are an all-volunteer institution, and they probably have more important things to funnel funding to than updating their YouTube channel.
Point being, it looks like porridge and I'm sorry, but at least the content is interesting.
The brownish color, which is actually from an evolutionary stage earlier than what's covered here, is because many early boards were milled of bakelite rather than electrodeposited onto a glass or fiberglas backplane. If you want to see some of what that might have been like, you can hop over to Usagi Electric. He uses CAD to mill boards, rather than the photochemical process described by Tektronix, but it's pretty much the same idea. He does a lot of it in pursuit of his mad obsession with building a vacuum tube computer here. (If you're curious, his logo says うさぎ電気, "Usagi Denki". "Usagi" is Japanese for rabbit or bunny -- there is one who appears at the end of some videos -- and the spelling of "denki" here specifically means electrics, as opposed to 電機, which is usually rendered electronics. It still pops up in the names of some engineering or technology firms, but generally only the really old ones.)
The wiggly nature of early boards is neatly explained by watching the drafting process, starting about three minutes into the video. It was originally done by hand. The rest of the half-hour video goes through the whole multi-stage process, but the gist is that when you lay out the board, you draw dark lines where you want the conductive traces to be on the final product. To get a consistent size, tape is used for "holes" and tape lines are uses for the traces. If you've ever used stripe tape in nail art, it was apparently something like that -- vinyl tape with a bit of stretch, so you could curve it around. It was a methodical sort of art form. Ever solved one of those "connect the same-color dots without crossing lines" puzzles? It's basically that. If you can't find a topologically-appropriate solution on a single plane, you can produce boards with traces on both the front and the back, as Tektronix does here, and these days you can actually bury traces in internal layers as well. It's just a pain and makes the cost go up exponentially.
The mention of "holes" is interesting. Early circuit boards were nothing but holes. Everything had legs and was soldered on from the underside. Today these are known as "through-hole mounted" components; the alternatives are "surface-mount" components, which are generally smaller and fiddlier to solder on by hand, but considerably easier to lay down and solder in place by machine. Surface-mount technology has been around since before this Tektronix piece, but remained NASA-grade esoterica until the automated assembly process became cost-effective in the 1990s. Today the conductive holes are referred to as "vias" and the little medal dots surface-mount things are soldered to are "pads".
I'll also note that they show the automatic soldering process for these boards late in the video. It involves skimming the boards across the surface of a pool of molten solder. Solder in the 1960s contained a lot of lead. I would not personally like to be in that room. Today a machine places little surface-mount doojiggers in place along with solder beads, and then melts it all very gently in a very hot oven until it all melds together, not unlike a pan of slightly too-runny cookies. If you do it right, the surface tension of the solder keeps it on the pads and out of the traces. This is particularly useful for placing CPUs, whose myriad tiny pins in a tight grid would be far too difficult to solder by hand, and the origin of "reflow" repairs for electronics that are exhibiting symptoms of flaky solder joints.
The "silkscreening" process here does not use silk, but originally it did -- it was invented in Asia, logically enough. The gist of it is that you take a piece of finely woven mesh, traditionally light silk but in modern times also metal or synthetic fiber, and you plug up all of the little holes in it in the areas where you don't want ink to get through it, usually with some sort of water-repellent substance. In the days of yore, you painted on some kind of sap or wax, but nowadays it's usually a light-sensitive plastic that's scraped across the whole mesh, topped with a stencil that is opaque where you want ink to flow, and exposed to UV light that sets the substance. The unset areas that were in shadow are rinsed clean, leaving the mesh permeable in those places. The ink emulsion is then applied to the printing surface beneath in the reverse process: Ink is spread across the mesh, then squeegeed through with enough force to push it through the holes in the weave and onto the surface beneath. The dots of ink bleed just enough to flow into one another, producing a solid area of pigment. The circuit board designs were originally drafted in black on a white background, then photographed and reduced to 1/4 their original size, and the film used as the stencil for the silkscreen.
Holes are drilled mostly by hand(!) in this clip, which is an error-prone process, as you can see from the Usagi Electrics guy. The worker uses what's called a pantograph drill. A pantograph is a device that translates motion from one place to another, often with a change in scale. Typically pantographs are mechanical in nature, based on the complimentary motion of opposite corners of a parallelogram, but you could make a pretty good argument that modern systems that accept movement inputs from a user and translate them elsewhere by computer are also members of the class. Robot-assisted surgery comes to mind. If you cared to have an even longer argument, you could also consider systems that scan items with laser photons in order to reproduce them on a lathe or CNC machine pantographs in spirit, if not in fact.
A visual or optical comparator is just a device that projects a magnified view of something up on a screen, along with a point, grid, or profile it needs to match, not unlike a microfiche viewer with a targeting reticule. They're still used in some areas, although software image processing is steadily gaining ground.
You'd be amazed at how many things still need a look-over by a human with a brain. The lack of human brains is how we got the sharply-angled board traces we have today, in fact. Computer-aided drafting was developed to a usable level in the 1980s, and predictably the people using it were mostly engineers. The kind of route-finding you do in those connect-the-dots puzzles, and that the electronics engineers did when drafting the boards, is one of those very slippery human things. You want to find the shortest path, to save on the precious metals you use as conductors, but the absolute shortest path (with reasonable tolerances) is often a very snaky curve that would require a large number of points to define. It's much simpler to work on a grid, hence the 45° and 90° angles -- this ensures that all trace paths can be defined exclusively by where their corners lie on a square coordinate system, and is much less calculation-intensive. This was a lot of what early graphics tablets (or digitizers) were used for, and some light pen systems.
Having watched my father do a lot of this as a kid, I gather that at least in modern CAD software, you can just pick things up and put them wherever you want, but that the autopathing gets very confused if you do it too much -- mostly it's better to let the computer figure out where the traces go and tell you if you want something impossible in 3D space. And if you screw up anyway, there's always blue wire.
Circuit boards don't have to be the ubiquitous green, either. That's just the color of the solder mask, a lacquer painted all over the parts of the board you don't want solder to stick to. It's mostly tradition at this point, but you can get boards in pretty much any color you like -- the second most common I've see is a dark navy blue, probably because copper traces and white silkscreening stand out best on those two colors. You're welcome to get neon purple, if you can find anyone offering it.
Comments
Post a Comment