GlytchMeister wrote:Dave wrote:Alkarii wrote:Nah, I'll stick with the corral trap, since it'll be easiest to make, and safest.
For people, I mean, not the pigs. Obviously we intend to kill them, so it doesn't need to be safe for the pigs.
Probably a good choice. The fancier you get with your design,
the more chance for something to go wrong.
TRUTH
You just earned some engineering points, Dave.
Heh.
Micromeritics (where i worked in the first half of the 80s) does high-precision materials evaluation instrumentation, including high-pressure liquid chromatography.
Get set for a sea story*.
They developed a new HPLC detector.
Digressing just a bit, for clarity for those not familiar with the concept, in HPLC, you pump the liquid you want to analyse at high pressure {duh} through a column packed with a material that adsorbs and releases the different constituents of the liquid at different rates, which causes them to spread along the column and emerge at different times from the far end. Since you know {or should} the rates at which the different components move along the column, you can determine how of each is there by monitoring the outflow of the solvent/carrier liquid involved, and detecting when and how much of other substances come out. {You can also slice the stainless-steel column lengthwise with a bandsaw and see where bands of the different materials remain in the packing, but we're not interested in that right now.}
Okay.
To detect the {frequently visually colourless} stuff as it flows out, you use an LED emitter that operates in bandwidths that the stuff isn't colourless in - usually, if not always, ultraviolet. You can select the precise wavelength you want by using a diffraction grid set at an angle to the beampath; since part of the procedure is based on the fact that different solutes absorb different wavelengths of UV more or less strongly, there is a high-precision mechanical vernier that lets you choose the exact angle of the grid to favour one wavelength or another. {Well, it was mechanical in 1986; quite probably digitally-controlled these days.}
Taking into account time, UV absorption numbers at different wavelengths, etc., you can get a pretty damn good idea of how much sugar, cola extract and cocaine is in your soft drink sample.
You probably have to make several runs using different wavelengths, to get the full breakdown.
"Well, by golly," thought the Micromeritics Brain Trust, "what if we could build a detector that could use more than one wavelength at once? That would speed things up. And we have access to new types of components - we could make it more sensitive, too!"
And the fun began.
The detectors used a photo-diode that generated a current when illuminated by the beam. The current was proportional to illumination strength, of course. The diodes were mounted in a block machined from brown Delrin, which is a REALLY good electrical insulator. The current is then fed to a current-to-voltage converter which isolates the measurement components from the detector, because loading effects of attempting to measure a current at that low level would cause the incorrect readings, big time.
The current-to-voltage converter uses {well, did use, thirty years ago} an
operational amplifier, whose input impedance is, for most practical purposes, infinite. Since no current flows into or out of the op amp input there is no loading on the diode's output. {This is, as i said, "for all practical purposes".}
SO, one of the IC makers had recently brought out Field Effect Transistor {FET}-input op amps, whose input impedance was one or more orders of magnitude greater than that of conventional-transistor-input op amps. This allowed them to make the current-to-voltage circuit more sensitive, because they could detect smaller currents, so they did.
Boy howdy, did they.
And the trouble started.
They had designed a current-to-voltage converter that was sensitive to current variations in the
femto-flippin'-amperes.
We discovered that, unlike previous designs, the photo-diode leads had to be soldered {i THINK previous designs used sockets, but the variable resistance of the lead-to-socket contact caused detectable current fluctuations, which varied with, among other things temperature and humidity}.
And you couldn't just do a conventional wire-joining solder joint, wrapping a wire around the diode peg and soldering; you had to lay a perfectly-straight tinned piece of wire-wrapping wire alongside the tinned diode lead, making sure that they were in contact for the entire length of the junction, then apply JUST Geno heat to join them.
Without bumps in the solder, because bumps - like bends and wraps - cause shot noise in the current.
And THEN they discovered that brown Delrin (remember the brown Delrin blocks the diodes were mounted in?), which effectively has infinite electrical resistance in applications up through kilovolts and probably even megavolts ... leaked enough current to throw off the readings. At effectively zero voltage, the carbon black {i think it is} that gives the stuff its colour leaked a few femtoamperes of current.
They had to make the mounting blocks out of virgin white Teflon.
==============
And, possibly because of that, possibly because of other "innovative" design choices, the damned things were unreliable. They sold two to Coca-Cola for its syrup lab in Hapeville, on the south side of Atlanta {based on the specs, before we'd ever actually turned one out, i suspect, remembering other fiascoes perpetrated by Marketing selling things it turned out we couldn't deliver}. Which wound up involving three or four machines and some field-service techs on a regular basis for i-don't know-how-long, because the things kept drifting out of spec and would have to be replaced with a freshly-calibrated one and hauled back to the plant thirty miles away, in Norcross, on the north side of Atlanta, every week or so.
====================
But it won a prize for innovative design from an engineering magazine that year.
================
* You quickly learn to tell the difference between fairy tales and sea stories because fairy tales begin "Once upon a time..." and sea stories begin "Now, this is no s**t, but..."