>>> A manufacturer of domestic toasters finds that a large (many
>>> thousands) consignment of toasters has a fault.

>>> Each toaster has a double pole switch which disconnects the
>>> element from both leads of the mains cord when the toaster is
>>> plugged in and not in use.  When the main "toasting handle"
>>> is depressed the two pole switch closes, passing mains power
>>> to each side of the element.  In the faulty batch, in about
>>> 5% to 10% of the toasters, one or other pole of the two pole
>>> switch is mechanically jammed on due to a fault in the design
>>> and manufacturing process.

>> Can we assume that the handles of the packed toasters
>> are all in the OFF position ?

> Yes.


Use relative, rather than absolute, values of capacitive or
inductive behavior at the plug prongs to assess failed switch.

I believe that the electrical characteristics of either the hot
or neutral lead in one power cord, when measured in relationship
to the ground lead, will be very close if the lead terminates at
a properly (i.e. open) switch.

If one pole of the switch is shorted, then the mated power cord
lead will have significant variance (due to connection of the
heating element) compared to the other lead in that cord.

Specificially, build a small device with a "wall outlet" into
which one toaster is plugged, a "start" button, and pass/fail
result light(s).  Test is as follows:

  0) worker plugs suspect toaster into test device

  1) worker presses start button

  2) check for DC continuity between hot and neutral pins to
     find switches with double fault -- both poles shorted.
     If found, illuminate fail light.

  3) measure capacitive / inductive / resonant frequency of
     hot lead in relationship to ground lead

  4) measure capacitive / inductive / resonant frequency of
     neutral lead in relationship to ground lead

  5) if result values in steps 3 and 4 are similar, then both
     switch poles are open; illuminate pass light.  If result
     values are not identical (i.e. dissimilar), then one pole
     of switch is shorted; illuminate fail light.

  6) when worker sees either light illuminate, they unplug
     toaster and put in proper pass or fail process queue

  7) after 3-5 seconds, test device turns off both lights.
     This step attempts to reduce false results if worker is
     distracted and assumes previous unit's results apply to
     a newly connected unit.  (Better that worker has to test
     a unit twice than a unit slips through without testing.)

Some experimentation is required to determine how close the
values have to be to be "similar" or "dissimilar" in step 5.
Test device only needs short term stability (allowing cheap
components that drift) and no absolute values (allowing high
variability between component values in different test devices).

Minimaly trained workers should be able to do quite a lot of
toasters per hour using the above device & procedure.

And a PIC would be well suited to drive such a test device.
Question now is will the non-recurring engineering cost to
develop & build a small number of testers be low enough, when
combined with labor costs, to justify testing the toasters.

                                                Lee Jones

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu