[sdiy] Pot life specifications

[David G. Dixon]

So, let's say a pot "fails" after 20,000 rotations or so.  What happens,
exactly?  Does the wiper start failing to make contact with the resistive
strip?  Does the resistive strip wear off the fiberglass substrate?  Does
the resistance change dramatically, and does this happen mostly in the
middle of the rotation, or at the ends, or only where the pot has spent most
of its life being set?  Or, does the pot just start generating noise when
you turn it?

Does it happen gradually or catastrophically?

Do the terminals fall off?  Does the shaft fall out?
 
In other words, how would you know that your pot is worn out?

I use mostly Alpha 24 mm pots.  Do 24 mm pots last longer than 16 mm pots or
12 mm pots?

BTW, I love the way they feel, and the fact that they cost a whopping $0.85
each in lots of 100 or more (mixed values) from Small Bear.  Since I don't
expect they will ever wear out, and if one does I can change it in five
minutes for less than a dollar, I don't see any reason ever to plump for
Bourns or Spectrol.


> > It's quite easy to get an indication of how good the pot (or encoder)
> life
> > is by this method;
> > 1. Take a pot and hold it in a vice with the shaft exiting upward.
> > 2. Using a retort stand or other method, suspend a dill above the pot,
> on
> > axis, and with the pot shaft in the drill chuck.
> > 3. Start the drill and leave to run for 10,000 revolutions or more.
> > 4. Mechanical wear can be additionally simulated by having the drill
> > slightly off axis causing an additional wobble.
> >
> 
> I don't think this is an accurate measure of pot life for a couple of
> reasons:
> 1) Most pots don't turn all the way around (360°), so you're going to
> need a limiting / reversing mechanism.
> 2) The heat generated by continuous motion would not be the same as
> generated by "normal" use. This heat would definitely shorten the life
> of the pot and not give you an accurate projection of real world
> longevity. An automated test like this would need to turn the pot,
> wait for a second or two, and then turn the pot back. You end up
> needing something with a timer and a (small) washing machine gear
> mechanism. I've been involved in testing like this at Apple, and you
> have to take all sorts of little details into account.