Any dropper can work, but resistors are poor
performers, as they reduce starting v and i, resulting in fans
failing to start when friction increases over time. Series diodes, if
you need a voltage other than 5v, are a better bet.
I have used series diodes, even made a 12 position rotary
switch to place additional diodes in series so I could test
different fans to determine what number of diodes would
produce the target RPM. After determining that and
comparing the result with use of a series resistor instead,
I find the resistor mechanically and functionally superior
if/when there is ever a difference.
I just don't ever see spin-up problems using the resistor
because if the current were limited that much to produce a
target RPM that spin-up were a problem, then most often that
target RPM is so low it had already introduced more
pulsating noise (if diodes or other voltage limiting were
used (even noisier than if it were ran at a slightly higher
RPM due to this)).
Today the only time I make use of the 12 position diode
switch is a quick test (since it has the appropriate
connectors to directly plug a 4 position molex plug into
it) to determine if a fan might run acceptibly at 5V, and
thus not needing any voltage or current limiting strategy
but rather taking it's molex pins and extracting and moving
the 12V pin to the 5V position for directly powering on 5V
rail. I tend to never power fans direct from 5V unless in a
rush, because this too produces more pulsating noise than if
a resistive limiter were used.
I use resistors in practically every system I set up and
never has the "friction increase" been a problem, including
systems over 10 years old. If the bearing is shot enough
that you find it can't start with the resistor, the friction
is bad enough the fan was about dead already and it would
only be a short while until no method would have it spin up.
The key then is not how it was throttled but to either
relube it if sleeve-bearing or replace with a better fan.
It is better not use a voltage reducer, to use a current
limiter instead. The difference is that at a certain point
of voltage reduction, the fan starts making a ticking sound
because with only voltage limiting there is a surge current
each time the (next) motor coil is energized, a sudden
increase in torque which is not just producing a new
additional noise but also harder on the bearings. The noise
is also at least as objectionable and probably moreso
because it is not a constant monotone but rather
tick-tick-tick rapidly. Granted, with some fans it is very
faint, but the resistive method is quieter.
With the resistive current limit instead the fan can run to
lower RPM and is gentler on the bearings, in addition to
being simplier to implement with one resistor than having to
tack together roughly 6 to 10 (!) diodes in series to get
the typical came-with-case fan down to a more reasonable
noise level. Not that I prefer the typical came-with-case
fans, but some can be acceptible so long as they're not the
cheapest sleeve-bearing junk and/or are relubed
periodically.
If you were only trying for a lesser voltage reduction, say
down to ~7V or above, either the current limiting resistor
or voltage reducing (diodes or linear or whatever) will work
acceptibly, as that is high enough that few fans would start
making a ticking sound and run with torque pulsations.
