Probably the simplest one is a half-wave rectifier that uses a diode to
"throw away" the negative half of the AC. This will cause 60 Hz flicker
that can be stopped somewhat with a really massive filter capacitor.
In other words, if this is what you have coming in
(use a fixed width font):
/\ /\ /\
\/ \/ \/
and you have a half-wave rectifier with no capacitor, this is what you
get coming out:
/\__/\__/\__
With a big capacitor, this might be improved to something like
/\--/\--/\--
i.e., not dropping completely to zero between the peaks.
The next simplest is a full-wave rectifier that "inverts" the negative
half of the AC. This will cause 120 Hz flicker that can be stopped with
a less massive filter capacitor. You might have
/\/\/\/\/\/\
with no filter capacitor and something like
/-^-^-^-^-^-
with a capacitor - i.e., not dropping much at all between the peaks.
These first two (full and half wave rectifiers) are probably most common
with "linear" power supplies that use a transformer operating at the
line frequency - the "heavy" ones. A "switching" power supply, with the
transfomer operating at line frequency - like most desktop and laptop
computers (the "light" ones), will _probably_ have less flicker in
general for a couple of reasons. One is that with the "peaks" of
current coming much closer together (anywhere from a few kHz on up, as
opposed to 50 to 120 Hz), they can be filtered with a much smaller and
cheaper capacitor. Another is that since the the manufacturer has
decided to spend a little more money on the switching supply (vs. a
linear supply) in the first place, they are probably a little more
inclined to include adequate filtering. It's entirely possible to build
a cheap, crappy switching power supply that causes lots of flicker in
the output, though.
Another possibility is to use a capacitor (instead of a transformer) to
reduce line voltage to something suitable for powering a LED. This
would also create a relatively lightweight supply. I am not sure what
the waveforms end up looking like for this option so I can't comment
much.
I suspect the only thing that light specifiers can do is to require that
the flicker index be below some value. Then they need to measure the
luminaires to make certain they comply.
Vic's suggestion is probably the only way to know for sure. A first
step might be to specify "full-wave rectification" and/or "switching
power supply", as those choices are slightly more likely to reduce the
amount of flicker you see. But, since it's possible to build a linear
power supply with full-wave rectification or a switching power supply
that cause the LEDs to flicker like mad, measuring is the only way to
know for sure.
If the connections between the power supply and the LEDs are accessible,
you can put an oscilloscope on them to see what's going on. You don't
need a fancy oscilloscope with lots of bandwidth for this - 1 MHz would
probably be fine. You _do_ need to have either a portable (non-line-
powered) scope, or understand how to use an isolation transformer to
prevent fireworks, to do this safely. This may not help much in the
specifying process but it might be interesting just to see how different
manufacturers do it.
I wonder if the phosphor on white LEDs tends to act as an additional
"capacitor" to filter out any flicker from the power supply.
Matt Roberds