Most light bulbs hum loudly when dimmed via a dimmer switch. A few are ok,
because they contain extra filament supports at critical positions.
My hanging (swag) kitchen tiffany-style light takes a G40-150w bulb. It
hums badly when dimmed.
(1) Is there a brand of G40-150 bulb that does not hum when dimmed?
(2) Alternatively, is there a small, in-line filter available, or a
filtered, table-top dimmer switch, that would create a smoother
short-duty-cycle output than the intermittent, alternating square-wave
created by a typical dimmer switch? Would that be safe for a household
lamp application? Would that eliminate the audible humming?
From my experience, some lamp types whine more than others. Typically the
longer the filament stanchions are, the more vulnerable to whine the lamp
is. I've also had the best success preventing whine with Lutron dimmers
We normally use it at full brightness. We dim it for special moods.
A "G40" is a 5" globe with a standard hosehold lamp screw base. It
requires a higher wattage for the filament to brightly light the globe
surface (inverse square law from filament to globe surface).
I don't think there's much you can do in a practical sense to "filter
out" the the non-sinusoidal waveform without engineering a special
purpose high chopping frequency dimmer, which would prolly waste a lot
much power in the form of heat.
BTW, it's not an alternating square wave, it's a sinusoidal waveform
which has a portion "removed" from every half cycle when you turn the
dimmer down from full brightnes. The lower you set it the less of the
original waveform is left, but the parts that are left follow the line
Before solid state stuff was a twinkle in Shockley's eye they used to
make wall mounted variac light dimmers for use in rich folks houses, and
I'd bet that they didn't make the light bulbs hum.
That's not been my experience.
I only have three dimmers for lights, but I've never had a lightbulb
hum because of one. Until a couple weeks ago, I was using a 150 watt
bulb in a desk lamp with a built-in dimmer right next to me.
Unfortunately, I knocked the lamp off the table and broke the bulb so
I don't have it, but I think it was the standard GE 150 watt bulb they
sell in supermarkets.
In the dining room I also have a 150 with a dimmer, and I'll check the
brand if you want me too. I only have to take off the cover. It's
probably not G-40 since I've never heard of that, so let me know if
you want me to check.
Otherwise, I'd say just try a different brand.
Since the output of the dimmer is still AC, that would dim it further.
by cutting the current in half, unless you used something like a
bridge rectifier (4 diodes arranged in a square) that is full wave.
(Sorry, maybe htat is what you mean to begin with.)
Then you'd be running yhour lightbulb on DC, and I would be very
intersted in how well that would work.
Filament singing is a common problem with phase control dimmers. I
design phase control dimmers for theatre and television use. The amount
of sing can be reduced by limiting the rise time when the switching
device switches on. With triacs and SCRs, this is normally done with a
choke in series with the device. Wall mount and other consumer dimmers
have relatively small chokes due to the amount of space available and
cost considerations. The choke is often a a "stick" of ferrite with
wire wrapped around it. We use toroid cores that are considerably
Some professional dimmers are now using IGBTs as the switching device.
These can be turned on more slowly, increasing risetime and limiting
filament sing. The slow turn-on increases power dissipation in the
device, but there are also losses in chokes used in thyristor dimmers,
so the total losses may be equivalent. These dimmers often vary the
rise time with heat sink temperature, speeding up the rise time as the
unit heats up, limiting total temperature rise.
Some IGBT dimmers also do "reverse phase control" where the lamp is
dimmed by an "early turn off" instead of a "late turn on." Fall time
control is used here to limit singing.
IGBTs are considerably more complex to drive when compared with SCRs
and triacs, so IGBT based dimmers are generally more expensive.
There are also some "true sine" dimmers that high frequency chop the
incoming AC, then filter out the high frequency. Due to complexity of
the high frequency drive of the IGBTs and the requirement for a high
power high frequency filter on each channel, these dimmers are also
more expensive than phase control.
As Jeff points out, Variac light dimmers (variable autotransformer)
output a variable voltage sine wave which keeps lamps from singing. My
high school auditorium had a dimmer system based on these. It had
really big handles you run up and down to control the lights.
Way way back, theatres used salt water for dimming. Electrodes would be
dipped in and out of salt water to bring the lamps up and down. I don't
suggest this at home!
If we are worrying about lights bulbs 'singing/humming' it makes one wonder
what other RFI (radio frequency interference) these somewhat unecessary
devices are causing? Why not use cheap bulbs and turn off a lamp or two?
And since many (most?) dimmers are manufactured 'to a price' is it correct
to assume that compromises are made?
This is the type of technology that I was hoping was available in a fairly
small package, for home use.
I was wondering if a simple capacitor filter would smooth the rise and
Thanks for the offer. But I do need a G50.
A G40 is a 5" diameter globe with a standard screw base. I could use a
plain 150w bulb, but it would not be as attactive. This is an tiffany-shade
type ceiling swag over the kitchen table.
A capacitor "attempts" to maintain a constant voltage (I=C dv/dt),
causing a current to keep the voltage the same. An inductor "attempts"
to maintain a constant current (V=L di/dt), causing a voltage to keep
the current the same. When the thyristor turns on, there's 0A going
through the inductor, so the full line voltage ends up across the
inductor. The current then ramps up at a rate inversely proportional to
the inductance (lamp resistance makes this an RL circuit, so it ends up
being an exponential ramp with time constant of L/R, but when the
thyristor first fires, the ramp starts out linear). This is exactly
what we want. The instantaneous current rise is slowed down to several
hundred microseconds of rise time. I can't think of how you'd be able
to limit the current slew rate with a capacitor. Sorry!
So the bulb itself shows. And it's 5 inches, so it might have a much
bigger filament than the minimum size 150 incandescent.
You might be better off removing the dimmer and buying the size you
most want to use.
I wanted the 150 for the dining area, because I only have one lamp, a
floor lamp, in my living area. It's always sort of awkward to get to
it, and sometimes near impossible, and it doesn't light up the whole
area. So I want a bright dining room bulb. If it weren't for that, I
think I'd use a 100W.
Years ago I tried to use a bridge and two 40mF 450V caps to banish some
nasty hum coming from two chandeliers, each with five 50W bulbs. Way too
much load for the caps. The hum remained and the caps were seconds away from
exploding due to their instant heat buildup. It might have worked with a
Thanks, everyone, for your replies, suggestions, and electronics
I haven't quite decided what solution I am going to explore long term, but,
for the moment, I am using the light undimmed.
I don't get it. I thought that if the capacity of a cap was exceeded,
it just filled up on one side, and after that the rest of the current
behaved as if there were no cap. In fact it occurs to me that in a
DC power supply circuit for a radio or tv, the only reason the cap
ever gets below full charge is that the load is *high* enough to draw
more than is currently, during low parts of the cycle, being provided
through the diodes, so it drains the cap.
During the high points, the peaks of the 120 cycles per second power
(after rectification) there is more than enough power and that's when
the the caps are refilled.
Lowering the load would mean the cap would fill up on one side, and
then just stay filled all the time.
Two 40mF sounds like a lot, but if it wasn't enough, it seems to me
there would have been no current in or out of the caps after the first
If 80mF was enough to filter, maybe the internal leads couldn't handle
the current in and out without getting hot, even though current in and
out is what caps do. Maybe that level of heat was within range.
And I would also think that nothing 110 volts could do, even
full-rectified to make it higher than 110, could damage a 450V cap.
I would also wonder if caps are necessary, since an incandescent bulb
with pulsing DC current would remain hot and giving light, despite the
pulsing. Don't electronic dimmers work by completely turning off the
current parts of the time? And yet all we see is a constant but dimmer
light. They don't use caps at all except maybe little ones to make
Posted and mailed because it's been almost 3 days and ahr is so busy,
I'm not sure anyone is reading this thread anymore. So I wanted the
poster to know I had replied.
In this case, the cap is way too small for the 250W load, and it is well
drained before the next 'fill up'.
Since the cap is pretty fully drained, there is a very high current at
each 'fill up' followed by high current to the load that drains the cap.
In addition to the leads, the capacitor has an internal resistance which
you might find specified as ESR (equivalent series resistance). Trying
to filter for light bulbs causes a relatively high current charging the
caps and then discharging to the lights. That will cause heating when
flowing through the capacitor's internal resistance. A reasonably sized
(larger) filter cap would have a far smaller ESR. But the dimmer may not
like the high peak currents to charge a large cap.
The AC voltage is a sinewave whose value is constantly changing. 110
volts is the RMS value - a form of average. The peak voltage is 1.4
times that - about 155 volts, which the cap would charge to with no
load. Far lower than 450V as you said.
Dimmers work by turning on late in each 'hump' in the sine wave. The
dimmer stays on until the next zero crossing. At full brightness the
dimmer turns on at the start of the 'hump'. At low brightness the dimmer
turns on late in the 'hump' and only the end of each 'hump' is there.
Rectifying just makes all the 'humps' positive instead of half being
positive and half being negative. The same basic waveform, which causes
the singing, is still there. Capacitor filtering would change the
waveform to DC. Series inductors, which some (all?) dimmers have also
changes the waveform, and if there is enough inductance will eliminate