Given that it's AC and not a 'steady' flow like DC, I suspect not. If it's part of a 3 way switch you could try it in another location. My only other thought is to go to a really low wattage, one more like the brightness you want.

Have you tried dimmable LEDs? They may be quieter, but otherwise it's the nature of the device. Unfortunately dimmers need to be kept cool so covering or muffling them isn't an option either.

The dimmer works by waiting a period of time after the zero crossing and then firing the triac which conducts until the next zero crossing. When you set the dimmer at half intensity, you have the sharpest rise time in that "turning on". This is what makes the filaments sing. A small toroid is added to slow this down both for filament sing and for other interference issues. Some dimmers are better at this than others.

It's not the DIMMER that's making the noise, it's the bulbs, alan.

Some bulbs (rough service perhaps) may be less noisy in this situation than others. However spending many years in the theater industry hanging and operating lights and repairing dimmer boards, I can tell you that some sing in incandescent bulbs on triac dimmers is inevitable.

Your best bet is to use smaller bulbs that you run at full dimmer settings.

What's the total number of bulbs in the fixture? Maybe you need a dimmer for more total watts. (eg. 1000w vs. 600w dimmer)

Try a different brand of bulbs.

I'll second what brickeyee said.
I had the same problem and just randomly trying different bulbs I eventually found some that didn't hum. Not the most efficient means of finding the right bulbs, but hopefully you'll be able to use the humming bulbs in some other non-dimmed fixtures.

The cheapest bulbs usually have very poor filament support and this contributes to the 'singing.'

It seems to be a lot less common with brand name bulbs.

The rough service bulbs were $6.00 per 2-pack.

I consider Reveal bulbs to be my "non-cheap" bulbs, but I just checked and they are made in China. (You may not be able to find these, btw. I bought them all up.)

Tonight, when the noisy day is over and I'm ready to hear quiet, I'll swap out some different ones.

Thanks to all.

Since the vibration is generated in the lamps, the luminaire and how it is mounted might have some role in the transmission of that vibration to air. Some luminaires might damp the sound better than others. I am willing to accept that the difference might be insignificant if the envelope of the lamp is directly transmitting most of the vibration to air.

You may wish to try two other style dimmers to see if either of them will help reduce your light bulbs from singing. One, is a Lutron dimmer for ELV (electronic low voltage) applications which can also be used on incandescent light bulbs. It is a trailing edge dimmer meaning it clips the waveform after the lamp voltage has already cycled up.

The other is one from the Lutron C-L line made specifically for dimmable CFL & LED light bulbs, and works with incandescent as well.

Neither make any claims to eliminate the singing problem in incandescent bulbs but, as part of their design, they both reduce some of the sharp transitions and harmonics that can cause problems in ELV power supplies, CFLs, and LEDs - which is also what's causing your problem.

Don't know if either will have any effect but that's what I would try. BTW, Lutron also makes an LDC (light debuzzing coil) which is wired in series to address this problem but I believe it runs around a couple hundred dollars.

Bill is right. THe trailing edge dimmer will help this. It's not so much the "trailing edge" aspect as the fact that unlike a TRIAC which tries to conduct all at once once it fires, the transistors in these dimmers can make the transition from on to off much slower.

I'm more skeptical about the HED dimmer, isn't that more of a dynamic adjustment of the TRIAC control to fix the low end drop outs?

Ron, I agree about the HED dimmer but since they designed it specifically for CFL/LED applications, you'd think the design would at least attempt to minimize the overshoots a little better than std. dimmers.

I looked on the Lutron website at their whitepapers for a technical description of their HED dimmers but couldn't find anything - still might be worth a try.

Why? There are two issues for CFL/LED/LowVoltage apps and the rapid rise time isn't one of them. As near as I can tell, other than the fact that it uses a neutral connection rather than expecting the bulb to be resistive, and the fact that it adjusts the dimmer linearity to handle the fact that CFL's take a decent jump to get them started.

Why? Since the input on CFLs is a bridge rectifier with something like a 4.7ufd filter capacitor hanging on it, it will draw something like 4-5 times it's normal operating current during the rise of each chopped cycle.

This is one of the problems of using switching power supplies on chopped AC or modified sine-wave inverters. If you look at the specs on the C-L dimmers they are derated to about 1/4 wattage for CFL as compared to incandescent.

With early CFLs, the rectified DC voltage was applied to one side of a high frequency switching transformer. The spikes and DC components caused by the nonliterary of TRIACS would cause various levels of saturation and overheating. This is the primary reason not to use "non-dimmable" CFLs on dimmers.

Modern LEDs use driver ICs that really don't care too much about the amount of AC ripple as they use constant current control in regulating the bulb. They usually can work from 90-265 VAC and actually work fine on DC as well.

The two problems with designing dimmable CFL/LED bulbs is dealing with the truncated waveform along with the overshoots/harmonics/DC component and then being able to interpret the phase angles of the dimmer into a control signal within the bulb. Fortunately driver ICs for both CFLs and LEDs have been developed and dimmer mfgs have worked towards the same goal.

" There are two issues for CFL/LED/LowVoltage apps and the rapid rise time isn't one of them. "

The chopped off low voltage (TRICS do not turn on till at least 0.7 V and many closer to 1.0 V) causes heating in inductors as does the delayed turn on ad sharp steps produced in the current waveform.

Even at 100% on time, the voltage loss is present on every AC cycle.

The delayed turn on also produces a sharp rising edge 9and by this means a lot of harmonics of the 60 Hz).

The impedance of an inductive element rises with frequency, meaning more power is dissipated in the part heating it up.

It is not hard to design AC-DC power supplies to cope with TRIAC switching, but it needs to be done if the supply needs to operate correctly.

CFLs have even more problems, since they both rectify the incoming power, and then use switching techniques to boost it up to many hundreds volts for starting, ad then keep it provide a lower (but still high) voltage ft operation at tens to hundreds of kilohertz frequency (instead of the power line 60 Hertz).

Dimming fluorescent lights has always been a problem.
Minor dimming (5 to 15) is not hard, but after that it gets worse and worse.

Leakage in the off device state also can produce odd effects on electronic ballasts.
They store up the small leakage charge until the voltage stored is high enough to start a strike to turn the device on.
Then the power is not adequate to keep the lamp on.
It appears as a quick starting flash at intervals depending the loads and leakage present.
Some LED switching supplies can respond in a similar way. while others just make the LED glow dimly all the time.

Switching to the trailing edge for cutting the current allows the parasitics in the circuit to limit the fall time of the current resulting in fewer harmonics being created (and less inductor heating).

The problem for the consumer is that the two devices (dimmers and lighting power supplies) have NOT always moved together.

A more complicated trigger circuit on a TRIAC can use the trailing edge, but they are far from trivial to design and take significantly more parts.
You have to pull current out of the gate to force the shutdown (and the negative gate current can depend on the load on the TRIAC, generally unknown and changing as the light dims).

Reporting back:

I exchanged one of the rough service bulbs for a reveal bulb - this dimmer controls two sconces - and the hum was reduced on both fixtures!

When I exchanged the second bulb, I realized that the hum, although now very low, was lower on the bulb that had been screwed in a little less tightly. So I loosened the other bulb. It's quiet enough now. I wonder if some of the vibration is now in the socket rather than the filament.
Thanks everybody.

The vibration is coming from the filament, and when the bulb is tightly screwed i it is more easily transmuted to the fixture the junction box, and even the wall.

You might also check how tight the fixture mounting screws are (including any mounting strap between the fixture canopy and the box itself), af even if the junction box is firmly attached to the structure (like a stud and not just the drywall).

If things fastening is tight small vibrations do not have enough energy to move the larger mass fastened securely together.