Power main distortion from TRIAC dimmers, what can I do to get rid of it??

[SharpEars]

Quote: TRIAC based dimmers are very efficient. TRIAC based dimmers are relatively inexpensive. TRIAC based dimmers flood your mains with TRIAC induced phase cutting noise!

The problem:

I have Lutron TRIAC based phase-cut dimmers all over my house, as do many people these days. These dimmers control dozens of halogen recessed lights and are often set to the 50% position. This floods my mains with 120 Hz transient noise from the TRIACs regulating power via phase-cutting. Here is what the noise looks like (courtesy of davehouston.org):



and an image of the TRIAC operation including the noise spikes (courtesy of hardforum.com):



This has not been a big issue for me to date, because most devices in the home are built to be immune to power line noise.

Recently, I have installed some active monitors (i.e., loudspeakers with the power regulation, amplifier circuitry and crossover circuitry built-in). These monitors do not do a good job of power noise filtering and are picking up the 120 Hz noise being generated by all of the dimmers around the house, causing their woofers to emit a light but annoying buzz at that frequency. Sadly, other than their power main noise intolerance, I love these monitors because they sound great and were reasonably priced.

If I turn all of the dimmers in the house off, the buzzing from the monitors stops. The more dimmers I have set to around 50%, the worse the buzzing from the monitors gets.

Other important info:

The active monitors are on their own 15 amp circuit that is not shared with the dimmers. However, the noise from the dimmers seems to be going beyond their respective circuits and crossing over to the circuit that the monitors are on (because every circuit in the house is on the same 200 amp main, ultimately). TRIAC phase cutting is efficient, but apparently it also floods your mains with transient noise introduced by the phase cutting that sensitive devices like my monitors pick up.

The question:

What I need is some sort of circuit I can build or device I can purchase that I can place between the power plugs from the active monitors and the noisy outlet to which they are connected. Power conditioners offer no help here, because although they filter out RFI (i.e., high frequency noise), they are useless for filtering out low frequency noise, such as the noise spikes that are being introduced into my power by the dimmers 120 times a second.

They sell coils that can be placed in series between the dimmer and the hot connector that are supposed to reduce (i.e., smooth out the spikes) of each dimmer, but I would have to install one of these for every dimmer, of which I have many, and I would need in-wall space for these things. These coils also buzz themselves, creating a whole new problem. The possibility of replacing all of the dimmers with non-TRIAC based solutions would be expensive and quite a PITA. 

I am hoping that there is a circuit I can build or device I can buy that I can place between the monitor power plugs and the wall mains that can filter out or at least decrease dimmer introduced noise (Smoothing it out some so that it becomes a non-issue is also an option). The monitors need up to 1000 W of power at 120 V, by the way, so the circuit would need to be able to accommodate this kind of load.

I seek your expert help in this matter, because although I understand the problem, I have no idea how to solve it. I have a basic knowledge of electronics fundamentals so feel free to get technical in your answer. However, when it comes to high voltage, I'd rather consult with you experts before building something ineffective or worse yet outright dangerous, since we are potentially dealing with some nine amps at 120 V.

[wraper]

You need at least to use mains filter like this for your audio equipment outlet. All audio devices must be connected to the output of it. Also it would be best to connect the same filter (or at least half of it as it is 2 filters in series basically) to the input of each dimmer.

[tggzzz]

You need to quickly check whether the impulsive noise is being transmitted to the monitor via the mains leads or some other route.

A quick and simple technique is to put themains lead through a toroidal ferrite filter. I've used them to completely suppress the click generated by a fridge switching on/off. See http://www.ukradioamateur.co.uk/full/html/c9-1-3.htm especially figure 9.3 Ideally this is done at the interference source, but by teh amplifier would have an effect too.

If that partially removes the problem then your next step would probably be a proper mains filter, specifically including common mode inductors.

If you can, do an FFT on the impulsive noise. You will find that it has frequency components far above the 120/240Hz shown in your oscillogram. RF attenuators will suppress those.

[Richard Crowley]

Professional dimmers (such as those used in the entertainment industry) feature di/dt filters (typically a largish torrodial inductor) to limit the current rise.



If you want to get into study of various design schemes:  http://sound.westhost.com/lamps/dimmers.html

[SharpEars]

You need to quickly check whether the impulsive noise is being transmitted to the monitor via the mains leads or some other route.

A quick and simple technique is to put themains lead through a toroidal ferrite filter. I've used them to completely suppress the click generated by a fridge switching on/off. See http://www.ukradioamateur.co.uk/full/html/c9-1-3.htm especially figure 9.3 Ideally this is done at the interference source, but by teh amplifier would have an effect too.

If that partially removes the problem then your next step would probably be a proper mains filter, specifically including common mode inductors.

If you can, do an FFT on the impulsive noise. You will find that it has frequency components far above the 120/240Hz shown in your oscillogram. RF attenuators will suppress those.

A toroidal ferrite filter seems like an inexpensive and easy to test solution, so I will give that a go once I find an inexpensive source for the toroidal ferrite ring(s).

Would something like this do: http://www.ebay.com/itm/4-x-TOROID-RING-FERRITE-CORE-EMI-RFI-CHOKE-FILTER-64-x-40-x-9mm-/331261704775?pt=LH_DefaultDomain_0&hash=item4d20bc7a47

I hope that the end result of this is not the same thing that a normal power noise suppression circuit does in a distribution panel (e.g.: http://www.tripplite.com/surge-protector-isobar-4-outlets-6-ft-cord-3330-joule-leds~ISOBAR4ULTRA/), because I've tried this route and it doesn't suppress the noise issue I am having.


Thanks

[SharpEars]

You need at least to use mains filter like this for your audio equipment outlet. All audio devices must be connected to the output of it. Also it would be best to connect the same filter (or at least half of it as it is 2 filters in series basically) to the input of each dimmer.

Can you please describe what this circuit does?

[tggzzz]

You need at least to use mains filter like this for your audio equipment outlet. All audio devices must be connected to the output of it. Also it would be best to connect the same filter (or at least half of it as it is 2 filters in series basically) to the input of each dimmer.

Can you please describe what this circuit does?

From left to right: 2*surge/overvoltage suppressor, capacitors filtering common and differential mode noise, common mode inductor, rinse and repeat. Standard EMI/EMC stuff described in many publications.

Don't make your own since the detailed construction is important: just buy the whole thing.

[Marco]

Shouldn't there be some non common mode inductors in that filter to attenuate the differential noise a bit more?

Or do they just intentionally make the coupled inductor leaky to get some differential mode filtering?

[SharpEars]

From left to right: 2*surge/overvoltage suppressor, capacitors filtering common and differential mode noise, common mode inductor, rinse and repeat. Standard EMI/EMC stuff described in many publications.

Don't make your own since the detailed construction is important: just buy the whole thing.

Where can I buy this thing? Could you please supply me with a link or a model number?

Perhaps something like this:

http://www.ebay.com/itm/Schaffner-FN2070-25-08-25A-EMI-Line-Filter-/161414361541?pt=LH_DefaultDomain_0&hash=item25950bb1c5

[Conrad Hoffman]

Great info above. In general, you always want to kill the noise at the source, rather than try to fix the problem downstream, which is never fully effective. Modern dimmers often omit the needed filtering. Get dimmers with filters. Yes, they usually cost a bit more.

[SharpEars]

Great info above. In general, you always want to kill the noise at the source, rather than try to fix the problem downstream, which is never fully effective. Modern dimmers often omit the needed filtering. Get dimmers with filters. Yes, they usually cost a bit more.

Show me in wall 600 W and 1000 W dimmers that include the proper filters, do they even exist in the sense that they could fit into the space of an an electric power switch?

[Marco]

AFAICS most of the distortion is differential ... against which those tiny ferrite chokes won't help much.

Big chunks of iron are needed.

[ajb]

If you're mostly just running the lights at 50%, why not use lower wattage bulbs?  That will cut down on the problem substantially, because not only will the currents be lower, but you won't be switching them at peak voltage.  You'll also get improved energy efficiency--all else being equal, a 100W incandescent bulb running at 50W produces less useful light than a 50W bulb running at 50W.  This is because the 100W bulb is running at a lower temperature and therefore its emission has shifted toward the red end of the spectrum and proportionally less of it falls in the visible range.  Running halogen bulbs at lower power levels also tends to diminish their lifespan, since the halogen process is not as effective at lower temperatures, so the filament erodes more quickly.

[bookaboo]

Rather than polish a turd why not look into a good quality dimmable LED setup?

[SharpEars]

If you're mostly just running the lights at 50%, why not use lower wattage bulbs?  That will cut down on the problem substantially, because not only will the currents be lower, but you won't be switching them at peak voltage.  You'll also get improved energy efficiency--all else being equal, a 100W incandescent bulb running at 50W produces less useful light than a 50W bulb running at 50W.  This is because the 100W bulb is running at a lower temperature and therefore its emission has shifted toward the red end of the spectrum and proportionally less of it falls in the visible range.  Running halogen bulbs at lower power levels also tends to diminish their lifespan, since the halogen process is not as effective at lower temperatures, so the filament erodes more quickly.

Because half the perceived light output of a 75W bulb would require 75W*22.5% actual(corresponding to 50% perceived)=17 W PAR38 halogen bulb which does not exist, especially given the fact that I have very specific bulbs in terms of light angle falloff, etc...

[SharpEars]

Rather than polish a turd why not look into a good quality dimmable LED setup?

Because then I will be playing the horrid LED CRI game and paying five times as much (per bulb). We have cheap power here (<$0.05 per kW/hr) and the lights, being dimmed as much as they are and being only on for a couple hours each day in the evening aren't costing me much power wise, thanks to those noisy but efficient TRIAC dimmers (I've gone full circle I know and there is much irony in my dilemma).

[SeanB]

If you are using 50W 12V downlighters you can get the lower light output simply by using 20W lamps instead. Not commonly stocked at retail outlets but always available from lighting distributors. About 20% less light but uses 40% of the power.

Older light dimmers had a large toroid inside the housing to do the interference reduction, but these were cost cutted out years ago as they were both a significant cost to add and made the unit too large to fit in some standard housings if there were more than 3 wires present or more than one conduit into the box.

[eneuro]

You need to quickly check whether the impulsive noise is being transmitted to the monitor via the mains leads or some other route.

When used simple dimmer circuit, but with RC snubber  and had radio tuned to AM LF RF below 1MHz, switching 230VAC iron to limit its heat power to <25% in this simple radio  it was significant audio change when was turning dimmer phase potentiometer 
I thought it might be RF noise from this circuit 5m away from radio, but maybe its 230VAC noise too 

If you want to get into study of various design schemes:  http://sound.westhost.com/lamps/dimmers.html

In this trailing edge dimmers

they say:

"C1 and L1 are again the RF interference suppression components. The rectifier is needed because MOSFETs cannot switch AC, only DC. "

And it is not true while... my AC mosfets switch easy switches AC mains 230VAC 

I used 1000Vds  3Ohm RDSON mosfets and it works fine and power losess with 40W bulb are about 0.2W on those 2 mosfets so they are air cold 
I'm using galvanic optoisolated  0.5A gate driver  with photocoupled MCU pins to drive this thing.
I can turn on/off 230VAC mains at any time, so these other types of dimmers are quite interesting to test and implement 

[SharpEars]

If you are using 50W 12V downlighters you can get the lower light output simply by using 20W lamps instead. Not commonly stocked at retail outlets but always available from lighting distributors. About 20% less light but uses 40% of the power.

Older light dimmers had a large toroid inside the housing to do the interference reduction, but these were cost cutted out years ago as they were both a significant cost to add and made the unit too large to fit in some standard housings if there were more than 3 wires present or more than one conduit into the box.

These are not LV lights, these are 120V 75 W lights times several dozen and three 600 W dimmers plus one 1000 W dimmer controlling the lot, and all this stuff across three 15 amp circuits. Even that is only counting the lights that are usually on, not all dimmers/lights I have that can be on, at which point things just become obscene.

Also, it is sad that the coils you referred to were removed from dimmers, because they would have at least reduced my problems by smoothing those 120 Hz spikes. Initially, they were removed and sold separately. Now, Lutron discontinued them and you have to fish to find them and I have nowhere to put them, in any case, because where would they go - the switch boxes are full already with the switches?

[NiHaoMike]

Use a diode to get half power?

[David_AVD]

Seems like you should find out how the noise is getting into the powered speakers.  Is the noise there when you disconnect the audio input to the speakers?

[Richard Head]

Whenever you have an interference problem there are a couple of choices you have. You can kill the noise at the source (better) or you can harden the equipment being interfered with.
I suspect in your case it'll be easier to harden the equipment being interfered with as there is generally very little extra room inside the dimmer space to add a suitable filter.
Incidently, the filter posted by Sharpears is totally unsuitable as it contains no differential mode filtering. Like Marco pointed out, the inductor will have to be a large iron cored type with a large inductance and current rating. The reason is that the inductor has to smooth out the massive current transitions (20A @ 1500W!), not just the steep edges. The only way is with a large inductor/cap L or Pi-network.
If you opt to harden the equipment being interfered with you can probably get away with a much smaller filter as the current draw will be much lower. I expect you could probably get away with just a single strategically place capacitor in the active speaker module.

[rs20]

And it is not true while... my AC mosfets switch easy switches AC mains 230VAC 

You've got two MOSFETs there, each of which is only seeing/switching DC. A clever arrangement reminiscent of an SSR, that exploits the (generally not very good*) body diodes in the MOSFETs to, in effect, rectify the current.

Almost all MOSFETs have body diodes, and can only switch current flowing against the body diode. Current can flow the other way, but it doesn't count because you can't switch/control that current.

* The body diodes of the SK1119, which I'm guessing is what you have there, have a forward voltage of 1.9 Volts. You could certainly make your circuit dissipate less heat if you added discrete diodes to "replace" the body diodes.

[eneuro]

You've got two MOSFETs there, each of which is only seeing/switching DC.

It does not matter, while this PCB as marked on its connectors can switch 230VAC  mains and there are NO mosfets body diodes loses when this switch is ON-RDSON bypasses them
I use SS495A Hall current sensor capable to monitor up to 25Amax current under load with 100mV/A response from sensor, which is maximum grid nominal current at home one phase 230VAC mains.

* The body diodes of the SK1119, which I'm guessing is what you have there, have a forward voltage of 1.9 Volts. You could certainly make your circuit dissipate less heat if you added discrete diodes to "replace" the body diodes.

If I wanted to switch higher load and higher average current I can put a few such AC mosfets switches in pararell as well as use mosfets with lower RDSON if needed.

[rs20]

You've got two MOSFETs there, each of which is only seeing/switching DC.

It does not matter, while this PCB as marked on its connectors can switch 230VAC  mains

Not sure if trolling, but the same can be said of the Richard's PCB. Just because a PCB with MOSFET can switch AC, does not mean that the statement "MOSFETs can switch AC" is true in any sense relevant to Richard's post. A single MOSFET, placed inline on an AC line, will act as a diode, even if you set V_GS to zero. That's a fail.

If I wanted to switch higher load and higher average current I can put a few such AC mosfets switches in pararell as well as use mosfets with lower RDSON if needed.

If RDSON*I is less than one volt, then most of your dissipation will be in the body diode, so RDSON is irrelevant here. Paralleling multiple body diodes has hardly any effect on the voltage drop of a body diode, since diodes have an exponential response. You'd be much better off using a single smaller, cheaper, lower Vf diode than multiple FETs in parallel. Also, the body diodes in a FET, unlike the FET itself, will undergo thermal runaway when paralleled, so what you propose is really a very bad idea.