What makes lights humm?

[mrpackethead]

This method of dimming doesn't reduce harmonics. It does however lengthen lamp life and will reduce filament hum.

Hi Ben, what makes teh lights hum when being switched by a traditional forward phased triac based dimmer?

[DynamoBen]

Hi Ben, what makes teh lights hum when being switched by a traditional forward phased triac based dimmer?

The hum you hear at a fixture or lamp is referred to as filament hum or ring. It comes from the filament slamming from one side to the other as it is switched on. There are two standard ways to lessen filament hum/sing. In a forward phase dimming system you use torrid coils to increase the rise time of the dimmer output. Instead of slamming the filament to one side it ramps up over time (300us). The other way is with reverse phase dimming. In this case the sine waves natural curve is used in place of a coil to provide rise time.

Harmonics are a totally different issue. Harmonics occur on the neutral coming from a forward or reverse phase dimmer system because the sine wave is being chopped up.

[mrpackethead]

From basic fundementals, i struggled to understand how reverse phase dimming coudl produce lower harmonics that forward.. Perhaps if you were controlling the fall time of the FET or IGBT on switch off and making it a bit longer, it might reduce the harmonic content.. However then you've got the transistor running in the so called 'analog' area and its going to get warm..

I've been reading up on the sine wave dimming system, and at least in theory it shoud'nt be too hard to implement.. I figure a that a pwm frequency in the order of three magnitudes higher than the mains frequency ( ie 50 or 60kHz ) should suffice, to switch a pair of igbts that are sitting across the mains. Generating a variable width pulse at those frequencys is well within the ability of a resonably simple microcontroller, either directly or indirectly.. A small passive filter network on the output should filter out the pwm, and leave me with a sine wave..

IGBT's are nice in the fact that they are really easy to drive from logic level devices, but igbts that will run at 50khz at resonable currents are a bit expensive.. however the result might be worth it..

What in my simplistic model have i forgoten.

[mrpackethead]

While I would love to answer your questions it would be thread hijacking. If the moderator would be kind enough to move this to a different location or start a new thread that might be helpful.

Hi ben! I lost track of the thread, and where it got too.. Now i found it over here.. Yes, it was thread highjacking and here is not a bad place..

I've been looking for a good text / reference material on teh whole topic of power electronics, its facinating.. We touched on it 15 no almost 20 years ago when i was at university, but its all a bit vauge in the head..

[mrpackethead]

Page 3 of this pdf, talks about the various methods of dimming, forward and reverse phase control, and sinewave dimming.. It shows graphs of the harmonics on the input.. It shows the forward control having the highest harmonics, followed by reverse and then sinewave..

Ben you suggested that this is'nt the case?

[DynamoBen]

No this is the case because you are talking about two different things. Power (feed) harmonics are different than lamp (filament) hum/sing.

Power or Feed Harmonics are caused by the sine wave being chopped up. It’s not until you go to sine wave dimming that this problem disappears. This hum can be heard at your breaker box.

The lamp/filament hum can be heard at a fixture. Usually it needs to be a larger lamp like a par or theatrical lamp. As you dim up or down you can here the fixtures sing in a quiet auditorium. The way to diminish this singing is to add a choke to a forward phase dimmer or use reverse phase dimming. With both solutions you are affecting the rise time from 0 for the lamp turn on. The side effect of adding additional rise time is longer lamp life because the filament isn't "slamming" from one side to the other.

Sine wave dimming remedies both problems and adds the ability to dim low voltage sources. Because sine wave dimming "compresses" the sine wave you can change the max peak to peak voltage to 12V even though you have an input voltage of 120V. Sine wave dimming works by sending a PWM signal to an IGBT at 32kHz. If you Google you may come up with more details, another place to look is Google’s patent search. ETC has several patents on this technology.

As far as building a sine wave dimmer it is not a trivial task. Its cost numerous large companies’ years of development and expense to create what they now have. BTW I used to work for the company that made that document. The theory of sine wave seems pretty easy to implement but I can assure you that it isn't. At one point there was a guy on the Propeller forums trying to do sine wave dimming he has yet to succeed.

[Wombat]

I Used to design mains power UPS inverters and noise from filiments etc was definiately there especially with quasi square wave inverters.

Running it with a sine wave is a solution but you can run into issues with the switching times on the devices. you need a good drive and sink circuit on the gate of the IGBT's or mosfets in order to switch them fast. Even then they still heat up.

It is possible to do but I think its overkill for just a lighting controller.
To generate a sine wave on multiple outputs is much more difficult than a plain PWM.

Back in those days I used to do it with an Eprom table with the sinewave lookups for each voltage step. but that was in the days before micros were commonplace and pics were cheap.

All done with descrete logic chips back then

Wombat

[rrowan]

I just assumed it was musical talent and the desire to be heard.

uh, sorry for the interruption.

Rick R.

[mrpackethead]

[/quote]
Back in those days I used to do it with an Eprom table with the sinewave lookups for each voltage step. but that was in the days before micros were commonplace and pics were cheap.

All done with descrete logic chips back then

Wombat[/QUOTE]

No need to create the Sine wave, if you have a reference one already ( aka the mains ).. Just chop it on and off at high frequency ( ~1000x faster than the mains )..

[mrpackethead]

Been looking at some variable voltage controllers for motors.. That might give me some inspiration, they seem to do it that way as well.

[Trepidati0n]

No need to create the Sine wave, if you have a reference one already ( aka the mains ).. Just chop it on and off at high frequency ( ~1000x faster than the mains )..

That is the brute force way of doing it. There are much more elegant ways of reducing AC voltage without a pure duty cycle method. Once you chop AC voltage you will reduce the fundamental however, you also create serious array of harmonics. By using methods that don't work on fixed duty cycles to create a fixed output...you can greatly reduce those harmonics. You can also reduce switching losses to by using method that don't use a fixed PWM frequency. So many ways....so much math to understand, but so damn fun.

[mrpackethead]

Yes, aggreed, that chopping the mains at high frequency with a variable duty cycle will create harmonics. It must do!.. However those harmonics are going to be at at high frequencys that are easy to filter out using a simple ( and low cost ) passive network, in fact the cable carrying the power to the lights may be sufficent, or just 5 turns through a ferrite core etc.

I'm wanting to dim loads of 1-2kW so, its a little different from your average blinky project where we are talking fractions of amps beting switched, so reducing harmonics and interferance is important.. but more importantly is that its really interesting.

[DynamoBen]

There is a reason 99% of the dimming on the market is either forward or reverse phase; I guess soon you will find out why ($$$). Personally I like forward phase only if you are willing to add the chokes. If you want to forgo the chokes then reverse phase is the way to do.

[mrpackethead]

There is a reason 99% of the dimming on the market is either forward or reverse phase; I guess soon you will find out why ($$$). Personally I like forward phase only if you are willing to add the chokes. If you want to forgo the chokes then reverse phase is the way to do.

$$$ may well be the case.. But once i get my head around it, and understand the problem, then it will be the standard cost reduction exercise..

[P. Short]

Just out of curiosity, what is the physical mechanism that causes the filaments to hum?

--

Phil

[DynamoBen]

Just out of curiosity, what is the physical mechanism that causes the filaments to hum?


With forward phase dimming you are suddenly applying power to the filament. A filament is suspended between two points. This sudden application of power causes the filament to "slam" one direction, at the next half cycle is "slams" the other direction. This oscillation is in the audible range hence the hum or sing. By increasing the rise time or using reverse phase dimming you are no longer applying instant power, its applied gradually over time which reduces the "slamming" and in turn eliminates the hum/sing.

[DynamoBen]

$$$ may well be the case.. But once i get my head around it, and understand the problem, then it will be the standard cost reduction exercise..

While sine wave dimming has its advantages it does have its disadvantages. The biggest is the parts count. Sine wave systems are complicated and have a higher failure rate due to their complexity. SCR based dimmers on the other hand have a lower failure rate because they are simple (they are also less expensive to repair).

Also properly designed sine wave dimmers respond slower than their SCR based counterparts. I'm not talking about something you would only see on a scope, its noticeable with the naked eye. To prevent cold inrush the system turns on the load gradually so the filament can slowly heat up. This happens fairly quickly but you will notice that they are a bit sluggish in their response.

Finally cooling, a sine wave system expends more heat than a similar SCR based system (without chokes). Keeping the components cool in a complex system like this is paramount.

In the end good luck. While I love sine wave dimmers when I'm working in a theater (not a concert) I wouldn't take it on as a hobby project. It breaks my #1 rule: "Is this too much work for too little effect?" :)

[deplanche]

Just out of curiosity, what is the physical mechanism that causes the filaments to hum?

--

Phil

They don't know the words to the song. :)

Yeah, I know, that was bad, but had to be said.

[DynamoBen]

They don't know the words to the song. :)

Yeah, I know, that was bad, but had to be said.

Nice, I had to know that was coming. ;)

[mrpackethead]

To prevent cold inrush the system turns on the load gradually so the filament can slowly heat up. This happens fairly quickly but you will notice that they are a bit sluggish in their response.


This woudl be a function of any dimming system.. If the filament needs to heat up slowy, then it needs to heat up slowly across any dimming system surely.. so are you suggesting that other dimmers are not 'properly' designed either.. I personally like to set the preheat as high as possible for expensive theatre lamps, so i don't thermally stress em out, when they are being used. ( at a $100 a throw a bit of power is worth it! )

I would have thought this was more a function of trying to protect the igbt from the current surge on turn on off a 'cold' bulb.. Some tungsten can draw 10x the current when its cold, over when its hot. Potententailly a problem but not insumountable.

>Finally cooling, a sine wave system expends more heat than a similar SCR based system >(without chokes). Keeping the components cool in a complex system like this is >paramount.

Keeping components within their thermal limits in any system is important if you want it to run reliably.. be it computers., audio amplifers, motor control or lighting dimmers..

PS.. DOn't need good luck.. jsut need determination, and a willingness to learn things, and not to accept that its 'just too hard'.. If strand or ETC can do it, then thats proof that its possible..

[Wombat]

If you are using high powered lamps which are out in the cold then you should be using a warming sequence which pre warms the lamps by putting a small dim value out just enough to heat the filliment that is not noticeable. This will reduce any inrush currents considerably and lengthen the life of the bulbs. if the lamps have been off in your sequence for long periods then a slow ramp up should be done at the first lighting of them so that they do not get a big shock which could possibly damage the filiments.

This is also what good stage lighting does.

Wombat

[DynamoBen]

PS.. DOn't need good luck.. jsut need determination, and a willingness to learn things, and not to accept that its 'just too hard'.. If strand or ETC can do it, then thats proof that its possible..

Obviously your determination out weighs my 20 years in the theatrical and concert lighting industry (both design and manufacturing). Not to mention the 8 years I worked for ETC, which by the way is the reason I know so much about DMX, ACN, and dimming.

[P. Short]

How does 'suddenly applying power' cause the filament to "slam" in one direction or another?

--

Phil

[DynamoBen]

How does 'suddenly applying power' cause the filament to "slam" in one direction or another?


A filament is coiled wire and is supported in a lamp/bulb by several thin wires. With a normal sine wave there is a weak magnetic charge that affects the filament and supports, but everything remains fairly steady. Think of the experiment in grade school where you wrapped a wire around a nail to make a magnet. However when you abruptly switch between an off state and an on state that instant application of current creates a strong magnetic charge which causes the lamp to slam one way. At the next half cycle it slams the other way, this is why most bulbs fail when you turn them on…the magnetic shock jars the filament and it breaks.

[mrpackethead]

Obviously your determination out weighs my 20 years in the theatrical and concert lighting industry (both design and manufacturing). Not to mention the 8 years I worked for ETC, which by the way is the reason I know so much about DMX, ACN, and dimming.

Perhaps in your 20 years you've just come to a point where you've accepted that it is too hard.

[P. Short]

Perhaps in your 20 years you've just come to a point where you've accepted that it is too hard.

Or is it more likely that these companies are operating under constraints that aren't necessarily present in the hobbyist arena? Or at least not acknowledged there? Such as meeting FCC/VDE emissions (radiated and conducted) requirements, UL, and so forth, and possibly working with a wider variety of loads than is being considered here.

--

Phil

[DynamoBen]

Perhaps in your 20 years you've just come to a point where you've accepted that it is too hard.

I guess you know me better than I know myself.

Don't take my word for it:

"While the concept of a PWM sine wave dimmer is relatively simple, building one that works reliably proved to be a difficult task. Some companies even shelved their sine wave R & D projects due to the cost of development. Pumping kilowatts of power through an IGBT that is switching at 25 to 35 kHz, while maintaining a smooth dimming curve, turns out to be an impressive juggling act. Switching a heavily loaded IGBT on and off 250 times per cycle makes it quite hot and changes its switching characteristics. As a result, sine wave dimmers fairly bristle with temperature, voltage and current sensors to monitor their activities and allow the CPU to keep all of its balls in the air."

Source - http://stage-directions.com/index.php?option=com_content&task=view&id=397&Itemid=39

Here are some more details about how this all works: http://www.google.com/patents?id=LOp3AAAAEBAJ&dq=sinewave+dimmer

[mrpackethead]

I guess you know me better than I know myself.
I doubt that is the case. Hense the use of the word 'perhaps'.. However you certainly are making it loud and clear that you think its hard, and not in the reach of a 'hobby' project. I am not suggesting it is easy, but at this time, do not think it is a challenge that is un-acheviable. Results will speak volumes in time, rather than words in a forum. I may well be wrong, but i'm going to give it a go.




Don't take my word for it:


I'm taking nothing for granted, and testing each bit of information i gather as 'untested' untill i have enough reasons to (a) know it to be correct, and (b) understand it to be correct.. That is very different from saying its 'wrong' untill i say its right, its just saying, i need to understand everything throughly before i jump to any assumption.

Your quote, while interesting was really just someones opinion, just like yours and mine, and so, needs to be treated as an 'opinion'. In this case, i tend to agree with him based on what i know today..

IGBT design has improved considerably in the last few years. Lighting dimmers are a very small subset of what IGBT's are used for.. There is much inspiration and usefull information to be gained from looking at AC motor control and UPS technology.. Which use similar techniques, albeit for different technologys.

Like i said early.. the proof will be in the delivery of somethign that actually works and works reliably..

[DynamoBen]

Like i said early.. the proof will be in the delivery of somethign that actually works and works reliably..

True and the intention for all of my posts were to give you an insider’s look at what’s involved. I have knowledge about the technology because I was exposed to it while it was being developed. In the areas of dimming and controls I have first hand knowledge and experience gained from the "inside." Think of me as a resource, not a stumbling block.