LEDs on dimmer - strange things

[PetrosA]

I recently installed four Sylvania 15W PAR30/LN LED lamps for a customer. There turned out to be a three way dimmer in another room that neither I or the home owner was aware of that caused issues with the LEDs turning off and on. Long story short, I consulted with Sylvania's tech support and got a list of compatible dimmers for these lamps. Because there are no compatible three way dimmers, the home owner agreed to let me eliminate the switch in the other room and install a compatible dimmer in the room with the lights. I picked a DVLV600P as it was readily available. Now the fun starts.

When the homeowner first called me back about this, I went and hooked up two multimeters (meter A inline between the hot and the switch for current, meter B measuring voltage) and did some data logging on my tablet with an additional math function of A*B to get watts with the original Ariadne 603P still installed. I saw that the lights were pulling 95-96W on full bright. If I started lowering the dimmer, the first 1/8" of the slide down would actually increase the wattage consumed by the lights, up to about 130W, then it would drop suddenly to about 75-80W and dim normally after that. I figured that might be typical for an incompatible dimmer. I also saw there were no serious voltage issues.

Today I did a control measurement with no dimmer, and found that the four 15W LEDs consume about 75W wired direct (18.75W each, which makes sense...). I then installed the Diva LV. With that dimmer they consume about 101W on full bright but it increases to nearly 150W as you start to lower the dimmer until you hit the sweet spot where it actually starts dimming the lamps! At the 150W spot, the lamps are really humming (17' ceiling and they were easy to hear). I left it at the 150W setting and after maybe 1 minute the wattage dropped to about 129V and stayed there for another 15 minutes till I changed the setting.

I explained to the customer that it's important to make sure the dimmer is either all the way up or dimmed to at least 75% or so otherwise the lamps may hum and they'll use more electric than they should. Obviously, this makes me sound like a crack head, but what's a guy to do?

Can any of you with high dollar power analyzers, some LEDs and a compatible dimmer confirm this phenomenon for me? Obviously it would be great if you had four 15W PAR30/LN Sylvanias and a DVLV600P dimmer, but any readings will be interesting to me at this point. I don't know if these readings are a shortcoming of my test setup, or whether LEDs and dimmers are really performing that strangely, and poorly. Any thoughts?

Voltage was measured between hot and ground with an Agilent U1211A, current was measured inline between hot and switch with an Agilent U1272A for a more exact current reading than my clamp meter can do. I can make .csv files available if anyone's interested.

 

[gar]

130322-2338 EST

First Volts*Amps does not equal power. The load you are dealing with is not a linear resistance. The dimmer is not an adjustable sine wave source.

Find a Kill-A-Watt EZ to use for your power measurements. Around SE Michigan they are about $30 at Home Depot. These work quite well with bad power factor which is the nature of your load. But the Kill-A-Watt EZ should be connected to measure the input to the dimmer.

Your dimmer is in some way going to modulate the input sine wave. Probably the dimmer uses a phase shift delay to turn on a Triac that turns off at the next current zero crossing, and this may be near the voltage zero crossing. A CFL or LED designed for dimming from a phase shift dimmer will probably sense the turn on phase shift to determine the amount of dimming.

The success of dimming a Fluorescent, CFL, or LED will be dependent on the type of dimmer and the design of the light. If you just had a variable DC current source, and a single LED or array with no electronics, then you could get very good dimming just like an incandescent right down to zero.

Fluorescent and LED lights have electronics between the AC input and the actual lighting element. The characteristics of the driver electronics in combination with the dimmer will determine the dimming performance. Generally I am not impressed.

See my web site http://beta-a2.com/EE-photos.html curves P14 thru P17 for CFL dimming. Using the same GE bulb and a 124 V AC source tonight with a phase shift dimmer that uses neutral I could adjust input power from about 25 W to 15 W. At about the 15 W level the light would go into a flasher mode. This dimmer is a delay to turn on type using a Triac.

I would suggest that you evaluate components and demonstrate to a customer what can be expected before the customer buys or has you buy any of the components.

.

 

[GoldDigger]

Your dimmer is in some way going to modulate the input sine wave. Probably the dimmer uses a phase shift delay to turn on a Triac that turns off at the next current zero crossing, and this may be near the voltage zero crossing. A CFL or LED designed for dimming from a phase shift dimmer will probably sense the turn on phase shift to determine the amount of dimming.

What concerns me is that PetrosA reported that the fixture got brighter at the same time the VA rose, through the first 1/4" of slider travel.

Either the bulbs were not really at full output with the slider full on, or else something about the waveform presented to the driver in the bulb is interfering with its own attempts to modulate the current through the LED. If it is using variable phase switching at 60/120 Hz just like the dimmer, there may be some complex interference which is messing up the attempt of the fixture's driver to control to a constant current.
If the driver is using rectification to DC and then PWM, then it is harder to explain the interaction.

The bottom line seems to be that those LED bulbs are apparently not in fact compatible with those dimmers, or else there is something unexpected in the wiring.

 

[gar]

130323-0900 EDT

I found my single LED bulb. Labeling very hard to read. Part of it is LED BR/DIM/830. Also seems to be labeled 8 W. This bulb is several years old and is only used for experiments.

The dimmer I am using is what I describe as a 3 lead dimmer (it actually has 4 leads plus a green ground). Nobody knows what I mean by a three lead dimmer. This is my terminology to distinguish the basic characteristics from a 2 lead dimmer.

A 2 lead dimmer is one that derives power for the dimmer electronics from current flowing thru the dimmer. Thus, it requires a moderately low resistance load. This would be a normal cheap phase shift dimmer.

My definition of a 3 wire dimmer is one that includes a neutral wire. This means the dimmer electronics has available a full 120 V power source independent of the dimmer load and dimmer setting. It also means that when set to a low output level and input AC power is removed and reapplied that the dimmer will produce an output comparable to before power was removed.

3 wire dimmers were originally made for use with dimmable fluorescent fixtures, and therefore have 4 wires. The fourth wire is a constant 120 V output to supply filament excitation to the fluorescent bulb for low brightness (low current when there is insufficient current for self heating).

This morning's LED experiment. At full brightness 8.8 W was the input. This could be reduced to 1.5 W with my 3 wire dimmer. That is the maximum phase shift I can get from this dimmer. Light output was monotonic with respect to power and dimmer setting, and varied uniformly with dimmer setting. The power at a given dimmer setting with the LED tended to fluctuate a little. This fluctuation does not occur with an incandescent bulb.

There are settings where the LED flickers.

.

 

[PetrosA]

What concerns me is that PetrosA reported that the fixture got brighter at the same time the VA rose, through the first 1/4" of slider travel.

Either the bulbs were not really at full output with the slider full on, or else something about the waveform presented to the driver in the bulb is interfering with its own attempts to modulate the current through the LED. If it is using variable phase switching at 60/120 Hz just like the dimmer, there may be some complex interference which is messing up the attempt of the fixture's driver to control to a constant current.
If the driver is using rectification to DC and then PWM, then it is harder to explain the interaction.

The bottom line seems to be that those LED bulbs are apparently not in fact compatible with those dimmers, or else there is something unexpected in the wiring.

I didn't say they got brighter, just the calculated wattage went up. If anything, they started flashing intermittently (very subtle effect).

Assuming Gar's point about the inadequacy of my gear to get an accurate reading for power is correct, I will have to get something like the Kill-A-Watt and try and recreate the test. It will definitely get costly to test any and all combinations of LEDs and dimmers that customers may want or need me to install. Regardless of that, there was a definite change in the performance of the LEDs when my gear showed it consuming 150W (noticeable hum, flickering), so there is definitely something going on that's not right in spite of the dimmer being "approved" by the lamp manufacturer.

 

[GoldDigger]

Regardless of that, there was a definite change in the performance of the LEDs when my gear showed it consuming 150W (noticeable hum, flickering), so there is definitely something going on that's not right in spite of the dimmer being "approved" by the lamp manufacturer.

There is the key. The similar perceived light level indicates that the average (not RMS) current through the LEDs themselves remains about the same. The increased hum indicates that there are sharp edges in the waveform, and possibly peak currents that are higher than they would be with no dimmer in the circuit at all. To me that indicates an instability exists between the attempts of the LED driver to maintain current at or near the device specification and the dimmer providing a chopped waveform. My guess is that using a variac to provide a sinusoidal waveform would eliminate this instability, but that is not practical. It is possible that this would also produce odd effects if the driver in the bulb tries to maintain constant light output over a "normal" range of input voltages!
That would result in a bulb that is not compatible with any dimmer.

PS: The increased VA your equipment measured is probably the result of the RMS current increasing while the light output is proportional to the average current and stays unchanged. Looking at it on a scope would tell the tale. Your ammeter probably does a very good job of displaying true RMS, even for non-sinusoidal current.

 

[ELA]

As Gar mentioned you need to appropriate equipment to measure true power due to the nonlinear nature of the load and with such as high crest factor as well.

However picking the correct dimmer might also be desirable.

Looks like you want the CL version:
http://www.lutron.com/en-US/Products/Pages/StandAloneControls/Dimmers-Switches/DivaDimmerSwitch/ModelNumbers.aspx

 

[GoldDigger]

Looks like you want the CL version:
http://www.lutron.com/en-US/Products/Pages/StandAloneControls/Dimmers-Switches/DivaDimmerSwitch/ModelNumbers.aspx

Nice catch! If Sylvania specifically called out the DVLV, it looks like their list was wrong. The LV stands for Low Voltage and is designed for controlling the input to a transformer for low voltage incandescent lighting.

 

[Electric-Light]

Dimming on two wires is the root of all evils. It is the most challenging thing that leads to flashing, flickering and buzzing.

The reason two wire is the preferred method is that it works with existing wall wiring.

I think the future is WiFi controlled setup, so the lamps connect straight to power and the control circuits have full reign over the control just like 0-10v setup without additional wires in walls.

 

[PetrosA]

If big name manufacturers are getting their lists wrong, then I give up. I normally use Philips without too many issues, but decided to try out the Sylvanias which are locally available. Now that I know there are no 3-way options, and the dimmers on their list are iffy at best, that's the last time I use them.

Regarding the CL dimmers - I've tried them now with WAC pucks and 24V LED strips, Philips LEDs and these Sylvanias and have had to replace with something else each time. It's aggravating that each time I've had issues with LEDs and dimmers, it's entailed phone tag between the manufacturer and Lutron with each passing the blame on to the other. I totally get why so many contractors aren't trying to sell LEDs to their customers.

 

[GoldDigger]

If big name manufacturers are getting their lists wrong, then I give up. I normally use Philips without too many issues, but decided to try out the Sylvanias which are locally available. Now that I know there are no 3-way options, and the dimmers on their list are iffy at best, that's the last time I use them.

The online catalog does lists the Dxxx600P, in particular the CL, dimmers as available in 3-way, just not with dimming control at both switches.

 

[PetrosA]

The online catalog does lists the Dxxx600P, in particular the CL, dimmers as available in 3-way, just not with dimming control at both switches.

I was sent a copy of this pdf by a tech support staffer:

http://assets.sylvania.com/assets/Documents/RETRO-DIM.5e22f3db-231c-480e-8fde-38d6868fe4ca.pdf

The lamp I used is covered in the first column on page 7 of the pdf. The DVCL 153P is NOT compatible with my lamps according to Sylvania. The only 3-way setup which is compatible is the Graphic Eye setup - "PHPM-WBX with DVF-103" - pricey.

I've been through this before and if a three-way version of a dimmer is not on a manufacturer's list, that means it either hasn't been tested with the lamp or it has and was found to be incompatible. Either way, you can't assume that a series of dimmers is compatible just because one of the switches in the series is.

 

[GoldDigger]

I was sent a copy of this pdf by a tech support staffer:

http://assets.sylvania.com/assets/Documents/RETRO-DIM.5e22f3db-231c-480e-8fde-38d6868fe4ca.pdf

The lamp I used is covered in the first column on page 7 of the pdf. The DVCL 153P is NOT compatible with my lamps according to Sylvania. The only 3-way setup which is compatible is the Graphic Eye setup - "PHPM-WBX with DVF-103" - pricey.

I've been through this before and if a three-way version of a dimmer is not on a manufacturer's list, that means it either hasn't been tested with the lamp or it has and was found to be incompatible. Either way, you can't assume that a series of dimmers is compatible just because one of the switches in the series is.

All that I was saying is that the Lutron data only qualifies the Diva CL suffix against CFLs and LEDs, and your Sylvania bulb is not on their list. But the only CL model is 3-way (153==> 150 watt, 3-way).
On the other hand, the LV suffix is not tested at all by Lutron for either CFLs or LEDs (except transformer fed??) and so from Lutron's point of view is not usable with the Sylvania bulb. And, as you noted the three-way is not listed by Sylvania although one might expect the DVLV-603P (3-way) to do just as well as the listed DVLV-600P (single). Which is to say, very poorly apparently.
I am inclined to agree with another poster that even with attempts by lamp and dimmer manufacturers to cross-qualify their products, the world of the two wire dimmer is a hard one to live in.

One other interesting note: The only Lutron 3-way dimmers on Sylvania's test list are styles for which ONLY 3-way are available.

 

[PetrosA]

All that I was saying is that the Lutron data only qualifies the Diva CL suffix against CFLs and LEDs, and your Sylvania bulb is not on their list. But the only CL model is 3-way (153==> 150 watt, 3-way).

I just looked over the Lutron CL list, and interestingly, the 2700K version of the lamp is on the compatibility list, while the 3000K lamp I used is not. Crazy.

 

[ELA]

Hello PetrosA,
It does all get very confusing with the introduction of CFL and LEDs and their compatibility with Dimmers. The manufacturers do not seem to be making it easy to deal with.

In my quick look I only saw that your dimmer was a low magnetic and as stated that would be more appropriate for a transformer. Looking again now I see that the CL model is also a two wire as others have pointed out. Also as they have mentioned a two wire dimmer is rarely a good idea for a CFL or LED. I do not understand why they would propose a two wire for use in any LED that would normally have a rectified/DC switcher input.

The problem is that devices with a rectified/DC switcher input often do not satisfy the minimum current requirement of the electronics in the dimmer. A Dimmer with a neutral takes care of this issue. I would recommend you look for a dimmer with a neutral that states it is compatible with LEDs ( and hopefully lists your LED).

Sometimes people will put a small incandescent in parallel with an LED on a two wire dimmer to get better results. Understood that may not be practical in most instances.

Best of luck to you.