Light Flicker

Status
Not open for further replies.
G

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
181209-2611 EST

ActionDave:

I think the light flicker problem should be studied. He is listed as an engineer and a contractor. What kind of engineer I don't know. If he is a mechanical engineer or other non-electrical, then we need some one local to him with suitable tools that is interested in looking at the problem to help. This is not an insignificant problem and shows the stupidity of the power company to properly look at the problem.

We had a very severe power problem a number of years ago, and the power company recorder did not identify a problem. I can not imagine why not. Voltage was fluctuating by at least 10 V at times. Later one afternoon it was very bad and we called DTE and they sent someone out. As he was waiting for another person, a lineman, he was outside and saw arcing at a couple transformer terminals. This is a two transformer wild leg open delta supply. Any reasonable recorder should have recorded this type of problem.

I can suggest ways to attack a study. The person that started the thread obviously has some electrical background, possibly more than many electricians.

.
 
I am also OK with reopening the thread, if necessary with a caution that the thread will only stay open if it sticks to theoretical exploration of the problem, including tests which can be done by an unlicensed person, but should not go into recommending corrective action, especially if those actions would properly be done by a licensed electrician.
 
I myself would want to know more about the Christmas light flicker problem with LEDs because it seems that LEDs are having loads of interference problems when used certain ways so perhaps we can learn better how to control them... like the one thread earlier that talked about self starting LEDs to keep them from being so hard on the eyes... now to find an easy way to wire the circuits to cover such a need if all the lighting is LEDs...
 
The only thing that I can even think to do on a problem like this is to consult with the manufacturer of the dimmers to see if they have had similar complaints. They may even know of a solution.

As an electrician I would swap the dimmers out for regular snap switches to make sure the problem is with the dimmers.

I'm thinking that the smaller voltage fluctuation which is perfectly acceptable to the power company may be causing problems with these dimmers.
 
181210-0859 EST

First what is needed is a definition of flicker.

growler:

You don't need to do any rewiring. Simple take one of the Cree bulbs put it in a socket that does not have a dimmer powering it and observe any flicker. This test needs to be done on both phases.

The degree of flicker will determine the next step.

At my home I see a slight flicker in an incandescent, not on a dimmer, when my neighbor's air conditioner starts. Noticeable, but not annoying. A relatively rapid cycling would probably be a problem.

.
 
Well isn't this a coincidence, I just happen to be going to Spearfish this week to a customer with the exact same problem. Any suggestions other than cash now plus mileage? Would my old Fluke 43B be up to the task?

eta: As I understand it LEDs take a pretty large amount of current at power up.
 
181210-1038 EST

I don't have time at this moment to run some tests.

However, I don't have a lot of LED bulbs, but I have had a few failures of Cree bulbs. On the other hand the Cree bulbs seem quite good in several respects. They seem to dim very well with either a phase shift dimmer or a variable sine wave (Variac). Also the Cree do not cause me major RFI problems. They appear to run around 50 kHz, and don't have spurious oscillations at other frequencies like some Feit bulbs.

.
 
Lots of ink in the trade magazines lately about how poor power quality affects LED lighting. I suggest that this is one of those instances. Really, nothing to do except try to find a dimmer that isn't effected- or get rid of the dimmer and replace it with a switch.

-Hal
 
181210-1718 EST

A quick experiment on a Cree 9.34 W 120 V bulb.

I made a series string of 4 15 ohm 1/2 W resistors and the Cree bulb.

The approximate drop per resistor is 1.3 V. I did not detect light change from an LED voltage change of 1.3 V, saw a slight change at 2.6 V, and quite a noticeable change at 5.2 V

.
 
181210-2110 EST

ptonsparky:

My Fluke 27 is not real fast on min and max hold, but I would use such an instrument as part of testing. I believe the 27 is at least in the 100 mS range. A scope would be good. I would see what an LED bulb did directly from the line with no dimmer.

If the dimmer is the biggest part of the problem, then try a true 3 terminal dimmer. Also try a Variac instead of a phase shift dimmer if the bulb will dim from a Variac.

.
 
Source resistance = 60 ohms ?

Source resistance = 60 ohms ?

gar said:
181210-1718 EST

A quick experiment on a Cree 9.34 W 120 V bulb.

I made a series string of 4 15 ohm 1/2 W resistors and the Cree bulb.

The approximate drop per resistor is 1.3 V. I did not detect light change from an LED voltage change of 1.3 V, saw a slight change at 2.6 V, and quite a noticeable change at 5.2 V

.
Click to expand...
Gar,
What does your experiment simulate? When are you expecting to have a 60 ohm source impedance from a 120V residential feed?

Your experiment limits max current available in an unrealistic way.
Why did you not use a variac?
 
181211-2108 EST

ELA:

To use a Variac I would need to use a light metering instrument. And this would be a long time steady state change. In my experiment I used the eyeball. The Variac would not accomplish much toward a short time change between two different low impedance sources.

Is my series resistance method a problem? Probably not for the test I wanted to perform. The 60 ohms only represented a 5 V change. What I wanted was a ballpark estimate of the voltage change that I would detect as a flicker. If I ran a well controlled experiment it is likely I can detect light changes smaller than produced by a 1 V change.

.
 
Status
Not open for further replies.
Top