Bulb flicker test

Status
Not open for further replies.

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230429-1706 EDT

Experiment performed with utility nominal 120 V 60 Hz supply.

The load was two bulbs, a 9.35 W CREE made in China with my date code of 160902-3-4, and a 7.5 W 120 V tungsten filament unit, in parallel. A series resistor was used to change voltage simultaneously to the two parallel bulbs with a relay contact shorting the resistor. This way there was no change in voltage other other other than between the two selected states.

The two times of the cycle were adjustable, but the long time was left at 3 seconds, and the other half cycle was set to either 0.12 second or 2 seconds. My eye ball was the visual indicator.

Using calculated current for each test, 16.85 W we get 0.14 A. . Then 0.14 A times my series resistors gives a good correlation on measured voltage drop. For 22 ohms this is 3.08 V. I measured 3.3 V. For 47 ohms this is 6.58 V, and I measured 6.6 V.

Visually I could barely detect the light change from a 3 V change, and it was harder to see it on the CREE.

Visually the almost 7 V change was only a little more detectable on the CREE, but slightly more noticeable on the incandescent.

Others should do their own experiments and report back.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230430-1145 EDT

Another C10:

This is really in response to another thread relative to all lamps changing intensity in unison.

When you have a complex system with many components it is important to study some small manageable part of the system to try to get an idea of how that part works.

So in this experiment I reduced the number of components to a very few. I could have run the test on a single CREE bulb, but I chose to add the incandescent directly in parallel with the CREE. This parallel combination should have no effect on the CREE operation because the parallel combination is essentially being driven by a voltage source.

I could have continued the experiment to an even larger voltage change, but this was adequate for the moment. I have previously done the experiment with much greater voltage changes.

An interesting aspect of my experiment was that it visually appeared that the visual response time of the CREE was slightly longer than the incandescent. To do a quantitative experiment on this aspect requires me setup a photo detector. This I did not do.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230430-1534

Go back and read the posts at:


These go into more detail.

.

.
 

Besoeker3

Senior Member
Location
UK
Occupation
Retired Electrical Engineer
230430-1534

Go back and read the posts at:


These go into more detail.

.

.
Don't you use LEDs these days"
 

tortuga

Code Historian
Location
Oregon
Occupation
Electrical Design
Visually I could barely detect the light change from a 3 V change, and it was harder to see it on the CREE.

Visually the almost 7 V change was only a little more detectable on the CREE, but slightly more noticeable on the incandescent.
Nice so this experiment is to see if 3Volts or 6Volts drop (or power fluctuation) produces noticeable flicker?
What size and type resistor did you use? A seems like you'd need a Ceramic?
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230505-1615 EDT

tortuga:

Power dissipation in the series dropping resistor is P = E * I . So for the larger voltage drop a 1 W carbon resistor is adequate. I used a 47 ohm resistor for the larger drop.

.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
2305061715 EDT

Any time I try to look at the original post on this light flicker problem my computer is essentially locked up. Very long time delay to look at the original thread.

The plots are impossile to read. There are some short spikes, many, and then there are longer periods with a descrete change. What is being recorded I don't know. Likely current or power.

That this is a whole house thing implies an external cause, or at least something in or before the main panel.

Clearly one needs to look at both voltage and current on the main wires entering the main panel.

I do not trust power company instruments used to look at and record at the main panel input. I have actual experience with their inability to detect a problem.

If the plots are of current, then the large increasing spikes could be charging current to a capacitor input filter.

The reason to look at a single bulb is to see if a small load is of any impotance. I do not expect a large short current pulse to come frm a CREE.

CREE seems to be quite different than other bulbs. More later.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230507-1328 EDT

The difference I see between CREE and other LEDs I have is very large narrow current spikes. These spikes repeat at an 0.7 to 1.0 MHz rate and are very narrow, oscillation of about 270 MHz within the spike.. The oscillation frequency maybe a result of external leads.

There is typically a large short duration oscillation at about a rep rate of 25 microsec, inside this repetition there is a slightly shorter pulse spaced 8 microsec from the 25 microsec pulse.

The amplitude of this superimposed group of pulsed oscillations is many times greater than a base 120 Hz component they are added to. To really understand what I have tried to say you need to see waveforms on an oscilloscope.

I need to recheck my time measurements. Something does not look correct.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
230507-1450 EST

Rechecking my data.

The 8 and 25 microsec are correct. 3.7 nanosec is correct but may not be the bulb,as much as external circuity, wires.

High frequency oscillation is about 5 times the magnitude of the 60 Hz base signal. The high frequency signal occurs during about 1/2 of the 120 Hz signal.

So what I see on the AM radio is harmonics of the base frequency.

.
 
Status
Not open for further replies.
Top