Dimmer voltage experiment

derekleffew

Resident Curmudgeon
Senior Team
Premium Member
Results from an experiment conducted in my garage. Equipment included Kill-A-Watt P3, Fluke 336, MicroTech DMX, triac-based 4x600W "shoebox" dimmer, SourceFour ERS, Osram HPL 575/115 (UCF) lamp.

Voltage measured at the dimmer input with no load: 121.9
Voltage measured at the dimmer output with no load: 118.3

Voltage measured at the dimmer input with lamp at 100%: 117.8 117.8
Output of dimmer measured as follows:
0% - not measurable
10% - 36.6
20% - 50.7
30% - 69.1
40% - 84.1
50% - 96.9
60% - 106.5
70% - 113.1
80% - 115.6
90% - 116.6
100% - 116.6
116.6

Questions:
1. Why the voltage drop on the input when a load is present?
2. Why such a flat "top end" to the dimming curve?
3. The difference, with a load, between input and output at full, (117.8-116.6) is 1.2. Can this dimmer be considered, {[1-(1.2/117.8)]*100} 99.0% efficient? [Seems unfair to me to penalize the dimmer (121.9-116.6)= ... 95.6% for losses due to the load.]

As always, [user]STEVETERRY[/user], et al, must wait at least one week before replying.
 

Attachments

  • dimmer.JPG
    dimmer.JPG
    45 KB · Views: 481
could I suggest that this experiment would be more informative with corresponding current and light output readings, I would have thought that very little difference occurs in the top 1/3 of the fade, but light reading would elucidate this and current readings would emphasize the resistance non linearity.
Just a thought.
 
1. The voltage drop on the input is either due to IR drop on the wires leading up to the dimmer or to wierd current waveforms confusing the meter.

2. It may be to compensate for the non-linear behaviour of the light and eye. It may also be the meter's reaction to the input waveform.

3. With the dimmer full on, the only losses are due to the triac's resistance and the (hopefully low) losses in the interconnects and choke.
 
Results from an experiment conducted in my garage. Equipment included Kill-A-Watt P3, Fluke 336, MicroTech DMX, triac-based 4x600W "shoebox" dimmer, SourceFour ERS, Osram HPL 575/115 (UCF) lamp.

Voltage measured at the dimmer input with no load: 121.9
Voltage measured at the dimmer output with no load: 118.3

Voltage measured at the dimmer input with lamp at 100%: 117.8 117.8
Output of dimmer measured as follows:
0% - not measurable
10% - 36.6
20% - 50.7
30% - 69.1
40% - 84.1
50% - 96.9
60% - 106.5
70% - 113.1
80% - 115.6
90% - 116.6
100% - 116.6
116.6

Questions:
1. Why the voltage drop on the input when a load is present?
2. Why such a flat "top end" to the dimming curve?
3. The difference, with a load, between input and output at full, (117.8-116.6) is 1.2. Can this dimmer be considered, {[1-(1.2/117.8)]*100} 99.0% efficient? [Seems unfair to me to penalize the dimmer (121.9-116.6)= ... 95.6% for losses due to the load.]

As always, STEVETERRY, et al, must wait at least one week before replying.

Are we yet at a date where I can reply?

BTW, you might want to eliminate colored fonts--they are almost impossible to read!
ST
 
Well, Mr. Terry, let's see...the original post is dated 11/16, and 16+7=23. Today is 11/25. So yes, the REPLY LAMP:twisted: is now lit, for all. And I believe the politically correct term today is "fonts of color"!;) The purpose of the multi-hued fonts was to add clarity. Are you not viewing CB the way it is intended--in Black and Gold "skin"?
 
Questions:
1. Why the voltage drop on the input when a load is present?

That is due to I-squared-R voltage drop on the branch circuit feeding the dimmer, and the feeder entering your house.

2. Why such a flat "top end" to the dimming curve?

You did not specify the type of dimmer used in the test other than "shoe box". I suggest: A. The dimmer is poorly designed and has garbage curve performance that is not even close to the ideal IES square-law curve, or B. If the dimmer has an analog firing circuit, it is out of calibration. There is no inherent characteristic of the feeder or the load that would cause this poor curve performance. BTW, all previous comments about meter inaccuracy being the cause of these measurements can be discounted--since you bought a good true-RMS-responding meter after someone gave you some good advice on that subject. :)


3. The difference, with a load, between input and output at full, (117.8-116.6) is 1.2. Can this dimmer be considered, {[1-(1.2/117.8)]*100} 99.0% efficient? [Seems unfair to me to penalize the dimmer (121.9-116.6)= ... 95.6% for losses due to the load.]

You are correct. There is only 1.2V of drop across the dimmer, or about 1% loss. The drop across the branch circuit wiring and feeder does not enter into this calculation. This low drop suggests to me that the dimmer has very small choke indeed (if it actually has one at all), and a very short risetime, else the drop would likely be higher. This translates into lamp noise, but with the tight filament of an HPL, it is probably still fairly quiet.


Finally, your measurement of the dimmer at full with no load is an invalid one. Thyristor dimmers need to have some load on them for an accurate measurement. A thyristor dimmer could read 120V on its output when it is actually turned off--due to small leakage current through the thyristor and the very high impedance of the meter input. A thought--is there any chance that your poor curve measurements were inadvertently taken with no load? This might explain the results.

ST
 
I note that in your photo, you have the Kill-A-Watt plugged into a dimmer output. Was this just a storage place for the photo, or were you actually using it for any measurements?

The Kill-A-Watt does not have a spec for crest factor, so we cannot be sure of its measurement accuracy on grossly non-sinusoidal voltage input such as a dimmer produces.


ST
 
...Was this just a storage place for the photo, or were you actually using it for any measurements?...
Both. There was negligible difference in voltage and current readings between the Kill A Watt™ and the Fluke 336. Surprised me too!

I intentionally did not identify the make and model of the dimmer pack. It is of far-Eastern origin, often available for $60-$120, and NOT ETC/Strand/Leviton, or EDI/Lehigh/CAE, brand.
 

Users who are viewing this thread

Back