Sunday, February 22, 2015

testing fluorescent fixtures - 40 watt

My first test in a series is this old florescent fixture with magnetic ballast and big T12 40 watt tubes which had been on my shop ceiling for years. It used 90 watts to give 60 footcandles illumination in my test jig at room temperature of 15C. The power factor was excellent. The ballast caused no FM interference. In spite of the leakage of potting compound, it had given an excellent 25 years of service.

Although this wasn't the first test that I did, I thought I would present it first since this is representative of old magnetic ballast and T12 fluorescent tube technology. I had four fixtures like this one on the main shop ceiling.

click any pic to enlarge

This is the condition of the test ballast, an old Canadian General Electric "Gold Label" 17A240E. It works but you can see that the potting compound has been leaking. At room temperature, the black potting material is hard. In fact, it was somewhat difficult to remove the fixture cover since the potting had effectively glued it shut.

I found a reference to this ballast which says that GE sold their ballast division to Alliance in the late 80s. On the back of my GE ballast, I found the code "9002" which might translate to Feb 1990. Elsewhere on the fixture, I found "4 90" which might mean April 1990. My readers will know my fondness for mature technology which stands the test of time. Although this fixture shows it's age, I am impressed that it seems to have given almost 25 years of good service!

The lamp holder posts in this fixture were dirty but in fair condition. There is no broken plastic and the contact fingers did not seem to be distorted or burnt. If I refurbish this fixture, I will likely clean these posts or replace them with new ones. I used them for this test.

It is worthwhile to examine these posts carefully if they are going to be reused since people who try to remove and replace tubes can easily damage these posts if they do not understand the correct way to do this.

The bulbs I used were from the fixture. They are older Sylvania 40 watt T12 Cool White tubes that show some blacking at the ends opposite the lables, shown in the bottom half of the picture.

I did three runs with this ballast at three different temperatures. This involved a run early in the morning when the shop was cold (room ambient temp about 1C). Then another later in the morning about an hour after the wood stove was lit (average temp abut 6C). Finally a third run mid afternoon when the stove had been running for several hours at an average room temperature of about 15C.

The graph shows that the light output (in footcandles) increases as the lamps run over the first ten minutes and then levels out at some steady value. As the temperature of the ambient air increases, the lamp output increases also.

I was pleasantly surprised that the lamps even lit at 1C since in my experience, some cold morning days in the shop, some of the lamps would not light up, or would not light up for some minutes.

The label on the GE claimed a power factor of 90%+. My measurements showed that the power factor was indeed greater than .90 within 5 minutes of startup. At room temperature of 15C, the power factor reached 0.97 after ten minutes. At the lowest temperature, power factor reached 0.93.

The label on the GE ballast claimed line current of 0.8A and this was certainly close except in the start up situation (less than 5 minutes operating) when it used slightly more, up to one amp.

The instantaneous power use was remarkably constant at the different tempertures and times, varying between 84 to 88 watts, as measured on the Kill A Watt meter.

My actual raw test data is available here as test10.pdf.

A troubling result of these tests was that the temperature of the old GE kept rising, up to 51.6C by the end of an hour from an ambient starting temperature of 13C. It was so hot in fact that more of the black potting compound had started to ooze out of the ballast by the end of the test and it was quite gooey in viscosity. Clearly there is something not quite right about this ballast and it will have to be scrapped. I wonder what it might be like after several hours operating in the summer, when the ambient temperature is more like 30C?!

Some older ballasts contained PCBs. I found this GE document which states that ballasts manufactured after January 1979 do not contain PCBs so this one does not. Further, it describes the potting compound as a mixture of asphalt and sand so the black ooze is no more hazardous than pavement.

One good thing about this old GE ballast is that it did not create any FM radio interference - none. This was even without having the fixture grounded which I found gave better interference results with some of the newer electronic ballasts.

In summary, this old technology magnetic ballast and T12 40 watt tubes, admittedly not new tubes, used about 90 watts to give 60 footcandles illumination in my test jig at room temperature of 15C. The power factor was excellent. The old ballast caused no FM interference. In spite of the leakage of potting compound, it had given an excellent 25 years of service. I did not attempt to measure anything about the color of the light produced.

The condition of this magnetic ballast should make us think carefully about products which claim to "retrofit" older magnetic and T12 lighting with newer versions by means of an adapter like this one. Better to assess the condition of what you have in case you discover ballasts like this one that really should be replaced?

I have a series of tests to post in the coming days comparing different fixtures as alternatives to this one. This old fixture will be the reference against which the others will be compared.

For background, this interesting article "The Fluorescent Lamp" with many pictures traces the history and workings of fluorescent lighting, from the Edison Tech Center.

Thank you for your interest.
George Plhak
Lion's Head, Ontario, Canada

Friday, February 13, 2015

testing fluorescent light fixtures - the test jig

Continuation of work done last year, described in this article.

Here at I have eight dual four foot ceiling mounted fluorescent fixtures. I want to add more light in the center of the main shop, probably two more fixtures, so ten altogether. The purpose of this work is to compare a variety of alternatives to come up with safe, modern high efficiency shop lighting while re-using as much as possible from the existing fixtures.


(click any picture to enlarge)

Because I wanted to do a series of comparisons, and it is not very convenient with the fixtures mounted on the ceiling, I built a simple test jig from wood scraps to temporarily hold a fixture to be tested at a standard height above a table (about 38").

The feet of the jig also can support a fixture off the table when I want to remove the cover to attach a thermocouple to the ballast or other changes. Some of these fixtures have sharp edges and protruding screws out of the backside that might scratch or damage the table surface. The two shelf brackets support the fixture when it is in the raised position yet allow easy changeover between fixtures and access to the bulbs.

A light meter will be positioned on the table surface directly below the center of the fixture.


I used a number of instruments to measure the energy into the lamps and the light output. None of these tools are particularly special or expensive. Since I would be making a series of comparisons, the absolute accuracy was not as important as consistent results and I would achieve those results by making careful measurements in the same way (as much as possible) between the various alternatives.

The light meter I use is a generic, made in China, bought on Ebay. Mine measures in footcandles and I used it on the most sensitive scale (200 footcandles full scale). I have just looked at "light meters" on Ebay and meters of this style and class sell for about CAD$25-50. Mine gives stable and consistent results and for my purposes, to know whether one source is relatively bighter than another, the absolute accuracy is not that important.

One footcandle is equal to one lumen per square foot or approximately 10.764 lux.

To help get consistent results, 1) I started the tests with a fresh 9 volt battery, 2) I placed a masking tape cross-hair on the table top so I put the meter head in the same place each time if I had to move it and 3) I tested the light meter before each fixture under an LED table lamp nearby (not shown) in the same position under the same bulb and ensured that I got the same reading each time before beginning the test run.

To measure factors about the power input to the fixture, I am using a "Kill A Watt" meter from P3. I have an older version that I have used for a number of years. The current model of what I have is a P4400 but mine seems to be the same.

The Kill A Watt plugs into a power outlet (in this case a Woods extension cord) and itself has a socket on the front into which I plug the cord to the fixture I am testing. The Kill A Watt makes its measurements on the electricity which passes through it to the load, the fixture I want to test.

I use the Kill A Watt to read the voltage, the current (in Amps), the Power (in watts) and the apparent power (in VA). The built in timer is also handy for telling how long the test lamp has been ON. I make a series of measurements at 5, 10, 15, 20, 25 and 60 minutes and after each measurement, I set the Kill A Watt to show elapsed time to help prompt me for the next measurement.

To measure temperature, I use a two input Fluke model 52II. Since there is a significant change in gas tube (fluorescent) light output with the temperature of the gas, it is useful to know the room ambient temperature. When the fixture first comes on, light output is less than when the gas warms up. The gas warms due to the electricity passing through it. At low temperatures, such as early morning in the shop during winter, the older fluorescent fixtures may not light at all.

The Fluke 52II uses two thermocouples to sense temperature. One of the thermocouples is shown in the picture, the small bead at the end of the wire. One thermocouple was loosely attached to a wood beam across the top of the test jig about the same height as the fixture, but not touching it, just hanging in the air. The other sensor I used to measure the temperature of the ballast in the fixture.

I included the auto transformer so that I could vary the voltage. I didn't do this during the regular test runs but experimented separately with varying the voltage on only some of the fixtures to see how sensitive they were to reduced voltage. It turned out that they were not very sensitive (well regulated).

The Woods extension cord which the Kill A Watt is plugged into is itself plugged into the auto transformer. The auto transformer is plugged into a regular grounded wall outlet in the shop.

In case I have confused you, here is a block diagram.

I have one more objective that I have not fully figured out yet. I want some measure of the color temperature or spectra of the lamps. By now we are probably all familiar with the desirability of "warm white" light as opposed to "cool white". Without an expensive spectrophotometer, such as one of these, I must improvise.

I think that by laying a sheet of white paper under the fixture, where the light meter normally sits and taking a photo in RAW mode with my excellent Canon G15 camera that I should be able to make some judgements or measurements of the color temperature with Photoshop or some other program that can read the RAW image. In RAW mode, the file produced by the camera is an uncompressed, unaltered record of the data produced at each pixel without any compression or compensation (like white balance).

If you have any ideas about how to do this, please leave me a comment below (I moderate the comments to prevent spam so they appear after I approve them) or write to me directly at george at ffwdm dot com.

This is the test jig in operation. I have tested a number of fixtures already with different types of gas and LED tubes and a variety of ballasts. I am recording and graphing the data and will present the results in upcoming posts.

Thanks for your interest!

George Plhak
Lion's Head, Ontario, Canada

BACK to the original article in this series