UVC Lamps, Part 2: Current, Voltage, and Power Measurements

by Frank · October 6, 2020

Introduction
Lamps and Lamp Assemblies
Experimental Method
Experimental Results

Introduction

Measurements were conducted on 6 watt rated lamps and 36 watt rated bi-tube lamps. These are commercial UVCG lamps. Two types of measurements were conducted. Wall plug measurements were made using a power meter that measures AC line voltage, AC line current, power factor, volt-amps, the real power, and line frequency . These measurements measure the ballast and lamp together. Tube voltage (arc discharge voltage) and tube current of the lamps were measured with an oscilloscope using current sensors and a high voltage probe. These measurements measure only the lamp except that results are affected by the type of ballast used to drive the lamp.

For the 6 watt rated lamps two types of ballasts were used: a magnetic ballast and an electronic ballast. For the 36 watt rated lamps only an electronic ballast was used. Lamp types studied were ozone producing and non-ozone producing. They differ according to the transmission property of their cylindrical tube envelope.

Lamps and Lamp Assemblies

A single 36 watt rated bi-tube lamp assembly was studied and is shown in Figure 1.

The electronic ballast is in the circular base. It is a low cost self-oscillating generic design operating at 34 KHz. The ballast does not have power factor correction. On the right is shown the assembly with the superstructure removed so that the full arc length in each tube can be seen. Arc length of each tube is 15 inches. Two different lamps were operated in this assembly: one ozone-producing and the other not.

Two different 6 watt rated assemblies were tested. The first uses a magnetic ballast and is seen in Figure 2.

The lamp with the ballast, which is at the far right, is mounted in a plastic housing. Both an ozone producing lamp and a non-ozone producing lamp were tested.

The second assembly uses an electronic ballast operating at 21 KHz. The ballast is self-oscillating and its design is low cost and is also not power factor correcting. It is shown in Figure 3.

In the figure the overall lamp length is indicated to be 8 inches. However, the actual arc length has been measured to be 6 inches (distance between electrode filaments). The ballast is enclosed in the base as shown. Figure 4 shows the ballast removed from the base and one can judge how low cost and simple its design is.

Experimental Method

AC Mains Power Meter

In Figure 5 is shown the ac mains power meter that was used to measure the assemblies.

The model number is P3 and this unit measures AC line voltage, AC line current, power factor, volt-amperes, the real power in watts, and line frequency.

Power Meter Pre-Testing

In order get some accuracy confidence concerning this meter some measurements were done on a two 4 foot T8 FO32/741 32W lamps driven with a known dimming ballast that has power factor correction along with the ability to dim from 100% to 50% lamp power. This ballast operates at 35 KHz and is at least 95% efficient . The results were:

Max: 119.8VAC, PF 0.98, 60.0 watts, 0.50 amps

Min: 119.8VAC, PF 0.96, 29.9 watts, 0.25 amps

These results were exactly what was expected for this ballast and they imply the power meter is providing accurate information.

Oscilloscope and Probes

In order to measure arc voltage and current an oscilloscope is required because one needs to capture both current and voltage waveforms as well as their phase. Oscilloscope is a Siglent Model SDS 1202X-E 2-channel 200MHz DSO. See Figure 6.

Since the arc voltage can be quite high especially just prior to arc ignition a 100X high voltage oscilloscope probe was used, Tektronix Tek P5100, shown in Figure 7.

In order to make the voltage measurements leads are attached appropriately and brought out to where the voltage probe can be easily attached.

In order to measure arc current current transformers were used. The magnetic ballast runs at line frequency so a large inductance current transformer was needed for it. See Figure 8 for details.

Take note that using a current transformer required cutting leads, threading the correct leads through the hole, and then re-soldering and insulating the leads. The BNC connector was added to connect to the oscilloscope using a 50 ohm coaxial cable.

The electronic ballasts run at frequencies of a few tens of kilohertz, so a different current transformer was required. See Figure 9 for its details. The burden resistor for this current transformer is a 50 ohm terminator attached at the oscilloscope input.

One last point regarding measuring the arc voltage using an oscilloscope is perhaps the most important point of which to be aware. Oscilloscopes always have their inputs grounded, that is, connected to earth ground. So, if one attaches ground to one side of the lamp and the voltage probe to the other side of the lamp the electronic ballast will definitely blow since it connects directly to the AC mains. Usually this damage is the diode bridge at its input. The scope’s ground must be isolated from earth ground. The recommended way to do this is to connect the scope’s power to an isolating transformer (warning: a so-called isolation transformer may not be isolating, that is, the primary and secondary are still connected together and “isolation” means voltage spikes are absorbed by the transformer core.) An alternative that is not generally recommended is to use a power plug with the ground lug cut-off so that the scope is no longer attached to earth ground.

Experimental Results

Photosensor Observation of UVC Output

The photosensor will not be described here but one was used initially to understand lamp behavior. The 36 watt bi-tube lamps were measured with a UVC photosensor and observed from start-up to 900 seconds. Figure 10 shows the result.

A couple of important takeaways from this data are:

It may take up to 10 minutes for the lamp output to stabilize. Upon warm-up the lamp output peaks and then decreases a bit before stabilizing.
It is notable in the graph that the ozone producing lamp measurements exhibit an erratic behavior, even when it reaches equilibrium. This behavior is attributable to the 254 nm UVC photons being absorbed as they are destroying the 185 nm created ozone in the air surrounding the lamp. A quote from the web site UVR: ” Ironically, UV light in the 240-315nm wavelength will break this third oxygen atom attachment above and convert it back to oxygen. The peak ozone destruction occurs at the 254nm wavelength. So, a UV-C lamp at the 253.7nm wavelength will actually destroy ozone!”

Convective air currents between the lamp tube and the sensor cause the erratic nature as 254 nm radiation passes through this veil of ozone. The same behavior is seen with the ozone-producing 6 watt lamp. That the UVC output seen by the sensor is lower than that of the non-ozone lamp is result of this absorption.

Power Meter Data for All Lamps

Table 1 contains all of the relevant lamp data including the measured power meter results for all lamps. The matrix of lamps tested is two 6 watt rated lamps and two 36 watt rated lamps where each rating has an ozone producing lamp and a non ozone producing lamp. The difference is the transmission properties of the tube envelope.

Analysis of this data makes a few things very clear.

The magnetic ballast drives the lamps significantly harder, that is, twice the current than which the electronic ballast drives them. This is also reflected in the power that the assembly dissipates. A typical efficiency for magnetic ballasts is 75% and for self-regulating electronic ballasts at least 90% can be expected. Therefore, the estimated power going into the lamps is shown in the table. The 36 watt rated lamps notably are being driven only at about 30 watts, the magnetic ballast is driving the 6 watt rated lamps beyond their rating, and the electronic ballast is driving them below their rating.
For a magnetic ballast the electrode losses are considerably higher than those of an electronic ballast and are estimated to be ~1.7 watts, irrespective of lamp length. Therefore the lamp power in the arc is about 6 watts. This is the power that generates UVC radiance.

36 watt Bi-tube Assembly – Measurement of Arc Power

The DSO was used to capture the waveforms of arc current and arc voltage for both the ozone lamp and the non-ozone lamp. The feature of the DSO to do math on the two waveforms allowed for real power to be displayed on the same sweep with the voltage and current. One of these traces is shown in Figure 11.

The purple trace is the arc current and the yellow trace is the arc voltage. These are muliplied together which generates the white trace that represents the instantaneous power dissipated in the arc. Note that current and voltage are in-phase as expected for an electronic ballast drive and the waveforms are substantially sinusoidal, but do have a noticeable charge-discharge shape. In this particular trace the power is asymmetrical. In Figure 12 one can see this is not the case and the waveforms are better sinusoids.

This behavior is explained by where on the 60 Hz input voltage waveform the traces in Figures 11 and 12 were at the instant the sweep was made. See the slow speed sweep in Figure 13.

In the figure one can see the power going up and down at a rate of 120 Hz. The diode bridge input capacitors of the ballast are not large enough to smooth out this phenomenon and is the cause.

The consequence of this is that to use this method to obtain the average arc power one must do a Monte Carlo approach. That is, one must randomly take traces, find the average power in the sweep, and then average these average power values over the set of sweeps taken. In order to do this the feature of the DSO where one can save the data to a “.CSV” file was used. The data was then imported into Excel and analyzed. Figure 14 shows one example of this analysis with plots and the numerical result for average power during the sweep. The plots include the instantaneous dynamic resistance of the discharge which can be easily calculated.

For the non-ozone lamp Table 2 shows the results for nine random traces.

These average to 29.63 watts. A similar result but for the ozone producing lamp the average is 31.28 watts.

These results compare very well with the estimated lamp power in Table 1. The result continues to support that the ozone producing lamp dissipates a bit more power in the arc.

RMS Arc Voltage and Current for 36 Watt Rated Lamp

Specifications for lamps normally contain an rms voltage and current values called the arc voltage and arc current. From the data above the average rms values were calculated:

no ozone lamp: 98 Vrms, 0.31 Arms

ozone lamp: 113 Vrms, 0.27 Arms

The voltage values are consistent with expectation. For example, a similar Phillips lamp is specified at 106 volts. However, the current values are lower given that the similar Phillips lamp is specified at 0.44 amperes. Lower is expected given that these lamps are driven at ~30 watts and below their rating.

6 Watt Magnetic Ballast Assembly – Measurement of Arc Power

For the two lamps, ozone producing and non-ozone producing, driven by the magnetic ballast, a somewhat different approach was used to obtain the arc power from scope traces. A magnetic ballast driving a lamp at line frequency of 60 Hz is expected to result in an arc voltage that is a square wave and the current sinusoidal and in phase with the square wave. This is substantially the case shown in Figure 15.

The current trace is a good sine wave and the voltage trace is substantially square in shape, at least good enough to represent it as a square wave without losing accuracy. Shown in the figure is the value of the burden resistor used with the current transformer as well as the calculation of the peak-to-peak current using its value. Subsequently arc power is calculated by algebraically multiplying a sine wave with this peak-to-peak current value times a square wave with a peak-to-peak voltage value of 80 volts and integrating over the time period of one cycle. The result is 6.1 watts rms. The results for both lamp types, ozone producing and non-ozone producing, resulted is substantially this same value.

6 Watt Electronic Ballast Assembly – Measurement of Arc Power

This assembly was not tested in detail in the manner as the 36 watt electronic ballast assembly where multiple measurements were made in Monte Carlo manner. Instead, since both current and voltage are substantially sinusoidal, the same algebraic approach just described was done and the results must therefore be treated as approximate.

Despite the power results show the ozone producing lamp has a lower arc power, the wall plug power shown in Table 1 shows results that are substantially the same.

Exitance Calculated From Measurements and Lamp Data

A quick summary of the all of the the previous measurements:

The 36 watt rated lamp is clearly being driven at about 30 watts of power into the lamp.
The 6 watt rated lamps are driven at about 5 watts lamp power by the electronic ballast and about 6 watts power into the lamps by the magnetic ballast. However it is important to note that 60 Hz drive of the discharge means that the arc extinguishes and re-strikes twice each cycle which causes meaningful electrode losses of about 1.7 watts. Electronic ballasts have much less electrode losses. Therefore, the arc power measured for the magnetic ballast includes these losses and must be subtracted from the measured arc power in order to obtain the radiant energy producing power in the arc.

One goal in this study all along has been to find the UVC radiant power output of these lamps given how they are being driven by the different ballasts. This power is called the exitance and is expressed in watts per square centimeter of lamp surface area of the glow. The measurements and lamp data together allow this to be done two different ways. The first is to simply use the wall plug-measured estimated lamp power and divide it by the surface area of the lamp that has the discharge glow. The second one is to use the arc discharge measurements of current and voltage to calculate what is called the wall loading. Wall loading is the power dissipated by the positive column in the discharge. The wall loading calculation is found in a brochure entitled “UVC_Germicidal_Lamps_Operation_Basics_LightSources_2016” and is best used for electronic ballasts where the electrode losses are much less than magnetic ballasts. Mathematically the two methods are:

D is the lamp internal diameter, ε is the conversion efficiency of the power in the arc discharge to UVC output energy, L_arc is the length of the arc discharge, P_wp is the estimated lamp power from the wall plug measurement, I_rms is the arc current in amps rms, and V_rms is the arc voltage in volts rms.

Weight measurements were made for the two different 6 watt rated lamps and together with the density of quartz a calculation demonstrated the wall thickness of the tubes to vary between 0.8 and 0.9 mm. The nominal outside diameter of a these tubes (T5) is 1.6 centimeters. Therefore for all lamps the internal diameter is taken to be D = 1.4 centimeters.

Efficiency of conversion to UVC radiance for these types of lamps can vary between 0.33 and 0.40 according to the brochure cited above and is found therein the table in Figure 16.

All of these lamps have wall loading less than 100 watts/cm2 so that one may assume the 0.40 factor applies. Therefore Table 5 compares only these results.

For the 36 watt rated non-ozone lamp and ozone lamp the wall plug result and the wall loading result are close to each other. For both of the 6 watt rated lamps the wall loading result is lower than the wall plug result and agreement, although poor, is not way off. To explain the discrepancy one must scrutinize the wall plug calculation more than to suspect the wall loading calculation. The physics that explain the wall loading calculation takes into account the voltage drop associated with the non-light-producing dark spaces in the discharge as well as adjusting the arc length to obtain more accurately the length of the positive column. Nothing equivalent is done in the wall plug calculation. For short lamps the effect of the dark spaces is more pronounced than for long lamps and failure to account for it easily leads to an exitance larger than it should be.

In Table 6 the power in the dark space is calculated and substracted from P_wp and the results added in red. Agreement for the 6 watt lamps is better but not perfect for the non-ozone lamp. For the 36 watt lamps the adjustment lowers the wall plug result roughly 10 percent below the wall loading result, but agreement overall remains reasonable given the accuracy of measurements together with the accuracy to which the expressions apply.