Routine Spectrophotometer Tests and Maintenance
4.1 Purpose of the Spectrophotometer Tests
It is possible that the spectral characteristics of a spectrophotometer may change with time in a number of ways so that the original calibration of the instrument will not apply. In order to detect such changes and make allowances for them, either by applying corrections to observational data or by making adjustments to the instrument, it is necessary periodically to perform four types of instrument tests--a mercury lamp test, a standard lamp test, a wedge calibration test, and a sensitivity test.
4.2 Frequency of the Tests
Mercury lamp and standard lamp tests should be performed at least once per month, preferably near the 28th day of the month, while wedge calibration tests should be performed at 3-month intervals. Ordinarily, it is sufficient to perform only one set of tests; however, if the test data should indicate that some change has occurred in the spectral characteristics of the instrument, it may be necessary to repeat one or more of the tests after certain corrective measures, to be described later, have been taken.
4.3 Recording of Test Data
Spectrophotometer test data should be recorded on a form similar to that shown in Figure 2. The forms should be numbered consecutively in order to denote the order in which the tests are made. Any adjustments made to the spectrophotometer, or maintenance work performed, should be described on the back of the forms.
4.4 Test Procedures
Spectrophotometer tests should be performed carefully according to instructions outlined in Sections 4.4.1 to 4.4.4. The instrument should be at temperature equilibrium when the tests are conducted.
4.4.1 Mercury Lamp Tests
The wavelengths falling on slits S2, S3 and S4 may change because of slow deformation of the spectrophotometer main frame casting or a shift of some of the optical components. To determine whether the Q-setting table in use is applicable (i.e., that ozone observations are being made on correct wavelengths), mercury lamp tests are performed. A mercury lamp supplied with the instrument can be fixed above the inlet window to illuminate slit S1. For routine checks it is sufficient to measure the value of Q1 when the effective mercury wavelength 3129 A.U. falls centrally on slit S2. Tests should be conducted at different temperatures in order to check on the temperature dependence of the Q-lever settings. The test procedure is as follows:(1) Light the mercury lamp, using a regulated voltage source (if possible), and leave it to warm up for 5 minutes.
(2) Place the ground quartz plate above slit S1.
(3) Turn on the spectrophotometer power supplies, and set the photomultiplier sensitivity step switch for minimum sensitivity.
(4) Place the mercury lamp over the inlet window to illuminate the ground quartz plate.
(5) Set the wavelength selector rod for SHORT wavelengths and turn the spectrophotometer dial to 300 degrees. (This means that effectively only slit S2 is open and that the microammeter deflection will be a measure of the light passing through it.)
(6) Set Q1 and Q2 levers for the mercury 3129 A.U. wavelength using the instrument's Table Settings of Q
(7) Adjust instrument sensitivity so that the microammeter reads approximately 14 to 18 microamperes. (If a reversing switch is not built into the microammeter, it may be necessary to reverse connections to the microammeter.) If the microammeter deflection is too great with the sensitivity switch set for minimum instrument sensitivity, the intensity of the light entering the spectrophotometer may be decreased by placing one or two small pieces of lens tissue paper over the ground quartz plate. Alternatively, a permanent light attenuator may be installed within the Hg lamp housing. The microammeter deflection should not be reduced by means of the microammeter shunt, since, if this is done, the amplifier may be overloaded and no longer linear.
(8) Read and record the temperature of the instrument to the nearest 0.1 degree.
(9) Adjust Q1 to give maximum microammeter reading.
(10) Move the Q1 lever upward to reduce the microammeter reading to one-half the maximum value. Read and record this value of Q1.
(11) Move the Q1 lever downward to give, again, the one-half maximum microammeter deflection, but on the other side of the maximum. Read the record Q1.
(12) The mean of the readings Q1 from (10) and (11) denotes the setting of the Q1 lever at which the mercury line 3129 A.U. will fall centrally on S2. Record this mean value.
(13) Repeat (9) to (12) four more times. Mean Q1 values should agree to within about 0.2 degree.
(14) Read and record the temperature of the instrument.
(15) Deduce and record the overall mean values of Q1.
(16) Using the Table of Settings of Q, read and record the setting of Q1 for the Hg-3129 A.U. line at the mean temperature of the instrument.
(17) Obtain the difference between the Q1 values from (15) and (16). The difference should be less than 0.3 degree. If it is found to be greater than 0.3 degree, repeat the mercury lamp test a day later, making certain that the spectrophotometer is at temperature equilibrium. If the large difference in Q values persists, it should be interpreted to mean that something has happened to the spectrophotometer so that the Table of Settings of Q used with the instrument is no longer valid. Try correcting the problem by removing the spectrophotometer cover and checking to see that the bridge containing the optical wedge is securely bolted down. (Prior to removing the cover, perform a standard lamp test (see Section 4.4.2)). If the problem persists, steps must be taken to correct the Table of Settings of Q in order to ensure that future ozone observations are made on correct wavelengths. Instructions for correcting an erroneous Q setting table are given in Appendix B.
4.4.2 Standard Lamp Tests
Standard lamp tests are performed to confirm that the level of calibration of the spectrophotometer has remained constant. Also, when a permanent change occurs in the spectral characteristics of the instrument, the lamp test data may be used to determine corrections to be applied to ozone data.
Tungsten-halogen lamps (e.g., 8.33 ampere, 24 volt, 200 watt lamps manufactured by G.E.C. Ltd., P.O. Box 17, East Lane, Wembley, Middlesex, England HA 9-7PG) are most suitable for use in performing standard lamp tests. Specifications for fabricating holders for the lamps may be obtained from the NOAA Air Resources Laboratory, Boulder, Colorado. (The specifications are based on a lamp holder design by R. A. Olafson of the Canadian Atmospheric Environment Service, Downsview, Ontario). It is recommended that these lamps be operated at 24.0 volts d.c. with the lamp voltage monitored accurately and held stable to within ±0.1 volt in order that spectrophotometer dial reading errors not exceed 0.1 degree. A limited number of standard lamp units (each comprising two lamps, a lamp holder, and a d.c. power supply with a digital voltmeter) are maintained in stock at the NOAA Air Resources Laboratory, Boulder, from which they are available through the WMO to ozone observatories that are unable to procure them elsewhere.
The tungsten-halogen lamps may also be operated at a current of 8.33 amperes a.c., with the current held constant to within ±0.03 ampere. This mode of operation requires the use of a high quality, expensive dynamometer for measuring the current. Other lamps are currently in use for conducting standard lamp tests, e.g., ultraviolet light transparent glass envelope tungsten lamps operated at 100 volts a.c., similar lamps operated at 200 volts a.c., etc.
Each spectrophotometer should be supplied with at least three standard lamps referred to in the following instructions as lamps A, B, C, etc. Prior to initial use, each lamp should be operated at rated voltage for 10 hours in order that its spectral characteristics become stabilized. Lamp tests should be performed from month to month using the same standard lamp, e.g., lamp A. Lamp B should be kept in reserve and used only occasionally as a check on test results obtained with lamp A. When lamp A burns out, lamp B should be used for all standard lamp tests, and lamp C employed for check tests, etc. Obtain a replacement for lamp A.
A standard lamp test should be conducted immediately after the mercury lamp test has been performed. It is important to position the lamp above the instrument in exactly the same way each time, particularly with reference to lamp filament orientation. The test procedure is as follows:
(1) Place the ground quartz plate above slit S1.
(2) Fix the standard lamp in correct position within the holder, and place the lamp unit over the spectrophotometer inlet window. A cover should be used to shield the lamp and ground quartz plate from other bright light, e.g., daylight. As a further precaution, standard lamp tests should always he conducted indoors rather than outside in broad daylight. Otherwise, the daylight may affect the lamp readings appreciably even with the lamp partially covered with the lamp cover.
(3) Adjust the lamp voltage or current to the correct value, using a regulated power source, if possible. Leave the lamp alight for at least 5 minutes.
(4) Push the S4 shutter rod all the way into the spectrophotometer base. The wavelength selector rod should be set to SHORT position when measurements are made on A, B, C, and D wavelengths, and to LONG position when C' wavelength measurements are made.
(5) Check the zero of the microammeter.
(6) Set Q1 and Q2 levers for A wavelengths according to the data given in the Table of Settings of Q.
(7) Increase the microammeter sensitivity by turning the shunt potentiometer fully clockwise. Increase instrument sensitivity as needed.
(8) Check and adjust the lamp voltage or current.
(9) Turn on the shutter motor and obtain a 30-second recording on the instrument's waxed chart or smoked disc. The spectrophotometer dial should be slowly oscillated during the measurement so that the microammeter needle deflects just to one side of zero, just to the other side of zero, etc.
(10) Set for B, C, C', and D wavelengths, in succession, checking the lamp voltage at each setting, and taking 30 second recordings.
(11) Repeat (6) to (10) two more times.
(12) Measure the traces and record the data.
(13) Compute the mean values of the dial readings for A, B, C, C' and D wavelengths.
(14) Compare the experimental mean data obtained with reference data for the same lamp given in the instrument's table entitled Reference Standard Lamp Data. Experimental values of RA, RB, RC, and RD should not differ from reference values by more than ±1.0 degree, and RC' values should agree to within ±2.5 degrees.
(15) If the experimental data do not agree with the reference data within the limits specified in (14), the following action should be taken:
(a) Wash the ground quartz plate with soap and water, dry it, and replace it in its holder.
(b) Check condition of the silica gel. Replace with dry silica gel, if needed.
(c) Perform standard lamp tests a day later using lamps A, B and C. If the discrepant standard lamp readings persist, proceed to instructions (d) to (h) below.
(d) Remove the spectrophotometer cover with instrument power off and in subdued light. To prevent accidental damage to the spectrophotometer photomultiplier tube, place a piece of black electrical tape over slit S5.
(e) Remove the bridge unit containing the optical wedge and inspect it for cleanliness. An air squirt should be employed to blow away dust and lint from the wedge. If additional cleaning is required, it is recommended that "dry" cleaning be attempted by breathing on the quartz surfaces and polishing with pure lens tissue paper. (Lens tissue paper commonly employed in cleaning eye glasses must not be used since it is impregnated with oil that may not be transparent to ultraviolet radiation.) Always use a fresh piece of lens tissue paper when touching a normally unexposed optical surface. Take care not to exert too much pressure on the optical wedge sections since they may break away from their holders if the glue holding them in place is brittle. If a persistent film remains, clean the optical surfaces with a piece of lens tissue paper lightly moistened with alcohol, and then dry clean. Care should be exercised not to use too much alcohol since it may affect the binding material used in construction of the optical wedge sections. Also, the black paint used on the instrument casting is slightly soluble in alcohol and may smear around the optical components.
(f) Examine other optical surfaces, e.g., the Q plates, lenses, mirrors, inlet window, etc. Clean the surfaces, if necessary, as in (e). Mirrors should NOT be cleaned with alcohol except as a last resort (see Section 5.9). The mirrors are front aluminized and are easily scratched. An air squirt may be used to rid the mirrors of dust and lint.
(g) Replace the spectrophotometer cover after first removing the tape from slit S5, and perform another set of standard lamp tests one or two days later using all standard lamps. (Use of the spare lamps will indicate whether a real change has occurred in the spectral characteristics of the instrument, or whether the spectral characteristics of the first lamp have altered.)
(h) If the standard lamp tests conclusively prove that the spectral characteristics of the spectrophotometer have changed, corrections to the instrument's NA, NB, NC, and ND tables must be determined; also, appropriate corrections must be applied to "back" ozone data. The procedure to be used in determining corrections to the N tables is outlined in Appendix E.
(16) It is expected that the need for more than one standard lamp test per month will arise rarely. Also, it should not be necessary to clean the optics according to instructions given above more often than about once per year. Removal of the instrument cover for cleaning of optical components should be minimized.
4.4.3 Wedge Calibration Tests
The object of the wedge calibration test is to determine whether the relative transmission along the optical wedge is changing. This is accomplished by alternately inserting a rhodiumized plate in and out of the beam of light passing through slit S3. If measurements are made using a standard lamp, there will be a difference in dial readings according to whether the plate is, or is not, in position. This difference, when converted to values of N, should always remain constant since it represents the transmission of the rhodiumized plate.
It is sufficient to make tests at one wavelength pair, and since the light is strongest at D wavelengths, these are used.
In order to make tests on those parts of the wedge which are most used, glass plates, X and Y, are placed between the lamp and slit S1, to give a convenient dial reading. These glass plates play no other part in the test. However, since their transmission changes with temperature as they are warmed up by the lamp, they should be in place some time before the test is conducted; otherwise, the dial readings will drift.
A special standard lamp should be reserved for use in conducting all wedge calibration tests in order to conserve the life of lamps normally employed for conducting standard lamp tests. The test procedure is as follows:
(1) Place the standard lamp mounted in its holder over the spectrophotometer inlet window. Do NOT use the ground quartz plate.
(2) Set Q1 and Q2 levers for D wavelengths.
(3) Place special glass plates X and Y under the lamp to cover the inlet window (place plate X on top).
(4) Switch on the standard lamp, adjust for correct lamp voltage or current, and leave the lamp alight for ten minutes. This allows the temperature of the lamp and glass plates to become nearly steady. (Canadian Dobson instruments are equipped with moveable brackets by means of which the glass plates are positioned in front of the spectrophotometer optical flats Q2. In this way, the plates are always maintained at instrument temperature.)
(5) Set the wavelength selector rod to position SHORT for the duration of the test.
(6) Adjust lamp voltage or current to the correct value.
(7) With the S4 shutter rod pushed all the way into the instrument (i.e., with the rhodiumized plate "out"), obtain a 30-second instrument recording.
(8) Move the S4 shutter rod to position RHODIUM PLATE (i.e., put rhodiumized plate "in") and make a record for 30 seconds. (If the position RHODIUM PLATE is not marked on the S4 shutter rod, set the rod to position OPAQUE.)
(9) Repeat (6) to (8) to get three records with the rhodiumized plate "out" and two with the rhodiumized plate "in". Check lamp voltage repeatedly.
(10) Record values of the instrument dial readings and compute mean values.
(11) Remove the lower glass plate Y and repeat (6) to (10). This will check the wedge over a different range. (The object of removing the bottom plate is that the top plate will have reached a steady temperature which will not change much when the lower plate is removed, whereas if the top plate is removed, the lower one would warm up.)
(12) Using the instrument's ND table, convert the mean dial readings corresponding to the "in" and "out" positions of the rhodiumized plate to ND values and, hence, to deltaN values for the thick (plates X and Y used) and thin (plate X used, only) portions of the wedge. The two deltaN values should always be equal.
(13) Compare the experimental deltaND values obtained with reference deltaND values for the instrument. The difference between experimental and reference data should not be greater than 1.0.
(14) If discrepant results are obtained in steps (12) and (13) take the following action:
(a) Check the optical wedge according to instructions given in Section 4.4.2 (15), and clean it, if necessary.
(b) Clean the rhodiumized plate, but do not remove it from its holder.
(c) Repeat the wedge calibration a day or two later.
(15) If the discrepancies persist, the implication is that the spectral characteristics of the optical wedge are changing (unless, of course, the opacity of the rhodiumized plate has altered). Such a fault in the instrument is serious, indicating that recalibration of the optical wedge may be needed. The method of recalibrating the optical wedge is described in Appendix C.
(16) Additional information on the altered spectral characteristics of the optical wedge or the rhodiumized plate may be obtained by repeating the wedge calibration test using A wavelengths. Changes in the spectral characteristics of optical components are often more pronounced for A than for D wavelengths.
4.4.4 Sensitivity Tests
The sensitivity test is performed in order to indicate whether the spectrophotometer sensitivity is adequate for ozone observations to be made with ease and precision. This is accomplished by finding the width of the record on the waxed chart when a weak light is used. Such a weak light is obtained from the standard lamp* used in performing wedge calibration tests. The test procedure is as follows:
[*This test can only be performed using tungsten filament standard lamps, since tungsten- halogen lamps cannot be operated at one-half voltage.]
(1) Remove the ground quartz plate from the spectrophotometer inlet window. (The ground quartz plate is NOT used.)
(2) Place the standard lamp in position and adjust the lamp voltage or current to one-half the normal value.
(3) Set Q1 and Q2 for C wavelengths.
(4) The S4 shutter rod should be pushed all the way into the instrument, while the wavelength selector rod should be moved to position SHORT.
(5) Find an optimum photomultiplier tube voltage setting in order to obtain the narrowest possible trace.
(6) Obtain one record on the instrument's waxed chart or smoked disc of about 2 minutes duration.
(7) Measure the trace width and record its value.
(8) The trace width should be less than 2.5 degrees (a preferred width is less than 1.5 degrees). Some idea of reduction in instrument sensitivity may be obtained by comparing experimental and reference data obtained during original calibration of the instrument.
(9) If the observed trace width is greater than 3.0 degrees (2.0 degrees in polar regions), measures must be taken to increase instrument sensitivity (see Section 5).
Note: For spectrophotometers equipped with tungsten-halogen standard lamps, instrument sensitivity changes with time may be monitored by measuring and recording the width of A-wavelength record traces obtained when monthly standard lamp tests are conducted.
Return to Table of Contents
Forward to Section 5. Instrument Maintenance and Repair