Pre-launch ground calibration of the UV-Grism

Relating the wavelength scale and sensitivity of the spectral orders

Report by Paul Kuin


Version: 1.1, converted to Sphinx 2013-06-11


In November 2002, the Swift UVOT underwent a ground calibration at the NASA GSFC. During this calibration, images and spectra were taken. The UV grism data discussed in this document are those taken for the redundant system which is the current operational system.

Observations and processing

The instrument was mounted on the optical bench of the test system with the artificial source at, or very near, the boresight (unless the telescope was rotated on the gimbal for special observations). The light source was a platinum/Neon lamp. A graph paper with the spectrum of the lamp with a resolution of 100Å/2cm was provided. Tests were taken with either a blank filter, or one of three narrow-band filters in the path of the beam that fed the telescope. The narrow band filters are known as NB193, NB214, and NB260. The document with the specifications of the band passes of the filters is not available to us. The effective bandpass of each band was reconstructed, using a combination of the observer notes provided, the zemax solution for the boresight, and tables of the Pt/Cr/Ne spectrum as provided by NIST on the WWW. The observations are available as event data in FITS files. The files were read in using an IDL routine to construct images, which are 2048x2048 pixels. The image X and Y were transposed in order to obtain the same geometry as the in-orbit data. The raw images were converted into detector coordinate images by appending headers to the raw images, and processing with the UVOT/ftools software (swiftxform).

In table 1, the observations of the current UV grism (redundant prism 1) that were used have been listed. Open items were not recorded.

Table 1. Observations used

obs # Path filename target metallic neutral density NDm# Narrow band filter λ lamp mA notes
0128 20021116_grism1_r/* 1037438536_0 5μ pinhole   NB214   on-axis
0129 20021116_grism1_r/* 1037440745_0 5μ pinhole   NB193   short time exposed with clear filter in addition to long exposure in NB193
0130 20021116_grism1_r/* 1037443124_0 5μ pinhole clear NB260 14 observer mistakenly notes a separation of the peaks λ2550 and λ2559 (These appear to be the λ2559+2550 blend and λ2531 instead)
0228 20021118_grism1_r/* 1037602492_0 5μ pinhole clear NB193 10 wheel position 200
0232 20021118_grism1_r/* 1037606076_0 5μ pinhole clear NB214 12 wheel position 200
0236 20021118_grism1_r/* 1037610501_0 5μ pinhole clear NB260   wheel position 200

Table 2: line identifications

Narrow band filter observed wavelength   line identifications (blends)
NB193 1883 (weak peak) 1883.0
NB193 1916 (main peak) 1911.7,1916.1
NB193 1930 (strong secondary peak) 1929.2,1930.0,1937.4
NB193 1971 (weak peak) 1969.7,1971.5,1978.8
NB193 2036 (weak peak) 2030.5,2032.4,2035.8,2036.5,2041.6
NB214 2175 (weak peak, blends with 2203) 2174.7
NB214 2203 (secondary peak) 2202.3, 2209.5
NB214 2246 (main peak) 2245
NB260 2631 (secondary peak) 2631
NB260 2660 (main peak) 2559.45,2650.8

In the following plot, the total counts observed in three observations (128,129,130) with three different filters have been combined and plotted as function of wavelength. Note that the spectral range probed is from about 1850 to 2800 Angstroms.

filter measurements

Figure 1. Combined in one figure are three observations with the measurements from each filter along the dispersion direction in the first order. The observations were taken in sequence with the source on-axis. The dotted lines show the individual filter contributions.

A contour plot shows the location of the flux on the image.Once again observations in three filters were combined in one image.

ground cal compared to zemax model result

Figure 2 shows the relative positions (from right to left) of the zero, first, and part of the second order, in the calibration observations. The zemax model calculations, with pixels scaled by 0.963, have been plotted as squares symbols at wavelengths ranging from 1900-2600 Å. The zemax model was shifted to fit the first order. As can be seen, the zero order shows a slight offset from the model position. This zemax model was calculated with a 3.8 degree correction.

Table 3 lists the (detector) image coordinates of the main peaks derived from examining the data in “DS9”. These have FITS coordinates defined by:

CTYPE1= 'DETX' /X coordinate type
CTYPE2= 'DETY' /Y coordinate type
CUNIT1 ='mm'   / X coordinate units
CUNIT2 ='mm'   / Y coordinate units

Table 3: Coordinates of peaks (DET image)

Filter obs Zeroth order First order Second order Third order
NB193 129 (1650.1,562.8) (1641.4,568.3) (1207.3,807.0) (1200.0,811.0) (781.3,1039.5) (768.5,1046.0) (351.0,1272.4) (339.0,1277.5)
NB214 128 (1605.9,592.9) (1603.0,594.0) (1108.8,864.0) (1095.3,871.8) (619.5,1129.0) (596.8,1142.0) (132.3,1394.0) (102.5,1413.3)
NB260 130 (1579.1,610.5) (987.4,933.0) (979.2,937.9) (395,1255) (390,1257.4) off detector
NB193 228 (1638.5,576.8) (1204.5,814.5) (1198,817.5) (796.5,1035.0) (780.0,1045.5) (360.5,1275.0) (345.0,1280.0)
NB214 232 (1604.5,600.0) (1598.9,602.1) (1105.5,871.5) (1093.0,878.5) (618.0,1136.5) (593.5,1150.0) (108,1411) (90,1421)
NB260 236 (1578.5,615.9) (1576,4,618.5) (984.1,940.5) (977.3,944.5) (395.4,1260.3) (388.8,1266.1) off detector

The table 4 lists the distances (along the dispersion) of the main lines of each narrow-band filter observation between the first and zeroth orders.

Table 4: measurements of order distances along dispersion direction (DET image)

filter obs line zero-first first-second first-third
NB193 0129 1916 505.7 485.3 975
NB193 0129 1930 503.7 419.4 979
NB214 0128 2203 566.2 556.4 1112
NB214 0128 2246 578.7 567.0 1131
NB193 0228 1916 502.0 463.8 961.5
NB193 0228 1930   476.1 970.3
NB214 0232 2203 568.1 554.9 1134.5
NB214 0232 2246 576.5 568.5 1140.3
NB260 0236 2631 677.3 670.0  
NB260 0236 2660 682.1 674.3  
NB260 0130 2631   674.3  
NB260 0130 2660   670.3  

A second order polynomial fit ( \(D(pix) = \Sum_i C_i \lambda^i,\ \lambda in Å\) ) to the distance from the first to the zeroth,second, and third orders as a function of wavelength was made. The coefficients are:

Table: Polynomial coefficients

. D(first-zero) D(first-second) D(first-third)
c0 241.6 +/- 52.2 -155.5 -2733
c1 0.0595 +/- 0.0463 0.375 3.136
c2 (4.00 +/- 1.01) 10-5 -2.374 10-5 -6.3 10-4

The result can be seen in Figure 3.

distance of first to zero order at borepoint from ground cal

Figure 3: the fit to the measured distances from the ground calibration.

The zemax model dispersion relative to 260nm in first order was plotted against the measured dispersion from the ground calibration. The corresponding zemax numbers, are presented in Table 5 for different scale factors. The scale factor of 0.963 +/- 0.003 was derived from fitting the wavelength scale to the first order spectra of WR52 and WR86. The distances between zero and first order for the 0.963 model are compared to those from the ground calibration in Figure 4. At wavelengths below 2000Å the zemax model is not reliable. It can be seen that the measurements for wavelengths larger than 2000Å are on the average four pixels less distant in the ground calibration. For comparision the zemax results, scaled by a factor of 0.957 have been included, which show a much better fit to the ground calibrated distances above 2000Å. A scale factor of 0.957 is two sigma from the result from fitting the first order wavelength scale, and is thus not inconsistent therewith. It must be kept in mind, that to derive the distance, the measured coordinates of the peaks were used. Since there is a slight offset of the actual zero order from the Zemax model, while the model distances include the offset, there is no reason to adjust the pixel scale factor from the 0.963 value.

Table 5. Zemax model distances along the dispersion for varying scale factors compared to the measured ground calibration

.ℷ(Å) 1.0 0.963 0.957 measured
1916 518 499 496 504
1930 522 502 499 504
2203 593 571 568 567
2246 604 582 579 578
2631 706 680 676 677
2660 714 687 683 682
zero to first order distance from ground cal compared to zemax

Figure 4: comparison ground calibration zero-first order distances to zemax with scale factor = 0.963.

The wavelength scale of observations might now be extended to the zero order, allowing to determine the wavelength of the uvotgraspcorr anchor point, and a photometric calibration for the zero order.

Distance of the first to second order as derived above from the ground calibration, was also compared to the zemax model results. Like for the zero-first order distances, the wavelengths below 2000Å do not fit very well with the zemax predictions, and the model with the scale factor of 0.963 fits quite well. It may be that the 1916Å is a misidentification, or that it is not very reliable since it is very noisy.

Table 6. Zemax model distances along the dispersion for varying scale factors compared to the measured ground calibration

λ(Å) 1.0 0.963 0.957 measured
1916 504.4 485.7 482.7 442 +/- 17
1930 508.1 489.3 486.2 480.7 +/- 3
2203 580.0 558.5 555.0 555.7 +/- 1
2246 591.4 569.5 565.9 567.8 +/- 0.6
2631 695.0 669.3 665.2 670.1 +/- 0.2
2660 703.1 677.1 672.9 672.5 +/- 2

The total counts in each filter give a good measurement of the relative sensitivity in the different orders, provided the coincidence loss is negligible. Inspecting the brightest features, no evidence for the kind of pattern we see in sky observations is seen. In fact, the source seems more extended. The source aperture may have been larger than a source on the sky depending on the distance of the pinhole to the instrument. It is therefore possible that some photons have been lost through the anto-coincidence code, which rejects detections that are within one frame in neighboring pixels. The 0130 observation appears to be approximately 2000 seconds long (no record was kept) , and the highest peak rate is about 27000 in the 260NB filter, spread out over 600-800 pixels. The maximum rate is therefore about 24000/2000/600 = 0.02 c/s/pixel. This is small enough that coincidence losses are negligible.

Table 7a. Measured relative sensitivities between orders: net counts by filter and observation

. NB193/0129 NB193/0228 NB214/0128 NB214/0232 NB260/0130 NB260/0236
zeroth order 240 319 367 183.4 10545 1900
first order 2572 4103 5010 2630 24232 4274
second order 2063.5 3250 2025 1158 5434 935
Zeroth/first 9.3+/-1.5 7.8+/-0.5 7.3+/-0.5 6.9+/-0.5 43.5+/-0.4 44 +/- 1.5
Second/first 80+/-2 79.2+/-1 40.4+/-1 44+/-2 22.4+/-0.5 22 +/- 1.5

Table 7b. Measured relative sensitivities between orders: net counts by filter and derived zeroth and second order effective areas (in cm2)

. NB193 NB214 NB260
Zeroth/first 8.6 +/- 1 7.1 +/- 0.4 44 +/- 0.5
Second/first 80 +/- 1 42 +/- 1 22 +/- 0.5
zero order effective area 0.51 0.76 5.6
first order effective area      
(cm2) (adopted from CALDB) 6.0 10.7 12.7
second order effective area 4.8 4.5 2.8

The extraction of the counts was done with a width of 25 pixels, which included most of the signal. The reason for such a wide extraction width is that the source is slightly extended due to the finite aperture. The effective areas were derived from the ratios and the effective area in the CALDB with latest update dated 2007-07-11, and are mainly included to give an idea of their size.

Point Spread Function

The point spread function in the zero and second orders was calculated with Zemax. A summary is given in table 8. The shape of the PSF is a peak with a horse-shoe like extension. It has been fitted with a 2-D gaussian, but I only list the radial widths. Listed are the full width at half maximum, as well as the maximum radius to which the PSF extends. The PSF needs to be studied further, in particular the extend along and across the dispersion, and the energy contained within certain limits which have to be defined intelligently. As such, this table should anly be taken as an indication of the character of the PSF variations.

Table 8. Summary of Zemax calculations of the PSF wavelength (Å)

wavelength zeroth order FWHM (pix) zeroth order maximum radius (pix) second order FWHM (pix) second order maximum radius (pix)
1900 9.0 19 5.9 12
2000 8.4 18 6.5 14
2100 7.7 17 7.7 16
2200 7.4 16.3 8.7 19
2300 7.1 15.7 10 22
2530 6.5 14.6 14 29
2900 5.9 13.6 22 46
4000 5.3 12.1 63 127
5500 5.0 11.5

Although below 1900Å the zero order dispersion is approaching one Å/pixel, the PSF of the spectrum is increasingly broadened. This is consistent with observations of WR52, which showed very broad bumps. A detailed comparison, using the wavelength scale from the data above, must still be made.

In the second order, the PSF becomes very large above 2500Å. A similar occurrence may happen in the first order above ~ 5000 Å.


  • The wavelength scale in the zero and second order has been determined relative to the first order, using the on-axis observations taken during the ground calibration in the 1900-2700Å range.
  • The relative sensitivity of the zeroth and second order with respect to the first order has been determined from the ground calibration measurements.
  • The point spread function in the zeroth and second order was determined from the Zemax optical model, and shows that at short wavelengths the lines in the zero order are the broadest, and the lines in second order are the broadest for the longest wavelengths.


  • “Atlas of the spectrum of a Platinum/Neon Hollow Cathode Lamp in the region 1130-4330Å,” J.E.Sansonetti, C.J Sansonetti, J.Reader, N. Aquista, A.M. Sansonetti, and R.Dragoset, J.Res.Natl.Inst.Stand.Tech. 97,1-212 (1992).
  • The ground test document: “Grism Resolution; Swift optical test procedure”. SWIFT-UVOT-145-R01, 10 Nov. 2002.
    • accompanying Spreadsheet 02_11_15_psf_data_sheet_final.xls.
    • accompanying CD-ROMs with data from the ground calibration.
    • accompanyin chart with spectral output of Pt/Cr/Ne Lamp, dd. 3-11-2002, Lamp S/N NY10258A.