Next: Programming Observations Up: Table of Contents Previous: Policies of the Guest Investigator Program

III. Description of the Princeton Experiment

A. The spacecraft

The satellite is similar to OAO-2 (described by Code et al. 1979), with some modifications designed to improve the guidance sensitivity when the spacecraft is in daylight, and the pointing stability. The Princeton equipment is described by Rogerson (1963) and Rogerson et al. (1973).

In brief, the Princeton Experiment Package (PEP) consists of a cassegrain telescope with an 80 cm primary mirror, a 7.5 cm secondary, and a Paschen-Runge spectrometer which utilizes a concave grating to focus the spectrum on a 1 meter Rowland circle (see Figure 1). Two movable carriages, each equipped with two photomultiplier tubes, scan the spectrum. These carriages, which are programmed to operate independently of each other, cover the wavelength regions shown in Table 1. The nominal bandpasses are also indicated. For brighter stars, the important region from 1500 to 1560 Å can be scanned with U1 in the first order.

B. Spacecraft pointing

The field of view utilized by the guidance Fine Error Sensor (FES) is 8 minutes of arc in diameter. The FES acts to keep the center of light in this field on the slit; therefore the target star should have no companion less than two magnitudes fainter within 5 minutes of arc.

If this restriction is relaxed, the main image will not be centered in the slit, and the photon count will be reduced accordingly. In the case of very bright sources for magnitude differences only slightly less than two, this loss of signal may be acceptable.

When the line joining the two stars is parallel to the slit, as will happen at short intervals during the year, the main image will be correctly centered, and if the distance between the stars is less than 30 arc seconds, light from both stars will go down the slit. Although it is feasible to command small spacecraft rolls to accomplish this alignment at other times, such maneuvers are generally avoided because of possible risks to the satellite. Several close systems have been observed in this way, including Alpha Cru AB and Sirius AB.

It is possible to introduce a small bias voltage in the FES, so that the center of light of the guidance field of view is as much as 5 arc sec off of the slit. This technique has been used successfully to center one component of a binary (Rho Oph A) on the slit, in a case where the magnitude difference was less than 0.5. Other special techniques have been applied to facilitate observations of stars in crowded fields.

The Inertial Reference Unit (IRU) can be used to point the telescope-spectrometer to within roughly ±40" of a given position on the sky, if a nearby bright star is available which can be used to update the FES. Use of the IRU appears to be a feasible technique for pointing at extended sources such as nebulae. It appears, however, that high-excitation, compact planetary nebulae are the only extended sources likely to be bright enough for detection with Copernicus, and these in general require a pointing accuracy of ±10" or better. It is hoped that the required accuracy will be achieved as a result of upcoming efforts to refine and calibrate pointing techniques using the IRU.

The spectrometer entrance slit is 39 arc seconds long and 0.3 arc seconds wide. The width can be changed to 1.2 arc seconds on ground command, increasingly the signal by about a factor of 2 for a stellar source, but this has generally been avoided for fear of failure in the slit transfer mechanism.

C. Guidance sensitivity

Before a Guest Investigator proposal is made, one should determine as accurately as possible that the target star is bright enough for the guidance system to settle on and to maintain attitude for the duration of the observations. This is estimated by predicting the output voltage of the Automatic Gain Control (AGC) which will occur when the Fine Error Sensor (FES) has settled on the target. When the AGC value is greater than 9.6 volts, the FES automatically turns off because the star is considered too faint to observe. In general for any star whose predicted AGC is greater than 8.6 volts the ability of the spacecraft to guide on the star will be tested before any observing time is committed.

It has been found that the most accurate and convenient index for predicting the AGC value for a star is its U magnitude, in the standard UBV system. The bandpass of the U filter matches the response function of the guidance system closely enough that in general no corrections are needed for spectral type or interstellar reddening. The AGC versus U calibration for the guidance configuration which is normally in use is shown in Figure 2 (upper curve). This can be used to predict AGC values for any star of spectral type earlier than about A3. U magnitudes for a very large number of stars can be found in the Photoelectric Catalogue of Blanco et al. (1968).

If no U magnitude is available, B and V can be used as approximate AGC predictors, but allowances must be made for both spectral class and reddening. Figure 3 shows the calibration curves for both indices and indicates the dependence on spectral type. AGC values derived from the B magnitude (upper plot in Figure 3) should be corrected for reddening by the addition of 0.8 E (B-V) volts. A correction of 2.0 E (B-V) volts should be added to AGC values derived from V magnitudes (lower plot). For example, an O star which has V = 1.6 and E (B-V) = 0.20 would have a predicted AGC of 3.0 + 2.0 (0.20) = 3.4 volts.

For stars of intermediate A and later spectral types, the actual AGC will be at least 1 V less than that predicted from U magnitudes because of a red peak in the guidance sensitivity spectrum which allows late-type stars to be acquired by the FES more easily than indicated by the UBV magnitudes. For example Theta Cas, an A7 star, has U = 4.61 (implying a predicted AGC of 7.7 volts) but an actual AGC of 6.6 volts. Alpha Boo, a K2III star, has U = 2.43 (implying AGC = 4.8 volts) and an actual AGC of 3.1 volts. In cases of doubt for faint stars, guidance tests will be performed before observing time is committed.

Recent tests of a heretofore unused guidance photomultiplier tube have allowed Copernicus to guide successfully on stars a full magnitude fainter than previously possible; i.e. with this guidance configuration, it is possible to guide on an unreddened early B star of visual magnitude 7.5. Only very hot and largely unreddened stars of such faintness in V are bright enough in the ultraviolet to produce a detectable signal in the Copernicus spectrometer, but even so, use of this configuration adds substantially to the number of feasible target stars which are available for study through use of long integration times. It is anticipated that the new guidance configuration will be used only periodically, so that degradation of its sensitivity can be minimized. The U magnitude vs. AGC calibration for this guidance channel is shown in Figure 2 (lower curve).

D. Experiment efficiency

In order to estimate the expected spectrometer count rate for a proposed target star, the efficiency curves shown in Figure 4 should be used. These curves, which give the 14 sec count rate as a percentage of the photon flux incident on the unocculted 2914 cm² of the primary mirror, are based on calibrated rocket data compiled by R.C. Bohlin (private communication) for the star Eta UMa. Shortward of 1200 Å, the efficiencies are based on model atmospheres calculations, there being no calibrated data available at present. The curves of Figure 4 take into account degradation which has occurred since launch, primarily shortward of 1100 Å. The degradation is wavelength dependent (greater at the shorter wavelengths), but the time dependence is not known in detail. Detectable changes in sensitivity over a 6 month interval should be expected.

Several calibration techniques are being applied by Guest Investigators and should soon be in the literature. In addition, the degradation of the optical system as a function of wavelength is monitored on a semi annual basis.

In view of the uncertainty in model absolute photon fluxes in the far ultraviolet, it is advisable to avoid total reliance on theoretical models in estimating predicted Copernicus count rates. If no rocket or other far UV observations are available, the typical count rates shown in Tables 2-5 for stars of a variety of spectral types may be useful. Corrections for distance, absolute magnitude, and interstellar extinction differences must of course be made. The UV extinction curve derived by a previous OAO experiment (Bless and Savage, 1972) is reproduced in Figure 5. Although there is substantial variation between different stars in the shape of the UV extinction curve, Figure 5 and the results of York et al. (1973) are helpful in estimation Copernicus count rates from Tables 2-5.

E. Photometric accuracy

Characteristics of the photometry are described by Rogerson et al. (1973), and will not be repeated here in any detail. Briefly, tubes U1 and U2 are normally photon noise limited over the time required to scan an interstellar line.

The count rate due to background particles is very large for V1 and V2, significantly reducing the photometric accuracy and usefulness of these phototubes, except for very bright stars. The dark count in U1 and U2 caused by these particles is generally insignificant except when the spacecraft is in the South Atlantic Anomaly, where observations are not normally made.

The chief contributor to the large fluctuations of dark count in the near-UV tubes is the presence of bursts of photoelectrons, produced by cosmic rays, which strike the photocells several times per second. It has recently been found that, if cumulative photon counts are stored every 1/8 second instead of once each 14 sec, this random noise can be greatly reduced by rejecting those 1/8-sec time frames which contain high counts due to cosmic rays.

In this way it is possible to reduce the rms fluctuation of dark count in tubes V1 and V2 from about 400 in 14 seconds to roughly N, where N is the total 14-second count (which contains a contribution of typically 3500 counts in 14 seconds, probably caused by phosphorescence following passage through the South Atlantic Anomaly). Because data storage in this mode occurs every 1/8 sec rather than every 14 sec, spacecraft memory becomes filled in 4 minutes, and less time per orbit can be used for observing than in the normal mode. This special observing technique is mainly for use with V1, the high-resolution near-UV tube. Proposals are invited for use of this mode for particularly important problems requiring accurate V1 measures over a narrow wavelength range. Scientific programming and final data presentation will be essentially the same as for the standard operating mode.

Particle background counts for standard 14-second integration are routinely predicted as a function of latitude and longitude at the time of observation. Figure 6 shows an example of the range in counts which may be encountered within a given orbit. A detailed description of the algorithms used is available on request. In addition, dark counts are taken about once every four orbits, and these can be used to provide time correction factors to the routinely predicted background counts. The predictions are based on data taken during the first eight months of operation, and may be systematically in error, due for instance to changes in solar activity.

For cases where particle background limits the usefulness of V1 or V2 data, observations can be restricted to particular orbits which provide minimum background counts (at a substantial reduction in observing efficiency), or the high time resolution mode described above may be used. For stars fainter than third magnitude routinely scheduled V1 and V2 observations are of minimal use.

The spacecraft pointing accuracy, which is generally about 0.02 seconds of arc during a period of a few minutes, is good enough so that photon counts during single scans of spectrum lines are usually not noticeably affected by guidance changes. Over a period of several hours, changes in spacecraft pointing can cause changes of 5 to 10% in the photon counts. To achieve the highest accuracy in scanning lines with U1, carriage 2 can be held motionless during each scan and the data from U2 can be used to smooth out fluctuations in U1 which are due to spacecraft pointing changes.

A more serious problem is stray light which enters photomultioliers U1 and U2 through vent holes in the phototube mounts. The stray light problems have been discussed by Rogerson, Spitzer, et al. (1973) and York et al. (1973). Details of the correction of U2 data for the stray light problem are available on request. The correction algorithm has been developed by Dr. Ralph Bohlin (GSFC) and is described by him (Bohlin, 1975). The U2 spectrum can be corrected to an accuracy of about sqrt[(.03)²+a²], where a is the fractional uncertainty introduced by guidance drifts, (generally, a < 0.1) providing about 60 Å of spectrum is available longward of the longest wavelength of interest for a particular program, and providing a continuing U2 scan is available over the region of interest. Hence, this additional region should be included in observations planned by a Guest Investigator. For cases where complete U2 spectra are available, the corrections can be made at Princeton, upon request.

The stray light on tube U1 can be eliminated by judiciously positioning the carriage 2 dipping mirror in front of the stray light hole for tube U1, as described in Section IV.C.6, below. For O and B stars, scattered light from the optical surfaces amounts to about 10% of the local continuum for 1000 Å < lambda < 1400 Å and <5% of the continuum for 1800 Å < lambda < 3200 Å. For the remaining accessible regions of the spectrum, the scattered light from other parts of the spectrum may be equal to or greater than the observed intensity of the stellar spectrum due to the low optical efficiency in these wavelength regions. For cooler stars, the scattered light becomes significant at somewhat longer wavelengths.

In most cases, the scattered light must be determined by noting the residual intensity at the bottom of saturated stellar or interstellar lines.