1. Fiber Basics
For many years, the fibers used in astronomy have been one of two types - often called "red" and "blue" fibers. Both are made of high quality fused silica with fused silica cladding doped to increase its index of refraction and keep the light contained within the core.
The "red" fibers transmit very well longward of about 500nm, but their internal attenuation rises rapidly in the blue and they are virtually useless below 450nm.
"Blue" fibers are doped with a small amount of OH which greatly improves the transmission at short wavelengths. These fibers are useful well into the UV, the short wavelength limit depending on how long the fibers are. Unfortunately, they have absorption bands in the IR, with a significant reduction in throughput over a narrow band between 700-730nm. Worse attenuation bands appear farther in the IR, beginning at ~850nm, making these fibers almost useless much beyond this wavelength. Transmission curves for fibers of these types are given by fiber manufacturers, such as www.polymicro.com
Internal attenuation in fibers is usually expressed in dB/km. A fiber with an internal attenuation of 10dB/km will reduce the intensity of a beam by half in 333M, while one with 100dB/km attenuation will absorb 50% of the light in 33M. In practical terms, since runs in astronomical fiber-fed spectrographs are seldom more than 30-40M, attenuations of less than 10dB/km are usually negligible, while losses of significantly greater than 100dB/km are unacceptable.
2. Other Losses in Fiber-fed Astronomical Spectrographs
Losses in fiber-fed spectrographs also come from beam spreading or "Focal Ratio Degradation" (FRD for short). All fibers increase the solid angle of slow beams to some extent. The best fibers commercially available can be used in optical systems as slow as f/8. In a high quality, well polished, unstressed modern fiber, 95%+ of the light within an f/8 beam will emerge contained within an f/6-f/7 cone.
In calculating transmissions of fiber-fed spectrographs, one must take into account an additional 7-8% of light loss from reflection at the fiber ends unless the fiber tips can be anti-reflection coated (usually impractical).Even when the fibers are carefully prepared, another 5-15% is generally lost in scattered light produced by polishing imperfections and slight misalignments. In the CTIO Hydra, ~5% of the light is lost at a connector which has been installed in the system to allow Hydra to be removed from the telescope without extracting the fibers.
Thus, even if fiber transmission were perfect, about 75% of the light hitting one end of the CTIO Hydra fibers can be expected to come out the other at an angle at which it can be captured by the spectrograph.
3. The Effect of Seeing
The seeing and accuracy of the fiber positioning further affects the amount of light falling into the fiber. According to Wolffe's standard model for calculating the energy in a seeing disc, 41% of the energy enters a pefectly centered fiber when the fwhm of the seeing disc is equal to the fiber diameter. 68% of the light enters when the seeing disc is 2/3 the diameter of the fiber and 85% will be captured when the seeing disc is half the diameter of the fiber. A comparison of the effect of fibers, slits and and Integral Field Unit (IFU) seeing is shown in this table of efficiencies of aperture types. Sometime during 2000, CTIO Hydra is expected to be equipped with changeable slit masks which will be placed in front of the fibers so that spectral resolution can be maintained even under conditions of less than ideal seeing.
4. CTIO Hydra Fibers
In CTIO Hydra, a new type of fiber has been used, made by Polymicro using a fused slica boule made by Heraus-Amersil. This fiber, sometimes called "STU" fiber combines the best features of red and blue fibers and transmits very well from UV to near IR. This Fiber attenuation graph was measured at CTIO on the Hydra fibers and shows that the attenuation is acceptable at all wavelengths over the spectrograph's design wavelength range from 330-1100nm. The short wavelength limit is determined by the UV cutoff of the glasses used in the ADC corrector. The long wavelength limit is set by the limit of useful sensitivity of silicon CCDs.
The internal transmission in the 36M of fibers in CTIO Hydra is approximately 52% at 340nm, 72% at 400nm, 85% at 500nm and >92% from 600nm-1100nm.
5. Estimating CTIO Hydra's Efficiency
The median seeing on the Blanco telescope is approximately .9 arcsec. CTIO Hydra's positioning is expected to be very good, given astrometric quality coordinates with an rms error of < .2 arcsec.
In practical terms, the large fibers will obviously always collect more light but the spectrograph resolution will be higher and the sky noise will be lower in the small (1.3 arcsec) fibers. Efficiency will be poor under conditions with seeing worse than 2 arcsec, which fortunately constitute a very small fraction of observing time at CTIO.
CTIO Hydra uses fibers 36M long. We can combine all this information to make the following estimate of the overall capture efficiency of the Hydra fibers as a function of seeing and wavelength, assuming perfect centering, 25% loss of light due to end losses and FRD. Attenuation is as measured at CTIO. This table suggests a rule of thumb to use in estimating throughput of Hydra. To first approximation, under average conditions it will be about 1/3 that of a classical spectrograph with a very wide slit.
|Small (200u) fibers
(Not available - reference only)
Large (300u) fibers
As can be seen from this table, there is no situation in this range of seeing where the small fibers intercept more light than the large fibers. At seeings of .5" or better, the small fibers would gather more light, but images of this quality are rarely, if ever seen at the R/C focus of the Blanco telescope.
In some cases, small fibers would be better because they will intercept less sky or because they will give better resolution in the spectrograph, which was the reason they were installed. However, the small fibers have turned out to be disappointing. They were brittle and a number of them broke on installation. The rest are much less uniform in transmission than the large fibers with only about half of the small fibers having reasonable transmission. They appear to suffer from some kind of manufacturing problem.
Because of their poor quality and limited usefulness, the small fibers have been withdrawn from service and are not currently available. The large fibers work better in virtually all circumstances. Unitl the new 400mm camera / Site 2kx4K CCD combination is available, the resolution attainable with the 300 micron fibers is all that can be reasonably expected. Once the new CCD has been commissioned, observers desiring higher spectral resolution can achieve it by the use of slit plates 200 and 100 microns in width which are available to be inserted in front of the fibers. These will work much better than the present 200 micron fibers under virtually all observing conditions.