ING Banner
Home > Astronomy > ISIS > Dichroic Response


Standard 5300 dichroic and blue mirror folder response

Unless otherwise requested, the default ISIS dichroic is the D5300. Those interested in using a different dichroic should clearly indicate this and justify its use in the observing time application, and should contact the ISIS instrument specialist if in doubt.

The following plots illustrate the response of the dichroics in the blue and red arms. The dichroic is mounted on a moving slide together with the plane mirror. Use of the mirror will therefore reflect all the incoming light to the blue arm, while red and blue arm observations can be performed with the dichroic in place.

Blue light is reflected from the front face of the dichroic, and the blue curve shows the dichroic response divided by that of the mirror (the plane mirror 'normalisation' level is thus dependent on the mirror throughput). Hence this is the actual relative throughput of the dichroic response compared with the mirror. Note that the response of the standard dichroic is better than that of the the plane mirror over a significant part of the wavelength coverage. This is a real effect which observers should bear in mind when doing blue-arm-only observations. Please also note that the ripples are almost negligible.

By removing the dichroic slide from the beam, light passes directly on to the red arm. Hence, the red response curve shown below is the dichroic (transmission) response divided by the 'straight-through' response. The curves are quite flat, and the slight ripples which are of the order 1-2% can be easily removed. There is a ~5% efficiency loss in the red arm when observing with the dichroic, as compared with red-arm-only operation. The 'crossover point' on the plot at ~5336 Å is for reference only, as the blue and red plots are normalised to different scales.

The throughput of the standard 5300 dichroic is better than that of the other dichroic units available (>95% in the red arm, and also in the blue according to predictions). Ripples, in particular in the blue arm, are much less pronounced. The cut-off region is also narrower, with the throughput in both arms below 70% in the 5200-5400 Å region. The dichroic reflectance is ~4% better than that of the blue mirror in the 3500-5100 Å region.


new dichoric response




The throughput of the blue mirror is larger than 95% according to the manufacturer. The relative throughput of the standard blue mirror with respect to that installed in the 5400-dichroic unit is shown in the figure below. Note that the throughput is >10% better.


New mirror response



Additional dichroics response curves

The following plots show the blue and red response of the rest of available ISIS dichroics. They are labelled according to their half-power point on the blue side (i.e. where response drops down to 50% in the normalised curve).

Blue light is reflected from the front face of the dichroic, and the blue curve shows the dichroic response divided by the response of the mirror (the plane mirror 'normalisation' level is thus different in each plot, as the actual mirror used in each case is the mirror within each dichroic unit). Hence, this is an actual relative throughput that you will notice in your data when you use a particular dichroic compared with its corresponding plane mirror. Apart from the ripples, three of the dichroics actually show better transmission than their respective blue mirrors over a significant wavelength region. As with the standard 5300-dichroic, this effect is real and has to be considered when performing blue-arm-only observations. The ripples can be a problem if flat-field normalisation is not carefully performed (see picture below). As the night observations are taken at different telescope position compared to the set of afternoon flat fields, due to the instrument flexure there is a shift of the spectrum on the detector. The way to correct for this shift is to take flat field exposure at the same telescope position as each observation during the night. This is quite time consuming, so some observers may decide to use only one arm of ISIS to eliminate ripples or to use the default dichroic 5300. On the example below a normalized flat field is displayed together with a ratio of two normalized flat fields taken at two different telescope positions. Obviously, the ratio of two flat fields is not flat. The images were taken with the dichroic 5700, grating R600B and the central wavelength of 4720 Å.

Effect of dichroic ripples
on flat-fielding

By removing the dichroic slide from the beam, light passes directly to the red arm. Hence, the red response shown here is the dichroic (transmission) response divided by the 'straight-through' response. The curves are quite flat, and the slight ripples, which are of the order of 1-2%, can be easily removed. There is a 10% efficiency loss (only 5% with the standard 5300-dichroic) in the red arm when observing with the dichroic, as compared with red-arm-only operation.

It should be noted that all dichroics exhibit a far-blue leak. Therefore, observations in the red arm at wavelengths >6000 Å demand the use of an appropriate order sorting filter, so that the second-order blue light scattered into the first order red end can be blocked. The 'Cross-over Point' marked on each plot is for reference only, as the the blue and red plots are normalised to different scales (i.e. the plane mirror and straight-through transmission, respectively).


5400 Dichroic
5700 Dichroic
6100 Dichroic
7500 Dichroic


Top | Back

Contact:  (ISIS Instrument Specialist)
Last modified: 06 September 2016