The next generation of massively multiplexed multi-object spectrographs (DESpec, SUMIRE, BigBOSS, 4MOST,
HECTOR) demand fast, efficient and affordable spectrographs, with higher resolutions (R = 3000-5000) than current
designs. Beam-size is a (relatively) free parameter in the design, but the properties of VPH gratings are such that, for
fixed resolution and wavelength coverage, the effect on beam-size on overall VPH efficiency is very small. For alltransmissive
cameras, this suggests modest beam-sizes (say 80-150mm) to minimize costs; while for cadioptric
(Schmidt-type) cameras, much larger beam-sizes (say 250mm+) are preferred to improve image quality and to minimize
obstruction losses. Schmidt designs have benefits in terms of image quality, camera speed and scattered light
performance, and recent advances such as MRF technology mean that the required aspherics are no longer a prohibitive
cost or risk.
The main objections to traditional Schmidt designs are the inaccessibility of the detector package, and the loss in
throughput caused by it being in the beam. With expected count rates and current read-noise technology, the gain in
camera speed allowed by Schmidt optics largely compensates for the additional obstruction losses. However, future
advances in readout technology may erase most of this compensation.
A new Schmidt/Maksutov-derived design is presented, which differs from previous designs in having the detector
package outside the camera, and adjacent to the spectrograph pupil. The telescope pupil already contains a hole at its
center, because of the obstruction from the telescope top-end. With a 250mm beam, it is possible to largely hide a 6cm ×
6cm detector package and its dewar within this hole. This means that the design achieves a very high efficiency,
competitive with transmissive designs. The optics are excellent, as least as good as classic Schmidt designs, allowing
F/1.25 or even faster cameras. The principal hardware has been costed at $300K per arm, making the design affordable.