Omega Prime - Introduction

Omega-Prime presently has broad-band filters for the J (1.2 micron), H (1.65 micron), and K (2.2 micron) bands, as well as the lower background 2 micron filters called Km (equivalent to K', 2.1 micron) and Ks (2.2 micron). In addition, there are presently 1% narrow-band filters at 2.122 micron (molecular hydrogen, v=1-0 S(1)), 2.144 micron (adjacent continuum), 2.166 micron (ionized hydrogen, Brackett gamma), and 2.295 (CO band) in the camera, and we expect to receive filters for 2.248 micron (molecular hydrogen, v=2-1 S(1)) and 1.644 micron (ionized iron, [FeII]). Further narrow-band filters can be accommodated, but must be specially ordered as they are 50mm in diameter.

To save you reading the whole thing, here's the bottom line. At present, we believe Omega-Prime at f/3.5 on the 3.5-m to be roughly as efficient as the MAGIC cameras at f/10 on the same telescope, in terms of background-limited detectivity for point and extended sources. Therefore, for any given source, it is appropriate to use the standard MAGIC sensitivity figures.

Of course, the big gain is that Omega-Prime has 4 times the field-of-view as MAGIC in its 0.8 arcsec/pixel mode (plus twice the spatial resolution), and 25 times the field-of-view as MAGIC in its 0.32 arcsec/pixel mode (with roughly equal seeing-limited resolution).

The Omega-Prime Gamble

We took this risk in order to create a very simple optical design for wide field imaging that could be realised relatively quickly and simply. Making an image of the telescope pupil requires taking the diverging f/3.5 beam of the telescope, making a collimated section for the pupil stop, then reimaging onto the detector at f/2.6. A lot of "beam bending" is required, and the optical would have had perhaps perhaps eight or ten lenses.

In the Omega-Prime design, we simply use a triplet of CaF2 and fused silica lenses to correct for the de facto distortion present at the prime focus of a Ritchey-Chretien design telescope, and also to provide some mild demagnification to the image scale of 0.4 arcsec/pixel.

With such an approach, you can show that every pixel on the detector array sees some of the area surrounding the primary mirror, and thus sees excess thermal emission from this warm black material. There is a sketch (obviously not to scale) that might (!) help explain this.

This should only be a problem at wavelengths longer than about 2.2 micron, where this 300K material starts radiating appreciably, and thus it is only the broad-band K filter (and its cousins Km and Ks) that should affected. Filters shortward of 2.2 micron such as J and H and the narrow-band filters, should not see any excess background in principle.

As the background gets higher, your sensitivity gets worse. The signal-to-noise for a given source in a given integration time is proportional to the square root of the background flux. Alternatively, the time taken to reach a given signal-to-noise on a given faint source is linearly proportional to the background flux. Therefore, by allowing extra background into the camera, we reduce its effectiveness.

However, there are several additional factors to consider. First, we can reduce the excess background by placing baffles in the camera system. Omega-Prime has two main baffle sets. One is inside the camera dewar itself and is cooled to 90K, thus adding no appreciable radiation at all within the 1-2.5 micron bandpass of the detector. This baffle extends roughly 40cm forward of the detector and provides the major part of the thermal background rejection. This baffle is shown in a schematic diagram of Omega-Prime.

Secondly, we have a warm annular baffle roughly another 40cm forward of the camera. Although this baffle is warm, it is in the form of a highly polished aluminium toroidal mirror. Where the detector can see this baffle, it then in turn just sees back into the cold dewar, again reducing the background excess. This external baffle lowers the background at K by another 25%.

Another major factor to consider is that not all the background in the K filter comes from thermal emission: roughly 50% comes from OH airglow in the atmosphere. Thus, if a thermal excess of 50% is seen by the camera, that is only effectively a 25% increase in the total background in that filter. Furthermore, the Km and Ks filters in Omega-Prime have shorter long wavelength cut-offs, deliberately designed to reject the thermal background which predominantly comes at the long wavelength end.

Finally, despite a reduced sensitivity per pixel, we have to remember that Omega-Prime has a much bigger detector array than MAGIC, and that therefore, overall the additional area coverage should more than make up for the reduced efficiency per pixel. A number of calculations were run to verify this conjecture, and on the basis of those, we went ahead and built Omega-Prime.

The real Omega-Prime

First, there are only three lenses on Omega-Prime, all low refractive index, while MAGIC has six lenses, several of which are high index (and therefore more prone to reflection losses). Second, there is only one mirror in the Omega-Prime optical train, the primary, whereas MAGIC uses the secondary as well. Third, because there is no Lyot stop in Omega-Prime, it cannot be undersized and misaligned. Finally, it is plausible that the detector quantum efficiency is higher in the new Rockwell arrays (although this is unconfirmed).

What effect does the higher throughput have? The signal-to-noise for a given source in a given time is linearly proportional to the source flux, whereas (as said above), it is proportional to the square root of the background flux. Therefore, the additional system throughput can in effect compensate for the additional background.

This appears to be the case. We have made preliminary measurements of the Omega-Prime throughputs and backgrounds through the various broad-band filters as follows. The same numbers of MAGIC are also given for comparison. CFZM stands for "counts for zero magnitude", and is the number of electrons per second that would be detected from a zeroth magnitude star. It is given here in units of billions (10^9) of electrons. The background is in units of magnitudes per square arcsecond. The final column gives a measure of the Omega-Prime "efficiency" compared to MAGIC. That is, for a given source in a given time, how does the Omega-Prime signal-to-noise compare with that of MAGIC? Numbers greater than 1 mean that Omega-Prime is more efficient on a per pixel basis; less than 1 means MAGIC is more efficient. If you want to calculate the time taken to reach a given signal-to-noise on a source, then you have to take the square of this number. Finally, note that this does not include the additional area coverage of Omega-Prime, as it is just for a single source.

Other aspects of the Omega-Prime status

We have a slight optics misalignment at the moment which results in the images having a small amount of coma. This seems to be limiting us to just under 1 arcsec FWHM resolution. We think we understand where the misalignment is - the whole camera seems to be off the telescope optical axis by a few millimetres. We have built an X-Y translation stage for the camera, which will be installed and tested when the science-grade array is available.

The system read-noise is roughly 25 electrons RMS at present. While this is half that of MAGIC, it is also nearly double what it should be for this new detector array. While the basic problem is well understood, the cause is not, despite many many hours of testing and debugging. We have plans in place to fix the problem, but have no real timescale as yet. There is also a low-level striping problem that seems to be directly related to the quality of the engineering grade array, but again, this is only seen at very low background levels. Virtually all observations with Omega-Prime will be heavily background limited, and the excess read-noise and striping issues will not be important.

The usable full well depth in the Omega-Prime detector is 160,000 electrons, roughly half that of MAGIC. Also, the minimum integration time is 0.8 seconds, compared to 0.05 seconds for MAGIC (this scales directly with the detector area, since both detectors are clocked at the same pixel speed). Therefore, Omega-Prime has a significantly fainter saturation limit than MAGIC. That is, in a broad-band K filter, Omega-Prime will saturate on stars brighter than (roughly) 9th magnitude in the minimum integration time. This may be important for your observations, and certainly requires that faint standard stars are available for calibration.

End note