As indicated above, at the bottom of the ISU, below the dichroic tertiary mirror, are the acquisition and guiding systems. These utilize the visible light transmitted by the dichroic and comprise, in order:

1. A large pupil-imaging lens, which allows off-axis guiding without vignetting, is mounted below the dichroic pickoff mirror. 
2. A precision X-Y screw-driven crosshead is mounted below the lens. Its absolute position encoder readouts are accurate to about 0.01mm The crosshead, manufactured by SKF, Germany, is very stiff: flexure is barely detectable even in ~horizon-to-horizon slews. 
3. An InterFace Unit (IFU) is supported by the crosshead. This can select between sending all the light straight down to the acquisition TV camera or diverting 95% of it at right angles to a Fast Guider system. It can also feed a second right-angled channel opposite the Fast Guider channel (the “Jose” channel). This capability allows more than one wavefront sensor at a time to be used, a facility which has been exploited to calibrate the Gemini Prime Focus WFS. 
4. The Fast Guider system re-images a guide star onto a very-low-noise CCD detector. The guide star position measurements are used to drive the secondary tip-tilt system to stabilize the image on the CCD. In a dark sky, sampling at 40 Hz, the system can guide on stars with V > 18. ^m6. 
5. The Fast Guider also allows Active correction of focus by a ~1 minute observation of a nearby star before commencing an observation. The system also allows seeing estimates to be derived from the RMS focus fluctuations. 
6. A low light level intensified vidicon TV camera is located in the “straight-through” channel below the cross-head and IFU. It is equipped with a focal reducer to increase the acquisition FOV. Note that if the fast-guider is operating, only 5% of the light gets to the TV camera, so you are losing 3.3 magnitudes of sensitivity. (We keep getting complaints about how poor the TV camera is: actually its performance is pretty average, it just that we just have better uses for most of the photons!) The properties of the TV system are as follows:
   • Acquisition FOV: 72″ x 54″ (without focal reducer, ~24″ x 18″) 
   • Faintest detectable star during Full Moon: V = 16^m±1 (dep. on distance from moon). 
   • ” ” ” ” Half Moon: V = 17^m±1 
   • ” ” ” ” New Moon : V = 19^m±1 The TV can also be used for slow guiding. Its bandwidth is much less than that of the CCD system and it is substantially less sensitive.

Back to Top-end and secondary mirror: tip-tilt system. 

Crosshead constraints and guider vignetting

The maximum crosshead travel is ±3.5 arcmin on the sky. However there are a number of other constraints on the selection of guide stars which must be borne in mind. The vignetting-free field, defined by the pupil-imaging lens, is circular.

The 10mm thick dichroic substrate (aka tertiary mirror) produces a 5″ refractive shift of the image of a star seen though it (e.g. on axis). This shift vanishes when the light does not pass through the substrate. A star within a few arcsec of a substrate edge will therefore be split into two images 5″ apart, so the edges must be avoided when choosing guide stars: two bands across the FOV, at right angles to the line towards the instrument in use, are thus vignetted by the edge of the substrate. 

Under the dichroic patch itself the light transmitted to the guider, etc, is attenuated by around 50%. 

This figure shows the exact dimensions of the areas of constraint which affect the selection of guide stars. 

Crosshead continuous motion constraints and offset guiding while observing moving objects

The telescope is often used to observe moving objects. This is relatively simple as long as the objects are bright enough to guide on. However, if the object is faint, it may be desirable to use an offset guide star for guiding. This situation requires the crosshead to continually move to compensate for the telescope motion and keep the guide star centered in the guide box. Tests conducted on the crosshead indicate that it can move at 0.04 mm/sec and still meet the tracking criterion. This converts to a differential tracking rate on the main telescope of RA = 0.00406/cos(Dec) sec of time/sec and Dec = 0.0609 arcsec/sec.

Tip-tilt image stabilization

This operation is done under control of the Fast Guider system. A relay lens re-images the telescope focal plane to form an image of a guide star on a low-noise CCD, in the correct corner to give fastest possible readout. This permits sampling rates of up to 200 Hz (though 100 Hz is the fastest normally used).

The CCD has a pixel field of view of 0.”314; in operation pixels are binned 3×3 to form 0.”942 superpixels. The guide star image is held at the intersection of four adjacent superpixels, which with their surrounding 12 superpixels form a high-performance centroiding sensor. Error signals from this guiding array are fed back to control the secondary mirror actuators. 

As currently implemented, a sophisticated Kalman filter in the feedback systems allows guiding at a 40 Hz sampling rate (~4 Hz closed-loop bandwith) on ~K0 stars down to V~18. ^m6. (The colour of the guide star matters because the unfiltered CCD detector senses at an effective wavelength somewhat redder than the V band.) 

Good guiding on a ~17. ^m0 star has been demonstrated ~40° away from a nearly-full moon in a clear sky. Guiding becomes unreliable when the guide pixels see much less than about 10 counts sample^-1 above the sky level (for sky levels of ~15 counts or less).

Other modes of the autoguider system

There are four selectable operating modes of the autoguider systems: acquisition, focus, autofocus or normal guide. In operation the control interface displays various parameter buttons and CCD output displays according to the current mode of operation. 

• Acquisition mode: This offers ~25″ FOV with 0.”314 pixels. The guide box is in the bottom left corner; the acquisition field center is offset (TBD) from that of the normal guide box. Currently this mode is rarely used, but could offer much better sensitivity than the TV (q.v.)
• Focus mode: this offers a ~9″(TBC) FOV, with 0.”314 pixels. Its centre is offset (TBD) from that of the normal guide box. This mode is semi-obsolete and used only for occasionally re-determining optimum guider focus settings. 
• Autofocus mode (see Telescope Focus). The Autofocus guide box is offset (-9″, -3″) (TBC) from the normal guide box. It employs an 8×8 array of superpixels (effectively 4 normal mode guide arrays). 
• Normal Guide mode: this employs a 4×4 array of 0.942″ superpixels (each binned 3×3) to form a centroiding detector, signals from which control the tip-tilt secondary. 

Chopping

The tip-tilt system can also be used for sky chopping in any direction. The maximum possible throw is ~34 arcsec, but in practice the maximum achievable throw at which control by the fast guider can be maintained is around 25″, while a maximum of 20″ is the norm. The chop waveform remains good at frequencies up to ~20 Hz (for small throws: there is some tradeoff of throw versus waveform quality). At higher frequencies the chop waveform deteriorates.

NB: The maximum attainable throw depends on circumstances: if the secondary control electronics at the top-end were last reset at a temperature too far above ambient (i.e. the weather has since turned colder) the maximum throw is reduced. There is also evidence that degradation of peizo properties at low operating temperatures also limits the maximum attainable throw, which is consequently somewhat smaller in winter than in summer.

The fast guider system stabilizes the image at both ends of the chop, but since guiding is done by the quadrant detector formed by four adjacent superpixels the stabilized image is always held at their shared corner. The chop throw is therefore necessarily quantified in units of the fast guider CCD superpixels. It is not currently possible to fine-tune the CCD focal plane scale, so the pitch of the superpixels will not in general match the pixel spacing on sky on the IR instrument in use, i.e. both images will not in general be similarly located relative to the detector pixels.

However, a chop throw of 7.”32 is accurately equivalent to 8 superpixels on the guider CCD and to 12 spatial pixels on the CGS4 detector array (when used with the 40l and 150l gratings). This throw will accurately locate both images on a pixel row in the same way, and should be employed for all point-source CGS4 spectroscopy with the indicated gratings which requires chopping. 

NOTES:

1. Chopping is now rarely used other than for mid-IR imaging.
2. For chop throws over ~14″ (p-to-p) the coma induced by the mirror tilt exceeds the image degradation tolerance specification employed in the UKIRT Upgrades Programme.