Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.mrao.cam.ac.uk/projects/OAS/publications/fulltext/MROI_SPIE2012_Young_TheMROIFastTipTiltCorrection.pdf Дата изменения: Fri Mar 1 16:36:01 2013 Дата индексирования: Sat Mar 1 03:36:15 2014 Кодировка: Поисковые слова: ccd

Copyright 2012 Society of Photo-Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.


The MROI fast tip-tilt correction and target acquisition system
John Younga , David Buschera , Martin Fishera , Christopher Hania , Alexander Reaa , Eugene Senetaa , Xiaowei Suna , Donald Wilsona , Allen Farrisb , Andres Olivaresb and Robert Selinab a Cavendish Lab oratory, University of Cambridge, JJ Thomson Avenue, Cambridge CB3 0HE, UK; b Magdalena Ridge Observatory, New Mexico Institute of Mining and Technology, 801 Leroy Place, Socorro, NM, 87801
ABSTRACT
The fast tip-tilt correction system for the Magdalena Ridge Observatory Interferometer (MROI) is b eing designed and fabricated by the University of Cambridge. The design of the system is currently at an advanced stage and the p erformance of its critical subsystems has b een veried in the lab oratory. The system has b een designed to meet a demanding set of sp ecications including satisfying all p erformance requirements in ambient temp eratures down to

-5 C

, maintaining the stability of the tip-tilt ducial over a

5 C

temp erature change without recourse We describ e the imp ortant

to an optical reference, and a target acquisition mo de with a

60

eld-of-view.

technical features of the system, which uses an Andor electron-multiplying CCD camera protected by a thermal enclosure, a transmissive optical system with mounts incorp orating passive thermal comp ensation, and custom control software running under Xenomai real-time Linux. We also rep ort results from lab oratory tests that demonstrate (a) the high stability of the custom optic mounts and (b) the low readout and compute latencies that will allow us to achieve a 40 Hz closed-lo op bandwidth on bright targets.
Keywords: interferometry, tip-tilt correction

1. INTRODUCTION
The fast tip-tilt correction system for the Magdalena Ridge Observatory Interferometer

1 is b eing designed and

built by the optical interferometry group at the Cavendish Lab oratory of the University of Cambridge. Each system will p erform closed-lo op correction of atmospheric tip-tilt (angle of arrival) uctuations on the light collected by an MROI Unit Telescop e (UT), and will also function as a narrow-eld target acquisition sensor. The preliminary design of the system has recently b een completed; as part of this eort several key asp ects of the design have b een prototyp ed. Reecting its dual role, the ocial name for the system is the MROI Fast Tip-Tilt correction and Narroweld Acquisition System (FTT/NAS). The interferometer system design

2 places an FTT/NAS on the Nasmyth

optical table at each Unit Telescop e (UT), where it is fed with light in the 3501000 nm waveband reected from a dichroic mirror. Other colours pass through the dichroic and are relayed to the b eam combining lab oratory. The purp ose of the FTT/NAS is to sense the lo cation of an image of a celestial target with resp ect to a predetermined ducial p oint (referred to here as the tip-tilt zero p oint), and initially to send p ointing corrections to the UT mount in order to move the target close to the zero p oint, and subsequently to detect and eliminate the smaller, rapidly varying tilt errors principally arising from atmospheric turbulence. This fast tip-tilt correction is accomplished by applying correction signals to a fast steering mirror (the UT secondary) and slowly-varying ooads to the telescop e mount. At the MROI, the lo cation of the zero p oint will b e realised each night, prior to observing, by sending co-aligned and parallel light b eams simultaneously to the interferometric instruments and to the UTs. The instruments in the b eam combining lab oratory will b e aligned with resp ect to the inward-propagating b eams, while at each UT the relevant outward propagating b eam will b e directed onto the FTT/NAS sensor after backreection from the FTT/NAS dichroic, retro-reection o a corner-cub e installed on the UT Nasmyth table and subsequent transmission through the FTT/NAS dichroic.


Time-variable osets may b e applied to the zero p oint determined at the start of the night, in order to comp ensate for dierential atmospheric refraction b etween the FTT sensor waveband and the fringe detection waveband, and/or to accommo date an o-axis tip-tilt reference ob ject.

1.1 Top-level requirements
The following brief list summarises some of the most critical of the FTT/NAS requirements:



Management of time varying osets due to atmospheric disp ersion and/or o-axis guiding; Supp orting the streaming of live diagnostic telemetry; Supp orting b oth acquisition and fast-guiding mo des; Realising the sensitivity (mV

= 16

for a red target in

0.7

seeing) desired for faint-source science;

Realising the zero-p oint stability requirements, esp ecially in an exp osed variable-temp erature environment; Meeting the thermal dissipation budget; Designing a system that is compatible with the space constraints present on the Nasmyth optical table.

All of these requirements, and others not mentioned here, have inuenced the design of the FTT/NA system. Although none of these requirements is individually unique, the combination has led to a design with some noteworthy features. In the remainder of this pap er we present our design, and describ e how the requirements have driven it (Sec. 2). The metho ds and results of the lab oratory validation carried out thus far are presented in Sec. 3. We then present our conclusions in Sec. 4, along with a brief summary of the development and integration tasks planned for the coming months.

2. DESIGN DESCRIPTION
Our design for the FTT/NA system is based around an o-the-shelf back-illuminated electron-multiplying CCD camera (Andor iXon X3 897), which oers fast readout, high quantum eciency and sub-electron eective read noise, all of which are needed to meet the stringent closed-lo op bandwidth and limiting magnitude requirements. Images from the FTT/NAS sensor will b e interrogated by a lo cal computer which will serve to b oth archive the data (either lo cally or via the central Interferometer Sup ervisory System (ISS)) and to send control demands to a fast steering mirror (the UT secondary mirror is used for this purp ose, and will b e supplied, complete with servo controller, by the UT vendor AMOS according to an MROI sp ecication). Information on the current state of the UT mount, and any ancillary information needed by the FTT/NAS will b e delivered via the ISS. The Andor iXon X3 897 EMCCD camera is able to meet all our derived requirements asso ciated with format, cost, cabling, mechanical stability, and heat dissipation. targets. A preliminary A However, a custom readout mo de will b e needed to achieve the high frame rate and low latency required to deliver a 40 Hz closed-lo op 3 dB bandwidth on bright

23 в 23

pixel custom mo de has b een implemented by Andor and tested in Cambridge.

32 в 32

pixel derivative of this mo de is now b eing develop ed by Andor, and oers the advantages of lower

pattern noise due to clo ck-induced charge, and a larger eld of view to accommo date eld rotation when using an o-axis tip-tilt reference star. We have chosen to use the same sensor for b oth target acquisition and fast tip-tilt correction, with a xed pixel scale. A suciently large-format camera (at least allows b oth accurate image centroiding and a

500 в 500

pixels), with suitable choice of image scale,

60 в 60

arc second eld-of-view for routine target acquisition.

The Andor EMCCD (as well as the alternative EMCCD cameras considered) is only guaranteed to op erate eectively at a temp erature ab ove enclosure at all times. A critical design goal for the FTT/NAS is the need to ensure that its optics, together with the dichroic and the sensor camera must remain suciently stable in tilt and displacement such that the tip-tilt zero p oint (guiding centre) do es not move by more than roughly half a micron on the detector surface over a night's observations .


0 C

and in a non-condensing environment, and so the FTT/NAS camera will

b e enclosed and a thermal control system supplied to maintain the temp erature and humidity inside the camera



The exact requirement is a variation of no more than 0.015 seconds of arc on the sky for a temp erature change of

5 C.


Figure 1. Left: A schematic diagram outlining the geometry of the FTT/NAS opto-mechanical layout. The b eam from the UT tertiary mirror enters from b ottom right and is intercepted by the dichroic to the right of the M4 mirror. The reected b eam passes through an ap o chromatic lens, and is fo cused onto the FTT/NAS camera sensor (at top) after reection o two fold mirrors. Right: 3-d rendering of the FTT/NAS as it will b e installed on the Nasmyth optical table. The featureless b oxes represent the current space envelop es allo cated to other hardware systems that will need to b e lo cated on the Nasmyth table.

To meet such high stability requirements we have chosen the p erhaps unusual approach of employing mounts without adjusters for the optical comp onents. This in turn will require the system to b e tolerant of fo cus changes so that fo cus adjustment is only required seasonally. These stability requirements demand low sensitivity to thermal changes and so thermal gradients across the comp onent mounts must b e minimised. This has led us to adopt aluminium rather than stainless steel or invar (which p erformed only slightly b etter than aluminium due to its p o or thermal conductivity, hence the additional material and fabrication costs were not warranted) for the mount material. Our design envisions the fast tip-tilt lo op b eing closed in software rather than with additional hardware, e.g. recongurable electronics such as FPGAs. This approach minimises the system's electrical p ower consumption. Basing our system on a standard PC has also allowed us to make the most extensive p ossible use of software libraries provided by the camera vendor. A xed frame rate of 1 kHz will b e used for all but the faintest targets (there is no noise p enalty for this due to the on-chip amplication) but the closed-lo op bandwidth will b e user-selectable by means of adjustable servo parameters that allow the degree of time-averaging of the correction signal to b e altered by the user/ISS. We have established that the Andor iXon X3 897 EMCCD can satisfy the derived requirements for a 40 Hz closed-lo op bandwidth with a custom CCD clo cking scheme using a an o-axis reference star for at least 300 seconds, b efore a brief ( is needed to rep osition the sub-frame.

32 в 32

pixel sub-frame read out at 1 kHz

frame rate. This sub-frame size will b e sucient, under worst-case conditions, to allow tip-tilt correction using

<1

second) interruption to fast tip-tilt mo de

2.1 Optical design and layout
The FTT/NAS optical train is depicted in the left panel of Figure 1. Its essential features can b e summarised as follows:



Interception of the 95 mm diameter collimated output b eam of the telescop e with a large dichroic splitter. This diverts the bluer' light to the FTT/NAS and allows the redder wavelengths to b e transmitted to the b eam combining lab oratory;




Fo cusing of the collimated b eam using a large diameter ap o chromatic lens. For this application we have traded-o the slight chromaticity of a lens-based solution with the much more demanding angular stability (в

20)

and installation tolerance (в9) needed for a non-chromatic o-axis parab ola;

Folding of the converging b eam path using two plane mirrors; Optimisation of the geometry of the folded path so as to keep the four principal optical comp onents as close together as p ossible these are all co-lo cated on a single sti baseplate and so as to lo cate this baseplate and the sensor head and enclosure as far away from the table edge as p ossible.

The selection of this layout was mainly determined by practical constraints on the space available on the optical table. These constraints included presence of additional systems (shown in the right panel of Figure 1), limited clearance ab ove the table, and a shallow angle of incidence on the dichroic mirror to allow its coating to b e optimised for measuring fringe data on p olarised targets in total intensity. We have chosen a custom 1.525 m fo cal length triplet lens to use as the fo cusing optic in the FTT/NAS. This will comprise three cemented elements made of common, easily worked glasses (N-BAK4, N-KZFS4, and N-LAF2). These have b een chosen so as to give excellent achromatic p erformance from 400 nm to 900 nm and a temp erature-dep endent fo cal length that will largely comp ensate for any thermal expansion/contraction of the steel top of the Nasmyth optical table. Only inter-seasonal fo cal changes will b e necessary (for example by utilising xed exchangeable spacers in p ositioning the camera mount).

2.2 Opto-mechanical design
The FTT-NAS opto-mechanical system will comprise two main assemblies; (i) a common baseplate assembly and (ii) the EMCCD camera and its thermal enclosure assembly. The layout of the two assemblies is shown in the right panel of Figure 1. Our design exploits the use of a single baseplate up on which the dichroic mirror, the fo cusing optic, and the two fold mirrors will b e co-mounted. The baseplate will mitigate, to rst order, the eects of any lo cal dierential tilts or deformations in the Nasmyth table induced during the night due to changes in temp erature. Such lo cal disturbances would lead to dierential movements and angular shifts of the FTT/NAS optical comp onents. The EMCCD camera will b e mounted on its own baseplate which covers a signicant area of the Nasmyth table so as to similarly reduce the eects of lo cal deformations of the table surface. The common baseplate (including the optics mounts) and the EMCCD camera mount will b e made of identical aluminium alloy and will b oth employ kinematic seat arrangements to allow for dierential expansion of the baseplates and the steel of the optical table. The mounting comp onents and materials of b oth assemblies will b e matched to ensure that the centre of the fo cusing optic and the EMCCD camera move together as the temp erature changes. As part of this strategy, the lens mount utilises material comp ensation to accommo date the dierence in thermal expansion co ecient b etween the glass lens and its aluminium mount the lens supp ort pins b eing made of a high-expansion p olymer and fabricated to a precise length so that the lens remains centred in the mount as the temp erature changes. This common baseplate approach also allows all the optical comp onents to b e p ositioned accurately relative to each other so that the installation inaccuracies will b e determined only by machining tolerances. These have b een kept well within the allowed image quality misalignment budget.

2.3 Thermal design
The vendor-sp ecied minimum guaranteed op erating temp erature for the FTT/NAS camera is condensing environment and its minimum survival temp erature is close to ambient temp erature but in any case not less than will b e insulated and heat removed from it. The co oling lo op intended for the electronics cabinets in the UT enclosure will b e used to remove heat from the FTT camera thermal enclosure. This solution has the distinct advantage that the uid that passes through the cold plates and then through the camera to remove the Peltier heat should never fall b elow

0 C

in a non-

-25 C



. Therefore for op erational and camera

safety reasons the FTT/NAS camera will b e lo cated within an enclosure in which the air will b e maintained

0 C

. To prevent overheating of the camera, and to

keep the surface temp eratures near the optical b eam within the required

2 C

of ambient, the camera enclosure

0 C

in normal

op eration. Consequently, there will b e no need for a variable co olant ow through the camera enclosure. The


Figure 2. Left: View of the FTT/NAS camera enclosure with the insulation panels removed.

The overall size of the

enclosure is 340 mm tall by 340 mm wide and 320 mm deep and has insulation panel thickness of 50 mm. The camera is surrounded on three sides by nned cold plates which are connected in series with the camera Peltier co oling connections. The lo ops of the cold pip es running through the cold plates are shown in section. Right: The FTT/NAS camera environmental control and monitoring scheme showing the connections b etween the camera enclosure lo cated on the Nasmyth table, the services interface panel lo cated close by, and the electronics interface lo cated with the FTT/NAS PC in the telescop e enclosure cabinet. Connections to/from the tip-tilt mirror controller, lo cated in the same electronics cabinet, are shown passing through the custom interface.

desired ow will b e set by a manually adjusted valve and monitored by a ow meter for safety and setting purp oses. The enclosure will b e constructed from rigid p olyiso cyanurate thermal insulation panels b onded to the outside of an aluminium frame formed from the cold plates used to remove heat from the enclosure. The outer surface of the insulation panels will b e covered with a protective skin and the whole assembly will b e supp orted from the optical table by low thermal conductivity spacers attached to the cold plate framework. The cold plates are purp ose designed and built so that the numb er of uid connections within the enclosure is minimised. Each plate will b e machined with straight channels to receive a nylon co oling tub e that passes through all of the cold plates in a single continuous run. Heat-sinks will b e connected to the plates to hold the tubing in place and force a go o d thermal connection to the cold plate. As can b e seen in Figure 2, the cold-plate assembly will b e mounted on a frame supp orted by four pillars b onded to a baseplate which is clamp ed to the table top. The frame, pillars and baseplate will b e made from a sp ecial low thermal conductivity glass bre comp osite and provide an eective thermal break. The camera will b e lo cated inside the enclosure on a mounting plate supp orted by pillars of low thermal conductivity glass bre comp osite that pass through the insulation to the camera mounting bracket outside. A tub e xed to the camera mounting plate will also pass through the insulation and on through a hole in the camera mount. A window will b e mounted so that air within the camera enclosure is prevented from escaping. The camera will b e xed only to the camera mount and will not b e in contact with the indep endently-supp orted enclosure. A dry air supply will b e connected to the enclosure to ensure that humidity levels are low. Thermal sensors will monitor the cold plate and enclosure air temp eratures and a humidity sensor will sense the enclosure air. A small heating element will b e mounted in the enclosure so that the internal temp erature can b e increased if it is to o cold for the camera to b e op erated.

2.4 Electronics design
The electronics for the FFT/NAS is lo cated in an electronics cabinet within the telescop e enclosure. The cabinet contains a rack-mount PC, which hosts the PCI cards for interfacing with the EMCCD camera and the Fast Tip-Tilt Actuator (FTTA) controller, and the following electronics and mo dules:


3U rack computer USB Ethernet Network USB

2U rack chassis I2C

Humidity

Temp

Temp

Labjack

Humidity I2C

Temp

Temp

Labjack

Flow rate Analog signal conditioning Enclosure heater Tip tilt mirror control Tip tilt mirror position

Control GUI Computer

Analog I/O

Logs Disk Logs Telescope steering correction Analysis GUI

Camera interface Camera power PCI bus

Camera power

Camera data/control

Figure 3. The prop osed software execution environment for the FTT/NAS, showing the rack-mount PC, the hardware devices interfaced to it, and network connections to external software comp onents.



The EMCCD Peltier p ower supply mo dule; Two Lab jack analogue/digital I/O mo dules; A custom electronics interface circuit b oard; A p ower supply for the custom electronics.

The electronics design is concerned mostly with interfacing the signals b etween the FTT/NAS PC and the FTTA servo controller but also with providing analogue inputs for the various sensors that are necessary to monitor the camera thermal enclosure and the services provided to it. Figure 2. For sensing and control low bandwidth interfacing will b e handled using two of the computer's USB p orts. Each p ort will connect with a Lab jack U3 to provide access to an I C bus and a variety of digital and analog input and output connections. One bus will handle a humidity sensor and three temp erature sensors asso ciated with the camera thermal enclosure and the other bus will handle sensors placed outside the enclosure and on the Nasmyth table. The analog inputs available on the Lab jack mo dules will connect via signal conditioning electronics on the custom circuit b oard to ow rate sensors used for monitoring the camera thermal enclosure co olant and dry air supplies. An analog output will control the camera thermal enclosure heater via an amplier mounted on the custom circuit b oard. A diagram of how the FTT/NAS electronics in the cabinet are interconnected with the camera and its services is shown in the right panel of

2

2.5 Software design
The software execution environment for the FTT/NAS is illustrated in Figure 3. The system software will run on a rack mounted Intel-style computer. The op erating system is exp ected to b e Linux 2.6.32 with Xenomai 2.6 kernel patches, but the source co de is compatible with more recent releases, including the forthcoming Xenomai 3. Besides the interfaces to the FTTA controller and environment sensors describ ed ab ove, the computer will have a Gigabit Ethernet connection to the MROI control ro om via a lo cal switch. This Ethernet link will b e the conduit for the interface with the Interferometer Sup ervisory System (ISS) and op erator GUIs.
2.5.1 Software architecture

The FTT/NAS software consists of two comp onents used to control the FTT/NAS hardware, a user interface software application, and a fourth comp onent used for oine visualization and analysis of previously-recorded monitor data.


The control software is partitioned into two comp onents b ecause of the need to maintain the environmental conditions within the FTT/NAS camera enclosure at all times, not just when the system is op erational. The comp onent resp onsible for controlling the camera environment and enabling/disabling the camera is called the environment controller, and the comp onent that p erforms the primary FTT/NAS functions such as target acquisition and fast tip-tilt correction is called the system controller. The environment controller will normally run continuously. The system controller runs when the interferometer is observing or preparing to observe. The system controller subscrib es to monitor data from the environment controller in order to detect whether the environment controller has given it p ermission to op erate the camera. In all other resp ects, the two systems function indep endently. The control/display GUI provides a graphical user interface for commanding the system and environment controllers and for live display of their monitor data (including camera images). It can also record monitor data to a set of FITS-format les when requested to do so by the op erator. The control/display GUI may b e used in one of two mo des: a standalone mo de in which it is fully functional, and a display-only mo de which can safely b e used when the FTT/NA system is under the control of the central ISS. The analysis GUI is used to visualize and pro cess monitor data previously recorded to FITS les by the control/display GUI. It provides a general diagnostic capability by means of functions for displaying image sequences, graphing scalar monitor data, and simple data analysis. The design of all comp onents of the FTT/NAS software is essentially complete. At the time of writing

(June 2012), preliminary versions of the system controller and control/display GUI have b een written and are undergoing testing. A precursor to the environment controller is complete and has b een used extensively for our opto-mechanical stability tests. Asp ects of the analysis GUI have b een successfully prototyp ed.
2.5.2 Software design and implementation

The op erating system is Xenomai ( ploited in this pro ject:

http://www.xenomai.org/

). It has two imp ortant prop erties that are ex-



It runs in hard real time. Consequently it reacts to hardware inputs and controls hardware outputs with minimal and predictable delays. This is a very useful prop erty for intro ducing a computational element to servo lo ops, as is needed in the FTT/NAS. It co exists with Linux. This means that the richness of the Linux system infrastructure is available to the software whenever hard real time p erformance is not required.

Both Linux and Xenomai have a user-space and a kernel space comp onent:



Kernel space contains the core system functionality, such as hardware interfaces and memory management. Co de running in kernel space has few restrictions on what it can do. User space is where the user interacts with the system. There are more restrictions on what co de is allowed to do in user space, but it is much harder for user space co de to crash the system. An interface is provided to allow applications to communicate across the barrier b etween user space and kernel space.

The system controller in particular makes extensive use of the user space and kernel space domains of b oth Xenomai and Linux. Further details of the design of the system controller and the other FTT/NAS software comp onents are given immediately b elow.

System controller

The system controller software p erforms acquisition functions, fast tip-tilt servo lo op This

closure and transmission of system diagnostic information (including camera images) over the network. software can b e broadly divided into two parts:



Co de that implements the real time fast tip-tilt servo lo op using Xenomai. This is implemented as one kernel space thread and one user space thread. Co de that implements supp orting functionality, including network control and monitoring, using several non-real-time threads in Linux.


FTT mirror Analog interface X

starlight

Camera

Y

Camera interface Control/status

X, Y Xenomai Kernel

Interrupt

Image data Linux kernel DMA buffer Andor kernel library Image data Control/ status Andor user space library Control/ status Application

Interrupt service routine

Analog driver

RT image

Find centroid Xenomai user space Buffer Find tip-tilt values

Figure 4. The implementation of the core servo lo op in Xenomai and its connection with non-real-time Linux co de.

The core function of the software is closure of the servo lo op b etween the EMCCD camera and the fast tip-tilt mirror. This must b e implemented within a hard real time op erating system in order to guarantee calculations meet deadlines on every servo cycle. Xenomai is used b ecause it is free, op en source, well integrated with Linux, and proven during development of the MROI delay line metrology system. The core software reads a camera image, calculates a centroid and then a correction, and sends the correction to the controller for the mirror. All this o ccurs in real time context. An interface is also provided to the remainder of the fast tip-tilt software, which runs in Linux. The implementation is illustrated in Figure 4. When the camera interface card reads out a camera image, it writes the data to computer memory using direct memory access (DMA) and then triggers an interrupt to notify the computer that the transfer is complete. This interrupt is intercepted by an interrupt service routine that has b een p orted to Xenomai from Andor's op en source driver, thereby providing real time access to image data while maintaining compatibility with the remainder of Andor's co de. The interrupt service routine copies the image data to a buer and blo cks a Xenomai user-space routine from reading the data until the copy is complete. When the user-space routine is allowed to read the data, it do es so immediately. It rstly reads the computer's NTP stabilised system clo ck to timestamp the data , then it calculates image centroid co ordinates, exploiting Xenomai's ability to use the computer's oating p oint hardware in real time context in user space. A p osition error is then calculated from the centroid, which is used as the error input for the fast tip-tilt servo. An appropriate correction is sent to a custom Xenomai driver for the analogue interface card, which applies scaled voltages to the FTTA controller inputs. Prototyp e centroiding and servo algorithms have b een tested in a simulation environment, and the results of these simulations suggest that a system with our estimated optical throughput (86% from dichroic to camera) and the eective read noise we have measured for the Andor iXon X3, can meet the 16th magnitude limiting sensitivity requirement in the sp ecied



0.7

seeing with an appropriate choice of closed-lo op bandwidth.

Meanwhile the image data and corrections are buered into an ordinary Linux application, where they are used for non-real-time tasks and logging. This buering serves as an interface b etween the Xenomai co de, which must meet strict deadlines, and the Linux co de, which only needs to b e able to keep up on average.


Xenomai V2.6 or later is needed to read the system clo ck in real time context and hence provide an accurate timestamp.

Linux user space

Image data


In addition to closing the fast tip-tilt servo lo op, the system controller software must also set up and initiate readout of the camera, send tracking ooads to the unit telescop e, execute commands and pro duce monitor data.

Interface (GSI)

3 used for communication with the ISS and an interface with the control/display GUI based on 4 messaging proto cols (dlmsg) develop ed in Cambridge for the MROI delay line control software. The system
controller will only accept commands that change the system state from one source, currently set via a ag at compile time.

Two communications interfaces are provided for monitoring and control.

These are the Generic System

Environment controller

The environment controller op erates indep endently from the system controller so It is a simple sensing application which reads and

that it can function even when the camera is not in use.

reacts to data on timescales of the order of one second. Hence the environment controller has no hard real time requirements and runs in ordinary Linux. Interfaces with the ISS and control/display GUI are provided in the same fashion as for the system controller. The application communicates with sensors and a heater via two USB-driven Lab jack U3 b oards. The sensors are either analogue voltage or I C typ es. The two Lab jacks provide two indep endent I C buses, which will b e

2

2

2 2 helpful if two I C devices are needed that have the same I C address.

Control/display GUI

The control/display GUI is a software application that provides an op erator interface

to the system controller and environment controller. The application is written in C and uses the GTK+ widget library. Images and scalar monitor data published by the systems are displayed to the op erator, and these The application can record the images and scalar data to The resulting les can b e read by the separate displays up date automatically as new data arrive.

a set of FITS les when requested to do so by the op erator. control/display GUI is also used to control the FTT system.

oine analysis GUI. If the system and environment controllers are not b eing directed by an ISS sup ervisor, the

Analysis GUI

The analysis GUI provides access to the FITS log les that can b e generated by the user of the

control/display GUI. The application can b e used to assess the images captured by the EMCCD camera during a particular recording or in plotting and analysis of guiding corrections. Other uses include:



The review and plotting of seeing conditions; The examination of image prole; The measurement of image scales and distances b etween ob jects in the eld of view; The review of the FTT/FLC system parameters.

The GUI is run under the Matlab environment. The main GUI deals with the selection and imp orting of FITS les or groups of recordings. Secondary windows may b e spawned to provide additional analysis functionality.

3. LABORATORY TESTS
The principal asp ects of the FTT/NAS that have b een tested exp erimentally to date are those related to camera p erformance, i.e. latency and functionality of readout, and opto-mechanical stability. Preliminary tests of the thermal p erformance of the camera enclosure are now complete (as of June 2012), and will b e followed by functional and p erformance tests of tip-tilt correction in closed-lo op. relevant requirements, test pro cedures and results obtained thus far. The following sub-sections outline the

3.1 Camera readout testing
As mentioned in Sec. 2 tests have already b een undertaken with an initial

23 в 23

pixel sub-array custom readout

provided by the EMCCD vendor. The results conrmed that the frame rate and latency requirements for the FTT/NAS could b e met with this small sub-frame with the camera op erating at high gain and with an eective readout noise of

0.25e-

.


Element Dichroic Fo cusing optic

Degree of freedom

Allo cation to global stability budget

Fold mirror # 1

Fold mirror # 2

Camera mount

x y x y z x y z x y z x y x y z z

0.047 0.045 0.47 µm 0.35 µm 140 µm 0.75 0.70 0.59 µm 0.090 0.049 0.31 µm 0.064 0.074 0.47 µm 0.35 µm 140 µm 2.32

Table 1. The stability budget allo cations for the selected optical layout for the FTT/NAS. In each case the optic must not move by more than this amount for a 5 C change in temp erature. The co-ordinate system used has the z-direction normal to the named optical comp onent, and the x- and y-directions p erp endicular to this. In all cases, the x-direction is p erp endicular to the surface of the Nasmyth optical table, and the gures represent displacements with resp ect to the symmetric expansion of the whole opto-mechanical layout.

Since then Andor have b een customising a larger (32

в 32

pixel) sub-array fast-readout mo de for us, and they

have conrmed that a 1 kHz frame rate is achievable, with approximately 1 ms latency b etween readout and pixel data arriving in RAM. This rough measurement is consistent with a mo del of the camera readout that we constructed in order to explain our own timing measurements of the preliminary

23 в 23

pixel mo de.

We have also measured the delay intro duced by the real-time software in resp onding to the interrupt generated by the Andor PCI card, and calculating and outputting a correction signal. For a dual core 3 GHz AMD Athlon I I computer, over many trials the maximum delay was measured to b e Putting these data together, we estimate that the latest of just over 43 Hz.

38 µs

.

32 в 32

pixel mo de will deliver a closed lo op bandwidth

3.2 Opto-mechanical testing
Our opto-mechanical testing is describ ed in more detail b elow. The rst stage of our testing fo cused on assessing to what extent our prop osed optical comp onent mounts were able to meet the stability requirements needed for the FTT/NAS in the presence of the temp erature changes exp ected at the Magdalena Ridge. More recently we have b een testing the overall optical stability of the full optical system in a so-called integrated test. This latter testing is not fully complete and so only preliminary results are presented here. Table 1 shows the error budget for the maximum allowable comp onent displacements that meet the top-level zero-p oint stability budget for our prop osed optical layout. These displacements are relative to the tip-tilt zero p oint (guiding centre) determined at the start of night by pro jecting an alignment source onto the FTT/NAS camera from the b eam combining lab oratory.
3.2.1 Individual comp onent testing

In order to test the thermal stability of the individual FTT/NAS comp onent mounts, a thermal enclosure was constructed within which each of the mounts (with the relevant optic attached) could b e intro duced. The chamb er was constructed of six cold plates, placed in pairs on the top and two side faces, which in conjunction with an aluminium baseplate made up a short tubular shell. The cold plates were supplied with temp erature-controlled water from a Grant RC350G chiller, which was placed in an adjacent lab oratory, and could b e used to either


warm up or co ol down the chamb er. The uid temp erature could b e controlled to roughly one degree Celsius, which proved more than adequate for our tests. By cycling the water temp erature according to a pre-determined schedule, the optical comp onent mounts could b e sub jected to environmental temp erature changes mimicking those that might o ccur during an observing night. This metho d of testing, however, is likely to have intro duced errors asso ciated with non-uniform co oling/warming of the mounts due to the forced air circulation in the chamb er (small fans were used) and p ossible uneven mixing of air in the chamb er. These inaccuracies are most likely to have o ccurred at the start of each co oling cycle, as aggressive co oling of the chamb er was taking place. The optical mounts were typically set up within the chamb er with p osition sensors attached such that any relative motion b etween the mount and the optic could b e determined. In addition, further sensors were used to monitor, for example, the chamb er temp erature, the temp eratures of dierent parts of each mount and the ambient temp erature outside the chamb er. The p osition sensors used were Linear Variable Displacement Transducers (LVDTs) which were typically able to measure displacements as small as 50 nm. Because LVDTs themselves exhibit a temp erature-dep endent resp onse, in all cases the measurements obtained with them were calibrated against known temp erature-dep endent motions (for example the expansion of a solid aluminium alloy blo ck) measured with identical sensor/mounting arrangements.

Mirror piston tests

A typical set of test results asso ciated with the piston stability of the dichroic/fold mirror

mounts, i.e. the extent to which changes in temp erature give rise to motion along the b eam propagation (z-) direction, is presented in Figure 5. This shows the uncalibrated z-motion of a fused silica mirror as a function of chamb er temp erature, in this case changing from roughly

13 C

to

23 C

, together with the uncalibrated

measurement of a reference blo ck. Also shown is the dierence in these two measurements: this represents the actual calibrated piston motion of the optic. This is very stable with temp erature, with only small variations of at most 100 nm p eak-to-p eak. This can b e compared with a piston stability requirement of roughly half a micron, i.e. a factor of ve larger, and conrms that the mounts are b ehaving as designed, easily meeting the FTT/NAS piston stability requirement.
1 24

22 LVDT reading / microns Temperature / degC

0 Data Temperature Calibration Temperature

20

18 Data -1 Calibration Difference 16

14

-2 0

0.5

1

1.5

2 2.5 Time / seconds

3

3.5

4

12 4.5 x 10
4

Figure 5. Raw (light and mid grey) and calibrated (black) piston uctuations measured in a typical dichroic/fold-mirror piston test. There is a small (quarter of a degree) dierence in the temp erature prole for the measurement and calibration exp eriments, with the temp erature in the test chamb er rising from roughly 13 C to 23 C in each case. Note that the raw uncalibrated measurements only show a variation that is a factor of a few greater than the allowed piston uctuations, and the calibrated measurements are a factor of ve smaller than the requirement.

Mirror tilt tests

A typical set of test results asso ciated with the tilt stability of the dichroic/fold mirror

mounts is presented in Figure 6. In this case the left hand panel shows the uncalibrated LVDT data in light grey, the LVDT calibration data in mid grey, and the dierence of these two, representing the actual tilt, in


black. The graph has b een scaled such that a linear displacement of 0.045 microns corresp onds to the stability requirement for the dichroic mirror mount (i.e. 45 milliarcseconds).
0.2 0.1 0 -0.1 -0.2 -0.3 Data -0.4 -0.5 0 Calibration run Difference 1 2 3 4 Time / seconds 5 6 x 10 7
4

0.1 0 -0.1 -0.2 -0.3 -0.4 Calibration run 1 -0.5 -0.6 0 Calibration run 2 Difference 0.5 1 1.5 2 2.5 3 3.5 4 4.5 x 10
4

LVDT reading / microns

LVDT reading / microns

Time / seconds

Figure 6. Left: measured (light and mid grey curves) and calibrated (black curve) LVDT tilt data expressed in microns for a dichroic/fold-mirror mount test. The allowed tilt variation expressed in these units for the temp erature change here is

45 nm

. Right: equivalent data for two calibration runs. In this case the recovered signal is exp ected to b e zero. The level

of residual uctuations in the black trace is indicative of the level b elow which calibration errors cannot b e eliminated. Overall, these data suggest that the prototyp e mount may b e compliant with its tilt stability requirement, but if it is not, it is unlikely to have exceeded it by more than a factor of two.

The data app ear to show a slowly decreasing trend with temp erature (this is again rising by roughly

10 C

over

the course of the ten-hour measurement time) of ab out 0.075 microns, roughly 70% greater than the FTT/NAS budget allows. In order to assess the reliability of this result, additional exp eriments were run where calibration data from a given exp eriment were calibrated using measurements from a calibration made on a dierent day. One such cross-calibrated dataset is shown in the right-hand panel of Figure 6. These data suggest that our test set-up is limited by calibration uncertainties to a level of no b etter than twice the angular shift allowed by the optical stability error budget. In summary, our exp erimental data demonstrate that the dichroic/fold mirror mounts may satisfy the tilt stability budget allo cation, and are very unlikely to have exceeded it by more than a factor of two.

Lens shear tests

The tolerances on the tilt and piston stability of the FTT/NAS lens mount are factors of

10100 times greater than for the dichroic/fold-mirror mounts, and so given that they share a common design approach, tests of these asp ects of the lens mount stability have not b een prioritised. Rather, it is the required half-micron stability (for

T = 5 C

) in x- and y-p osition of the lens, p erp endicular to the direction of b eam

propagation, that is likely to present the greatest design challenge. Our test for this motion utilised a pair of LVDT prob es mounted along a diameter of the lens and monitoring the centration of the lens within its aluminium mount while the chamb er temp erature was changed (see left panel of Figure 7). The right panel of Figure 7 shows the results for a lens shear test where the lens supp ort pins were delib erately fabricated with an incorrect length. Similar tests using other pairs of incorrectly fabricated pins allowed us to determine the correct length needed to give no lens shear as

16.1 ± 0.8 mm.

Quite separately

we measured the CTE of the pins in a lab oratory test set-up. These, and other rep eat data, were then used to indep endently compute the pin length needed to keep the lens centred. This gave a value of FTT/NAS requirements. A summary of the results of our comp onent mount tests is presented in Table 2.


16.5 ± 0.9 mm,

consistent with the LVDT predictions, and conrming that our material comp ensation strategy can meet the

In the case of lens shear, the value listed is the movement predicted assuming a 1 mm error in the comp ensating pin

length based on the movements seen with pins of other lengths.


Figure 7. Left: A face-on view of the triplet lens mount showing the LVDT measurement prob es in place and the p olymer pins that supp ort the lens in the radial direction. Right: FTT/NAS Lens shear (solid line) and chamb er temp erature (dashed) as a function of time during a thermal cycle test. In this example, the p olymer pins supp orting the lens were delib erately fabricated with the incorrect length and so the lens do es not remain centred in its mount but moves by roughly 1 µm/ C. Element Degree of freedom Dichroic/mirror mount Dichroic/mirror mount Lens mount Piston stability Tilt stability Shear stability Measured motion 100 nm Required stability Comments

100 nm 250 nm

< 500 nm 45 nm 350 nm

Compliant Compliant within factor Compliant 2

Table 2. Comparison of the FTT/NAS comp onent stability test results and the p erformance needed to meet the top-level requirements. The dichroic/fold mirror mount tilt stability requirement has b een converted into a linear measure.

.

3.2.2 Integrated testing

The fo cus of our most recent opto-mechanical testing has b een to attempt to validate the full optical train of the FTT/NAS. To this end a much larger thermal chamb er has b een constructed which allows for the full optics baseplate, with optics, to b e mounted on an optical table in our lab oratory and then b e temp erature cycled. The interfaces b etween the baseplate and the optical table are identical in concept to those designed for the FTT/NAS but have b een fabricated to a slightly lower level of precision. The design of the chamb er, depicted schematically in the left panel of Figure 8, and shown part assembled in the photograph in the right panel, allows for a 9 mm diameter collimated b eam from a HeNe laser to b e injected along the optical train and b e measured on exit from the optical system via a small op en p ort 3. This measurement is p erformed by a CCD camera mounted to the optical table externally and close to its b oundary. Similarly, the input b eam can also b e measured after (i) passing through the dichroic at lo cation R; (ii) passing through a semi-silvered fold mirror 1; or (iii) passing through a semi-silvered fold-mirror 2. The rationale for this approach is that, by simultaneously measuring the output b eam and, for example, the b eam exiting at p ort R, one can in principle remove the eects of any instabilities in the laser b eam injection direction from motions seen in the output b eam. Simultaneous measurement of the b eams exiting at the other calibration p orts may b e used to assess other typ es of systematic errors and can help identify any unstable comp onents. The typical sequence for a test was that the chamb er was held at ro om temp erature for several hours, and then chilled to roughly

10 C

b elow ambient over three hours, during which time the output b eam as well as one

of the calibration output b eams were monitored. Thereafter, the chamb er temp erature was allowed to return to the ambient level over a longer p erio d. A typical integrated test result is shown in Figure 9. This shows the calibrated exit b eam motion

measured in camera pixels, where motions of up to 0.3 pixels are allowed plotted as a function of time. The chamb er temp erature is shown by the dotted trace, and fell by roughly

4 C

over the course of the rst three

hours of the run. In these data the sp ot p osition variations have b een corrected for b oth the observed motions


3 2 1

R
Figure 8. Left: The optical arrangement used for the FTT/NAS integrated test. A collimated laser b eam enters from the right, part reects o the dichroic (at b ottom left), passes through the lens and then o the two semi-silvered fold mirrors b efore exiting at p ort 3. Ports R, 1 and 2 the b eam to b e interrogated after travelling successively longer p ortions of the optical train. Right: A view of the FTT/NAS optics mounted on their baseplate and enclosed by a partially assembled thermal chamb er.

of a calibration b eam exiting at p ort 1, and the shear in the vertical direction asso ciated with the material comp ensation of the lens mount. This is b ecause the camera and lens system b eing used to monitor the exit b eam motion is not sub ject to the temp erature changes exp erienced by the lens mount itself. The most noticeable features of Figure 9 are the extreme motions of the exit b eam direction at certain times during the exp eriment. Exhaustive tests have conrmed that these large amplitude motions are exclusively asso ciated with abrupt changes in the temp erature dierence b etween the top and b ottom skins of the optical table on which the tests are b eing undertaken. It app ears that at key times during the exp eriments, notably (a) when the co oling cycle is initiated and (b) when active co oling of the chamb er is stopp ed and the temp erature of the thermal enclosure is allowed to increase slowly back to ambient, the optical table undergo es a thermal sho ck and the calibration measurements are compromised. These times are identied by the horizontal arrows in Figure 9 and last no more than an hour.
0.4 0.3 Temperature Change / degC 0.2 0.1 0 - .1 0 - .2 0 - .3 0 - .4 0 - .5 0
5000 Time / s 10000 15000

5 4 3 Spot motion / px Temp change / degC 2 1 0 -1 -2 -3 -4 0 X motion Y motion Temp change

Table Top Table Bottom Chamber Temperature

2.3

2.4

2.5

2.6 2.7 Time / s

2.8

2.9

x 10

4

Figure 9. Left: Typical results from an FTT/NAS integrated test showing the calibrated motion in x- and y- for a laser b eam that has propagated the full optical train. The grey band indicates times at which the data-logger failed. In this exp eriment the chamb er temp erature has b een dropp ed by 4 C and the corresp onding allowed image motion is 0.3 pixels. The horizontal arrows show the times at which the largest excursions are seen, and during which the

T

b etween the top

and b ottom skins of the optical table jumps rapidly. Outside these p erio ds the observed b eam motion is approximately at the level of twice the FTT/NAS requirement. Right: The measured temp erature of the top and b ottom skins of the optical table during a warm-up. In these data the water chiller for the thermal chamb er was adjusted (at t which is matched by a much more gradual increase in the upp er skin temp erature.

25, 000

seconds)

so as to allow the chamb er to slowly rise to ambient conditions. Note the abrupt change in the lower skin temp erature


We conclude that our data have identied a warping of the top and b ottom optical table surfaces, and resulting changes in the p ositional and angular orientation of the laser b eam and the FTT/NAS baseplate (since the lo cations of its kinematic seats are altered). This eect is not fully comp ensated by our calibration strategy. We b elieve that there is also a contribution from warping of the baseplate due to a changing temp erature dierential b etween its upp er and lower surfaces. At other times in the course of the tests, when the dierential skin temp erature of the optical table is stable,
but importantly while the chamber temperature is changing by several degrees Celsius, the calibrated sp ot motion

is very stable. During these p erio ds the observed motion is approximately at the level of only twice the FTT/NAS requirement. This is consistent with the results of our individual comp onent mount tests, and indicative that we are very close to meeting the overall FTT/NAS zero-p oint stability requirement. We are intending to continue our integrated tests, with the goals of monitoring any table surface motions directly, reducing any co oling-induced temp erature sho cking of the optical table, and improving the temp erature equilibration of the common baseplate.

4. CONCLUSIONS
We have presented a comprehensive design for the FTT/NA system together with test results which allow us to predict the exp ected opto-mechanical stability and closed-lo op bandwidth. There are ambiguities in interpreting the integrated test results we have obtained thus far, but we b elieve the results are consistent with the comp onent test results which indicate that we will exceed the zero-p oint stability requirement by a factor of roughly 2. If left uncorrected, such zero-p oint drifts would have a small impact on the limiting sensitivity (equivalent to 6% throughput loss) and calibration accuracy of MROI. However we will correct these drifts using the continuous tilt/shear correction system that has now b een incorp orated into the MROI system design, and our measurements conrm that they are slow enough to b e fully corrected. Having passed a Preliminary Design Review in May 2012, the Cambridge team is pro ceeding with renement and manufacture of our current design. Pro curement and manufacture of the opto-mechanical comp onents will b e done in parallel with continued software development. These activities will come together in the setting up of a closed-lo op lab oratory demonstration that will correct articially-generated fast tip-tilt p erturbations. We exp ect to b egin the closed-lo op lab oratory tests within a 6 month timescale, and subsequently to commence integration activites on the Magdalena Ridge during 2013.

ACKNOWLEDGMENTS
The Magdalena Ridge Observatory (MRO) interferometer is hosted by the New Mexico Institute of Mining and Technology (NMIMT) in So corro, NM, USA, in collab oration with the University of Cambridge, UK. MRO is funded by Agreement No. N00173-01-2-C902 with the Naval Research Lab oratory and an institutional revenue b ond issued by NMIMT. We wish to also acknowledge funding for Cambridge University sta from STFC in the UK.

REFERENCES
[ 1 ] Creech-Eakman, M. J., Romero, V. D., Payne, I., Hani, C. A., Buscher, D. F., Farris, A. R., Jurgenson, C. A., Santoro, F. G., Selina, R. J., and Young, J. S., The Magdalena Ridge Observatory interferometer: a status up date, Proc. SPIE 8445 (2012). Pap er 8445-23, these pro ceedings. [ 2 ] Buscher, D. F., Bakker, E. J., Coleman, T. A., Creech-Eakman, M. J., Hani, C. A., Jurgenson, C. A., Klinglesmith, I I I, D. A., Parameswariah, C. B., and Young, J. S., The Magdalena Ridge Observatory Interferometer: a high-sensitivity imaging array, Proc. SPIE 6307, 11 (2006). [ 3 ] Farris, A., Klinglesmith I I I, D. A., Seamons, J., Torres, N., Buscher, D. F., and Young, J. S., Software architecture of the Magdalena Ridge Observatory Interferometer, Proc. SPIE 7740, 77400R (2010). [ 4 ] Young, J. S., Boysen, R. C., Buscher, D. F., Fisher, M., and Seneta, E. B., Software and control for the Magdalena Ridge Observatory interferometer delay lines, Proc. SPIE 7013, 70134C (2008).