1: \documentclass[12pt,preprint2]{aastex}

2: 

3: \usepackage{overpic}

4: \usepackage{multirow}

5: \bibliographystyle{apj}

6: 

7: \slugcomment{To appear in the Astrophysical Journal}

8: \shorttitle{BLAST}

9: \shortauthors{The Blast collaboration}

10: 

11: \newcommand{\degree}{\ensuremath{^\circ}}

12: \newcommand{\foot}{\ensuremath{^{\prime}}}

13: \newcommand{\inch}{\ensuremath{^{\prime\prime}}}

14: \newcommand{\blast}{{BLAST}}

15: \newcommand{\blastfs}{{BLAST03}}

16: \newcommand{\blastkir}{{BLAST05}}

17: \newcommand{\blastmcm}{{BLAST06}}

18: \newcommand{\qvec}[2]    {\left(\begin{array}{c}

19:                                 #1 \\

20:                                 #2

21:                         \end{array}\right)}

22: 

23: \begin{document}

24: 

25: \title{The Balloon-borne Large Aperture Submillimeter Telescope: \blast}

26: 

27: \author{E.~Pascale,\altaffilmark{1,2}

28: 	P.~A.~R.~Ade,\altaffilmark{2}

29: 	J.~J.~Bock,\altaffilmark{3,4}

30:         E.~L.~Chapin,\altaffilmark{5}

31:         J.~Chung,\altaffilmark{1,5}

32: 	M.~J.~Devlin,\altaffilmark{6}

33: 	S~Dicker,\altaffilmark{6}

34: 	M.~Griffin,\altaffilmark{2}

35: 	J.~O.~Gundersen,\altaffilmark{7}

36:         M.~Halpern,\altaffilmark{5}

37:         P.~C.~Hargrave,\altaffilmark{2}

38: 	D.~H.~Hughes,\altaffilmark{8}

39: 	J.~Klein,\altaffilmark{6}

40: 	C.~J.~MacTavish,\altaffilmark{1}

41: 	G.~Marsden,\altaffilmark{5}

42:         P.~G.~Martin,\altaffilmark{9,10}

43: 	T.~G.~Martin,\altaffilmark{1}

44: 	P.~Mauskopf,\altaffilmark{2}

45: 	C.~B.~Netterfield,\altaffilmark{1,10}

46:         L.~Olmi,\altaffilmark{11,12}

47: 	G.~Patanchon,\altaffilmark{5,13}

48: 	M.~Rex,\altaffilmark{6}

49:         D.~Scott,\altaffilmark{5}

50: 	C.~Semisch,\altaffilmark{6}

51: 	N.~Thomas,\altaffilmark{7}

52: 	M.~D.~P.~Truch,\altaffilmark{14}

53: 	C.~Tucker,\altaffilmark{2}

54:         G.~S.~Tucker,\altaffilmark{14}

55: 	M.~P.~Viero,\altaffilmark{10}

56: 	D.~V.~Wiebe\altaffilmark{1}}

57: 

58: 

59: 

60: 

61: 

62: \altaffiltext{1}{Department of Physics, University of Toronto, 60 St.~George Street, Toronto, ON M5S~1A7, Canada; {\url{enzo@physics.utoronto.ca}}}

63: 

64: \altaffiltext{2}{Department of Physics \& Astronomy, Cardiff University, 5 The Parade, Cardiff, CF24~3AA, UK}

65: 

66: \altaffiltext{3}{Jet Propulsion Laboratory, Pasadena, CA 91109-8099}

67: 

68: \altaffiltext{4}{Observational Cosmology, MS 59-33, California Institute of Technology, Pasadena, CA 91125}

69: 

70: \altaffiltext{5}{Department of Physics \& Astronomy, University of British Columbia,

71: 		 6224 Agricultural Road, Vancouver, BC V6T~1Z1, Canada}

72: 

73: \altaffiltext{6}{Department of Physics and Astronomy, University of Pennsylvania, 209 South 33rd Street, Philadelphia, PA 19104}

74: 

75: \altaffiltext{7}{Department of Physics, University of Miami, 1320 Campo Sano Drive, Coral Gables, FL 33146}

76: 

77: \altaffiltext{8}{Instituto Nacional de Astrof\'isica \'Optica y Electr\'onica (INAOE), Aptdo. Postal 51 y 72000 Puebla, Mexico}

78: 

79: \altaffiltext{9}{Canadian Institute for Theoretical Astrophysics, University of Toronto, 60 St.~George Street, Toronto, ON M5S~3H8, Canada}

80: 

81: \altaffiltext{10}{Department of Astronomy \& Astrophysics, University of Toronto, 50 St.~George Street, Toronto, ON M5S~3H4, Canada}

82: 

83: \altaffiltext{11}{Istituto di Radioastronomia, Largo E. Fermi 5, I-50125, Firenze, Italy}

84: 

85: \altaffiltext{12}{University of Puerto Rico, Rio Piedras Campus, Physics Dept., Box 23343, UPR station, San Juan, Puerto Rico}

86: 

87: \altaffiltext{13}{Laboratoire APC, 10, rue Alice Domon et L{\'e}onie Duquet 75205 Paris, France}

88: 

89: \altaffiltext{14}{Department of Physics, Brown University, 182 Hope Street, Providence, RI 02912}

90: 

91: 

92: 

93: 

94: 

95: 

96: \begin{abstract}

97: The Balloon-borne Large Aperture Submillimeter Telescope (\blast) is a

98: sub-orbital surveying experiment designed to study the evolutionary

99: history and processes of star formation in local galaxies (including

100: the Milky Way) and galaxies at cosmological distances. The BLAST

101: continuum camera, which consists of 270 detectors distributed between

102: 3 arrays, observes simultaneously in broad-band (30\%)

103: spectral-windows at 250, 350, and 500\,\micron. The optical design is

104: based on a 2\,m diameter telescope, providing a diffraction-limited

105: resolution of 30\arcsec\ at 250\,\micron. The gondola pointing system

106: enables raster mapping of arbitrary geometry, with a repeatable

107: positional accuracy of ${\sim}$\,30\arcsec; post-flight pointing

108: reconstruction to $\lesssim 5$\arcsec\ rms is achieved.  The on-board

109: telescope control software permits autonomous execution of a

110: pre-selected set of maps, with the option of manual override.  In this

111: paper we describe the primary characteristics and measured in-flight

112: performance of \blast\@. \blast\ performed a test-flight in 2003 and

113: has since made two scientifically productive long-duration balloon

114: flights: a 100-hour flight from ESRANGE (Kiruna), Sweden to Victoria

115: Island, northern Canada in June 2005; and a 250-hour,

116: circumpolar-flight from McMurdo Station, Antarctica, in December 2006.

117: \end{abstract}

118: 

119: 

120: 

121: 

122: \keywords{submillimeter --- galaxies: evolution --- stars: formation --- instrumentation: miscellaneous --- balloons}

123: 

124: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

125: \section{Introduction}     \label{sec:intro}

126: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

127: 

128: We have built and flown a balloon-borne submillimeter observatory,

129: designed to study the star formation in our Galaxy, in nearby resolved

130: galaxies, and in high-redshift starburst galaxies by observing at

131: several hundred microns, a wavelength range which is not available

132: from the ground.

133: 

134: The Balloon-borne Large Aperture Submillimeter Telescope (\blast) is a

135: stratospheric balloon-borne 2\,m telescope which observes the sky with

136: bolometric detectors operating in three 30\% wide bands at 250, 350,

137: and 500\,\micron.  The diffraction-limited optics are designed to

138: provide \blast\ with a resolution of $30\arcsec$, $42\arcsec$, and

139: $60\arcsec$ at the three wavebands, respectively.  The detectors and

140: cold optics are adapted from those to be used on the SPIRE instrument

141: on {\sl Herschel} \citep{grif03}.

142: 

143: \blast\ addresses important Galactic and cosmological questions

144: regarding the formation and evolution of stars, galaxies and clusters

145: \citep{devlin04} by providing the first large-area

146: (${\sim}$\,0.8--200\,deg$^2$) surveys of unique spectral-coverage,

147: angular resolution and sensitivity.

148: 

149: The primary scientific goals of \blast\ are: (i) to conduct

150: confusion-limited and shallower wide-area extragalactic surveys to

151: constrain the redshift-distribution, star formation history, and

152: evolution of optically-obscured luminous galaxies by measuring

153: photometric-redshifts \citep[derived from the \blast\ colors and

154: complementary data,][]{hughes02}; (ii) to study the spatial clustering

155: of this population; (iii) to improve our understanding of the earliest

156: stages of star formation by determining the physical properties and

157: mass-function of cold pre-stellar cores and the efficiency of star

158: formation within different Galactic environments; and (iv) to

159: investigate the nature and structure of the interstellar medium by

160: making the highest resolution maps to date of diffuse Galactic

161: emission at these wavelengths.

162: 

163: 

164: The \blast\ bands bracket the peak of the thermal radiation emitted by

165: dust at temperatures of around 10\,K\@.  Assuming that the temperature

166: range of dust in submillimeter galaxies is ${\sim}$\,30--60\,K,

167: \blast\ can explore redshifts from 1 to 5 \citep{hughes02}.  The

168: primary advantage of \blast\ over existing submillimeter bolometer

169: arrays such as {SCUBA} and {SHARC} is its enhanced sensitivity at

170: wavelengths $\le 500$\,\micron\ (where the far-IR background peaks) due

171: to the increased atmospheric transmission at balloon altitudes. With

172: sensitivities of 250\,mJy\,s$^{1/2}$, the \blast\ mapping speed is 10

173: times faster than the design goal for {SCUBA}\,2 at 500\,\micron\ 

174: \citep{holland06} and more than 100 times the mapping speed of the

175: pioneering flights of the FIR balloon-borne telescope {PRONAOS}

176: \citep{lam98}.  \blast\ is complemented at shorter wavelengths by

177: surveys from {\it IRAS}, {\it ISO}, {\it Spitzer} and {\it Akari}, and

178: at longer wavelengths by SCUBA~2, LMT, and ALMA, to constrain spectral

179: energy distributions\@. {\it Spitzer\/}'s higher resolution will also

180: provide accurate astrometry for many of the \blast\ sources.  \blast\

181: complements the {\it Herschel}\ satellite by testing detectors and

182: filters which are similar to those to be used in the SPIRE instrument

183: \citep{grif03}. Furthermore, the results from \blast\ will be

184: available early enough to influence the design (depth and area) of

185: future {SPIRE} surveys.  In addition, \blast\ will complement the

186: large-area spectroscopic Galactic surveys of {\it SWAS}, provide

187: submillimeter targets for the Fabry-Perot spectrograph {SAFIRE} on

188: {\it SOFIA}, and impact the design of the scientific case for the next

189: generation of submillimeter surveys from space (e.g.\ {\it SPICA} and

190: {\it SAFIR}, and later {\it SPIRIT} and {\it SPECS}).

191: 

192: \blast\ has made three flights to date. A 24-hour test flight

193: (\blastfs) from Fort Sumner, NM, in September 2003 demonstrated the

194: performance of the instrument sub-systems.

195: 

196: \blast\ made its first science flight (\blastkir) in June of 2005,

197: flying on a $1.1 \times 10^6$\,m$^3$ balloon from the Swedish Space

198: Corporation base of ESRANGE, near Kiruna, Sweden to Victoria Island in

199: northern Canada.  During this 4.5-day flight at an average altitude of

200: 38\,km, the instrument performed well, except for degraded optical

201: performance. This was possibly due to a failure of structural elements

202: in the carbon fiber mirror. The result was decreased resolution, which

203: had a significant impact on our ability to achieve our extragalactic

204: science goals\@.  \blast\ acquired 100 hours of data on Galactic

205: targets, providing some of the first arcminute resolution images at

206: these wavelengths \citep{Chapin2008}.  In this flight, surveys of five

207: star forming regions were conducted, including the well-studied

208: Cygnus-X field, three regions of intermediate/high-velocity cirrus,

209: the Cassiopeia-A supernova remnant, and several individual bright

210: targets.

211: 

212: \blast\ made its second science flight (\blastmcm) in December, 2006,

213: flying from the Williams Field Long Duration Balloon (LDB) facility

214: near McMurdo Station in Antarctica.  In this flight, the instrument

215: met its performance goals in resolution, sensitivity, and pointing.

216: The flight was terminated after 250 hours of data were acquired.

217: Unfortunately, an anomaly with the parachute separation system

218: prevented the parachute from being released from the gondola after

219: landing.  The parachute dragged \blast\ for 200\,km before it came to

220: rest in a crevasse field on the Antarctic Plateau.  While most of the

221: instrument was destroyed, the hard drives containing the data were

222: found nearby, and the complete dataset recovered.  In this flight,

223: \blast\ conducted shallow (10\,deg$^2$) and deep, confusion-limited

224: (0.8\,deg$^2$) extragalactic surveys of the Chandra Deep Field South

225: (CDF-S) as well as a 10\,deg$^2$ region near the South Ecliptic Pole.

226: A Galactic survey in Vela mapped 52\,deg$^2$ and 200\,deg$^2$ of sky

227: in deep and shallow observations, respectively.  Surveys of several

228: Galactic and extragalactic selected targets were also conducted.

229: 

230: The telescope and optics are described in Section~\ref{sec:telescope},

231: while Sections~\ref{sec:rec}

232: and~\ref{sec:cryo} focus on detector arrays and

233: cryogenics.  Sections~\ref{sec:gondola} through \ref{sec:psens} provide detailed information regarding the

234: gondola, command and control, and pointing systems.  Sections \ref{sec:thermal} through \ref{sec:performance} provide

235: overviews of the unique thermal and power requirements, and a

236: description of the pointing performance.

237: 

238: 

239: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

240: \section{Telescope} \label{sec:telescope}

241: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

242: 

243: The basic \blast\ optical configuration is shown in

244: Figure~\ref{fig:zeemax_design}\@.  Incoming radiation is collected by

245: the telescope (M1 and M2 in the figure) which is located on the inner

246: frame of the gondola (described in Section~\ref{sec:gondola}), along

247: with the cryogenic receiver (Section \ref{sec:cryo})\@. The telescope

248: focus is located about 20\,cm behind the optical surface of the

249: primary.  Cold (1.5\,K) re-imaging optics (M3, M4, and M5) in the

250: receiver relay the sky image to the detector focal planes. A baffled

251: Lyot stop (M4) is located at the position of an image of the primary

252: mirror.  This blocks stray radiation due to scattering and

253: diffraction.  After M5 the radiation is divided into three spectral

254: bands through the use of two low-pass edge dichroic filters

255: \citep{ade06}.  An aperture at the center of the Lyot stop is made to

256: match the one at the center of the primary mirror to reduce loading.

257: This same hole accommodates a commandable thermal source, which

258: provides a stable and repeatable optical signal (calibration lamp) to

259: monitor detector responsivity drifts \citep{hargrave06}.  The optical

260: design is optimized to deliver a diffraction-limited image of the sky

261: at the detector focal planes, with 60 and 30\arcsec\ beams at

262: 500 and 250\,\micron, respectively. The telescope designs of

263: \blastkir\ and \blastmcm\ are discussed in this section and the

264: relevant parameters can be found in Table~\ref{table:blast_optics}.

265: \begin{figure*}

266:         \centering

267: 	\includegraphics[width=0.8\linewidth]{f1.eps}

268: \caption{The optical layout of the \blastkir\ telescope and receiver

269: is shown on the left and the 1.5\,K optics, located within the

270: cryostat, are shown in expanded view on the right.  The image of the

271: sky formed at the input aperture is re-imaged onto the bolometer

272: detector array at the focal plane.  The mirror M4 serves as a Lyot

273: stop defining the illumination of the primary mirror for each element

274: of the bolometer array.  The three wavelength bands are separated by a

275: pair of dichroic beamsplitters (not shown) which are arranged in a

276: direction perpendicular to the plane, between M5 and the focal plane.}

277: \label{fig:zeemax_design}

278: \end{figure*}

279: 

280: \subsection{Focal Plane}

281: 

282: The illumination of the Lyot stop and the main reflector depends

283: primarily on the properties of the feed-horns in the detector array.

284: \blast\ uses $2f\lambda$ spaced, smooth-walled conical feeds which are

285: similar to the SPIRE feeds. Their design and measured characteristics

286: have been described by~\citet{chatt03}, \citet{rownd03}, and

287: \citet{griffin02}.  The optical system design is based upon the

288: feed-horn nominal $f/5$ focal ratio and requires the chief-rays to be

289: perpendicular to the aperture of the horn, with all the beams

290: overlapping at the Lyot stop.  Knowledge of the fundamental propagated

291: electromagnetic mode (TE$_{11}$) is used in a

292: Zemax\footnote{\url{http://www.zemax.com/}} Physical Optics module to verify

293: the final design.

294: 

295: \subsection{\blastkir\ Optical Design}	

296: 

297: For the 2005 flight, the design incorporated a 2\,m carbon fiber

298: spherical primary mirror with a mass of 32\,kg and a surface rms of

299: 2.4\,\micron.  It was designed and built by Composite Optics

300: Incorporated\footnote{Composite Optics Incorporated (COI), 9617

301: Distribution Avenue, San Diego, CA 92121.} as a technology prototype

302: for the 3.5\,m {\em Herschel\/} telescope.  An aluminum correcting

303: secondary mirror, with a diameter of 50\,cm, was designed to give a

304: diffraction-limited performance over a $14'\times7'$ field of view

305: (FOV) at the telescope focus.  The estimated antenna efficiency is

306: $\ge 80\%$; losses come from a combination of the rms surface

307: roughness of the primary and the quality of the re-imaging optics. Two

308: of the re-imaging cold elements, M3 and M5, are ellipsoidal mirrors;

309: M2 and the other cold element, M4, compensate for most of the

310: spherical aberration of the primary, whereas M3 and M5 contribute to

311: achieve the correct focal ratio.  The perimeter of the primary mirror

312: was structurally weak, forcing M2 to be suspended by a carbon fiber

313: structure attached through the hole in the primary mirror.  This

314: geometry resulted in ${\sim}$\,12\% blockage of the beam.

315: A detailed discussion  about \blastkir\ optical design and optimization can be found in~\citet{olmi01, olmi02}. 

316: %

317: The 2\,m primary mirror was

318: under-illuminated by setting the entrance pupil diameter to 188\,cm

319: and the reflector was strongly tapered, resulting in a field taper of

320: about $-15$\,dB. A Strehl ratio $> 0.96$ was achieved over the whole focal

321: plane, and far field beams had a first sidelobe at $-10$\,dB. A detailed analysis of the relation  between Strehl ratio and aperture efficiency, which is more commonly quoted for filled-aperture antennas, can be found in \citet{olmi07}.

322: 

323: \subsection{\blastmcm\ Optical Design}

324: In the 2006 Antarctic flight, \blast\ flew a Ritchey-Chr\'etien

325: telescope, with a 1.8\,m diameter, aluminum primary mirror with a mass

326: of 114\,kg.  Originally, the mirror was intended to be used only for

327: the test flight in 2003.  However, after the destruction of the carbon

328: fiber mirror on landing of \blastkir, the aluminum mirror was

329: re-machined to improve the quality of the surface from 8\,\micron\ to

330: better than 4\,\micron\ rms\@.  This machining was performed by the

331: Precision Engineering Group at Lawrence Livermore National Laboratory.

332: A 40\,cm diameter aluminum secondary was suspended by four carbon

333: fiber struts\footnote{ Innovative Composite Engineering, 139

334:   E. Columbia River Way, Bingen, WA 98605.}  (with a zero linear

335: thermal expansion coefficient) attached to the perimeter of the

336: primary mirror. This reduced the obscuration to ${\sim}$\,7\% compared to

337: the \blastkir\ design.  Simulations of in--flight detector loading

338: have shown a reduction of $\gtrsim 40$\,\% for this solution,

339: relative  to the  \blastkir\ configuration. 

340: % 

341: % 

342: In an attempt to further reduce loading, angled reflecting baffles

343: were installed on the inner surface of each strut to deflect most of

344: the obscured beam to the relatively cold sky.

345: % 

346: % 

347: % 

348: The cold re-imaging optics form an ideal Offner relay. In this

349: configuration, M3, M4, and M5 are all spherical and share a common

350: center of curvature; M3 and M5 are concave while M4 is convex. In order

351: to achieve a diffraction-limited beam of 60\arcsec\ at 500\,\micron\ and

352: 30\arcsec\ at 250\,\micron\ with a smaller diameter primary compared to

353: \blastkir, M1 was more aggressively illuminated. This was made possible

354: by reducing the diameter of the Lyot stop by about 12\%, leading to a

355: field taper of $-7.5$\,dB on M1 and $-17$\,dB sidelobes in the far field

356: beams.

357: 

358: \subsection{\blastmcm\ Focussing System}

359: 

360: The relative distance between the primary and the secondary mirrors

361: has a tolerance of one wavelength before significant image degradation

362: is introduced.  Thermal modeling indicated that diurnal temperature

363: fluctuations at balloon altitudes of the aluminum M1 and M2 mirrors

364: could have been as large as 10\degree\/C. The required correction in

365: the relative distance between M1 and M2 was calculated to be

366: 50\,\micron/\degree\/C.

367: 

368: To correct for the thermal motion, we implemented a motorized system

369: for in-flight refocusing of the secondary.  M2 was coupled to its

370: mounting structure via 17-7PH stainless steel leaf springs. Three

371: stepper motor actuators\footnote{Ultra Motion, 22355 Route 48, \#21,

372: Cutchogue, NY 11935.} (2.12\,mm\,rev$^{-1}$ lead screw) drive the

373: mirror back and forth to correct for the focus position, as well as

374: adding tip/tilt capability to initially align M2 to M1.  Accurate

375: positioning was achieved with differential optical encoders ($\simeq$1

376: count \micron$^{-1}$ resolution) factory mounted to the motors, and

377: three DC Linear Variable Differential Transformers\footnote{Macro

378: Sensors, 7300 US Route 130 North, Building 22, Pennsauken, NJ 08110,

379: 0.6\,\micron\ or better \mbox{repeatability}.} (LVDT)\@.  This system

380: allowed for $\pm$5\,mm of motion about the nominal telescope focus,

381: which was sufficient for accurate positioning.  The flight computer

382: commands each motor individually by communicating with the stepper

383: motor controllers over a RS485 bus.

384: 

385: Pre-flight focusing of the system was hampered by the fact that water

386: vapor absorption makes ground-based submillimeter observations in the

387: far-field of the telescope (about 10\,km) impossible.  To mitigate

388: this, the secondary mirror was physically offset to 3 positions so

389: that the focal plane was reimaged at 50, 100, and 150\,m from the

390: telescope.  The focusing system was then used to fine-tune the focus

391: at each position.  These results were extrapolated to determine the

392: far-field focus for the flight.

393: 

394: In flight, the focus of the system was verified and had to be manually

395: adjusted by maximizing the response when scanning over bright point

396: sources: a correction of 300\,\micron\ to the position set at launch

397: was required.  Subsequently, the temperatures of the primary and

398: secondary were monitored and the secondary automatically repositioned

399: to account for thermal variations that would produce a focus

400: displacement of 100\,\micron\ or more. The primary and secondary

401: mirror diurnal temperature fluctuations were $\pm1.5^\circ$C for most

402: of the flight.

403: 

404: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

405: \subsection{Array Bandpass Characteristics}

406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

407: 

408: Low-pass edge dichroic filters~\citep{ade06} split the incoming

409: radiation emerging from M5.  The first dichroic filter reflects

410: wavelengths shorter than 300\,\micron\ and transmits longer

411: wavelengths.  The bandpass for the 250\,\micron\ array is further

412: defined by a filter directly in front of the array, which reflects

413: wavelengths shorter than 215\,\micron, and by the waveguide cutoff at

414: the exit of each of the feedhorns.  For the 350\,\micron\ and

415: 500\,\micron\ arrays, in each case, the band is defined at the short

416: wavelength end by the dichroic transmission and at the long

417: wavelength end by the waveguide cutoff. Each band has a 30\% width.

418: The filter stack frequency performance was evaluated with Fourier

419: Transform Spectroscopy. Bandpasses are shown in

420: Figure~\ref{fig:band_pass}.

421: \begin{figure*}

422:         \centering

423: 	\includegraphics[angle=-90, width=\linewidth]{f2.eps}

424: 	\caption{Relative spectral response of the three channels \blast\ channels. The \blastkir\ bands (top) have been measured with the filter stack in the cryostat. The \blastmcm\ (bottom)  bands are obtained by multiplying the profiles measured for each filter individually; the feed waveguide cut--off is assumed at the longer wavelength. A Fourier Transform Spectrometer has been used in both cases.}

425: \label{fig:band_pass}

426: \end{figure*}

427: 

428: 

429: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

430: \section{Receiver} \label{sec:rec}

431: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

432: 

433: The \blast\ focal plane consists of arrays of 149, 88, and 43

434: detectors at 250, 350, and 500\,\micron, respectively.  The arrays are

435: cooled to a temperature of $270$\,mK.  Each array element is a silicon

436: nitride micromesh ``spider web" bolometer \citep{bock96, bock98}.  The

437: detector arrays, the feeds, and the mounting scheme are based on the

438: SPIRE instrument design \citep{turner01}.  Each array hosts a number

439: of diagnostic channels: two dark bolometers (channels that have been

440: capped to avoid illumination), two thermistors, and one resistor.  The

441: detector parameters are summarized in Table~\ref{table:det_params}.

442: 

443: 

444: The detectors and the readout electronics are designed so that the

445: sensitivity is ultimately limited by the photon ``BLIP'' noise. To

446: achieve this performance, other contributions to the bolometer noise

447: have to be controlled. A bolometer is fundamentally limited by phonon

448: noise in the thermal link with the heat sink. The Noise Equivalent

449: Power (NEP) \citep{bock96} is NEP~$= \gamma\sqrt{4k_{\mbox{B}} T^2

450: G}$, where $G$ is the thermal conductance, $T$ is the bath temperature

451: and $\gamma$ takes into account the properties of the thermal link

452: between the detector and the bath. For a given background load $Q$,

453: the maximum sensitivity is achieved when

454: $G \sim Q/T$~\citep{mather_1984}.  The detectors have been

455: optimized for loads of 55, 40, and 30 pW at 250, 350, and

456: 500\,\micron, respectively.

457: 

458: The detector transfer function is measured using cosmic ray hits, as described

459: in~\citet{crill03}. We have found typical time constants of $2$\,ms with a dispersion of 15\% across the arrays.

460: 

461: An additional advantage to the ``spider web'' bolometer

462: is the relatively small cross-section to cosmic rays.  Balloon flights

463: at high-latitudes have about 100 times the mid-latitude cosmic ray

464: flux. The detected cosmic ray flux was 1.5, 2, and 6 events per bolometer per minute during both science  flights at 

465: 250, 350, and 500\,\micron, respectively.

466: 

467: 

468: \subsection{Readout Electronics}

469: 	

470: A schematic diagram of the readout electronics is shown in

471: Figure~\ref{fig:det_ro}.  Each bolometer is pre-amplified with a

472: Siliconix U401 differential JFET with 5--7\,nV\,Hz$^{-1/2}$ noise at

473: $\nu > 100\,\mbox{Hz}$. Compact 24 channel JFET modules are integrated

474: into the design of the cryostat, allowing us to sink their power

475: dissipation ($240\,\mu$W per pair) to the vapor-cooled shield (see

476: Section \ref{sec:cryo}) and decrease the load on the helium bath by

477: approximately 60\,mW. The JFETs are operated at a temperature of

478: ${\sim}$\,145\,K.  Their output is then amplified using the Analog

479: AD624 instrumentation amplifier (5\,nV\,Hz$^{-1/2}$) with a 100\,Hz

480: wide band-pass filter, centered at 208\,Hz\@.

481: \begin{figure*}

482:         \centering

483: 	\includegraphics[angle=-90, width=\linewidth]{f3.eps}

484: 	\caption{A schematic diagram of the detector readout

485: 	electronics. A 200.32\,Hz sine wave biases each bolometer

486: 	across two $7$\,M$\Omega$ load resistors (RL). JFET pairs

487: 	buffer the detector signal to the warm electronics. The two

488: 	signals are then differenced and digitized. A digital lock-in

489: 	rectifies and filters the output sending it to the flight

490: 	computers over the BLASTbus.}

491: 	\label{fig:det_ro}

492: \end{figure*}

493: %

494: To mitigate the effects of $1/f$ noise, the detectors are AC biased with a sine

495: wave at 200.32\,Hz. A reference square wave at this frequency is

496: generated by dividing down the main 32\,MHz clock serving the readout

497: electronics, making the bias synchronous with the sampling. This

498: signal is then split into three paths (one per band), stabilized to a

499: voltage amplitude that is selectable from 0 to 300\,mV with a resolution

500: of 7 bits, and filtered into a sine wave using low-noise operational

501: amplifiers (Analog OP470). The bias is delivered to the detectors and

502: a reference is sent back to the Data Acquisition System (DAS) where

503: a digital lock-in is implemented on the DSP (Analog ADSP21062) in the

504: readout electronics. The bolometric AC signals and the bias reference

505: are digitized with a 24\,bit $\Sigma\Delta$ (Burr-Brown ADS1252) fed by

506: a low-noise dual instrumental amplifier (Burr-Brown INA2128) and

507: sampled at 10\,kHz. The reference is phase-locked and a numerical sine

508: wave is generated in-phase at the bias frequency (the clock is

509: shared). The rectified signal is then low-passed with a 4 stage

510: boxcar filter having a first null at 50.08\,Hz, and decimated to match the

511: DAS sampling frequency (100.16\,Hz).

512:  

513: The resulting flat-phase filter is well approximated by a Gaussian

514: and is computationally efficient in the time domain, allowing up to 25

515: independent channels to be locked-in on a single DSP.  Only the real

516: part of the signal is sampled, hence the relative phase between signal

517: and reference must be set manually. Since the phase difference is a

518: function of the bolometer impedance, it needs to be adjusted in flight

519: along with bias levels.  This is achieved by maximizing the signal

520: from the pulsed calibration lamp at the Lyot stop.

521: 

522: The overall data acquisition electronics noise is kept below the

523: estimated photon noise and  

524: provides  stability to low frequencies ($<30$\,mHz), which allows the

525: sky to be observed 

526: in a slowly-scanned mode. 

527: 

528: \subsection{Detector Noise}

529: 

530: The instrument is designed to be limited by photon noise.  The error

531: budget, which drove the telescope and readout electronics design,

532: accounted for (i) the photon noise arising from astrophysical

533: backgrounds and emissivity of the optical surfaces, (ii) the bolometer

534: noise budget, and (iii) the readout noise. A Noise Equivalent Flux

535: Density (NEFD) of 250\,mJy\,s$^{1/2}$ was expected from this analysis.

536: 

537: The voltage noise is determined by taking the Power Spectral Density

538: (PSD) of the time stream acquired in flight and it is shown in

539: Figure~\ref{fig:noise} for a representative 250\,\micron\ detector

540: during the \blastkir\ flight. Before taking the PSD, each time stream

541: is processed as described in \citet{truch2007} and

542: \citet{patanchon2007}, which involves deglitching, deconvolution of

543: the bolometer and electronics transfer functions, and removal of

544: common mode signals, synchronous to each detector in a array.

545: 

546: The voltage noise is well described by a brown noise model. The white

547: noise level is 11\,nV\,Hz$^{-1/2}$ and the 1/f ``knee'' is at

548: ${\sim}$\,35\,mHz. Using the responsivities in \citet{truch2007}, we

549: obtain a noise of ${\sim}$\,15\,MJy\,sr$^{-1}$\,Hz$^{-1/2}$. A similar

550: analysis at the longer wavelengths gives ${\sim}$\,7 and

551: 4\,MJy\,sr$^{-1}$\,Hz$^{-1/2}$ at $350$ and 500\,\micron,

552: respectively.  Within a single array the pixel noise varies by

553: 20\%. 

554: 

555: Measured noise levels are consistent with the design target for the NEFD

556: for nominal beam sizes.

557: 

558: \begin{figure*}

559:         \centering

560:         \includegraphics[angle=-90, width=\linewidth]

561:         {f4.eps} 

562: 

563: 	\caption{A 250\,\micron\ detector noise power spectrum density (PSD).

564: The time stream was acquired during the \blastkir\ flight and analyzed as explained in \citet{patanchon2007}:  (i) the time stream has been deglitched, (ii) the detector and electronics transfer functions have been deconvolved from the data, and (iii) a common mode signal, synchronous to each  channel in a array, has been removed in the time domain.

565: The PSD is shown before any gain. The dashed line is a brown noise fit to the PSD which shows a white noise level at 11\,nV\,Hz$^{-1/2}$ and a 1/f ``knee'' at 35\,mHz (indicated by the vertical dashed line).}

566: \label{fig:noise}

567: \end{figure*}

568: 

569: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

570: \section{Cryogenics} \label{sec:cryo}

571: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

572: %CAN: overview

573: 

574: The cryostat houses most of the receiver system (Figure

575: \ref{cryostat}) and was fabricated by Precision

576: Cryogenics.\footnote{\url{http://www.precisioncryo.com/}}  It is constructed

577: of aluminum and G10 woven fiberglass reinforced resin epoxy and is

578: maintained at vacuum to prevent thermal convection.  Each thermal

579: stage is light-tight except the optical path, which is thermally

580: protected with infrared blocking filters \citep{truch-thesis}.

581: \begin{figure*}

582:         \centering

583: 	\includegraphics[width=0.9\linewidth]{f5.eps}

584: 	\caption{Cutaway model of the \blast\ receiver showing the

585: 	optics box.  The $^3$He refrigerator is omitted for clarity.

586: 	The cryostat is held to the telescope structure via bolts in 

587: 	the top flange and jack screws around the perimeter near the base.}

588: 	\label{cryostat}

589: \end{figure*}

590: 

591: Liquid baths of nitrogen and helium maintain the temperatures of the

592: 77\,K and 4.2\,K stages.  In flight, an absolute pressure

593: regulator\footnote{Tavco, Inc, 20500 Prairie St., Chatsworth, CA.} is

594: used to maintain approximately 1\,atm above each bath.  A vapor-cooled

595: shield (VCS) is located between the nitrogen and helium stages which

596: is cooled by boil-off gas from the helium bath and reduces the thermal

597: loading on the helium stage.  Radiative and conductive loading is

598: 7.2\,W on the nitrogen tank and 71\,mW on the helium tank.  An

599: additional effective load of 9\,mW and 7\,mW on the helium tank is due

600: to the liquid drawn into the pumped-pot by the capillary (discussed

601: below), and heating required to cycle the $^3$He refrigerator,

602: respectively.  The cryostat holds 43\,L of nitrogen and 32\,L of

603: helium and has an 11.5 day hold time.

604: 

605: A liquid He pumped-pot maintains the optics box

606: at 1.5\,K.  A long, thin capillary connects the helium bath to the

607: pumped-pot with a volume of ${\sim}$\,100\,mL.  A pump line between the

608: pot and the exterior of the cryostat keeps the pot at near vacuum,

609: maintaining the temperature at 1.5\,K.  The pressure difference between

610: the helium bath and the pot forces helium into the pot.

611: On the ground, a vacuum pump is used to evacuate the pump line; at

612: float altitude, the pump line is open to the atmosphere, which is at

613: a pressure $< 0.01$\,atm\@.

614: 

615: The bolometers and feed horns are maintained at $< 300$\,mK by the

616: closed-cycle $^3$He refrigerator. The $^3$He is condensed by the

617: 1.5\,K pumped-pot and collects in the $^3$He cold stage.

618: The $^3$He refrigerator is

619: able to extract 5\,J of energy and needs to be recycled every 48--60 hours.

620: The recycling process takes less than 2.5 hours.

621: 

622: The cryostat performed as expected during both science flights.  The

623: nitrogen and helium baths maintained constant temperature and the VCS

624: fluctuated by less than 3\,K due to the $^3$He refrigerator cycles.

625: The pumped-pot maintained a temperature below 1.8\,K with fluctuations

626: less than 10\,mK\@.  During $^3$He refrigerator cycles the pumped-pot

627: temperature rose as high as 2.5\,K\@.  The bolometer temperature was

628: maintained below 300\,mK with less than 1\,mK fluctuations on hour

629: time scales.  The first science flight was too short to test hold

630: time.  In the second 12.5 day flight, the helium tank was exhausted

631: 11.5 days after reaching float altitude.  The nitrogen tank was not

632: exhausted before termination.

633: \begin{figure*}

634:         \centering

635: 	\includegraphics[width=\linewidth]{f6.eps}

636: 	\caption{Front and side schematic drawings of the BLAST gondola.

637:   A 1\,m tall Emperor penguin is shown for scale. The height of the pivot is set by

638:  the size of the launch vehicle. The width of the gondola {\it

639:  without\/} deploying the solar panel arrays matches the width of the

640:  laboratory doors. The inner frame, which can be pointed in elevation,

641:  consists of the two star cameras, the telescope and its light baffle,

642:  the receiver cryostat and associated electronics.  The telescope baffle

643:  shown at right, which was used in 2005, was replaced by a smaller

644:  system in 2006.  The CSBF solar panels and transmitting antennae are

645:  suspended below the structure shown here. The lines marked 25\degree\

646:  and 60\degree\ show the useful range of orientation of the optic axis.

647:  The dot-dashed line at the right originating above the pivot shows a

648:  20\degree\ avoidance zone required to avoid accidental contact at

649:  launch.  The dashed parallelogram at the top in the right hand diagram,

650:  and to the left in the left hand diagram shows the shape of an

651:  extension to the Sun shields added for the Antarctic flight to allow

652:  observations to be made further from the anti-sun direction. }

653: 	\label{fig:gondola}

654: \end{figure*}%

655: 

656: 

657: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

658: \section{Gondola} \label{sec:gondola}

659: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

660: 

661: The \blast\ gondola provides a pointed platform for the telescope and

662: the attachment point to the balloon flight train. The gondola was

663: designed and built by Empire Dynamic Dynamic

664: Structures,\footnote{\url{http://www.empireds.com/}} based on initial

665: designs from members of the \blast\ team.  Schematics of the gondola

666: are shown in Figures \ref{fig:gondola} and \ref{fig:gondola_dressed}.

667: The breakdown of the mass of various components is given in

668: Table~\ref{tab:mass}.

669: 

670: \subsection{Requirements}

671: The gondola design is driven by the pointing requirements of the

672: science case.  The elevation range of the inner frame, including the

673: 2\,m primary mirror and the ${\sim}$\,200\,kg cryostat, is

674: 25--60\degree. The entire gondola can rotate to any azimuth angle.

675: The in-flight pointing is accurate to ${\sim}$\,30\arcsec\ and

676: post-flight pointing reconstruction is accurate to better than

677: $5\arcsec$. The telescope's primary observing mode is to scan in

678: azimuth at ${\sim}$\,0.1\degree\,s$^{-1}$.  All of the mirrors are

679: shaded from solar radiation.

680: 

681: As a consequence of the pointing requirements, the feedback rate of

682: the control system is ${\sim}$\,10\,Hz. To accommodate this, the

683: gondola is designed to be rigid, with a minimum resonance frequency

684: greater than $14.4\,$Hz. All mechanical tolerances are set to minimize

685: backlash. All bearings are low friction, low stiction and treated with

686: low-temperature grease (see Section~\ref{SEC:ATTITUDECONTROLL}). The center-of-mass of the inner frame is on

687: the rotational axis (the elevation axis) so that translations of the

688: gondola (by wind or from the balloon) do not generate torques which

689: re-orient the telescope.

690: 

691: A target weight of ${\sim}$\,2000\,kg was set for the entire system

692: and the moment of inertia was estimated to be

693: ${\sim}$\,4500\,kg\,m$^2$. The system has been engineered to meet the

694: flight criteria set by the Columbia Scientific Balloon Facility

695: (CSBF)\@.

696: 

697: 

698: \subsection{Layout}

699: The gondola frame consists of three components: an outer frame,

700: suspended from the balloon flight train by cables and a pivot motor;

701: an inner frame which is attached to the outer frame at two points

702: along a horizontal axis; and a set of Sun shields that attach to the

703: outer frame. The frame is constructed of light-weight aluminum tubing

704: and I-beam. \blast\ incorporates large-diameter torque motors for all

705: motion control.

706: 

707: \subsubsection{Major Components}

708: The outer frame consists of a horizontal surface, 2 yokes to support

709: the inner frame, and four legs. The outer frame is pointed in azimuth

710: using a reaction wheel and pivot.  It also provides mounts for various

711: electronics boxes, including the flight computers, the 

712: CSBF electronics,

713: various pointing sensors, and flight batteries.

714: 

715: The inner frame is made from thin-wall aluminum box beam. It supports the

716: telescope, cryostat, detector read-out electronics, gyroscopes and

717: star cameras. It is attached to the outer frame at two points,

718: defining an axis of rotation.

719: 

720: The back, sides and bottom of the gondola are surrounded by Sun

721: shields, allowing the telescope to point as low as 25\degree\ in

722: elevation. A lightweight aluminum frame encloses the telescope and is

723: covered with extruded polystyrene panels wrapped in aluminized Mylar. The original

724: design shielded the secondary mirror from the Sun at 30\degree\ above

725: the horizon, the highest the Sun rises during an Antarctic flight, with

726: the telescope pointing directly away from the Sun. Additional panels

727: were added for the \blastmcm\ flight to improve the shading when the

728: telescope is not pointed anti-Sun. The telescope is also protected

729: from solar radiation scattered from the ground. As well as the opening

730: for the telescope, there are forward-facing openings in the Sun shields

731: to allow for cooling of the electronics by radiation (see Figure~\ref{fig:gondola_dressed}).

732: \begin{figure*}

733:         \centering

734:  	\includegraphics[width=\linewidth]{f7.eps}

735:  	\caption{A CAD model of the gondola, showing \blast\ fully

736:         assembled in the \blastmcm\ configuration. The electronics can cool

737: 	radiatively through the triangular gaps

738: 	which are visible in the Sun shields.  These gaps always point away

739: 	from the Sun in flight.}

740:  	\label{fig:gondola_dressed}

741:  \end{figure*}%

742: \subsubsection{Attitude Control}\label{SEC:ATTITUDECONTROLL}

743: The telescope's attitude is controlled by three torque motors: the

744: reaction wheel, pivot and elevation motors. The motors are rare-earth,

745: permanent-magnet, direct-drive DC torque motors\footnote{Kollmorgen QT-6205.}

746: with a peak torque of 13.6\,N\,m.  They are controlled by 50\,A

747: pulse-width 

748: modulated servo amplifiers.\footnote{Advanced Motion Controls,

749: model 12A8K.}  The housings, made of hardened steel, are custom

750: designed and incorporate the bearings. The original design called for

751: a dry Teflon bearing lubricant, but we found that low-temperature

752: bearing grease\footnote{Dow

753: Corning Molykote\,$^{\mbox{\tiny\textregistered}}$ 33.}

754:  provides lower friction and stiction.

755: 

756: The telescope is controlled in azimuth by a 1.5\,m reaction wheel,

757: made of 7.6\,cm thick aluminum honeycomb,

758: and forty-eight 0.9\,kg brass disks mounted around the

759: perimeter to maximize the ratio of moment of inertia to mass. 

760: The reaction wheel is mounted at the center of the outer frame with its rotation

761: axis going through the pivot.

762: Torquing the gondola against the reaction wheel controls azimuth pointing.

763: 

764: The gondola is subject to external torque from wind shear acting on

765: the gondola  and from balloon rotation acting through the flight

766: train.   The second term is minimized, but not entirely eliminated,

767:  by a vertical axis pivot at the attachment  of the gondola to the

768: flight train (see Figure~\ref{fig:gondola}).  Since the flywheel motor

769: saturates at ${\sim}$\,50\,rpm, the additional torque motor at the

770: pivot is needed to reduce excess angular momentum, keeping the

771: rotation of the reaction wheel within reasonable limits.  

772: Figure~\ref{fig:motion} demonstrates that this nested servo system,

773: where the reaction wheel is driven to maintain telescope orientation

774: and the pivot is driven to maintain low reaction wheel speed,

775: successfully decouples the telescope from external torques.

776: \begin{figure*}

777:         \centering

778:  	\includegraphics[width=0.5\linewidth]{f8.eps}

779: \caption{Isolation of vertical torques.  All panels show data from the

780: same 12 minute segment of the 2006 flight, when the telescope was

781: executing a routine azimuthal scan.  The third panel from the top

782: shows a trapezoidal angular velocity profile corresponding to constant

783: angular velocity scans connected by constant angular accelerations.  

784: This panel also includes a plot of the desired angular speeds,

785: completely buried by the measured signal.   The second panel from the

786: top shows that azimuthal speed is governed by accelerating and

787: decelerating the speed of the reaction wheel.  A torque at the pivot

788: proportional to reaction wheel speed reaches threshold and overcomes

789: the stiction in the pivot when the reaction wheel speed is near 0.3

790: revolutions per second.  One such event is marked by the vertical

791: dashed line in all panels.  The bottom panel contains a plot of the

792: difference between measured gyroscope signal and desired azimuthal

793: speed.  Notice that there are small spikes visible corresponding to

794: the corners in the scan speed but that there is no azimuthal motion of

795: the telescope itself coincident with the release of stiction in the

796: pivot.}

797: \label{fig:motion}

798: \end{figure*}

799: 

800: The elevation of the inner frame is controlled by a motor mounted on

801: one side of the inner frame at the attachment point to the outer

802: frame.  A free bearing provides the connection point on the other

803: side. The motor housing incorporates a 16-bit shaft

804: encoder\footnote{Gurley Precision Instruments A25SB.}  which measures

805: the orientation of the inner frame relative to the outer frame. A

806: fluid transfer system, consisting of a pump and two tanks containing

807: ${\sim}$\,10\,L of a high-density non-toxic fluid,\footnote{Dynalene

808: HC-40.} corrects for long time-scale balance offsets induced by the

809: boil-off of the cryogens and to the redistribution of cryogens due to

810: changes in elevation.

811: 

812: The balloon environment introduces small oscillations with a period of

813: approximately 25\,s in the pitch and roll of the gondola. Typical

814: amplitudes of ${\sim}$\,2\arcmin\ are observed. A roll damping system

815: consisting of a motor and a small 30\,cm diameter reaction wheel is

816: mounted perpendicular to the flywheel on the outer frame, with the

817: goal of dissipating roll modes.  The reaction wheel is made of

818: aluminum and weighs ${\sim}$\,2\,kg, most of which is at the

819: perimeter. Oscillations in pitch are removed by controlling the

820: elevation of the inner frame.  The oscillations in pitch and roll

821: where measured to be 50\% smaller compared to a case when both

822: controllers were off.

823: 

824: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

825: \section{Command and Control} \label{sec:control}

826: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

827: 

828: \blast\ is designed to operate autonomously without need of ground

829: commanding.  Telescope control is provided by a pair of redundant

830: flight computers with Intel Celeron processors at 366 and 850\,MHz,

831: running Slackware\footnote{\url{http://www.slackware.com/}} Linux 9.2

832: with Kernel version 2.6.8\@. The computers are kept at near

833: atmospheric pressure to allow the hard drives to function properly and

834: to provide the appropriate thermal environment for the CPUs.

835: 

836: The flight computers run a single, monolithic program, the ``master control

837: program'' (mcp), written in C,  which performs primary control of all aspects of the

838: telescope including in--flight pointing solution,  motion, commanding,

839: telemetry, data archiving, and thermal control.  

840: 

841: The communication link between the telescope and the ground is

842: provided by CSBF through a number of line-of-sight (LOS) transmitters and

843: satellite links. The ground station computers interface with CSBF's ground station equipment to send commands to the payload and display

844: down-linked data. The ground station software uses

845: kst\footnote{\url{http://kst.kde.org/}} and other Linux applications

846: developed by \blast\ team members.

847: 

848: \subsection{Gondola Electronics}

849: The flight computers are built upon a passive PCI-bus backplane and

850: contain a single board computer, a four port 8250 serial port

851: extender card, and a custom made PCI controller card, which

852: interfaces major components as well as the LOS data

853: transmitter. This card interfaces with the

854: BLASTbus, a proprietary RS485 bus with three differential pairs (data,

855: clock, strobe).  The BLASTbus protocol is a poll and response

856: architecture composed of a 32-bit request followed by a 32-bit

857: response containing 16 bits of data.  The BLASTbus clock runs at 4\,MHz,

858: giving an effective bandwidth on the BLASTbus of 1\,Mbit\,s$^{-1}$.

859: 

860: The BLASTbus connects the flight computers to the Attitude Control

861: System (ACS) and DAS.  Each DSP card in these systems 

862: monitors the

863: BLASTbus. They provide data on request and receive command input

864: from the flight computers via the BLASTbus.

865: 

866: Data are marshaled on the BLASTbus into 100\,Hz frames.  Bolometers are

867: polled once per frame, as are the gyroscopes and other high-precision

868: pointing data.  The majority of housekeeping signals do not need to be

869: polled at 100\,Hz; these ``slow data'' signals are polled at 5\,Hz with

870: groups of twenty multiplexed into a single 100\,Hz channel.  The 100\,Hz

871: frames provide the basic data structure for data archival. The frames

872: are written to disk and transmitted by the line-of-sight 1\,Mbit\,s$^{-1}$ 

873: transmitter.

874: 

875: In addition to the BLASTbus, the flight computers also collect data

876: via 8250-style serial connections to the differential GPS, the SIP

877: computers, the lock motor, and the secondary actuators.  Ethernet

878: provides further connectivity to the star camera computers and Sun

879: sensor computer.  In order to provide data synchronization, the

880: auxiliary channels are written back to the BLASTbus and the star

881: cameras are triggered via the BLASTbus.

882: 

883: 

884: \subsection{Pointing Control}

885: \blast\ is a scanning experiment.  The primary scan-mode involves an

886: azimuthal raster coupled with a slow elevation drift or discrete

887: elevation steps.  

888: 

889: Detector response times, $\tau$, together with $1/f$ noise and star camera

890: integration times set limits on the azimuthal scan rate, $v_{{\rm az}}$.  

891: Two samples per FWHM are required to fully sample a Gaussian beam.

892: The maximum angular rate allowed by the detector $\tau$

893: is therefore

894: \[

895: 	v_{{\rm az}} = \frac{\mbox{\scriptsize FWHM}/2}{2\pi\,\tau} 

896: \]

897: or about 0.3\degree\,s$^{-1}$ for the 30\arcsec\ beam at 250\,\micron\ 

898: and $\tau\sim 2$\,ms.  Mid-scan smearing in the star cameras also

899: limits azimuthal scan speeds to about 0.1\degree\,s$^{-1}$. A scan

900: rate larger than this is acceptable only for relatively small scans

901: where star camera solutions at the turn-around are more frequent. Low

902: frequency ($1/f)$ noise sets the largest angular mode that can be

903: constrained to $v_{{\rm az}}/f_0 \lesssim 3$\degree\ at

904: 0.1\degree\,s$^{-1}$ ($f_0$ is the $1/f$ knee at 35\,mHz).

905: Therefore, \blast\ scans most of the time at $v_{{\rm az}} \simeq

906: 0.1$\degree\,s$^{-1}$, occasionally going twice as fast on small

907: ($\lesssim 0.3$\degree\ wide) maps.  Azimuthal acceleration is limited

908: to 0.1\degree\,s$^{-2}$ by the control electronics, resulting in,

909: typically, 2\,s for an azimuthal turn-around.

910: 

911: Elevation drift speeds of 10\arcsec\,s$^{-1}$ or elevation steps of

912: 40--100\arcsec\ in adjacent azimuth scans provide adequate

913: spatial sampling of the sky by the detector arrays.

914: 

915: Three primary scan modes are implemented (Figure~\ref{fig:scan_modes}): ``cap'', a circle centered on a

916: target RA/Dec; ``box'', a rectangle in azimuth and elevation centered on a

917: target RA/Dec; and ``quad'', an arbitrary quadrilateral specified by its

918: four corners in RA and Dec. 

919: \begin{figure*}

920:         \centering

921:        	\includegraphics[angle=-90, keepaspectratio, width=\linewidth]{f9.eps}

922: 	\caption{Idealized representations of \blast's three scan modes: from left to right a ``cap'', a ``box'', and a ``quad''. In practice, the gyroscopes are used to control the speed and orientation of each single scan, while star camera--based orientation solutions at the endpoints are used to prevent gyroscope errors from accumulating.  }

923: 	\label{fig:scan_modes}

924: \end{figure*}

925: 

926: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

927: \section{Pointing Sensors} \label{sec:psens}

928: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

929: 

930: The primary pointing sensors for \blast\ are a pair of CCD-based star

931: cameras and Fiber-Optic rate gyroscopes. The star cameras provide

932: absolute pointing and the gyros provide velocity information which can

933: be integrated to allow interpolation of the gondola's attitude between

934: star camera solutions. Coarse attitude determination is provided by

935: several additional sensors: in elevation, there is an encoder on the

936: elevation axis and a tilt sensor on the inner frame; in azimuth, there

937: is a Sun sensor, a differential GPS unit, and a magnetometer.  The

938: system provides post-flight pointing reconstruction to $\lesssim

939: 5\arcsec$\ rms.

940:  

941: 

942: \subsection{Star Cameras}

943: 

944: Four primary factors drive the design of the star cameras: (i) an

945: absolute pointing accuracy of ${\sim}$\,5\arcsec\ is required to

946: over-sample the diffraction-limited size of the 250\,\micron\ beam;

947: (ii) integration times (and hence efficiency) have to be short enough

948: to avoid significant smearing at the maximum normal scan

949: angular-velocity of the gondola (0.1\degree\,s$^{-1}$); (iii) the

950: system must {\em always\/} detect stars to calibrate gyroscope drift;

951: and (iv) the frequency of the solutions must be high enough to control

952: the $1/f$ random walk noise in the integrated gyroscopes.

953: 

954: To meet these requirements, we incorporated two star cameras for

955: redundancy, and to enable increased positional accuracy in post-flight

956: processing.  A detailed description of the cameras can be found

957: in~\citet{rex06}.

958: The cameras each use a Nikon lens with a 200\,mm focal length and a

959: 100\,mm aperture to produce a 2.5\degree\ $\times$ 2.0\degree\ FOV

960: with 7\arcsec\ pixels.  Each camera is controlled by its own PC/104,

961: 300\,MHz Celeron computer which calculates pointing solutions for the

962: gondola at a rate of ${\sim}$\,1.5\,Hz.  The computers command the CCD

963: cameras via FireWire, control the focus and aperture size using

964: stepper motors via a serial port, and regulate the temperature of the

965: camera using a small USB DAQ module.  The entire system is contained

966: in a pressure vessel to allow the operation of the hard drives,

967: control the thermal environment, and maintain mechanical rigidity.

968: Since the cameras operate during the day, the dominant source of noise

969: is from the sky background, despite the altitude of the experiment. A

970: Nikon R60 red filter is used to attenuate the background. In addition,

971: a 1.2\,m long cylindrical baffle is attached to the front of each

972: camera to reduce stray-light contamination beyond 10\degree\ from the

973: optical axis.

974: 

975: The software controlling the camera in flight provides real-time

976: pointing information by analyzing the star patterns in the CCD frames.

977: The pointing algorithm locates blobs in the camera image, rejecting

978: the known bad pixels.  The best fit positions of star candidates are

979: then used by a pattern recognition algorithm to identify a unique

980: constellation matching the observed angular separations in a star

981: catalog \citep[Guide Star Catalog 1.1, see][]{lasker87}.  The

982: magnitude limit of the catalog is chosen manually and no brightness

983: information for the stars is otherwise used. The algorithm is aided by

984: an approximate pointing solution from the flight computer, required to

985: be accurate to ${\sim}$\,5\degree\ (see Section~\ref{sec:CS}) to

986: reduce the number of candidate star identifications. A visual

987: magnitude limit of 9 was required to obtain sufficient completeness. A

988: ``Lost in Space'' algorithm based on the Pyramid technique

989: \citep{mort01} was also implemented to be used in case the approximate

990: solution was found to be unreliable (which never happened during the

991: \blast\ flights).

992: 

993: Once the CCD object centroids are matched to stars with known

994: coordinates $(\alpha,\delta)$, the pointing solution is calculated,

995: parametrized by the celestial coordinates of the center pixel,

996: $\alpha_0$ and $\delta_0$, and the roll of the camera, $r$. A model in

997: which the image is assumed to be a perfect gnomic tangent-plane

998: projection, with the tangent point at $\alpha_0$ and $\delta_0$, is

999: used to project each star RA and Dec into the plane of the CCD.  The

1000: rms distance between the CCD and model star coordinates is then

1001: minimized using an iterative Newton solver with respect to the three

1002: model parameters. This procedure produces pointing solutions with

1003: uncertainties of ${\sim}$\,3.5\arcsec\ and ${\sim}$\,200\arcsec\ for

1004: the position of the tangent point and of the roll, respectively.  A

1005: post-flight comparison of simultaneous pointing solutions from both

1006: cameras results in an rms uncertainty of $<$\,2\arcsec.

1007: 

1008: \subsection{Gyroscopes}

1009: Fiber-Optic rate Gyroscopes (FOG) are used to extrapolate the star

1010: camera absolute attitude to provide pointing information at each

1011: detector time sample. Two redundant sets of three FOGs are

1012: mechanically arranged to measure the gondola angular velocity along

1013: orthogonal axes. One of these sets uses ECORE 2000 analog gyroscopes

1014: while the other uses DSP 3000 digital output gyroscopes from the same

1015: company.\footnote{KVH Industries, Inc.} They have an angle random walk

1016: of 5\arcsec\,s$^{-1/2}$ and 4\arcsec\,s$^{-1/2}$, respectively.  Since

1017: FOGs are sensitive to magnetic fields \citep{bohm}, they were wrapped

1018: in a 250\,\micron\ thick $\mu$-metal sheet. This reduced the signal

1019: induced from the Earth's magnetic field by a factor ${\sim}$\,10. The

1020: gyroscope assembly is temperature controlled to mitigate bias drifts.

1021: For \blastkir\, both sets worked flawlessly.  In \blastmcm\ two out of

1022: the three digital gyroscopes stopped working temporarily, possibly due

1023: to a cosmic ray interaction with the gyro electronics. A power cycle

1024: of the gyroscope itself restored function.  Post-flight attitude

1025: reconstruction of the 2005 and 2006 data shows that the FOGs performed

1026: at near to their specified sensitivity.

1027: 

1028: \subsection{Coarse Sensors} \label{sec:CS}

1029: Two distinct Sun sensors were used on \blast\ for coarse azimuth determination.

1030: For \blastfs\ and \blastkir, a linear CCD behind a thin slit was used.  It was replaced by an array of photo-diodes

1031: arranged in a azimuthally oriented ring for \blastmcm.  

1032: The photo-diode Sun sensor was smaller, lighter, and consumed less power 

1033: (5\,W compared to 50\,W) than the CCD-based sensor.  

1034: 

1035: The photo-diode Sun sensor is a small, all-in-one unit.  A PC/104

1036: embedded computer system is mounted below a dodecagonal ring.  Twelve

1037: photo-diodes are mounted radially around the ring, and the ring is

1038: mounted with its axis parallel to the azimuth axis.  The intensities

1039: of light incident on the photo-diode with the highest intensity and on

1040: its 2 nearest-neighbors are compared and fit to a $\cos \theta$

1041: function to determine the azimuth relative to the Sun.  The flight

1042: computer calculates the azimuth of the Sun to determine the azimuth of

1043: the gondola.  The CCD Sun sensor, used during the first science

1044: flight, had a short-time, relative precision of 2\arcmin, but an

1045: overall absolute accuracy of 5\degree\@.  The photo-diode Sun sensor,

1046: used during the second science flight, had a precision of 4\arcmin,

1047: with an absolute accuracy of 5\degree\@.  Both units provide azimuthal

1048: data at $>5$\,Hz.

1049: 

1050: \blast's second coarse azimuth sensor was an ADU5 differential GPS

1051: unit.\footnote{Magellan Navigation, Inc.}  The four required antennae

1052: were installed on booms above the Sun shields to minimize multi-path

1053: reflections from the gondola (see Figure~\ref{fig:gondola}).  This

1054: orientation was predicted to provide better than 10\arcmin\ rms

1055: absolute pointing.  However, for \blastkir, the GPS stopped providing

1056: azimuthal solutions before the instrument reached float altitude and

1057: never recovered.  For \blastmcm, the GPS did not provide solutions

1058: until 110 hours after launch.  It then began providing solutions and

1059: the unit operated with a precision from 0.1\degree\ to~0.2\degree\

1060: rms.  The cause of the irregular reliability of the GPS is uncertain,

1061: but may be due to a lack of thermal rigidity in the antenna mounts.

1062: The position, altitude, velocity, and time data provided by the GPS

1063: unit worked well at all times.

1064: 

1065: \blast's third and most reliable coarse azimuth sensor was a 3-axis,

1066: flux-gate magnetometer\footnote{Applied Physics, model 534.} which was

1067: used to determine the gondola's attitude relative to the Earth's

1068: magnetic field.  The magnetic field orientation was determined using

1069: the World Magnetic Model.\footnote{WMM-2005, National

1070: Geospatial-Intelligence Agency (NGA).}  Even though \blast\ passed

1071: close to the magnetic pole, the model was accurate enough and the

1072: gondola pendulations small enough that the magnetometer-based pointing

1073: solution was good to 5\degree\ peak to peak in both science flights.

1074: 

1075: 

1076: \section{Thermal Environment} \label{sec:thermal}

1077: The LDB thermal environment is characterized by continuous but

1078: variable solar illumination combined with small coupling to the

1079: atmosphere.  Extensive shielding is used to regulate the temperature

1080: of the instrument.  Following the design of BOOMERANG \citep{crill03},

1081: \blast\ is surrounded by shields comprised of 24\,\micron\ thick

1082: aluminized Mylar applied to 2.5\,cm thick polystyrene open-celled

1083: foam.  The Mylar side faces out on all surfaces which may be exposed

1084: to the Sun.  While bare aluminum surfaces can be expected to reach

1085: temperatures in excess of $130^{\circ}$C in the LDB float environment,

1086: Sun-facing aluminized Mylar surfaces reach temperatures of around

1087: $45^{\circ}$C due to their high ratio of thermal IR emissivity to

1088: visible light absorptivity.  For telescope azimuth ranges of

1089: $\pm$60\degree\ from anti-Sun, the shielding blocks direct radiation

1090: from reaching the optics and all electronics, except for the pivot

1091: motor controller and Sun sensor. The actual angles of Sun avoidance

1092: have a weak dependence on the elevation of the telescope.

1093: 

1094: The majority of the electronics dissipate their heat radiatively.  A

1095: high IR emissivity is achieved by painting electronics enclosures with

1096: white paint.\footnote{We used a variety of generic spray-on epoxy

1097: enamel.}  Painting the interiors of boxes white aids in radiative

1098: transfer between the electronics and the box.  This is effective for

1099: all boxes, except for the detector readout electronics boxes, each of

1100: which dissipates about 100\,W.  To cool these boxes, emissive vertical

1101: plates were installed between each card to couple heat from the

1102: electronics to the top and bottom walls of the boxes.  A closed-cycle

1103: fluid cooling loop was used to distribute the heat to the gondola

1104: frame.

1105: 

1106: The thermal strategy was effective in keeping the temperatures of

1107: essentially all electronics well within their operational range.  A

1108: shield added to \blastmcm\ to keep the Sun from illuminating the

1109: secondary mirror reflected solar radiation back onto the pivot motor

1110: controller.  This increased the average temperature of the pivot by up

1111: to $20^{\circ}$C compared to what was experienced in \blastkir\@.  As

1112: a result, two non-essential observations where shortened when the

1113: motor controller temperature exceeded $62.5^{\circ}$C.

1114: 

1115: For \blastmcm, the primary mirror was thermally isolated from the

1116: inner frame with G10 spacers. Approximately twelve layers of Mylar

1117: Super-insulation\footnote{Cryolam\,$^{\mbox{\tiny\textregistered}}$}

1118: were added in the gap between the mirror and the frame. This

1119: stabilized the in-flight temperature, which exhibited a diurnal

1120: excursion of about $\pm$1.5\degree\/C\@.

1121: 

1122: See Table~\ref{tab:thermal} and Figure~\ref{fig:thermal} for a list of temperatures achieved during the various \blast\ flights.%

1123: \begin{figure*}

1124:         \centering

1125: 	\includegraphics[angle=-90, width=\linewidth]{f10.eps}

1126: 	\caption{ Temperatures during the 2005 (top panel) and 2006

1127: (bottom) flights.  Several gondola temperatures are plotted for the

1128: period starting just before launch and extending through the first few

1129: days from both science flights.  Dramatic cooling in the troposphere,

1130: reheating in the stratosphere and diurnal variations due to variation

1131: of the solar elevation angle are visible in all traces.  The curves

1132: are a: the pressure vessel containing the data acquisition computers,

1133: b: the gondola inner frame, c: the elevation axis motor and d: the

1134: primary mirror.  The temperature difference between high power

1135: consumption electronics (a) and more passive components (b and c) is

1136: only slightly larger at float than it is on the ground.  The primary

1137: mirror has the best geometry for cooling radiatively and therefore

1138: attains the coldest temperature.  Notice the relatively large

1139: variations in the primary mirror temperature in the 2005 flight which

1140: caused us to recognize the need for in-flight focussing capability

1141: combined with better insulation of the primary.}

1142: \label{fig:thermal}

1143: \end{figure*}

1144: 

1145: \section{Power System}

1146: Solar panels provide power to the flight electronics.  They are

1147: mounted at the back of the telescope on the support structure for the

1148: Sun shields (see Figure \ref{fig:gondola}). The panels face the Sun

1149: from only one side and are radiatively cooled to the sky from the

1150: other side.

1151: 

1152: During normal operation, \blast\ requires 500\,W, divided between two

1153: completely isolated systems: the receiver, and the rest of the gondola

1154: (see Table~\ref{table:power}). During line-of-sight operations, an

1155: additional 150\,W are required to power the transmitters. Power is

1156: provided by the solar power system\footnote{MEER Instruments, Palomar

1157: Mountain, CA, http://www.meer.com/ and SunCat Solar, Phoenix, AZ.}

1158: which charges NiMH batteries.\footnote{Cobasys Model 9500.} The arrays

1159: provide 1250\,W for normal incidence at float. The arrays are thus

1160: able to provide full power to the experiment up to a Sun-array angle

1161: of $>60^{\circ}$. The batteries are essentially only used during

1162: launch and ascent.

1163: 

1164: As a NiMH battery reaches its top-off voltage, its charge efficiency

1165: decreases, leading to heating of the battery. Additionally, the topoff

1166: voltage of NiMH batteries decreases with increased temperature. This

1167: combination can lead to thermal runaway if the batteries are charged

1168: with a fixed top-off voltage.  For \blast, the charge controllers were

1169: modified to allow the fight computer to reduce the top-off voltage as

1170: the battery temperature increases, permitting full battery charge

1171: without thermal runaway.

1172: 

1173: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1174: \section{Flight Planning}

1175: 

1176: \blast\ functions autonomously using schedule files that consist of a

1177: sequential list of observations or actions as a function of the local

1178: sidereal time.  This system is robust against temporary system

1179: failures because the telescope only needs to know the current time and

1180: location to resume operation upon recovery.  Using a local sidereal

1181: clock rather than a clock fixed in some time zone, it is possible to

1182: account for purely astronomical visibility constraints (such as the RA

1183: of the Sun and of the astronomical targets) using a static

1184: description. However, the amount of time the telescope spends on a

1185: given observation varies depending on telescope longitudinal drifts.

1186: 

1187: Flight schedules are calculated on the ground, in advance of the

1188: launch using best-estimates for the launch time, flight path and

1189: length.  Given the uncertainties in these flight parameters and the

1190: complexity of the observations (many small maps scattered across the

1191: sky, of which only a subset are available at a given instant), a

1192: schedule file generator was developed to automate the process.  This

1193: system interprets a high-level description of the desired scientific

1194: observations, such as ``map a circular region centered at a given RA

1195: and DEC with a given radius''.  It then accounts for a number of

1196: parameters such as: the launch date and time, projected duration of

1197: flight, the start longitude and latitude, a guess for the termination

1198: longitude, a latitude range over which the payload may drift, and the

1199: position of the Sun and the Moon.  With this information, the scheduler

1200: generates an optimized list of consecutive actions.  At each instant,

1201: observations that are possible under the current visibility

1202: constraints are scheduled in order of scientific priority.  At regular

1203: intervals, calibration observations are assigned maximum priority, so

1204: that sensitivity, beam shape, and pointing variations may be tracked

1205: throughout the flight.  Once a schedule file is generated, a simple

1206: model for the power usage of the system and the charge rate of the

1207: solar panels as a function of Sun incidence angle is used to

1208: calculate the available battery charge over time. The schedule is

1209: modified by hand if this model indicates that the batteries will have

1210: less than 50\% of the total charge capacity at any point in the

1211: flight.  Schedules are then tested using a simulator that mimics the

1212: scanning motion of the telescope to determine the coverage (effective

1213: integration time across the maps) and cross-linking among each of the

1214: maps for a given flight trajectory.

1215: 

1216: Six schedule files are generated for each launch opportunity.  The

1217: latitude of the gondola can change by as much as ${\sim}$\,15\degree\

1218: during a flight, hence three different schedules are created in

1219: latitude bands that are 6\degree\ wide, with 1\degree\ of overlap.

1220: The gondola uses the time and payload position reported by the GPS to

1221: decide which schedule file to use.  Two sets of these three schedules

1222: are made: the first set assumes the instrument is working with the

1223: target sensitivities; the second assumes degradations of the telescope

1224: beam size by a factor of $\sqrt{2}$, and sensitivity by a factor of 2.

1225: At the beginning of the flight, the sensitivities are estimated from

1226: scans across calibrators.  Based on this information the ground

1227: station team can decide which of the two sets of schedule files the

1228: instrument should use, and switch between the two using a single

1229: command.

1230: 

1231: 

1232: 

1233: 

1234: \section{Pointing Reconstruction} \label{sec:performance}

1235: Post-flight pointing reconstruction estimates the rotation (attitude)

1236: of the gondola with respect to the celestial sphere as a function of

1237: time, providing RA, Dec, and rotation angle information at each

1238: detector sample.  The star camera provides absolute attitude on an

1239: unevenly sampled time grid (${\sim}$\,1.5\,s), with an accuracy of $<

1240: 2$\arcsec\ rms.  Each solution is sampled at a known phase with

1241: respect to the detectors.  The detector and gyroscope sampling are

1242: synchronized; therefore, the integration of the gyroscopes gives one

1243: estimate of the gondola attitude. The star camera is used to correct

1244: the random walk drift induced by the integrated gyroscope noise and to

1245: give an estimate of the integration constant.  \blast\ implements a

1246: Kalman filter approach to estimate the attitude, expressed as the

1247: state quaternion $q(t)$, a 4-dimensional quantity describing the solid

1248: body rotation of the gondola.  The non-linear state model representing

1249: the gondola is defined as

1250: \begin{equation}

1251:         \left\{\begin{array}{l}

1252:         \vec{b}_{n+1} = \vec{b}_n + {\vec {w_b}}_n\\

1253:         q_{n+1} =  \qvec{1}{\frac{1}{2}\left(\vec\omega_n +  \vec{b}_n + \vec{w_\omega}_n \right) {\scriptstyle\Delta T}}q_n\\

1254:         \end{array}\right.

1255: \end{equation}

1256: where $\vec\omega_n$ is the gyro angular velocity with

1257: $\vec{w_\omega}_n$ its noise, $\vec{b}_n$ is the gyro bias,

1258: $\vec{w_b}_n$ is the filter process noise, and $\Delta T$ is the time

1259: resolution of the system (10\,ms).  The filter is run forward and

1260: backward in time and the two solutions are weighted together, using

1261: the Kalman covariance matrix output from the filter as the

1262: weight. Using just one star camera and the digital gyros, the final

1263: averaged attitude is better than $5$\arcsec\ rms; one example is

1264: shown in Figure~\ref{fig:psol_a} using data from

1265: the \blastkir\ flight. The achieved precision is sufficient for \blast\

1266: beam sizes, but can be improved once the second set of gyroscopes and the additional 

1267: star camera

1268: are included in the solution. 

1269: \begin{figure*}

1270:         \centering

1271: 	\includegraphics[angle=-90, width=\linewidth]{f11.eps}

1272: 	\caption{An example of pointing reconstruction from the

1273: 	\blastkir\ data. In the upper panels, the solid blue lines

1274: 	represent the reconstructed pointing solution obtained by

1275: 	integrating one set of gyroscopes onto one of the two star

1276: 	cameras.  Dots represent the positions reported by the other

1277: 	star camera, which is not used in the solution. The lower

1278: 	panels show the residuals as yaw (${\sim}\,\Delta {\rm

1279: 	RA}\times\cos{\rm (Dec)}$) and pitch (coincident with Dec in

1280: 	this particular case).  The yaw residuals are slightly wider,

1281: 	since the telescope scans in azimuth, which is close to RA at

1282: 	the \blastkir\ latitudes. The projection of the gondola

1283: 	angular velocity along the yaw axis is always larger than the

1284: 	pitch axis for the elevation range of the telescope

1285: 	(25\degree--60\degree), and hence the star centroids in star

1286: 	camera images have larger measurement errors in that

1287: 	direction.  The overall error of the pointing solution can be

1288: 	obtained by summing in quadrature the standard deviations of

1289: 	the residuals, yielding $\simeq 4.2$\arcsec\ rms. Improvements

1290: 	to this result are discussed in Section~\ref{sec:performance}}

1291: 	\label{fig:psol_a}.

1292: \end{figure*}%

1293: An improvement of $\sqrt{2}$ to 2 better can

1294: be reasonably expected.

1295: The pointing solution is referenced to the star camera reference frame and needs

1296: to be rotated into the submillimeter array coordinate frame. 

1297: Bright optical and submillimeter point sources

1298: are used to evaluate the rotation quaternion $Q$. 

1299: This correction is ideally static, but, in

1300: practice, was found to be a weak function of time. As shown in

1301: Figure~\ref{fig:gondola}, the star cameras are mounted at the top of

1302: the inner frame. A  structural deformation of the frame can modify the

1303: relative orientation between the cameras and the submillimeter beam.

1304: Therefore, several pointing calibrators are observed each time the

1305: geometry of the scan changes. This effect can be as large as

1306: 35\arcsec\ and has an 80\% correlation with the elevation, and a 25\%

1307: correlation with the temperature of inner frame. Although it is

1308: possible to model the effect based on the gondola thermometry and

1309: attitude, $Q$ is such a weak function of time that this was found unnecessary.

1310: 

1311: \section{Conclusion}

1312: 

1313: \blast\ is an instrument designed to probe the

1314: local and high-z Universe at short submillimeter wavelengths.  Over the course of two

1315: science flights, \blast\ has addressed a broad range of Galactic and

1316: extragalactic topics. The balloon environment requires a novel combination

1317: of ground and satellite technology and techniques.  This paper is

1318: intended to serve as a reference for a suite of science papers from

1319: the \blastkir\ and \blastmcm\ flights.

1320: \\

1321: 

1322: The \blast\ collaboration acknowledges the support of

1323: NASA through grant numbers NAG5-12785, NAG5-13301 and NNGO-6GI11G, the

1324: Canadian Space Agency (CSA), the Science and Technology Facilities Council (STFC),

1325: Canada's Natural Sciences and Engineering Research Council (NSERC),

1326: the Canada Foundation for Innovation,

1327: the Ontario Innovation Trust,

1328: the Puerto Rico Space Grant Consortium, the Fondo Istitucional

1329: para la Investigacion of the University of Puerto Rico, 

1330: and the National Science Foundation Office of Polar Programs;

1331: C.~B. Netterfield also acknowledges support from the Canadian Institute for Advanced Research. 

1332: L.~Olmi would like to acknowledge Pietro Bolli for his help with Physical 

1333: Optics simulations during the testing phase of the \blastmcm\ telescope.

1334: We would also like to thank the Columbia Scientific Balloon Facility

1335: (CSBF) staff for their outstanding work, the Precision Machining Group at

1336: Lawrence Livermore Laboratory, the support received from Empire Dynamic Structures in the design and construction of the gondola,

1337: Daniele Mortari for helpful discussions in the development of the Pyramid code, 

1338: Dan Swetz for buliding the Fourier transform spectrometer,

1339: and  Luke Bruneaux, Kyle Lepage,  Danica Marsden, Vjera Miovic, and James Watt for their contribution to the project.

1340: 

1341: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1342: 

1343: \bibliography{ms}

1344: 

1345: \clearpage

1346: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1347: \begin{deluxetable}{lcc}

1348: \tablewidth{0pt}

1349: \small

1350: \tablecaption{\blast\ optics design summary. \label{table:blast_optics}}

1351: \tablehead{ \colhead{Element} & \colhead{\blastkir} & \colhead{\blastmcm} }

1352: \startdata

1353: 		Primary Mirror Diameter 	& 2\,m 	& 1.8\,m \\

1354: 		Effective Focal Length   		& 10\,m	& 9\,m   \\

1355: 		Field Taper      	& --15\,db	& --7.5\,db     \\

1356: 		Obscuration     	& 14\%	& 7\%     \\

1357: 		FWHM 250\,\micron 	& 40\arcsec	& 30\arcsec	\\

1358: 		FWHM 350\,\micron 	& 58\arcsec	& 42\arcsec	\\

1359: 		FWHM 500\,\micron 	& 75\arcsec	& 60\arcsec	\\ 

1360: \enddata

1361: \tablecomments{The \blastkir\ instrument used a larger mirror which was under-illuminated. In order to maintain the point-source detectability, the \blastmcm\ telescope was more aggressively illuminated. The total transmission of the optical system was evaluated with optical CAD software and using the known efficiency of each smooth-walled feed horn-coupled pixel.}

1362: \end{deluxetable}%

1363: 

1364: 

1365: \begin{deluxetable}{lccc}

1366: \tablewidth{0pt} 

1367: \small 

1368: \tablecaption{Summary of relevant detector characteristics. \label{table:det_params}}

1369: \tablehead{ & \colhead{250\,\micron} & \colhead{350\,\micron}  & \colhead{500\,\micron}}

1370: \startdata

1371: 		Light detectors	& 139 & 88 & 43 \\

1372: 		Dark pixels	&  2   & 2   &  2  \\

1373: 		Resistors       &  1   & 1   &  1  \\

1374: 		Thermistors    &  2   & 2   &  2 \\

1375: 		G (pW/K)       &  880   &  640  & 480  \\ 

1376: 		$\Delta (K)$  	& \multicolumn{3}{c}{50}    \\

1377: 		$R_0$ ($\Omega$)     	& \multicolumn{3}{c}{55}    \\ 

1378: 		$R_{\rm L}$ (M$\Omega$)     	& \multicolumn{3}{c}{$7 + 7$}   \\ 

1379: 		$\tau$ (ms)		& \multicolumn{3}{c}{$2$}         \\

1380: 		Temperature (mK)        & \multicolumn{3}{c}{270}\\

1381: 		NEP (W\,Hz$^{-1/2}$)    & \multicolumn{3}{c}{$3 \times 10^{-17}$}\\

1382: \enddata

1383: \tablecomments{Bolometer parameters are given for a typical array pixel. $\Delta$ and $R_{\rm o}$ define the bolometric model, $R(T) = R_0\,e^{\sqrt{\Delta/T}}$, while the optical time constant, $\tau$, and the thermal conductance, $G$,  define its thermal behavior. The NEP is computed at 1\,Hz.}

1384: \end{deluxetable}

1385: 

1386: \begin{deluxetable}{lc}

1387: \tablewidth{0pt} 

1388: \small 

1389: \tablecaption{Weights of the main components on the payload.\label{tab:mass}}

1390: \tablehead{  &  \colhead{Weight} \\

1391: 	\colhead{Component} & \colhead{(kg)} }

1392: \startdata

1393: {\bf Inner Frame}  	& {\bf 705} \\ 

1394: \hspace{.25in}Frame               	& 120  \\

1395: \hspace{.25in}Mirror             	& 115  \\

1396: \hspace{.25in}Cryostat (empty)    	& 215  \\

1397: \hspace{.25in}Electronics        	& 65 \\

1398: \hspace{.25in}Cryogens            	& 40   \\

1399: {\bf Outer Frame}            		& {\bf 1100} \\

1400: \hspace{.25in}Batteries               	& 80  \\

1401: \hspace{.25in}Solar Panels             	& 30  \\

1402: \hspace{.25in}Electronics              	& 110  \\

1403: \hspace{.25in}Sun Shields              	& 195  \\

1404: \hspace{.25in}Frame                 	& 195  \\

1405: {\bf CSBF Hardware \& Electronics}    	& {\bf 215}\\[2ex]

1406: {\bf\large Total Weight at Launch}     	& {\bf\large 2020} 

1407: \enddata

1408: \tablecomments{Bold face values refer to the \blastmcm\ configuration and include the items shown here as well as all the other elements on the gondola. The total weight at launch was measured at the hook of the launch vehicle before take-off, and the moment of inertia was ${\sim}$\,4500\,kg\,m$^2$.}

1409: \end{deluxetable}

1410: 

1411: 

1412: 

1413: \begin{deluxetable}{lrcrcrccc}

1414: \tablewidth{0pt} 

1415: \small 

1416: \tablecaption{\blast\ temperatures \label{tab:thermal} (\degree\/C).}

1417: \tablehead{  & \multicolumn{2}{c}{\blastfs} &  \multicolumn{2}{c}{\blastkir}& \multicolumn{2}{c}{\blastmcm} & \\

1418: & \colhead{Trop} & \colhead{Diurnal} & \colhead{Trop} & \colhead{Diurnal} & \colhead{Trop} & \colhead{Diurnal} & \colhead{Model}

1419: }

1420: \startdata

1421: ACS 5V DCDC\tablenotemark{a} & 25& 51 to

1422: 23& 15& 62 to 46& 3.6& 41 to 36& ... \\

1423: ACS Case\tablenotemark{b}& ... & ... & --30& 21 to 1& --21& 16 to 10& 28 to 13 \\

1424: PV Case\tablenotemark{c}& ... & ... & --21& 21 to 13& --21& 17 to 12& 18 to 4 \\

1425: PV Air& --5& 26 to 1& --7& 35 to 24& --6& 24 to 29& ... \\

1426: Preamp Case\tablenotemark{d}& --20& 10 to --10& ... & 31 to 19& --5& 25 to 21& 25 to 19 \\

1427: Inner Frame& --16&-2 to --27& --22& 14 to --2& --32& --2 to --5& ... \\

1428: Batteries\tablenotemark{f}& 10&17 to 10&-3& 21 to 10& --6& 19 to 17& 16 to 0 \\

1429: Solar Panels\tablenotemark{g}& ... & ... & --30& 66 to 5& --29& 79 to 58& 85 \\

1430: Mylar Shields &--30& 17 to --55& --34& 18 to --4& --38& 20 to 10\tablenotemark{h}& ...\\

1431: Primary Mirror& --24& --8 to --38& ... & 8 to --15& --17& --6 to --9& --25 to --10\\

1432: Secondary Mirror& --33& 0 to --70& ... & ... & --26& --13 to --16\tablenotemark{i}& ... \\

1433: Pivot&

1434: --33& 15 to --50& --38& 20 to --5& --35& 57 to 36\tablenotemark{j}& ... \\

1435: Outer Frame\tablenotemark{k}& --11&

1436: --1.4 to --37& --30& 5 to --13& --24& 2 to --3& 25 to --5 \\

1437: Star Camera& ... & ... & --11& 15 to 1& --12& 6 to 1& 18 to 5 \\

1438: Sun Sensor\tablenotemark{l}& 0& 31& --23&

1439: 43 to 30& ... & ... & ... \\

1440: 

1441: \enddata

1442: \tablecomments{\blast\ temperatures in the three flights, and as predicted by a Thermal Desktop (http://www.thermaldesktop.com/) model for \blastmcm\@.  The model predicted larger diurnal variations than observed, partially because \blastmcm\ followed a very southerly track, where the Sun elevation varied less than the simulation's track.  The \textit{Trop} column gives the minimum temperature the component reached during ascent through the Tropopause.  The \textit{Diurnal} column gives the daily range of temperatures.  In some cases, there was slowly increasing temperature over and above the diurnal variation due to UV degradation of the Mylar shields.}

1443: \tablenotetext{a}{The inside of the case was painted white for \blastmcm.}

1444: \tablenotetext{b}{Partially painted white.}

1445: \tablenotetext{c}{Painted white.}

1446: \tablenotetext{d}{Painted white + cooling loop.}

1447: \tablenotetext{f}{Packaged in a foam box.}

1448: \tablenotetext{g}{The model neglects current draw.}

1449: \tablenotetext{h}{Increased by 20\degree\/C over 12 days.}

1450: \tablenotetext{i}{Increased by 5\degree\/C over 12 days.}

1451: \tablenotetext{j}{Increased by 4\degree\/C over 12 days}

1452: \tablenotetext{k}{Measured at the lock motor near the elevation bearing.}

1453: \tablenotetext{l}{With cooling pump to heat exchanger.}

1454: \end{deluxetable}

1455: 

1456: 

1457: 

1458: 

1459: \begin{deluxetable}{lc}

1460: \tablewidth{0pt} 

1461: \small 

1462: \tablecaption{The steady--state power budget for \blast\@. \label{table:power}}

1463: \tablehead{ \colhead{Component} & \colhead{Power Consumption}  \\

1464: & \colhead{(W)}}

1465: \startdata

1466: \multicolumn{2}{l}{\bf ACS SIDE} \\

1467:  \hspace{.25in}Pointing Motors\tablenotemark{a}     &  85\\

1468:   \hspace{.25in}Flight Computers\tablenotemark{b}   &  55\\

1469:   \hspace{.25in}Star Cameras       &  55 \\

1470:   \hspace{.25in}ACS\tablenotemark{c}                &  30  \\

1471:   \hspace{.25in}Gyro Heaters\tablenotemark{d}       &  30\\

1472:   \hspace{.25in}Aux. Systems       &  15  \\

1473:   \hspace{.25in}Differential GPS   &   5\\

1474:   \hspace{.25in}BLAST06 Sun Sensor &   5\\

1475:   \hspace{.25in}LOS Transmitters   & 150 \\[2ex]

1476: ACS TOTAL             & 430\\[3ex]

1477: 

1478: \multicolumn{2}{l}{\bf DAS SIDE} \\

1479:   \hspace{.25in}DAS                   &113\\

1480:   \hspace{.25in}Preamps \& Cryostat    &113\\[2ex]

1481: DAS TOTAL               &225\\

1482: \enddata

1483: \tablecomments{

1484:  When slewing, the pointing

1485:   motors may have short-term excursions up to 560 Watts, nearly doubling

1486:   \blast's power consumption.  The LOS Transmitters are only on at the

1487:   start of the flight and do not contribute to the steady-state budget.}

1488: \tablenotetext{a}{Steady state}

1489: \tablenotetext{b}{Includes everything else in the Pressure Vessel}

1490: \tablenotetext{c}{ACS crate only}

1491: \tablenotetext{d}{Mean value}

1492: \end{deluxetable}

1493: 

1494: 

1495: 

1496:  \end{document}

1497: