OSU Logo The Ohio State University
College of Mathematical & Physical Sciences
Department of Astronomy

MODS Preliminary Design Review

PDR Committee Questions and MODS Team Responses

Updated: 2001 June 1

The MODS Preliminary Design Review for the LBT Science Advisory Committee will be held on Monday 2001 June 11. In advance of the meeting, the SAC chair, Tom Herbst, forwarded questions from the review committee. The questions are reproduced below, along with the responses of the MODS team. In some cases our full response will be presented at the PDR proper.

Note: The numbering scheme relates to broad categories, and not the sections in the PDR document.

The respondents on the PDR committee (so far) and the letters used to identify their comments are:

GB: Gary Bernstein
WS: Walter Seifert
GS: Gary Schmidt
GH: Gary Hill
TH: Tom Herbst
DF: Dan Fabricant
GV: Giampaolo Vettolani

Glossary of acronyms and abbreviations used below:

A&G: Acquisition & Guiding
EE: Encircled Energy
FCS: Flexure Compensation System
IFU: Integral Field Unit
LBTPO: LBT Project Office
LBTOSWG: LBT Optical Spectrograph Working Group


Contents

1. Optics
1.0 General
1.1 Field of View & Image Quality
1.2 MOS & IFU
1.3 Gratings
1.4 Camera
1.5 ADC
1.6 Alignment & Tolerancing
1.7 Calibration
2. Mechanical
2.0 General
2.1 Flexure Control System (FCS)
2.2 Thermal
3. Detectors
4. Software
5. Management Issues


1. Optics

1.0 General

1.0.1 (GH)
Describe the scattered light analysis, in particular with respect to red light in the blue arm. The scattered light analysis should be presented at the PDR.

Response:
Scattered light analysis requires a full design including baffling which is not yet ready at this time. As for red/blue cross-over, the instrument architecture allows rather complete cross-baffles, and is not one of the more difficult areas for baffling.

1.0.2 (TH) pg 14
Doesn't a slit four pixels and 0.6 arcsec wide raise the danger of the astronomical object acting as its own slit when the seeing gets very good? This can cause undesired wavelength shifts in the resulting spectra.

Response:
The simplest solution is to use one of the smaller slits (and get benefit of greater sky rejection), or just ignore it. For spectrophotometry, this is almost a virtue as you get less light loss. For precision radial velocities it is an issue, but in general one resorts to other techniques (e.g., image scrambling) when high radial velocity precision is required. High-precision radial velocities was not a design requirement of this instrument as developed by the LBTOSWG.

1.0.3 (TH) pg 22
The document mentions using a thin field lens in order to avoid defects associated with small bubbles or inclusions near the focus. What about dust?

Response:
The field lens is not sufficiently close to the focal plane that dust is a severe issue. Nonetheless, keeping this optic clean will be part of the maintenance procedures for MODS.

1.0.4 (TH) pg 22
Removing the dichroic will also introduce a defocus.

Response:
The collimator is designed to be refocused to compensate. Typical focus parameters for in and out positions will be determined during commissioning and included as tables in the observing documentation and included as settable parameters in the instrument control software.

1.0.5 (WS)
Will there be a possibility to implement polarization optics into MODS?

Response:
No.

1.0.6 (WS) pg 15, 1.5
In Table 1.1 I don't understand the meaning of the column 'goal'. E.g. the image quality given as goal cannot be reached at all with the present optics (see Table 2.1, p.23).

Response:
In general the instrument meets all of the requirements, but some of the more ambitious goals were not met by the adopted optical design.

1.0.7 (WS)
What is the distortion of the optics? What are the extreme values (not symmetric)? How does the distortion pattern change with instrument temperature (critical for preparation of MOS observations)?

Response:
Here is a map of the optical distortion introduced by the MODS optics. The distortion pattern is roughly radially symmetric. To first order, the magnification (ratio of collimator to camera focal lengths) is the same as they are both made of the same material (Pyrex), and respond to temperature in the same way (i.e., while the radii of curvature changes, the ratio is the same).

Accurate analysis of the distortion of any images used as templates for slit masks is required. We plan to include a reference mask in each cassette to aid in measuring any significant variation in the distortion.

1.0.8 (WS) pg 21, 2.3.8
How severe are the ghost images arising from reflections between detector and field flattener?

Response:
We will compute this and show the results at the PDR.

1.0.9 (WS) pg 24, 2.4.4
How much lateral color is present in the system? One can crudely estimate from the Figures in Appendix D, but some numbers or a plot would be helpful.

Response:
Here is a graph of the lateral color (PDF). It is unimportant for spectroscopy, but it is a significant component of the image quality in broad-band imaging modes.

1.0.10 (WS) pg 36, 2.7
Could you provide the manufacturer for that AR coating better than 98% from 320-1000nm on Silica? We are designing an optics which should also have extreme wavelength coverage and we have the problem to find a suitable coating on Fused Silica...thanks.

Response:
In MODS this is only an issue for the field lens which is common to both channels. The optics on the individual channels will be optimized for the relevant (smaller) blue and red ranges, respectively. For the field lens we are considering using SolGel over MgF for this element, and hope to achieve the performance described by Stilburn (2000, SPIE, 4008, 1361).

1.0.11 (GV) section 1.5
Goals and requirements. Why is the throughput requirement 50 per cent for the range 350-900 and not 320-900 . The UV is one the qualifying point of MODS. I fully agree on the goal, but if it is not reached I would like to see a stringent throughput requirement for the 320-350 range. If there is no requirement one can go to transmission optics directly!

Response:
We agree that the UV response of MODS should be as high as possible. We believe that our design is consistent with this goal.

[Contents]


1.1 Field of View & Image Quality

1.1.1 (DF)
I think that the current optical design is aiming too low in image quality. MODS will be competing in a tough arena against very capable instruments at the largest telescopes. Many competing spectrographs have far larger fields of view. MODS's strengths appear to be broad spectral spectral and high efficiency. I would be far more comfortable if MODS could offer superb image quality across the modest 6 arcminute diameter field. Table 2.1 on page 23 shows that the performance is entirely driven by the collimator. My advice is to take another hard look at alternative designs even if they are more complex. The Magellan telescope is delivering 80% power into less than 0.2 arcsec, and this may get better yet. A spectrograph that delivers 1 arcsec (80% EE) at the edges of its field of view doesn't seem competitive.

Response:
We believe that the highest priority objective for MODS is sensitivity to faint objects. Thus, we tend to consider high throughput our most important objective in the design; adding additional optical elements to improve image quality over a large field is not a high priority. We note that over a 4 arcminute field the instrument will deliver reasonable images. Since the sky can be rejected at the slit, this corresponds to a small loss in sensitivity in excellent seeing.

Furthermore, we note that over a relatively small field (roughly 1-2 arcminutes) the optical design performs very well. In particular, with refocus the instrument can make use of images as good as 0.3 arcseconds FWHM without excessive degradation of the image quality. See figures D.10 and D.11 for example.

We wish to note that 0.2 arcsecond 80% encircled energy corresponds to diffraction limited images at ~5 microns or Strehl ratios of roughly 30% at K on Magellan. Gemini achieves ~10% Strehl and 80% encircled energy of ~0.5 arcseconds at K with a 36 actuator AO system. This performance is excellent, and we would like to hear more about Magellan at the PDR.

1.1.2 (GH)
Encircled energy curves (EE) for the various fields and wavelengths would be useful in evaluating the optical performance. The EE(80%) or D80 values are quite large (1 arcsec) by the corners of the primary science field (4x4 arcmin): is this acceptable for the science goals? What about the corners of the 6x6 arcmin extended field? The polychromatic performance with the gratings should be presented.

Response:
There are extensive performance curves presented in the PDR document (see especially Appendix D).

1.1.3 (WS) pg 14/15, 1.5
I think that for faint object spectroscopy and imaging, the present optical quality of the instrument is limiting the performance. With D80<0.54 arcsec (Table 2.1) within the 4 arcmin FOV, the resulting image for 0.6 arcsec seeing will degrade to something like 0.8 arcsec, i.e. 33% increase of D80. And even for 1.0 arcsec seeing, the degradation will be more than 10%. What do you mean by the statement in the text on p.14 'images will not be substantially degraded by the instrument'?

Response:
An 80% encircled energy diameter of 0.54 arcseconds corresponds roughly to a FWHM of less than ~0.3 arcseconds. Thus, the degradation of 0.6 arcsecond (FWHM) seeing is small (less than ~10%).

Note that for a typical seeing profile (a Gaussian core with extended wings), 0.9 arcsec 80% encircled energy diameter corresponds to ~0.6 arcsecond FWHM.

[Contents]


1.2 MOS & IFU

1.2.1 (GB)
Slit masks (p. 48): Has the GMOS material/method been demonstrated to meet the specifications required for MODS? This is an area with potentially long lead times and R/D requirements.

Response:
The GMOS strategy seems adequate for MODS, we are watching their progress very carefully.

1.2.2 (GH)
Two-telescope operation modes are powerful, but require very precise alignment of the two spectrographs on the sky, and very high repeatability of slit mask insertion. The commonly used mode for multi-object masks is to allow some RA,dec,rotation adjustment of the telescope/instrument on the sky in order to align the mask with all the objects. In MODS the requirement is that each mask be accurately inserted to within some tight constraints on position such that the setup for one automatically aligns the second spectrograph. One could imagine allowing independent rotation of each, but the rotation axes would have to be very accurately aligned and the operational overhead would be large. This issue needs to be captured into a specification for insertion accuracy/repeatability for the slit unit (below).

Response:
Since the two LBT primary mirrors can be pointed independently (to separations up to ~1 arcminute) and the rotators are also independent, the alignment of the two spectrographs can be done independently. We do not believe that precise slit mask insertion will be required.

1.2.3 (TH)
Has an IFU option been preserved. Is it still possible and, if so, how?

Response:
Space above the focal plane has been reserved for an IFU option, but an IFU is not part of the instrument baseline configuration.

1.2.4 (WS) pg 7, 1
What kind of IFU do you have in mind to be implemented into the instrument as a possible upgrade? With fiber IFU's there will be the problem of the focal ratio mismatching (needs post-optics which reduce the efficiency), and a slicer IFU needs a significant amount of space which has to be available.

Response:
See answer to 1.2.3

1.2.5 (WS) pg 48, 3.3.1
How will spherical masks be produced? I can imagine that the cost is quite high. Wouldn't also cylindrical masks be sufficient with regard to the lower image quality in the extended field?

Response:
There are a variety of techniques for making them (e.g., graphite epoxy vacuum formed onto a mandrel). We can discuss blank mask creation at the PDR. We feel that the masks should be spherical so as to match the shape of the telescope focal plane for the full FOV of MODS.

1.2.6 (GV)
Multislit mode. Quantify the reduction of the spectral coverage due to the object position in the field. Which is the area for which full (within 10 per cent) spectral coverage is possible? Being a two channel instrument this can cause difficulties in combining spectra in the red and blue (gaps in between).

Response:
You lose roughly 500 pixels per arcminute away from the center of the field (out of 8000 pixels total). For R=2000 mode, we can get most of the spectral region over the entire 6x6' FOV.

1.2.7 (GV)
Slits and Masks: The discussion of these two fundamental components do not exist. At the meeting it must be presented a convincing study of their feasibility. Two crucial points are indicated below: 1. Slits machining precision is fundamental for sky subtraction at the instrument limit Which is the goal set for slit edge quality ? With a nodding technique when observing and slits with roughness less than 1 per cent one gets half mag more as limit in the red where sky is dramatic (and fringing of the CCD also) 2. Masks (slide 25). What is intended for mask substrate? A substrate (plastic or what ?) deposited on the mask material (carbon fiber ) in order to transform an otherwise flat mask into one with the same radius of curvature of the focal plane? Is this proved feasible ? Is it a simple industrial process or a special manifacturing process (cost can explode) ? To my understanding cutting a spherical mask is not trivial at all. This process needs clarification.: feasibility, cost etc

Response:
We will include a suite of long slit masks that will stay resident permanently in the cassette insertion mechanism. These long slit masks can be made almost arbitrarily precise using any one of a variety of techniques. They will not just be blank masks with long slits machined into them.

We can discuss our preliminary plans to create the multi-slit masks at the PDR meeting. We currently envision the masks to be pre-formed graphite-epoxy shells machined with a 3-axis laser system. There are several processes that can create these shells, and we do not think they will be a cost driver. The laser machining unit may be a cost driver, and we hope to share part of the cost with the LUCIFER project.

[Contents]


1.3 Gratings

1.3.1 (DF) pg 26, section 2.6
Don't underestimate the required time scale for ruling new large gratings. These gratings may well cost $100K as well.

Response:
We are not using new rulings for first light (see section 2.6 in the PDR document), but will be using new rulings for higher resolution modes to be deployed later in the project.

1.3.2 (GB)
Gratings: (p. 43): It is stated here that gratings will be ruled onto Zerodur, which presumes you'll be having RGL produce new master gratings. Why then was your analysis restricted to catalog rulings? New masters could be another long lead-time item; I don't know how long this takes to get from RGL.

Response:
To clarify, we will use catalog gratings for the R=2000 modes for first light (see section 2.6), whereas later high-res (R=8000) modes will most likely require new master rulings. We mispoke on page 43, we meant to say "gratings will be replicated onto solid Zerodur substrates", not directly ruled onto Zerodur proper.

1.3.3 (GH)
I'd appreciate a table of grating properties.

Response:
See PDR document section 2.6, table 2.2

1.3.4 (GS)
Gratings are said to be "ruled onto solid Zerodur blanks" (p43). Yet the examples shown are Richardson Large Astronomical (replica) gratings. What is the actual plan?

Response:
See response to 1.3.1

1.3.5 (TH) pg 26
I could not help noticing that the MODS implementation plan restricts itself to a few available gratings - new rulings are well understood - while at the same time planning for a totally new CCD development. Also, on page 43, the document states that the team plans to use gratings ruled onto solid Zerodur while this section mentions replicas.

Response:
See response to question 1.3.2 and section 3 below

1.3.6 (WS) pg 25, 2.5.2
What grating parameters have been used to produce Fig. 2.5?

Response:
The reference low-resolution gratings currently available from RGL.

1.3.7 (GV)
Gratings (page 27) My overall impression is that new rulings are necessary. No high dispersion in the blue and the transmissions look quite odd. Anyway these are gratings to be used in Littrow configuration. I think this is a point to clarify fairly well at the meeting.

Response:
We will clarify this point at the meeting.

1.3.8 (GV)
Throughput. Taking also into account the above considerations, it is shown a curve for the throughput with "predicted gratings". What does it mean. It is necessary that curves with the predicted throughput with the real gratings in the Richardson lab catalogue is shown.

Response:
By "predicted" performance we mean the output of GSOLVER for the stock RGL gratings used in the correct configuration for MODS.

[Contents]


1.4 Camera

1.4.1 (GH)
The camera corrector is potentially difficult to manufacture and align. Please discuss the trade-offs that lead to the choice of a singlet corrector with asphere rather than a doublet corrector with spherical surfaces. Describe the manufacturing, testing, and alignment strategy for the corrector element.

Response:
See PDR document (especially section 2.2).

1.4.2 (GH)
Is there a null-test for the off-axis section of the corrector? Is the plan to manufacture several cameras at once and hence manufacture the corrector as an on-axis system before cutting out several correctors from it? More details are needed on this.

Response:
Several optical manufacturers have indicated their willingness to produce the corrector to our specifications. Two correctors can be made from a single blank as an on-axis element; testing can be done either with profilometry or, if necessary, using null-tests.

1.4.3 (TH) pg 21
Were filters included in the optical design process for the camera?

Response:
The filter substrate material was included in the design, and all filters will be made to the same optical thickness.

[Contents]


1.5 ADC

1.5.1 (TH) pg 19
Does the configuration of the LBT instrument mount allow for 900 mm in front of the focal plane for an atmospheric dispersion corrector? This may conflict with the auto guider system. Also, doesn't locating in the ADC so far from the pupil cause undesirable pupil blurring and/or shifts?

Response:
Yes, there is enough room. We are not using the Potsdam/Steward AGW system, we are using our own on-board A&G system because of the way that our grating turrets and mask insertion mechanism "crowd" the space above the MODS focal plane.

1.5.2 (WS) pg 19, 2.3.1
An ADC of 200 diameter will cover a FOV of 4 arcmin diameter, not 4x4 arcmin. What is the thickness of the ADC? It will require rather large refocussing of the telescope. Please clarify with John.

Response:
Yes, the ADC requires refocusing of the telescope, we believe the focus travel is sufficient.

[Contents]


1.6 Alignment & Tolerancing

1.6.1 (GB)
Optical tolerances (p. 37): equivalent r0's certainly don't ADD in quadrature as the combined effect of two surfaces has equiv. r0 which is smaller than either surface alone. (just an editing comment here...)

Response:
Thanks.

1.6.2 (GH)
I would like to see a detailed tolerance analysis presented at the PDR along with a discussion of risk areas and plans to mitigate risk. This applies in particular to the camera. Tolerances will be tighter for use with AO, and this should be factored into the analysis.

Response:
This was discussed in the PDR document in section 2. With a suitable focus the inner 1 arcmin FOV delivers AO-ready images.

1.6.3 (TH) pg 37
I am unfamiliar with the r0 technique of specifying and tolerancing optical components. Does this method include separate consideration, sensitivity testing, and tolerancing of tilts, decentrations, refractive index variations, etc.? Also, I am troubled by the absence of the ADC, field lens, and other flat surfaces in the error budget. Doesn't a tolerance and error analysis include tests of the sensitivity to errors in the radius of curvature? A flat surface also has an (infinite) radius. Tilts and lack of parallelism will also certainly have an effect.

Response:
We will discuss this at the PDR.

1.6.4 (WS) pg 38, 2.8
What is meant by the column 'Optical Support' in Tables 2.3 and 2.4?

Response:
This is the change in r0 due to the gravity induced distortion of the optic as supported as indicated by preliminary FEA of the optics supports.

1.6.5 (WS) pg 38/39, 2.9
Which optical component was misaligned by 0.5mm to produce Fig. 2.16? If there are several components displaced and/or tilted, then it will be a bit difficult to perform the iterative alignment only from the PSF itself. But I think that the mechanical accuracy for aligning the system is much better than 1mm. A sensitivity table should be provided to give the information about critical components.

Response:
There is an error in the PDR document. The appropriate section and figure caption should read:

2.9 Optical Alignment Strategy

The accurate alignment of the optical components of MODS is critical to the successful operation of the instrument. We plan to fully configure and align the instrument in the lab (using the high bay space available for instrument assembly). The camera optics can be aligned separately. The procedure will begin with careful mechanical alignment of the structure and optics. We estimate that standard techniques should allow the optics to be aligned to less than 1 mm. We will then illuminate system with an expanded collimated He/Ne laser source and measure the resulting on-axis images using a lab CCD camera. This CCD camera will over-sample the images and provide an accurate representation of the on-axis monochromatic point-spread-function. We will then compare this on-axis point-spread-function with CodeV predictions. The left panel of Figure 2.16 shows the PSF of a perfectly adjusted camera for the above conditions. This will actually be an out of adjustment condition for wide field spectrographic illumination. A smaller overall error function results when the camera is adjusted to produce an on-axis PSF corresponding to the right panel of Figure 2.16. This change corresponds to an added decenter of the corrector lens of 0.99 mm and a refocus of the camera mirror of 0.258 mm. We have used this technique to align a variety of optical systems in the past (e.g., offner relays, collimator-camera lens systems, etc.).

Revised caption:

Figure 2.16: Best on-axis monochromatic PSF (left). Optics re-aligned (right) for best polychromatic wide field performance. The PSF of the re-aligned optics has an 80% encircled energy diameter of <30-m.

[Contents]


1.7 Calibration

1.7.1 (GB)
Guiding/Calibration (p 49): What are the plans for guiding and for calibration lamps?

Response:
The acquisition and guiding plan is still preliminary as we are still developing the A&G system concept.

We have prepared a Draft Calibration Protocol for MODS to provide us with the requirements for the calibration system. It was also circulated to the LBTOSWG for comment. This aspect of the MODS design is still in development.

1.7.2 (TH) pg 39
Where does the team plan to install a calibration unit?

Response:
Plans for the calibration unit are still under development. There is adequate room for the calibration unit above the slit.

1.7.3 (WS) pg 39, 2.10
From the experience with FORS we know that it needs quite an effort to provide a good calibration illumination for a large field MOS spectrograph. Thus the design of the unit should proceed as soon as possible in order to find a suitable place for it.

Response:
The calibration unit design is proceeding, and we have left sufficient room for it above the slit. Final design must of course await completion of the full structural design, but room for the system is being factored in.

1.7.4 (GV)
The calibrations of the instrument (lamps etc are not discussed)

Response:
See previous answers.

[Contents]


2. Mechanical

2.0 General

2.0.1 (GH)
I'd like to see a detailed description of each of the optics supports at the PDR.

Response:
See the PDR document, specifically sections 3.2.2.

2.0.2 (GH)
Please describe each of the mechanisms in detail at the PDR. I am particularly concerned about the repeatability of the slit mask insertion mechanism. The 3-D solids representation seems to imply that the mask is cantilevered out significantly. I suspect that some kind of registration scheme is envisioned, but this was not discussed in the overheads. My concern is absolute registration repeatability of the masks, since non-repeatability will compromise the interesting 2-spectrograph modes (as noted above in question 1.2.2). It is also important for calibration of long-slit observations.

Response:
See the PDR document, Section 3 for the descriptions. As for mask registration, the mechanisms we are using can deliver 5-10 micron repeatability. Since the plate scale of the LBT 8.4m f/15 Gregorian focal plane is ~1.6 arcsec/mm, 5-10 microns corresponds to ~0.008-0.016 arcsec.

2.0.3 (GH)
The grating turret is complicated with compound motions (select and tilt). I'm pleased to hear that there is a prototype of the tilt mechanism, and I would like to see it at the PDR, along with test results from the prototype.

Response:
A tour of the lab will be part of the PDR on June 11. In the meantime, see MODS Views for current pictures of this unit.

2.0.4 (GH)
Does the change in mass of liquid nitrogen through the night effect the alignment of the instrument?

Response:
This is not an important contributor to the alignment error budget.

2.0.5 (TH) pg 21
In figures 2.2 and 3.9, there seems to be very little space for the vacuum cryostat needed by the detector. Isn't there a danger of vignetting?

Response:
No, the cryostat is not in the beam.

2.0.6 (TH) pg 52
You seem to plan for only 7.5 l of liquid nitrogen. Why not more? Also, have you considered the implications of the boil-off gas on instrument "seeing" and re-filling the dewar on telescope operations? The instrument may be hard to reach.

Response:
Only the CCD and IR alignment sensor are cooled by LN2. We estimate that 7.5 liters of LN2 is adequate to keep these devices at operating temperatures for >24 hours, so filling of the dewar will be required no more than once per day.

2.0.7 (TH) pg 81
I would be very interested to hear more about the telescope simulator for flexure testing.

Response:
It will be a full two-axis test fixture, with the instrument supported exactly as it will be on the telescope.

2.0.8 (WS)
Which type of shutter will be used? Where will it be located?

Response:
The shutter will be a two-vane cantilever shutter located at the camera entrance aperture. A detailed design will be presented at the PDR.

2.0.9 (WS) pg 53, 3.3.8
Is there really enough space for the dewar? Was the weight of the fully loaded dewar included in the FE calculations for the image motion?

Response:
Yes, there is enough room for the dewar. It is not part of the camera structure and therefore has no impact on the motion of the images.

2.0.10 (GV) section 1.5
Stability of 0.5 pixel as requirement is definitely too high. The goal is perfect but if not reached the instrument is a failure because in 4 hours exposures (without recentering) half object light go outside the slit (assuming 4 pixels 0.6 arcsec). If the seeing is brought to 0.3 through some device the situation is even worse!

Response:
The 0.5 pixel per hour specification is for the flexure of the instrument and motion of the image on the science detector, not for the pointing/tracking precision of the telescope. Note that simple addition of the flexure over four hours is almost certainly incorrect, since the gravity load on the instrument cannot change by more than a factor of two over this time. Furthermore, individual exposures will not typicall exceed 1 hour due to concerns about cosmic rays. A series of images will usually be acquired.

[Contents]


2.1 Flexure Control System (FCS)

2.1.1 (GB)
Flexure Control (p 46): Have you considered in any detail how the IR source will be integrated into the focal plane, given that the focal plane will be filled with interchangeable slit masks? The motion of the IR spot centroid will provide feedback for image motion due to flexure. What about defocus, and aberrations due to tilting of elements - do the FEA models show these effects to be negligible from flexure? I don't see how one could easily use the IR spots to compensate for such effects, if they are significant.

Response:
The IR source will be located outside the science field, and there is adequate room for it at that location. The FCS will not be sensitive to defocus or aberrations and optical modeling shows that these effects are negligible. There will be collimator focus adjustment based on temperature using a look-up table, whereas the camera is athermal to first order.

2.1.2 (GH)
Flexure control is a serious issue in an instrument of this size. The IR-bypass grating idea is potentially an excellent solution. Since it is innovative, I'd like to see a detailed presentation of the open and closed loop flexure compensation strategies at the PDR. Particular questions are: _ is there focus/image shape distortion associated with temperature changes and flexure and with the compensation for the flexure? _ If so, can it be compensated? _ Where is the IR reference beam injected into the optical train? If it is post slit, will the exclusion of the slit mechanism and surrounding structure effect the ultimate accuracy of the feedback? _ where is the IR detector located in the focal plane, relative to the optical CCD?

Response:
See previous response.

2.1.3 (GH)
Collimator and camera focus are other mechanisms that need detailed discussion: is it expected that the IR- beam flexure compensation will provide feedback for aligning the whole optical system following configuration changes? The information provided mentions requiring different collimator/camera focus combinations for different spectrograph configurations. I'm concerned about maintaining alignment of the optics under constant adjustment of positions.

Response:
The FCS will be designed so that it will not "walk" away from correct system alignment. Note that we do expect that the FCS will provide the useful ancillary benefit of allowing for precise configuration alignment (i.e. grating position reproducibility).

2.1.4 (GS)
The Flexure Control System is an interesting concept, but difficult to evaluate as to its performance. Can the IR laser be centroided to the required accuracy? Proper collimation of the camera mirror and corrector are stated to be critical to its performance, yet the FCS system plans to remove image motion by tilting camera mirror. If image motion is not introduced by misalignments in the camera, but instead by grating or collimator motions, this will result in misalignment of the camera. Have these effects been evaluated quatitatively? I am always leery of active alignment concepts using reference beams, after that grand scheme of maintaining alignment in the original MMT was found to be unworkable.

Response:
An IR laser emits adequate power so that centroiding should be extremely accurate (<0.1 pixels). An analysis with CodeV shows that the small misalignments of the system introduced to compensate for image motion have negligible effect on the image quality.

2.1.5 (TH) pg 47
Doesn't using an infrared detector for the FCS require that the camera operate well to 1.5 microns? Also, there seems to be very little space in the camera focal plane for any additional hardware. How will this work with the dichroics?

Response:
The camera works adequately well at 1.5microns. The footprint of the IR spot on the dichroic is very small and its location (outside the science beam) will be coated so that particular location acts as a simple beamsliptter.

2.1.6 (TH) pg 56
Will the IR spots provide defocus information?

Response:
No.

2.1.7 (WS) pg 46, 3.2.5
If I understand correctly then such a IR system for the FCS cannot work for both channels of MODS at the same time? Are the flexures in both channels very similar? Keep in mind that there is also flexure in the CCD/IR sensor mounting. This can be in the order of several microns. Thus I would not expect to reach a performance of better than 0.5pixel. But this should be completely ok for 0.6 arcsec = 4 pixel slit width and even for a 0.3 arcsec slit.

Response:
See response to 2.1.5. The CCD and IR sensor are mounted on the same block of metal in the detector system and there is no relative motion.

2.1.8 (GV)
The flexure compensation system is a crucial point for this spectrograph. A prototype must be shown to be working before the instrument is built. If it does not this spectrograph does not work as a whole allowing exposures of a few minutes only without recentering.

Response:
We recognize the importance of the FCS and a prototype system is currently under development.

[Contents]


2.2 Thermal

2.2.1 (DF) pg 40, section 3
This is a very large instrument with lots of motors. Will the spectrograph be used as an imager with the associated frequent filter changes? I would undertake a thermal study to try to understand temperature gradients and their effect on stability. I would be nervous about counting on a flexure control system to solve all problems.

Response:
All motors are idle during exposures. The duty cycle of the highest use mechanisms (probably the filter wheels when the instrument is in use as an imager) is still quite small. The thermal loading due to the motors is negligible.

2.2.2 (GB)
Structure (pp 45-46): Is there some estimate as to whether the use of stainless steel in the primary structure will make the thermal defocus of the collimator sufficiently small? Aside from the camera supports I didn't catch much discussion of thermal stabilization of the structure.

Response:
The FCS will compensate for temperature-induced image motion by design, but not focus, which we'll do using a look-up table.

2.2.3 (GV)
There is no thermal analysis. As steel as a coefficient of 2 to 8 microns per meter per degree and the average cooling at Mount Graham is 0.25 deg per hour (5-6 degrees in a night) it would be necessary to know the behaviour of the instrument, the critical points, and how they think to compensate (with the active flexure compensation system I guess).

Response:
See previous response.

[Contents]


3. Detectors

3.0.1 (GB)
Detectors (pp 58ff): The choice of CCD device will have to be made very soon and the discussion here is quite open-ended. The "ideal" detector has 4k format, 32 amps per chip, skipper, and frame transfer - a long way from anything currently produced, as noted. Starting a new design and mask set would make for a very long lead time and high risk - do the desired features really justify this expense and risk? For example, under exactly what circumstances will read noise below the commonly-available 4e- really be of help? If this is only needed for faint sources + AO + high resolution, then such detectors won't be of help until a laser-guide-star system is operating, which could be well into the future. Likewise, what circumstances require a 64-channel readout? The project should investigate carefully whether a CCD development project is required, and this decision needs to be made now or very soon to have a 2004 deployment.

Response:
As pointed out in section 5.2.2, there are many CCD detector options for MODS. The ideal detectors we discuss are most useful in low background modes (high resolution, narrow slit, etc.) and we believe there is a large amount of interesting science that could be done with detectors that are available "off-the-shelf". We hope to have made concrete detector selections by mid-2002.

3.0.2 (GH)
How do the science objectives drive the choice of CCD format? In particular, what are the drivers to use frame transfer and skipper amp technologies? I don't see a high speed spectroscopy or imaging mode discussed, so presumably the goal is to reach the lowest read noise. This is probably needed for the blue arm, but maybe not for the red.

Response:
See our response to question 3.0.1

3.0.3 (GH)
The frame transfer requires a lot of CCD area that is not used for imaging in other modes. Is an even wider field mode (with poorer images) available that could use this extra area effectively, when sky noise dominates the exposures?

Response:
No. The FOV is limited by the available aperture in the focal plane and the relatively slow focal ratio of the telescope (f/15).

3.0.4 (GH)
With frame transfer plus skippers, and nodding of the telescope, an interesting observing mode is enabled, similar to observing in the IR, since the read-noise penalty of splitting the exposure into many parts is eliminated. This mode would allow the object(s) to be nodded back-and-forth along the slit(s), obtaining a perfect set of interspersed sky exposures, thus eliminating systematics in sky subtraction in the red part of the spectrum.

Response:
We agree that the availability of the "ideal" detector described in the PDR document would likely lead to many interesting and innovative observing modes.

3.0.5 (GH)
What trades were needed in the optical design to accommodate the extra area of the CCDs used for frame transfer?

Response:
The detector is located slightly more off axis than is needed by the 6x6' FOV to accommodate the extra frame transfer area. This slightly degrades the camera performance, but is insignificant to the overall optical performance of the system.

3.0.6 (GH)
I see significant risk to budget and schedule from developing (in particular) the 4x8K monolithic device. Even 4x4K is not a readily available format and yield may be quite low. The fall-back option of two 2x4K devices butted length-wise to produce 2x8K with an accompanying increase in the speed of the camera, should be explored and presented as a fall-back. 4x4K without frame transfer seems not to be a good fall-back choice because the yield will likely be lower (i.e. cost more) and with the current design a lot of that CCD area will not be used.

Response:
This was addressed in the PDR document, Section 5. Our baseline plan is to deploy the first-light MODS with available CCDs, whereas the frame-transfer CCDs will require more time and be deployed later.

3.0.7 (GH)
I would like to see data from the C3D CCD presented at the PDR to demonstrate the amplifier technology. The Lesser QE curve is presented for room temperature. Please show the equivalent for the operating temperature as the red QE will probably be significantly lower. Describe the HiRo technology and discuss the trade-offs for the various technology choices along with a risk assessment.

Response:
CCD quantum efficiency is slightly lower at realistic operating temperatures. We can discuss the possible availability of detectors with excellent red response at the PDR.

3.0.8 (GH)
The CCD electronics section needs more details. It is too cryptic in the single overhead (of course).

Response:
See section 5 of the PDR document.

3.0.9 (TH) pg 59
A motorized mask for frame transfer seems like quite a difficult cryo-mechanism, particularly given that there is some almost no space for mechanisms near the focal plane. Will there also be a shutter?

Response:
The shutter is located at the camera aperture. We agree that the mask is a difficult mechanism given the space constraints, but we have designed and deployed similar cryo-mechanisms in the past.

3.0.10 (TH) pg 61
Are these really room temperature QE curves? What happens at operating temperature?

Response:
The red-wavelength cutoff of the detector moves towards the blue slightly as the operating temperature decreases.

3.0.11 (GV)
I would like to see also the informative plot which is usually shown for instrument evaluation namely: exposure time versus magnitude both for point sources and extended sources to reach a give signal to noise (typically 5, 10 and 20) at 5 sigmas (in the continuum for spectroscopy) This for all observing modes. A similar plot for one observing mode was shown previously (Slide 9) but now has been dropped

Response:
We will make these for UBVRI imaging and R=2000 spectroscopy and have them ready at the PDR. These are modes for which we have sufficient information to create these curves at this time.

[Contents]


4. Software

4.0.1 (GB)
Software (p. 65): is it the intent to have RS-232 communication between DOS PC's as the low-level elements of MODS? I would be concerned about maintaining a viable DOS system for the next decade given that it's already several years obsolete.

Response:
MODS will indeed use RS-232 standard serial connections for low-level command and control communications between DOS PCs. However, we use dedicated high-bandwidth links for raw data transport from the CCDs (either fiber optics or dual-ported SCSI disks).

We expect that DOS will remain available and useful for a long time to come. Far from being obsolete, DOS is found on virtually every desktop machine (do a search for "command.com" on any Windows 95/98/2000 PC and execute it). For us, DOS is particularly appropriate to the instrument computers in MODS, which, despite their physical resemblance to desktop PCs, have more in common with embedded systems: they run a single real-time program that performs the same set of activities continuously for the entire life of the instrument. In the embedded systems world, DOS remains popular for its small footprint, meager system requirements, and exclusive real-time tasking.

4.0.2 (TH) pg 57
How many total computers will MODS have and how will they interact?

Response:
Although the formal number of computers used in MODS appears to be large, most are better thought of as embedded controllers (similar to those found in modern washing machines). These will typically interact over RS-232 lines at very low data exchange rates. There will be one computer per CCD detector system (4 for the complete MODS), one computer to act as the "instrument server" for each MODS (see section 6 of the PDR document). The observer's console will be an LBT-provided observer's workstation that will run our interface software to the instrument servers.

4.0.3 (WS) pg 65ff, 6
I did not find a description on the preparation software for the MOS observations, i.e. setting up the MOS field, selecting the targets, producing the masks. A suitable and easy to use software for defining the masks is absolutely necessary for a facility instrument and the efficient use of the observing time.

Response:
Agreed. We gave a preliminary sketch of one of the observing planning tools in Section 6.3.3 of the PDR document. Our experience has shown that observing procedures are best worked out in detail by actual practice as we learn how to observe with MODS and LBT during the Phase-I commissioning activities. Observing preparation and planning tools will be delivered with the first full 2-channel instrument a year after commissioning.

[Contents]


5. Management Issues

5.0.1 (DF) pg 58, section 5
Beginning new CCD foundry runs in 2002 to produce detectors for an instrument that will be integrated in 2003 seems very risky.

Response:
We agree, the first deployment will likely use currently available detectors for the initial implementation of the instrument.

5.0.2 (DF) pg 77, section 7
I see an ambitious schedule, many new mechanisms, and two mechanical engineers. These mechanisms will require lots of thought, prototypes, and detailed FE analysis. Two years and two engineers for the whole works seems very light.

Response:
The mechanism count for MODS is only roughly twice what we have deployed on several previous instruments. Furthermore, many of the mechanisms are based on designs from previous instruments. We agree that the schedule is demanding, however.

5.0.3 (GB)
Plan (p. 15): it is not clear what phases of the MODS deployment are currently funded, or how much of the full twin-double spectrograph elements will be or needs to be fabricated. I see the optics bid package specifies all four channels, and the plan will require two structures and two mask changers plus at least 3 grating & filter changers. So just the first two phases will require nearly the full design effort and nearly the full fabrication effort - is this funded?

Response:
The financing plan for MODS will be discussed at the PDR.

5.0.4 (TH) pg 48
Does the LBT Project really plan to provide software to control laser machining of multi slit masks?

Response:
By "project" we meant "MODS project".

5.0.5 (TH) pg 77
I agree. David is very smart :)

Response:
David wishes to tell his side of the story, and will do so at the PDR on June 11.

5.0.6 (TH) pg 93
You have requested quotes for optics for both instruments, yet only one is funded. Can you estimate the non repeating cost savings of building both at once?

Response:
The large optics can be made in pairs from a single substrate, so the additional costs are relatively small for this particular purchase. Both instrument (and indeed most of both channels of each instrument) are identical, so additional design time is negligible. Our implementation plan calls for the construction of both structures and some additional mechanisms. The major cost of building the second MODS is in gratings, detectors, and additional optics (flats, dichroic, etc.).

5.0.7 (WS) pg 80, 7.5
There is nothing mentioned about testing of the optical performance and the characterization of the instrument performance in the lab and at the telescope during commissioning.

Response:
The instrument optics will be thoroughly tested in the lab before shipment to Mt. Graham.

5.0.8 (WS) pg 81, 7.5.3
Even though scientific observations are valuable for exploring the instrument capabilities, a detailed testing of the characteristics and performance of the instrument at the telescope are of at least same priority. All modes, grating parameters, filters, image motion etc. should be tested under observing conditions. 'Scientific useful' targets are well suited only for a few of those tests. There should be a clear programmatic division between commissioning of the instrument and the science verification.

Response:
This is clearly a philosophical issue and could be discussed at great length. In a sense, any "instrument capability" that is not useful for science observations does not need to be tested! Obviously, there will need to be a systematic exploration of the instrument performance and we anticipate ample engineering time for these system tests.

5.0.9 (GV)
I strongly recommend an acceptance review by external reviewers before shipment to the Observatory. In this review the full functionality must be demonstrated through tests agreed in advance between the builders and the panel of reviewers

Response:
This is a good idea, we will discuss this with the LBTPO.

5.0.10 (GV)
Nowhere is mentioned the spare parts problem

Response:
We plan to provide a complete set of spares for MODS. This will include spare electronic components, spare mechanism components, and, as appropriate, spare optical components.

5.0.11 (GV)
I do not know if there are peculiar safety regulations at Mount Graham, but no safety is discussed.

Response:
We are unfamiliar with specific safety concerns on Mt. Graham, but we plan to work with the LBTPO to ensure compliance with appropriate regulations.

[Contents]


Return to: [ MODS Project Page | OSU LBT Page | OSU Astronomy Home Page ]
Updated: 2001 June 1 [rwp]