Author Topic: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD  (Read 32183 times)

Offline Star One

  • Senior Member
  • *****
  • Posts: 7480
  • UK
  • Liked: 1202
  • Likes Given: 168
SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #60 on: 07/28/2017 06:43 PM »
http://spacenews.com/cameras-on-nasa-exoplanet-spacecraft-slightly-out-of-focus/

Quote
The TESS team thinks there will be a 10 percent cut in terms of the number of planets that they expect to be able to detect.
Quote
Despite the reduction, Boss said TESS scientists believe they will still be able to meet the mission’s primary science requirements, and thus there is no need to fix the cameras.

Quote
Jeff Foust @jeff_foust
Updated our article on TESS spacecraft camera focus issue with additional NASA comment playing down the problem:

https://mobile.twitter.com/jeff_foust/status/890951536141975553

From the revised article.

Quote
Chou added July 28 that the out-of-focus area is limited to the outer edges of the image, and that “recent testing shows that the camera focus towards the image center is better than originally designed.”
« Last Edit: 07/28/2017 06:45 PM by Star One »

Offline Semmel

  • Full Member
  • ****
  • Posts: 958
  • Germany
  • Liked: 672
  • Likes Given: 1955
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #61 on: 07/28/2017 07:25 PM »
Well, that makes sense. I am working on scientific instruments, not on TESS though. It is common that the focal surface of some optical system is curved (spherical) but the detector usually flat. There are curved detectors but they are experimental and not used (yet). So imagine a spherical focal surface intersected by a flat detector surface. If you minimize the offset between both surfaces everywhere, both the center and the edges of the flat detector surface are somewhat out of focus. The optimal focus is on a ring somewhere in between. If the optics is such that the edges are more out of focus, by geometry, the center is more in focus, meaning the flat detector surface is closer to the spherical optical focal surface.
For this reason, usually the detector or optics can be pistoned just a little bit to adjust the focus for operational conditions. This is done manually in commissioning and not touched later. Unless the conditions change of course. I am surprised that TESS does not have such a system.

Offline jgoldader

  • Full Member
  • ****
  • Posts: 640
  • Liked: 203
  • Likes Given: 124
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #62 on: 07/29/2017 12:16 PM »
There's a ppt about camera construction and testing here at NTRS:
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20160010628.pdf

The cameras are f/1.4, so I expect focusing is going to be dodgy to begin with, with significant field curvature.  As Semmel noted above, the focal "plane" is curved, but the CCDs are not.  The best you can get is some optimal focus, which in the perfect world would cause only minimal PSF (point-spread function, basically the shape of a point source, like a star) variation across the field.  All this can be modeled (see the NTRS paper above, there are PSF models on p. 28) and surely, they thought the optical system design was adequate.  From the SpaceNews article, it sounds like the CCDs might be too far forward or backward on the optical axis (the remark about glue crystallization makes me think of expansion or contraction).

The reason for reduced effectiveness in finding planets is that if you spread the PSF over more pixels, you add noise to the photometry, because each pixel contributes noise.  More noise means you are less sensitive to smaller variations in brightness.  So, near the center of the FOV, all could be well, but the signal-to-noise ratios would be lower than expected farther from the center of the field.

The article says the mission should meet its primary goals, so it's not immediately clear if the PSF is actually out-of-spec, or just not as good as hoped for.  It isn't obvious that anybody even messed up; it could be that the sample tests and such didn't scale up as expected based on some reference formula.  In any case, the final word won't come until TESS is in space and has outgassed and cooled.
Recovering astronomer

Offline Semmel

  • Full Member
  • ****
  • Posts: 958
  • Germany
  • Liked: 672
  • Likes Given: 1955
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #63 on: 07/29/2017 08:24 PM »
I didn't check the optical system. F/1.4 is pretty fast indeed. For those who don't know that is the ratio of rays convergence to focus. So imagine a defocus if 5 microns, than the PSF would become larger by 5/1.4=3.5 microns. With such a fast beam, the surface roughness of the CCD becomes problematic. I can't talk about what the surface of our CCDs looks l like, but the lowest to highest point can be more than 10 microns hight difference. This alone can mean that the size of the PSF varies by more than a pixel in diameter over the field, even with perfect optics.

Online Comga

  • Senior Member
  • *****
  • Posts: 3829
  • Liked: 1196
  • Likes Given: 1017
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #64 on: 07/30/2017 05:17 AM »
Well, that makes sense. I am working on scientific instruments, not on TESS though. It is common that the focal surface of some optical system is curved (spherical) but the detector usually flat. There are curved detectors but they are experimental and not used (yet). So imagine a spherical focal surface intersected by a flat detector surface. If you minimize the offset between both surfaces everywhere, both the center and the edges of the flat detector surface are somewhat out of focus. The optimal focus is on a ring somewhere in between. If the optics is such that the edges are more out of focus, by geometry, the center is more in focus, meaning the flat detector surface is closer to the spherical optical focal surface.
For this reason, usually the detector or optics can be pistoned just a little bit to adjust the focus for operational conditions. This is done manually in commissioning and not touched later. Unless the conditions change of course. I am surprised that TESS does not have such a system.
There is no curvature of field in the design
(The first two authors are Primeau and Chrisp.  They are terrific optical designers and optimized several orders beyond field curvature.)
That the center is better than expected and the edges worse is probably a complex issue of tolerancing.
And "crystallization" may be a simplification of having the bond material pass its glass transition temperature warmer or colder than expected. Beyond that point the coefficient of thermal expansion changes significantly.
Passive focus of very cold instruments is really tricky. Even for JPL.
What kind of wastrels would dump a perfectly good booster in the ocean after just one use?

Online jg

  • Full Member
  • **
  • Posts: 239
  • Liked: 88
  • Likes Given: 5
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #65 on: 07/30/2017 04:27 PM »
Well, that makes sense. I am working on scientific instruments, not on TESS though. It is common that the focal surface of some optical system is curved (spherical) but the detector usually flat. There are curved detectors but they are experimental and not used (yet). So imagine a spherical focal surface intersected by a flat detector surface. If you minimize the offset between both surfaces everywhere, both the center and the edges of the flat detector surface are somewhat out of focus. The optimal focus is on a ring somewhere in between. If the optics is such that the edges are more out of focus, by geometry, the center is more in focus, meaning the flat detector surface is closer to the spherical optical focal surface.
For this reason, usually the detector or optics can be pistoned just a little bit to adjust the focus for operational conditions. This is done manually in commissioning and not touched later. Unless the conditions change of course. I am surprised that TESS does not have such a system.
There is no curvature of field in the design
(The first two authors are Primeau and Chrisp.  They are terrific optical designers and optimized several orders beyond field curvature.)
That the center is better than expected and the edges worse is probably a complex issue of tolerancing.
And "crystallization" may be a simplification of having the bond material pass its glass transition temperature warmer or colder than expected. Beyond that point the coefficient of thermal expansion changes significantly.
Passive focus of very cold instruments is really tricky. Even for JPL.
Or the corrector plates aren't up to snuff.  F1.4 is extremely fast and making the corrector plates for that kind of optical design really difficult.

Offline ccdengr

  • Member
  • Posts: 66
  • Liked: 20
  • Likes Given: 8
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #66 on: 07/30/2017 06:27 PM »
Or the corrector plates aren't up to snuff.
This is an all-refractive lens with seven elements, two aspherical.

If you read the report you'll see that the assembly flow is very complex and finicky.  If I had been reviewing this design I'd have been very skeptical that a high level of athermalization was a realistic expectation -- the results don't look too bad to me.

Online jg

  • Full Member
  • **
  • Posts: 239
  • Liked: 88
  • Likes Given: 5
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #67 on: 07/30/2017 06:43 PM »
Or the corrector plates aren't up to snuff.
This is an all-refractive lens with seven elements, two aspherical.

If you read the report you'll see that the assembly flow is very complex and finicky.  If I had been reviewing this design I'd have been very skeptical that a high level of athermalization was a realistic expectation -- the results don't look too bad to me.

Even more routine f/2.0 systems are really hard to get right. 

The ATLAS design, a Wright Schmidt f/2.0 system, which are being used with flat 10Kx10K CCD chips, found their original corrector plates were not well enough made, and have been replacing them.  http://www.fallingstar.com/specifications.php This caused focus problems (rather, you could never get the size images you needed).

I am a member of the Amateur Telescope Makers of Boston.  One of our members is an optical technician who fabricates a lot of precision optics for satellites and ground systems, and I chatted with him about ATLAS's problems. Making big pieces of aspheric glass (with the surfaces different), is really quite hard according to him.  F/1.4 is much more of a nightmare.  If they have "good enough" optics, great. 

ATLAS didn't have "good enough", so had to have them remade, as it makes a factor of five in ATLAS' asteroid detection rate.


Offline as58

  • Full Member
  • ****
  • Posts: 699
  • Liked: 224
  • Likes Given: 155
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #68 on: 07/30/2017 07:25 PM »
There's more detail in TESS status update from the recent APAC meeting: https://smd-prod.s3.amazonaws.com/science-red/s3fs-public/atoms/files/Ricker_TESS_APAC_July%202017_v5.pdf

Offline LouScheffer

  • Full Member
  • ****
  • Posts: 1493
  • Liked: 1630
  • Likes Given: 183
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #69 on: 07/30/2017 09:44 PM »

There is no curvature of field in the design

I think a better statement would be "as far as practical, the focal plane has no curvature".  If you look at the graphs of lens performance (see TESS status, page 10) you will see that if you get best focus in the middle, best focus at 6o degrees off axis needs a 20 micron shift, then 40 microns at 12 o, then back to 20 microns at 16o.  This is pretty typical - best focus typically looks like a low order polynomial in distance from the center.   

The fundamental problem is that a flat focal surface is a very "unnatural" configuration.  Each surface, when varied, produces a smooth radial function of best focus.  But the function is different for every surface, so all you can do is get them to cancel approximately.  This also induces tough constraints on other surfaces as well.

To see how hard optical systems need to work to approximate a flat field is, consider this paper.  By allowing a curved focal plane, with otherwise the same specs, the reduce a design with 14 lenses, 2 aspheres, and 10 types of glass, to a design with 9 lenses, no aspheres, and 3 types of glass.

Or intuitively, how does your eye get such good results for a single lens made of jelly?  A lot of it is because your retina is a strongly curved focal surface.


Online jg

  • Full Member
  • **
  • Posts: 239
  • Liked: 88
  • Likes Given: 5
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #70 on: 07/30/2017 09:51 PM »

There is no curvature of field in the design

I think a better statement would be "as far as practical, the focal plane has no curvature".  If you look at the graphs of lens performance (see TESS status, page 10) you will see that if you get best focus in the middle, best focus at 6o degrees off axis needs a 20 micron shift, then 40 microns at 12 o, then back to 20 microns at 16o.  This is pretty typical - best focus typically looks like a low order polynomial in distance from the center.   

The fundamental problem is that a flat focal surface is a very "unnatural" configuration.  Each surface, when varied, produces a smooth radial function of best focus.  But the function is different for every surface, so all you can do is get them to cancel approximately.  This also induces tough constraints on other surfaces as well.

To see how hard optical systems need to work to approximate a flat field is, consider this paper.  By allowing a curved focal plane, with otherwise the same specs, the reduce a design with 14 lenses, 2 aspheres, and 10 types of glass, to a design with 9 lenses, no aspheres, and 3 types of glass.

Or intuitively, how does your eye get such good results for a single lens made of jelly?  A lot of it is because your retina is a strongly curved focal surface.

Another example of flat image plane, but at very high cost, is the Baker/Schmidt design of LSST.

https://www.lsst.org/about/tel-site/optical_design

It is *really* hard to get a wide flat focal plane of large chunks of sky.

Online Comga

  • Senior Member
  • *****
  • Posts: 3829
  • Liked: 1196
  • Likes Given: 1017
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #71 on: 07/31/2017 03:23 AM »

There is no curvature of field in the design

I think a better statement would be "as far as practical, the focal plane has no curvature".  If you look at the graphs of lens performance (see TESS status, page 10) you will see that if you get best focus in the middle, best focus at 6o degrees off axis needs a 20 micron shift, then 40 microns at 12 o, then back to 20 microns at 16o.  This is pretty typical - best focus typically looks like a low order polynomial in distance from the center.   

The fundamental problem is that a flat focal surface is a very "unnatural" configuration.  Each surface, when varied, produces a smooth radial function of best focus.  But the function is different for every surface, so all you can do is get them to cancel approximately.  This also induces tough constraints on other surfaces as well.

To see how hard optical systems need to work to approximate a flat field is, consider this paper.  By allowing a curved focal plane, with otherwise the same specs, the reduce a design with 14 lenses, 2 aspheres, and 10 types of glass, to a design with 9 lenses, no aspheres, and 3 types of glass.

Or intuitively, how does your eye get such good results for a single lens made of jelly?  A lot of it is because your retina is a strongly curved focal surface.

OK
There isn't just simple Petzval curvature of field.
As you said, it looks like a complex curve that swings backward.
And the second report did indeed say "crystallization" of the bond material.
But I repeat: Passive focus at very cold temperatures is hard, even for MIT.
Curved focal planes are even harder.
edit: Or you can approximate it like the Kepler photometer.
But we are wandering from TESS
« Last Edit: 07/31/2017 03:25 AM by Comga »
What kind of wastrels would dump a perfectly good booster in the ocean after just one use?

Offline catdlr

  • Member
  • Senior Member
  • *****
  • Posts: 4299
  • Marina del Rey, California, USA
  • Liked: 1286
  • Likes Given: 733
Re: SpaceX Falcon 9 : TESS : March 2018 : GENERAL THREAD
« Reply #72 on: 08/05/2017 03:22 AM »
TESS Camera Mounting Timelapse

NASA.gov Video
Published on Aug 4, 2017


The four TESS cameras being mounted to the camera plate in preparation for integration to spacecraft at Orbital ATK in Dulles, Virginia.

https://www.youtube.com/watch?v=SOj1PCtG2yk?t=001

Tony De La Rosa

Tags: