-
LISA follow-on (Earth's solar orbit scale interferometer) by LouScheffer on 09 May, 2023 00:44
-
Here is preliminary work for a follow-on to the LISA gravitational wave interferometer. It gains sensitivity with longer arms - a huge equilateral triangle the size of the Earth's orbit around the Sun. One spacecraft at each of the leading and trailing Lagrange points 60 degrees behind and ahead of Earth, and the other at the Lagrange point opposite Earth.
In a remarkable display of conservatism, they assume the launcher available for this mission in 2045 or so would be an Ariane 64.
Almost no new technology is required - just careful design and money.
-
#1 by skizzo on 09 May, 2023 01:42
-
2045? Jesus
-
#2 by Zed_Noir on 09 May, 2023 02:20
-
.....
In a remarkable display of conservatism, they assume the launcher available for this mission in 2045 or so would be an Ariane 64.....
-
#3 by tyrred on 09 May, 2023 02:32
-
Sounds amazing... I imagine station-keeping of the individual components at the separate Lagrange points to maintain precise laser lock is not a trivial problem. Has anything similar been proven out yet in space?
-
#4 by VSECOTSPE on 09 May, 2023 03:15
-
#5 by Hobbes-22 on 09 May, 2023 05:37
-
Sounds amazing... I imagine station-keeping of the individual components at the separate Lagrange points to maintain precise laser lock is not a trivial problem. Has anything similar been proven out yet in space?
LISA Pathfinder proved the concept using two test masses that were free-floating in a single spacecraft. LISA is planned to be launched in 2037, with three spacecraft arranged in an equilateral triangle with sides 2.5 million km long. -
#6 by redliox on 09 May, 2023 07:18
-
Could something similar be done with an optical interferometer?
-
#7 by LouScheffer on 09 May, 2023 12:06
-
Could something similar be done with an optical interferometer?
Both this proposal, and the earlier LISA (likely 2030s) are huge optical interferometers. -
#8 by LouScheffer on 09 May, 2023 12:31
-
Sounds amazing... I imagine station-keeping of the individual components at the separate Lagrange points to maintain precise laser lock is not a trivial problem. Has anything similar been proven out yet in space?
The satellites do not station-keep, except to keep their test mass centered in their housing. That's because it's critical that they are in perfect free fall. Any other station-keeping would screw up the experiment. The free-fall needs to be maintained to extreme accuracy. This was tested with the LISA pathfinder.
With the ends in free-fall, the arm lengths are always varying somewhat. So they don't maintain lock the same way the earthbound interferometers do. Instead each satellite interferes its own laser (which has been bounced off the other two) with the other two satellite's lasers (they are doing the same thing). The combination of all these measurements reveals the path lengths with (very) sub-wavelength precision. The Grace follow-on mission tested inter-spacecraft interferometry, though at much smaller scale.
The combination of these two methods means very special orbits are required. The relative motion of the spacecraft must be small enough that the doppler shift between any two beams is less than 10 MHz. Otherwise they cannot measure the phase to the needed degree of accuracy. So both this mission and LISA use Earth-like orbits around the Sun that are far from Earth. LISA trails the Earth by about 20 degrees, and the proposed missions uses the Lagrange points about 60 degrees from the Earth. -
#9 by RON_P on 09 May, 2023 17:16
-
#10 by LouScheffer on 09 May, 2023 17:49
-
Here's the physics arXiv article if you want more detail.
-
#11 by eeergo on 09 May, 2023 17:52
-
2045? Jesus
LISA is currently NET 2037 or so, which is quite a dilated timespan for a mission currently transitioning to Phase B2, but required due to the complexity of a 2.5 million kilometer interferometer requiring a pretty substantial price tag and required mature technology developments.
A proposal for LFO (or LISA-Max, as I see they call it in the linked article) can't be reasonably foreseen for a short time period after LISA. Even if no new technology were to be needed, and a relatively straightforward "LISA on steroids" was possible, you'd want to try out LISA itself for a few years -taking into account transit, commissioning, and a long enough period of observations including data analysis- to make sure it actually works as intended before committing to building "the same but bigger".
As for the claim almost no new technology will be needed... well, color me skeptical on that. There are many subsystems in LISA that are already at the bleeding edge of what's possible with today's technology - in particular, the sensors dealing with constellation acquisition that must detect a picowatt-scale signal emitted by the distant spacecraft in order to lock on to it and establish the interferometric signal. Going 2 orders of magnitude up on the arm length scale (2.5 million km in LISA to 250 million km in LFO) means the emitted signal would roughly be 10000x lower, i.e. on the order of hundreds of attowatts.
Not an expert on sensitive detectors, but that seems a stretch, so very possibly more emitted power would be needed, which comes with all sorts of associated challenges for the S/C bus, starting with the very obvious issue of power and its generation, alongside the related issue of mass and size, which in turn strains both maneuvering propellant and crucially consumables for the stationkeeping/drag-free thrusters. All of this assuming the interferometer itself can be kept approximately as it is currently foreseen for LISA and doesn't bring its own sets of scalability issues. The optical tracking mechanisms themselves also would probably be subject to a requirement to have a greater angular range (in LISA it is about 3º) to compensate for the breathing angle of the constellation, leading to further disturbances for the interferometer, just because the locally-perturbing forces in an Earth-like orbit must be larger than in an Earth-trailing/leading one, but I would have to check that because that behavior may be non-intuitive. -
#12 by Robotbeat on 09 May, 2023 17:59
-
Super impressive concept. I love it when there’s an opportunity of just taking a design parameter (interferometer baseline) and cranking it by orders of magnitude. “What if the entire orbit of the Earth was our baseline?”
…this sort of idea is also how we measured the parallax to the nearest stars & figured out how vast the universe actually is…
And the concept would work for deeper space, too… we could got to 10AU if we wanted… or heck, with nuclear power and advanced propulsion we could go to 100AU or even 1000AU with better propulsion… and the same interferometric approach (precise stationkeeping with test masses, etc) could be applied to imaging as well. -
#13 by LouScheffer on 09 May, 2023 19:48
-
As for the claim almost no new technology will be needed... well, color me skeptical on that.
The known issues, at least, are covered in the arXiv paper.QuoteThere are many subsystems in LISA that are already at the bleeding edge of what's possible with today's technology - in particular, the sensors dealing with constellation acquisition that must detect a picowatt-scale signal emitted by the distant spacecraft in order to lock on to it and establish the interferometric signal. Going 2 orders of magnitude up on the arm length scale (2.5 million km in LISA to 250 million km in LFO) means the emitted signal would roughly be 10000x lower.
Fortunately, this loss decreases as the 4th power of the mirror diameter. (The transmitter sends a tighter beam, and the receiver catches a larger fraction). So going from 30 cm mirrors (LISA) to 1 meter mirrors gains back a factor of 123. The larger mirrors are more massive, which drives the size and mass of the satellites. With these larger mirrors, they think that LISA's power (1W) will suffice.QuoteThe optical tracking mechanisms themselves also would probably be subject to a requirement to have a greater angular range (in LISA it is about 3º) to compensate for the breathing angle of the constellation, leading to further disturbances for the interferometer, just because the locally-perturbing forces in an Earth-like orbit must be larger than in an Earth-trailing/leading one, but I would have to check that because that behavior may be non-intuitive.
Surprisingly, the Earth Lagrange points selected give an orbit that is more stable than LISAs. Their angles are constant at 60 +- 0.6 degrees over the whole mission, and the doppler shift of the beams is only half that of LISA's. -
#14 by eeergo on 09 May, 2023 20:36
-
As for the claim almost no new technology will be needed... well, color me skeptical on that.
The known issues, at least, are covered in the arXiv paper.QuoteThere are many subsystems in LISA that are already at the bleeding edge of what's possible with today's technology - in particular, the sensors dealing with constellation acquisition that must detect a picowatt-scale signal emitted by the distant spacecraft in order to lock on to it and establish the interferometric signal. Going 2 orders of magnitude up on the arm length scale (2.5 million km in LISA to 250 million km in LFO) means the emitted signal would roughly be 10000x lower.
Fortunately, this loss decreases as the 4th power of the mirror diameter. (The transmitter sends a tighter beam, and the receiver catches a larger fraction). So going from 30 cm mirrors (LISA) to 1 meter mirrors gains back a factor of 123. The larger mirrors are more massive, which drives the size and mass of the satellites. With these larger mirrors, they think that LISA's power (1W) will suffice.QuoteThe optical tracking mechanisms themselves also would probably be subject to a requirement to have a greater angular range (in LISA it is about 3º) to compensate for the breathing angle of the constellation, leading to further disturbances for the interferometer, just because the locally-perturbing forces in an Earth-like orbit must be larger than in an Earth-trailing/leading one, but I would have to check that because that behavior may be non-intuitive.
Surprisingly, the Earth Lagrange points selected give an orbit that is more stable than LISAs. Their angles are constant at 60 +- 0.6 degrees over the whole mission, and the doppler shift of the beams is only half that of LISA's.
The paper deals with science operations, and the concept of operations at that. It gives no real answer as to how those figures will be achieved technically.
As I mentioned, it may be the interferometry is feasible without huge modifications to LISA's - I'm not an expert on this and will take the paper's findings at face value. I was talking about "accessory" functions that are just as critical for the mission though, such as achieving constellation lock in the first place, or reacquiring it after a disengagement. That's where the extremely feeble power levels are an issue. In LISA, there is an intermediate step after attitude stabilization with the star trackers (relatively coarse accuracy for what's needed) and acquisition of interferometer lock. It dictates the need for a sensor that is both sensitive enough to the distant spacecraft, fast enough in its integration that it will record the "blink" of the distant spacecraft's beam passing over it, and have a sufficiently wide field of view in order to minimize the maneuvering to search for it. That's where the pW-->aW issue is, I believe, the most difficult to solve without a non-trivial development.
Crucially, you mention the expectation to send a narrower, more collimated beam in order to lose less power to dispersion to counter the 1/r2 losses (along with tripling the mirror size, which is itself a relevant undertaking that will impose the need for new developments to account for more propellant needed to counteract solar radiation pressure, as well as power, since the drag-free conditions needed for the interferometer to work are dependent on an extremely low thrust propulsion system that runs continuously and without "jolts"). This is positive for nominal interferometric operations, but actually detrimental for constellation acquisition: the "blink" will generally be more feeble unless it hits the sensor head-on, and the acquisition maneuver longer to account for the reduced angular size the beam will illuminate.
Believe it or not, most aspects of the mission are dictated by these "ancillary" support systems rather than by the core scientific instrument itself.
The need to have 3º of movement in LISA's Movable Optical Sensor Assemblies (MOSAs) does not stem from just the expectable breathing angle variations due to orbital mechanics between the three spacecraft, but also operational considerations. It might be that the expectable variation is smaller in a larger orbit than for LISA's (+/- 0.6º in the paper vs 1º for LISA), but this can change as the mission design evolves: current requirements call for 1.5º, which doubles in contingency scenarios where a MOSA is locked in place and the other needs to pick up the slack, leading to the 3º I mentioned. Anyway, good to know Sun-Earth Lagrange points are expected to be more stable to such constellations. -
#15 by LouScheffer on 09 May, 2023 23:59
-
[...] I was talking about "accessory" functions that are just as critical for the mission though, such as achieving constellation lock in the first place, or reacquiring it after a disengagement. That's where the extremely feeble power levels are an issue. In LISA, there is an intermediate step after attitude stabilization with the star trackers (relatively coarse accuracy for what's needed) and acquisition of interferometer lock. It dictates the need for a sensor that is both sensitive enough to the distant spacecraft, fast enough in its integration that it will record the "blink" of the distant spacecraft's beam passing over it, and have a sufficiently wide field of view in order to minimize the maneuvering to search for it. That's where the pW-->aW issue is, I believe, the most difficult to solve without a non-trivial development.
It's not clear to me that a next generation LISA will have so much trouble acquiring the other spacecraft. Using a 1 meter mirror with 1 micron light means a beam divergence of about a micro-radian, and similar requirements for pointing the receiver. Using Delta-DOR navigation can locate the spacecraft in the plane of the sky with nano-radian class precision. Ranging is good to a few meters. Thus the position of each spacecraft is known very well (well within a kilometer size box). From this computing the direction to point is straightforward, and should be known to much better than a micro-radian.
But how does the spacecraft know which way it is pointing? This requires star trackers, and old star trackers were not all that accurate (a microsecond or more). But modern ones are much better - here's one from ESA good to 0.1 micro-arcsecond. Here's an older tracker from Ball aerospace that's good to 0.2 micro-arcsecond.
Now of course errors stack up, and aligning the star tracker with the telescope is not trivial, and so on. But the search space for acquisition should not be large if state of the art measurements and star trackers are used.QuoteThe need to have 3º of movement in LISA's Movable Optical Sensor Assemblies (MOSAs) does not stem from just the expectable breathing angle variations due to orbital mechanics between the three spacecraft, but also operational considerations. It might be that the expectable variation is smaller in a larger orbit than for LISA's (+/- 0.6º in the paper vs 1º for LISA), but this can change as the mission design evolves: current requirements call for 1.5º, which doubles in contingency scenarios where a MOSA is locked in place and the other needs to pick up the slack, leading to the 3º I mentioned.
This is a good point. These things will be impossible to service, and need long data sets to be useful. So contingency operations need to be carefully considered, and may result in more difficult specifications to meet. -
#16 by eeergo on 10 May, 2023 06:44
-
[...] I was talking about "accessory" functions that are just as critical for the mission though, such as achieving constellation lock in the first place, or reacquiring it after a disengagement. That's where the extremely feeble power levels are an issue. In LISA, there is an intermediate step after attitude stabilization with the star trackers (relatively coarse accuracy for what's needed) and acquisition of interferometer lock. It dictates the need for a sensor that is both sensitive enough to the distant spacecraft, fast enough in its integration that it will record the "blink" of the distant spacecraft's beam passing over it, and have a sufficiently wide field of view in order to minimize the maneuvering to search for it. That's where the pW-->aW issue is, I believe, the most difficult to solve without a non-trivial development.
It's not clear to me that a next generation LISA will have so much trouble acquiring the other spacecraft. Using a 1 meter mirror with 1 micron light means a beam divergence of about a micro-radian, and similar requirements for pointing the receiver. Using Delta-DOR navigation can locate the spacecraft in the plane of the sky with nano-radian class precision. Ranging is good to a few meters. Thus the position of each spacecraft is known very well (well within a kilometer size box). From this computing the direction to point is straightforward, and should be known to much better than a micro-radian.
But how does the spacecraft know which way it is pointing? This requires star trackers, and old star trackers were not all that accurate (a microsecond or more). But modern ones are much better - here's one from ESA good to 0.1 micro-arcsecond. Here's an older tracker from Ball aerospace that's good to 0.2 micro-arcsecond.
Now of course errors stack up, and aligning the star tracker with the telescope is not trivial, and so on. But the search space for acquisition should not be large if state of the art measurements and star trackers are used.
The first star tracker you list (ASTRO-XP) is the current formulation baseline for LISA, as is also evident from the infographic at the top of the page. You may be confusing arcseconds with microradians though. The stated quantity of 0.1 *arcseconds* (not micro) is equivalent to about half a *microradian*. This is not true for the 3-sigma requirement imposed for LISA however, where ASTRO-XP reaches a 5 urad absolute performance error, or 10x the star tracker's "class". This stacks up as you say with other errors, such as the STR-to-TEL alignment, which is estimated at a further 25 urad due to the precision achievable through calibrations.
However, these are still subdominant contributors to the pointing error, which is foreseen to be dominated by the ground orbit tracking estimates propagating into the S/C's deviation from the true line of sight (around 50 urad, conservatively). The fact that you can locate a spacecraft from the ground with nrad precision *from Earth* doesn't translate into two spacecraft a few hundred million kilometers away from Earth being able to point to each other with such precision - not by a long shot.
So LISA is currently looking at a pointing error slightly under 0.1 milli-radians (i.e. ~80 microradians) when none of the spacecraft has seen any of the distant ones. Note beam divergence for LISA is expected to be about the same as what you postulate for the LISAmax proposal (on the order of a urad). -
#17 by LouScheffer on 11 May, 2023 02:07
-
This stacks up as you say with other errors, such as the STR-to-TEL alignment, which is estimated at a further 25 urad due to the precision achievable through calibrations.
I'd imagine this could be calibrated out if required. Perhaps, after4 launch, scan a 50 urad square known to contain a single star. Once you find it you can calibrate the star-tracker to telescope connection.QuoteHowever, these are still subdominant contributors to the pointing error, which is foreseen to be dominated by the ground orbit tracking estimates propagating into the S/C's deviation from the true line of sight (around 50 urad, conservatively). The fact that you can locate a spacecraft from the ground with nrad precision *from Earth* doesn't translate into two spacecraft a few hundred million kilometers away from Earth being able to point to each other with such precision - not by a long shot.
This is the part I don't get. If you can apply full DSN accuracy, you should get nano-arcsec accuracy in the plane of the sky. That's a few hundred meters at the distances we are talking about here. Uncertainties in Earth rotation matter little since the angles are referenced to distance quasars during the tracking. Range is even better determined - a few meters. So both spacecraft should be located within a km in the same JPL barycentric reference frame.
This should determine the pointing angles to much less than a urad. A 1 km error from 300 million km is about 0.003 urad.
Now there may well be practical problems arguing against this. These super-accurate angles are a pain (they need simultaneous observations at two or more stations, plus lots of processing), so JPL may not be very interested in scheduling them if there is any alternative. If they don't use them, then the angular position is much worse.
Also this may be a system level complexity tradeoff. If you need to do a scan anyway (say to calibrate the star tracker, so you can use precise pointing) you might as well scan to locate the other spacecraft. Also, if you do that, you don't need to ask the DSN for super-accurate tracking. And you only do this once at the start of the mission, so taking a little longer to scan a bigger field, rather than some complex calibration/pointing might be a better tradeoff.QuoteSo LISA is currently looking at a pointing error slightly under 0.1 milli-radians (i.e. ~80 microradians) when none of the spacecraft has seen any of the distant ones. Note beam divergence for LISA is expected to be about the same as what you postulate for the LISAmax proposal (on the order of a urad).
You seem super familiar with how LISA plans to find the other stations. I agree that just doing the same procedure with 100x greater spacing won't be practical. Assuming a 3x bigger telescope, the EIRP is 10x as much, but the path loss is a factor of 10,000 worse. If it's a signal-to-noise problem, then detecting a source that's 1000x weaker requires scanning a million times slower, which is surely not practical.
However, I think the pointing could be improved considerably using existing technology, *if there was a need to do so*. LISA does not have this need, but LISAmax likely will. -
#18 by eeergo on 12 May, 2023 12:00
-
This stacks up as you say with other errors, such as the STR-to-TEL alignment, which is estimated at a further 25 urad due to the precision achievable through calibrations.
I'd imagine this could be calibrated out if required. Perhaps, after4 launch, scan a 50 urad square known to contain a single star. Once you find it you can calibrate the star-tracker to telescope connection.
This is *after* calibration. There are several assumptions to be made here, but believe me - it has been MC'd to death and won't get too much under this figure. Also, this is absolute knowledge - there are thermoelastic deformations on top of that which can reach 10-20x that value.QuoteThis is the part I don't get. If you can apply full DSN accuracy, you should get nano-arcsec accuracy in the plane of the sky. That's a few hundred meters at the distances we are talking about here. Uncertainties in Earth rotation matter little since the angles are referenced to distance quasars during the tracking. Range is even better determined - a few meters. So both spacecraft should be located within a km in the same JPL barycentric reference frame.
This should determine the pointing angles to much less than a urad. A 1 km error from 300 million km is about 0.003 urad.
Now there may well be practical problems arguing against this. These super-accurate angles are a pain (they need simultaneous observations at two or more stations, plus lots of processing), so JPL may not be very interested in scheduling them if there is any alternative. If they don't use them, then the angular position is much worse.
Also this may be a system level complexity tradeoff. If you need to do a scan anyway (say to calibrate the star tracker, so you can use precise pointing) you might as well scan to locate the other spacecraft. Also, if you do that, you don't need to ask the DSN for super-accurate tracking. And you only do this once at the start of the mission, so taking a little longer to scan a bigger field, rather than some complex calibration/pointing might be a better tradeoff.
Can't really tell you how this was derived although it does include both tracking and orbit propagation (there will be a delay between the tracking and ranging taking place, the processing necessary to derive the km-sized angular uncertainty, and the uplink to tell the spacecraft where to look), in the case of LISA for a smaller distance by a factor of 50x.
It might well be that some of the practicalities you mention are at play there, although I suspect compounding the nrad-class achievable pointing precision to the relative pointing between two distant objects is at play here. Pointing to a km-sized area at 300M km is a few nrads of angular dispersion... but you'd be doing so from anywhere within a km-sized area within which you don't know where you are too.QuoteQuoteSo LISA is currently looking at a pointing error slightly under 0.1 milli-radians (i.e. ~80 microradians) when none of the spacecraft has seen any of the distant ones. Note beam divergence for LISA is expected to be about the same as what you postulate for the LISAmax proposal (on the order of a urad).
You seem super familiar with how LISA plans to find the other stations. I agree that just doing the same procedure with 100x greater spacing won't be practical. Assuming a 3x bigger telescope, the EIRP is 10x as much, but the path loss is a factor of 10,000 worse. If it's a signal-to-noise problem, then detecting a source that's 1000x weaker requires scanning a million times slower, which is surely not practical.
However, I think the pointing could be improved considerably using existing technology, *if there was a need to do so*. LISA does not have this need, but LISAmax likely will.
I am indeed familiar with the constellation acquisition process, although I'm by no means an expert on it - so can't really judge authoritatively on the possibility that it's impossible with current technology. However, and given the uncertainties expected for LISA (which did and are requiring technology development beyond the state of the art), my sixth sense gives me the impression claiming "no new technology needs to be developed for LISAmax" is overstretching optimism.
.