So if you have been paying attention to the last 2 earnings reports, when asked about the status of Blue Walker 3, you will have heard @AbelAvellan say that "assembly and testing of Blue Walker 3" is going well.
If you are like me, you would have liked a bit more detail.
2/n
If you are like me, you would have liked a bit more detail.
2/n
Even though I have a couple of decades of experience in aerospace and defense, I dug deeper on this subject - after all, I am a software guy, not a hardware guy.
The purpose of this thread is to share what I have learned with #SpaceMob.
Before I get started, I do need to
3/n
The purpose of this thread is to share what I have learned with #SpaceMob.
Before I get started, I do need to
3/n
point out that I have no access to inside information. Everything that you are about to read is based on knowledge from having worked in the industry, research on current best practices, and reading between the lines to figure a few things out.
With that out of the way,
4/n
With that out of the way,
4/n
here goes.
Though the main focus of this thread is on something called Assembly, Integration and Test (AIT), it's necessary to give a bit of background about what precedes AIT.
The Mission:
Everything involved in the design of a satellite begins with the mission. If you
5/n
Though the main focus of this thread is on something called Assembly, Integration and Test (AIT), it's necessary to give a bit of background about what precedes AIT.
The Mission:
Everything involved in the design of a satellite begins with the mission. If you
5/n
recall the David Marshack video, David mentioned that one of the key points in the design of what AST is building was Abel's insistence that the system had to work with ordinary, unmodified cell phones.
That's a key factor in the mission for BW3 and the BlueBird1 satellites,
6/n
That's a key factor in the mission for BW3 and the BlueBird1 satellites,
6/n
but it's not the only factor that has to be considered.
All of the functions necessary to perform the mission have to be taken into consideration.
In addition to being able to talk to unmodified cell phones, part of the mission for AST satellites involve latency of
7/n
All of the functions necessary to perform the mission have to be taken into consideration.
In addition to being able to talk to unmodified cell phones, part of the mission for AST satellites involve latency of
7/n
communications, bandwidth per user, capacity per satellite and other metrics that impact the Quality of Service (QoS) for end users.
But that's only one set of requirements. In the case of SpaceMobile, the metrics that determine the QoS led to a decision to build satellites
8/n
But that's only one set of requirements. In the case of SpaceMobile, the metrics that determine the QoS led to a decision to build satellites
8/n
that will operate in a Low Earth Orbit (LEO) as opposed to a Geosynchronous orbit.
That's an important consideration that leads to numerous other design decisions - everything from the solar arrays necessary to recharge the batteries and/or capacitors that power the
9/n
That's an important consideration that leads to numerous other design decisions - everything from the solar arrays necessary to recharge the batteries and/or capacitors that power the
9/n
satellite, to the design of cooling and insulation systems that protect the satellites from harsh conditions in space.
Beyond the orbit, the mission itself dictated antenna design, which led to the overall size of the space craft and the necessity to make BW3 and BB1
10/n
Beyond the orbit, the mission itself dictated antenna design, which led to the overall size of the space craft and the necessity to make BW3 and BB1
10/n
satellites based on a folding design
Engineers designing the satellites have to break things down functionally into things that are necessary to
1) Execute the mission
2) Maintain the health and safety of the satellite
3) Provide command and control between space and ground
11/n
Engineers designing the satellites have to break things down functionally into things that are necessary to
1) Execute the mission
2) Maintain the health and safety of the satellite
3) Provide command and control between space and ground
11/n
This was a very brief look at what takes place before the first engineering drawings are made, but decisions made at this stage of the game have a ripple effect on everything else that follows.
Once the high level functional requirements are decomposed, the architecture
12/n
Once the high level functional requirements are decomposed, the architecture
12/n
of the satellite is designed.
Engineers typically use block diagrams to specify physical and functional components of a system, define interfaces between the components, and identify gaps between the functional requirements and the initial architectural design.
13/n
Engineers typically use block diagrams to specify physical and functional components of a system, define interfaces between the components, and identify gaps between the functional requirements and the initial architectural design.
13/n
Design engineers are leading the conversations at this stage of the game, but AIT specialists are usually involved so that their take on the feasibility of the design can be assessed. (The further to the left of the timeline that you discover a problem, the shorter the
14/n
14/n
overall timeline is, and the less it will cost to produce a system)
Once the overall architecture is defined, deep dives will be performed in each area of the overall system.
There are always tradeoffs between performance, reliability and cost. But these are tempered
15/n
Once the overall architecture is defined, deep dives will be performed in each area of the overall system.
There are always tradeoffs between performance, reliability and cost. But these are tempered
15/n
by the nature of the mission.
Before I get into AIT, I want to step back to tell everyone why I went into detail about the design of satellites.
Each satellite is unique.
So as much as I wish that it wasn't the case, we can't take numbers from how long it takes to
16/n
Before I get into AIT, I want to step back to tell everyone why I went into detail about the design of satellites.
Each satellite is unique.
So as much as I wish that it wasn't the case, we can't take numbers from how long it takes to
16/n
perform certain tests on one satellite design, and apply them to another.
So with that out of the way, let's talk some AIT.
First, it's a very intensive process. That probably won't come as a surprise - if there is a problem in the build or design of a space craft, you
17/n
So with that out of the way, let's talk some AIT.
First, it's a very intensive process. That probably won't come as a surprise - if there is a problem in the build or design of a space craft, you
17/n
to find out about it while it is still on the ground. Often times, there is no way to repair something once it has been launched. (There have been exceptions, for example the Hubble space telescope had the equivalent of glasses made for it to correct for defects in the
18/n
18/n
build of the mirror. But for the most part, satellites that don't work well are lost.
If you look at the assembly of a satellite, there is nothing that is done that is too surprising. But the integration and testing of satellites may surprise people.
Again, looking at a
19/n
If you look at the assembly of a satellite, there is nothing that is done that is too surprising. But the integration and testing of satellites may surprise people.
Again, looking at a
19/n
generic spacecraft, there are a lot of subsystems. Most are either well understood, or easy to understand.
The assembly includes things like :
Mechanical components - are all the parts built within tolerances and do they work well together?
Electrical Systems
20/n
The assembly includes things like :
Mechanical components - are all the parts built within tolerances and do they work well together?
Electrical Systems
20/n
Thermal Control
Propulsion
Altitude Determination and Control System (ADCS)
Telemetry and Command (TTC)
Fault Management
On board Software
All things that you would probably expect.
But if you haven't worked in the field, you may be surprised about the integration and
21/n
Propulsion
Altitude Determination and Control System (ADCS)
Telemetry and Command (TTC)
Fault Management
On board Software
All things that you would probably expect.
But if you haven't worked in the field, you may be surprised about the integration and
21/n
testing processes.
The following screen shot is taken from a NASA document on Space Flight System Design and Environmental Test.
You can find that document at nasa.gov/sites/default/…
22/n
The following screen shot is taken from a NASA document on Space Flight System Design and Environmental Test.
You can find that document at nasa.gov/sites/default/…
22/n
As you can see, this snippet only applies to the Environmental Test Requirements for a spacecraft.
But it illustrates the point that I want to make in this thread.
In a tweet on Feb. 4th, Abel mentioned that the ControlSat for BW3 was going was going to ship for vibration
23/n
But it illustrates the point that I want to make in this thread.
In a tweet on Feb. 4th, Abel mentioned that the ControlSat for BW3 was going was going to ship for vibration
23/n
and thermal vacuum testing (more on environmental testing later)
24/n
https://twitter.com/AbelAvellan/status/1489804826695413761
24/n
Well, the ControlSat falls under the category of Unit/Component from the NASA chart I posted earlier.
As you can see from that chart, it would have to be tested at the Unit level, and at the Spacecraft level.
While I don't have the design data for BW3 or the BB1 satellites
25/n
As you can see from that chart, it would have to be tested at the Unit level, and at the Spacecraft level.
While I don't have the design data for BW3 or the BB1 satellites
25/n
I do know that satellites operating in a Low Earth Orbit can expect to see temperature swings between -65c and 125c during their orbit around earth.
And given that something operating at the altitude that the BlueBird1 satellites will be operating at will orbit the earth
26/n
And given that something operating at the altitude that the BlueBird1 satellites will be operating at will orbit the earth
26/n
once every 90 minutes.
That means that these satellites have to go though some pretty wide temperature swings in a short amount of time.
I can't tell you how long it will take to do the thermal and vacuum testing for the ControlSat of BW3, but I would GUESS that it will
27/n
That means that these satellites have to go though some pretty wide temperature swings in a short amount of time.
I can't tell you how long it will take to do the thermal and vacuum testing for the ControlSat of BW3, but I would GUESS that it will
27/n
take several weeks.
So if you are double checking the calendar and counting the days between Feb 4th and the end of the Summer 2022 launch window that has been announced and thinking to yourself "the ControlSat needs to be tested as a unit for several weeks, plus a bunch
28/n
So if you are double checking the calendar and counting the days between Feb 4th and the end of the Summer 2022 launch window that has been announced and thinking to yourself "the ControlSat needs to be tested as a unit for several weeks, plus a bunch
28/n
of other things have to be tested as units, then everything needs to be assembled and tested again - there is no way that we will make the launch window", take a breath - I have some good news for you.
The document I linked to earlier in this thread describes the level of
29/n
The document I linked to earlier in this thread describes the level of
29/n
rigor applied to the assembly, integration and testing applied to space craft since the beginning of the space era.
But the need to put up more space craft in less time has led to a change of thought and in best practices in the space industry.
This is an industry that
30/n
But the need to put up more space craft in less time has led to a change of thought and in best practices in the space industry.
This is an industry that
30/n
traditionally felt that if you changed the design by even a small amount - like changing the type of screw used on a single component, then you had to re-qualify the entire space craft.
But the times, they are a changin.
Lean manufacturing techniques are being employed
31/n
But the times, they are a changin.
Lean manufacturing techniques are being employed
31/n
Drawings that I will be using in other parts of this thread come from the Master's thesis of Captain Lisa Baghal of the US Air Force.
You can find the complete thesis at apps.dtic.mil/sti/pdfs/ADA52…
Though the document was written in March 2010, you can read between the lines
32/n
You can find the complete thesis at apps.dtic.mil/sti/pdfs/ADA52…
Though the document was written in March 2010, you can read between the lines
32/n
in tweets that Abel has shared with us and see that AST is using what Captain Baghal calls Rapid AIT.
Traditional AIT is the process by which subsystems are integrated together to form a system.
Traditional AIT is:
Documented
Formal
Sequential
The intent of traditional
33/n
Traditional AIT is the process by which subsystems are integrated together to form a system.
Traditional AIT is:
Documented
Formal
Sequential
The intent of traditional
33/n
AIT is :
1) To identify unanticipated interactions among substems
2) To define failure modes and recovery procedures
3) To spot faulty workmanship while the space craft is still on the ground
4) To prevent "infant mortality" (the early death of a satellite)
34/n
1) To identify unanticipated interactions among substems
2) To define failure modes and recovery procedures
3) To spot faulty workmanship while the space craft is still on the ground
4) To prevent "infant mortality" (the early death of a satellite)
34/n
So if all those things are the goals of AIT, what are the goals of Rapid AIT?
Both techniques share the same goals.
The difference is in how those goals are achieved.
There are two different scenarios that need to be considered.
Some space craft, the James Webb
35/n
Both techniques share the same goals.
The difference is in how those goals are achieved.
There are two different scenarios that need to be considered.
Some space craft, the James Webb
35/n
telescope is an example - are one-offs. There is only one of it's kind that will ever be built.
Other space craft have many copies of the same design built. BlueBird1 satellites are a good example. Let's not consider the second set of 168 satellites that will begin being
36/n
Other space craft have many copies of the same design built. BlueBird1 satellites are a good example. Let's not consider the second set of 168 satellites that will begin being
36/n
deployed in 2027. And let's also ignore that the initial BB1 satellites will be using FPGAs instead of ASICs.
Unless there is a reason to change the design, the same basic design can be duplicated for BB1 #1 though BB1 #168.
Though AST has already learned a lot of lessens
37/n
Unless there is a reason to change the design, the same basic design can be duplicated for BB1 #1 though BB1 #168.
Though AST has already learned a lot of lessens
37/n
that will apply to the BB1 satellites from the build of BW3, the more BB1 satellites it builds, the more it will learn.
Using a philosophy of lean manufacturing, it can use these lessens learned to cut the time and expense of building BB1 satellites as time passes.
38/n
Using a philosophy of lean manufacturing, it can use these lessens learned to cut the time and expense of building BB1 satellites as time passes.
38/n
The following screen shot is from Captain Baghal's thesis on Rapid AIT.
The data refers to the build process of Orbital Science's ORBCOMM program - a constellation of 36 satellites.
QM refers to "Qualifying Module" - the first 2 satellites they built.
Subsequent
39/n
The data refers to the build process of Orbital Science's ORBCOMM program - a constellation of 36 satellites.
QM refers to "Qualifying Module" - the first 2 satellites they built.
Subsequent
39/n
satellite builds can be seen on the lines that follow.
Note how in subsequent builds, the number of tests continually fell.
This has a very real impact on how fast you can deploy satellites, and how much it costs to build a satellite.
40/n
Note how in subsequent builds, the number of tests continually fell.
This has a very real impact on how fast you can deploy satellites, and how much it costs to build a satellite.
40/n
Now lets look at the process flow of another constellation - it just so happens to be a constellation that is far from a favorite of SpaceMob - GlobalStar.
The first generation of GlobalStar satellites consisted of 52 satellites that were launched between 1998 and 2000.
41/n
The first generation of GlobalStar satellites consisted of 52 satellites that were launched between 1998 and 2000.
41/n
This is a system flowchart of the initial satellites used to qualify the build.
42/n
42/n
After they gained experience in building the satellites, the revised workflow for the remainder of the build looked like this:
Note the radical reduction in the number of steps needed to build a satellite.
43/n
Note the radical reduction in the number of steps needed to build a satellite.
43/n
While the benefits of continuous improvement on the time and expense needed to deploy an individual space craft in a family of identical spacecraft are obvious, Rapid AIT also offers advantages in building either a unique spacecraft, or in building the first of many
44/n
44/n
identical spacecraft.
The following is a screen shot from Captain Baghal's thesis on the tenants of Rapid AIT
45/n
The following is a screen shot from Captain Baghal's thesis on the tenants of Rapid AIT
45/n
Again, I have no direct knowledge of the test plan that AST has in place, (but Abel has been kind enough to provide us clues on what they are doing), but there are certain bullet points from that screen shot that jump out at me
Specifically
Develop modular spacecraft design
46/n
Specifically
Develop modular spacecraft design
46/n
The use of microns to build BW3 and BB1's absolutely screams adherence to modular design.
Test only functionality rather than performance in system level tests - I can't prove this, but the fact that the ControlSat was sent out for testing the weekend of 2/4 combined with
47/n
Test only functionality rather than performance in system level tests - I can't prove this, but the fact that the ControlSat was sent out for testing the weekend of 2/4 combined with
47/n
launch window being confirmed as Summer '22 strongly implies that the more thorough testing is done at the component or unit levels, and not after assembly is completed.
Design facility to minimize hazardous spacecraft movement operations - in the two earnings calls to date
48/n
Design facility to minimize hazardous spacecraft movement operations - in the two earnings calls to date
48/n
, Abel has emphasized the work they are doing on what he calls "industrialization" (I would call it productionalizing, but the point is the same) Given that emphasis, I find it hard to imagine that they don't have a well thoughout work flow through the factory floor.
49/n
49/n
Use online systems to control and catalog procedures, drawings, and test data
I need to research this software more, but iBASEt is the vendor of the Manufacturing Execution Software that AST is using (Hat tip to @DaMadMonk_ for pointing it out - ibaset.com/ibaset-provide… )
50/n
I need to research this software more, but iBASEt is the vendor of the Manufacturing Execution Software that AST is using (Hat tip to @DaMadMonk_ for pointing it out - ibaset.com/ibaset-provide… )
50/n
Automate functional tests to repeatability and consistency - we have heard about automation in both earnings calls so far - it's obviously important to Abel.
Reduce complexity of spacecraft design as much as possible - as @CatSE___ApeX___ likes to point out, the AST way
51/n
Reduce complexity of spacecraft design as much as possible - as @CatSE___ApeX___ likes to point out, the AST way
51/n
is to eliminate complexity wherever possible (for example, the stored energy concept behind the deployment mechanism) and to transfer complexity from space to the ground when complexity can't be eliminated.
Those are the things that are obvious to me, as an outsider, that
52/n
Those are the things that are obvious to me, as an outsider, that
52/n
are being done to support Rapid AIT and responsive spacecraft.
But I have gotten tips from an insider - Abel himself.
Outside of the timing of the vibration and thermal vacuum testing, Abel was kind enough to share this tweet with us all:
53/n
But I have gotten tips from an insider - Abel himself.
Outside of the timing of the vibration and thermal vacuum testing, Abel was kind enough to share this tweet with us all:
https://twitter.com/AbelAvellan/status/1487455914210832402
53/n
I don't know how long the original folding process took. And obviously AST is testing the deployment/unfurling process extensively, but the reason that you put a lot of engineering effort, tweaking and testing into the folding process has less to do with the testing of
53/n
53/n
BlueWalker 3 before launch.
Yeah, sure, reducing the time it takes to fold the satellite does save some time between testing rounds of the unfurling process.
But it is glaringly obvious to me that the tweaking and engineering is more about the upcoming BlueBird1s than
54/n
Yeah, sure, reducing the time it takes to fold the satellite does save some time between testing rounds of the unfurling process.
But it is glaringly obvious to me that the tweaking and engineering is more about the upcoming BlueBird1s than
54/n
it is about saving some time between test iterations of the unfurling process. The engineering dollars spent to improve the process of folding the satellite probably exceed any savings on the testing side.
55/n
55/n
@threadreaderapp unroll
• • •
Missing some Tweet in this thread? You can try to
force a refresh
