Case study: Trivial Pursuit

Episode 6 has the topic of “how to waste money and human ingenuity”. Today we will look at one of these cases on how if academia sets themselves targets that are not far enough beyond the state of the art then industry will overtake them in innovation to an extend that in the end they are left with catching up. As an example we will look at how star trackers have been developed in Germany and how even after there were already very successful industry products academia continued to spend effort in developing these capabilities. The STELLA and subsequent star tracker developments from university of Würzburg is such a case. The German Aerospace Centre (DLR) spent in total almost 850,000 EUR to develop and test a technology that by the time they were finished long since existed as an established product in the market. Lets have a look!

Small Satellite Star Trackers

Star trackers for small satellites have been around since many years. In fact the first small satellite to fly a star tracker was TUBSAT A in 1991. TUBSAT A was the first university small satellite in Germany build by the Team of Prof Renner at TU Berlin. Since then many miniature star trackers have been developed by academia and industry. For example the team of Prof. Renner alone has built 5 generations of star trackers until 2007.

the image shows the TUBSAT star tracker first flown on TUBSAT A in 1991. The right part of the image shows flight results from this mission.

Star Trackers for small satellites have been pioneered in Germany. 5 Generations were built at TU Berlin from 1991 to 2009

Industrial built star trackers became more common place since the early 2000s with multiple commercial offerings from companies like DTU (DK), SSTL (UK), Jena Optronic (Ger) and Vectronic Aerospace (GER) suitable for small satellites.

Commercial small satellite star trackers were common since the early 2000s. Germany alone had 3 companies that successfully those devices.

When the trend went into giving Cubesats more advanced ADCS performance star trackers were on the wish list of many. Unfortunately, the existing solutions were a bit on the large side for 1U to 3U platforms. For example the ST100 which was available from 2009 and which Berlin Space Technologies inherited from one of its predecessor companies was 650g with baffle and thus too large for a Cubesat.

Miniature Star Trackers

For cubesat one would require a more miniature star tracker. In order to decide whether this is a task that could be done by industry (simple implementation, close to state of the art) or precious university R&D ressources have to be used (complex implementation, low TRL) it is good to look not only towards the state of the art but also towards best practice in other industries. To enable such a comparison it is good to visualize what a star tracker is made of.

What makes a star tracker?

A star tracker is basically a just a lens & camera with an attached processor:

Camera & Optic: Camera senses the light of the stars and turns it into electrons in pixels. This matrix pixel values is the star image this is given over to the processor.

Processor: extracts events and separates potential stars from noise. Then by comparing the pattern with a star catalogue calculates the orientation.

Since all algorithms and general features of a star trackers state of the art in Germany since the mid 1990s (pioneered by TUBSAT star trackers) it is merely a question of how small can you build the camera and the processor.

Since star tracker algorithms were known industry wide since latest the early 2000s a miniature star tracker is only limited by how small you can build camera/lens unit and processor unit

As space due to their conservative nature is usually behind the latest technical developments it is useful to look for other industries to identify the state of the art.

Non Space Miniaturisation

No alt text provided for this image

What can be seen is that miniature CMOS and ubiquitous ARM processors created a run for miniaturisation the smallest camera and system on module (SOM). Above chart shows a few of the smallest designs of its time. For the 2009/10 timeframe the target size should probably be 25x25mm for sensor board and 50x50mm for the processor board as one could imagine a scenario someone would “McGyver” a solution based on µEye-LE and PicoCom4.

How Complex is the task

A second indicator on whether or not a design task is worthy an R&D grant is to look how far beyond the state of the art the task is. This is important to make sure that your research is not overtaken by industry as they are usually much faster for all tasks that they can do.

Berlin Space Technologies – ST100

When it comes to miniaturisation the ST100 was behind the industrial state of the art. That said, it was never built to be small. It was built for the TUBSAT style satellite it originated from and did not even try to be a small and lightweight. This is easily visible in the choice of the components.

  • Lens: COTS with space heritage (120g)
  • Large baffle fitting the lens (130g)
  • heavy shielding around electronics (300g)
  • Electronics (100g)

If BST would have wanted to build lightweight star tracker based on ST100 technology it would probably have looked similar to what I would call a no-innovation successor.

With almost zero effort one could have built a smaller version of the ST100 star tracker which would not have been very different in size from the Stella star tracker.

For this you have to take a smaller COTS lens like the later ST200 (-105g) and reduce the housing to 1mm wall thickness (-265g) then with very little effort and literally only mechanical modifications to an existing design you get to a star tracker that is 150g / 210g without / with baffle. In the end however BST followed an opportunity to collaborate with Hyperion technologies to create ST200, the worlds smallest star tracker.

No alt text provided for this image

Berlin Space Technologies – ST200 (2011)

True miniaturisation very close to (camera 25x25mm) and in part exceeding the state of the art (processor 25x25mm) was achieved thus by the ST200 of Berlin Space Technologies in 2011. The ST200 was developed by the industrial collaboration between Hyperion Technologies and Berlin Space Technologies. Hyperion was responsible for the miniature electronics, which to this date is one of their specialities and BST was responsible for system design and star tracker implementation based on 20 year star tracker design in Berlin. From the perspective of a star tracker design that has started in 2009 this would obviously not have been visible however the state of the art in non space miniaturisation should have indicated that this was within reach.

ST16 Sinclair Interplanetary (2011)

However, even if we ignore BST for a moment, latest by 2010 companies were working on bringing Cubesat capable star trackers to the market. Most notable was Sinclair Interplanetary who had their first miniature star tracker in space in 2013 [1].

No alt text provided for this image

The ST16 from Sinclair Interplanetary is one of the best selling miniature star trackers. So much so that in previous years the annual production often sold out in February. If the ST200 had an aspiration and role model for success it would be the ST16.

What can be observed is that the Sinclair ST16 is reasonably small. Likely one 50x50mm PCB including a camera module in in line with industrial state of the Art of 2010.

Therefore it can be seen that a thorough analysis of the state of the art would have allowed the the University and or the grant giver (German Aerospace Centre) to apply their resources on a more research worthy topic. Unfortunately this has not happened and so the colleagues in Würzburg were constantly behind the time trying to catch up.

About Stella

Stella was a research project by the University of Würzburg aims for the “development and qualification of a miniature star tracker for pico and nano satellites”. It was funded by the German Aerospace Centre (DLR) under reference number (FKZ 50RM0901). The project started in 2009 and concluded in 2012 with the delivery of one flight model [2]. This flight model was launched into space on the Technosat satellite of TU Berlin in 2017.

No alt text provided for this image

60x60x100mm & 170g (with baffle) the Stella star tracker 500,000 KEUR were spent to barely touched the state of the art.

The team that built the STELLA star tracker at University Würzburg is very particular about it’s features [2]. However, if you look closely you can find that these features were actually standard for most or all star trackers before STELLA was built.

Taste this new ice cream flavour: Its cold! In other words STELLA offers nothing new.

No alt text provided for this image

In addition to design a “miniature” star tracker aim of the STELLA research grant I imagine it was the hope of the German Aerospace Centre to foster local research which down the line could help Germany to become a relevant player in the market for miniature star trackers. Unfortunately, the project achieved neither.

Success, or the lack of it.

At the end of the project the University of Würzburg had build a star tracker that was 4x as heavy and 7x as big as the industrial state of the art. The numbers look a bit better when the baffle is included but only due to the fact that the baffle is only based on physics and cannot be miniaturized the same way electronics can.

In the Stella final report the ST200 is mentioned. More precisely doubt is expressed that it is so small that it cannot possibly work… [11]

No alt text provided for this image

Looking at both star trackers using the table above, I am not sure I had expressed the same sentiments as the team from Würzburg did in their final report.

In 2017, the Stella star tracker finally reached space. The ST200 been delivered 37 times and flown 12 times by that point.

The Stella star tracker was finally flown on the TechnoSat mission of TU Berlin in 2017. By this time the ST200 had already been delivered 37 times and flown 12 times. All of this was achieved even though the team of BST and Hyperion did not receive a single Euro of R&D grants. Started last, came in first. This should serve as another indicator of why demanding research topics are required.

There is some catching up to do…

What do you do if you figure out that industry has overtaken you in your own field of research. You step down and look for another topic that you can apply your skills on right?

If you have been defeated on your own field of research, apply yourself to a better topic.

This was however not how the team in Würzburg was thinking. When I met them on one of the Würzburg Cubesat Workshops I was rather surprised when a poster drew my attention to the development of a post Stella star tracker.

Considering that the Stella team had previously [11] claimed the ST200 is impossibly small they were quick to propose to DLR the development of their own version

The so called PicoStar was aiming on something that looked very familiar. It was in size roughly the ST200. When I asked whether this again would be done using a grant (and that this would be a waste of R&D resources) I was assured by Stella project team leader that this was entirely internally funded. To make sure that this stayed this way I informed the DLR about my findings and asked that they would not continue to fund capabilities that already exist in industry. You may therefore imagine my surprise when I found out that PicoStar evidentially still got funding under the name AROS (FKZ 50RM1522).

No alt text provided for this image

Meet AROS, the project to reimagine what Industry is already capable of

The aim of the project is to automate the design optimisation process of miniature star trackers. Or in the words of the project developers [6]:

No alt text provided for this image

From the amount of description you may get the impression that the optimisation tool is a mighty sword: complex to build, it will strike hard and conquer the unknown.

No alt text provided for this image

Therefore you will be excused to find the project schedule [7] confusing. Long story short, the main task of this 325KEUR development program was done inside the red dotted line.

Can you see the main task of the project?

It’s inside that red dotted line!

But wait there is more…, according to the AROS final report [8], despite what the schedule might tell you, it is not the hardware design that is the main task of the project.

According to the AROS final report, hardware design was not the main task.

Yet it was – evidentially from the schedule

It gets even more confusing when you look into the master thesis [9] which contains at least in some form the selection of components for the AROS star trackers and a hardware design until EM level.

If the AROS design software was supposed to do the star tracker system design automatically, why is there a master thesis which was started before AROS was approved that this all the leg work?

No alt text provided for this image

Wasn’t this selection the supposed to be done by the AROS software and not by a student? Does that remind anyone else of these 18th century “chess computers”?

No alt text provided for this image

Very confusing. I mean shouldn’t at least the final report be consistent?

How does OP1 (star tracker developed in AROS) compare to industry

The team worked hard and in the end they build a star tracker that with baffle is smaller than an ST200. Excellent. That said, the OP1 baffle is miniature. For protecting against sunlight it likely won’t do anything.

My prediction is that the OP1 baffle will have a sun exclusion angle of close to 90° boresight, my guess not better than 80°. Anybody care to prove me wrong with in orbit data from Sonate? I’ll wager a beer 🙂

On the other hand if I take a ST200 without baffle and attach the same size baffle as OP1 then very likely I come to the same size as OP1… oh!

No alt text provided for this image

I think that means that after spending 325KEUR and additional 4 years, the team in Würzburg finally has build a star tracker as small as that which Industry has successfully been selling more than 140 times, since 2012.

I have one question though…

If the software that was supposed to be developed in AROS is designed to come up with the best solution for any scenario correct? Then what was the scenario that OP1 was supposed to serve. From the selection of lens and in particular the baffle my feeling is that – if the software really exists – somebody cranked it all the way to “as small as possible”.

No alt text provided for this image

Not only is that an unwise choice it is also not necessary to go any smaller as the ST200 which exists since 2011. As evidence I would like to provide the flight proven iADCS-100 from Berlin Space Technologies. The iADCS-100 is a joint development with Hyperion Technologies and was introduced to the world alongside the in 2011 [4]. It is a full ADCS solution for 1-3U satellites which includes star tracker, reactions wheel, magnetorquer as well as mems magnetometer and gyro. The iADCS has successfully flown on multiple missions including the high profile ESA OPSSAT. To state in 2015 that no such solution exists is obviously wrong and should not have provided them with a grant.

What could have done different?

First of all, at the time STELLA was funded by the German Aerospace Centre star trackers for small satellites were nothing new. Not globally and certainly not in Germany. The first small satellite star tracker had been developed in Germany and flown in 1991. There were commercial offers for small satellite star trackers latest since 2001. Granted these models were not small enough to fit on a Cubesat but that was mainly a question of demand not a question of an immature technology. As you have seen with minimal modifications (only mechanical) it would have been possible to achieve what Stella was aiming for just by using the existing ST100 technology. In my eyes this underlines the fact that to develop something like Stella was a trivial task, certainly not something that would require a research grant.

In 2009 developing a miniature star tracker was not a topic worthy of a grant .

That said and giving the team the benefit of a doubt, once researchers in Würzburg had made contact with an industry team which outperformed them significantly, the project should have been stopped. At the very least the AROS project should not have happened – but it did.

To throw good money after bad and fund AROS in 2015 was a mistake.

Therefore I think it is necessary to figure out ways how to improve the situation.

Recommendations to academia

Dear researchers, let’s get this straight, nobody wants to take away your research. I am the first to admit that high tech research at universities is necessary – that is for all topics that could otherwise not be done. You are being paid by public funds which in means you are in service of and obligation to society. Your obligation is to be the tip of the spear when it comes to new technologies. Secondly you have to understand that by the way you are operating you will always be slower to implement something than somebody who does not have to write proposals, project reports or publications – this is industry. Don’t be the guy in academia that invents things which industry has done 5 years ago. Therefore I dare you to do better. Aim higher, don’t play trivial pursuit.

Recommendations to Grant giving institutions

Dear DLR, dear ESA, dear European commission don’t waste precious resources lifetime and money by giving grants to research topics which are not bold enough. Encourage your researchers to focus on the bleeding edge of technology. Considering from what I have seen in the Förderkatalog for German grants there is significant room for improvement.

Clear rules for what happens when a project falls behind the state of the art

In addition you need to define rules what happens if a research team is overtaken by the state of the art. My suggestion would be kill the project but allow the researchers to come up with new ideas. If you want to be more elaborate use carrot and stick. Self reported falling behind – no penalties but failed to report reduction in points for future grant applications.

Buy available products from industry instead of reinventing them in academia

Secondly, to support your industry provide more business opportunities. Including the $120K that were spent on the Canadian ST16 almost 1 MEUR wasted in this particular trivial pursuit and the only thing to show are 3-4 flight models of star trackers that were behind the times. Considering the competitive market price of star trackers there would still be plenty of money to go around for actual innovation at the universities if the star trackers would have been bought on the market.

How can you help:

This text is part of a series of articles in which the author sets the framework to start a discussion about the wrongs of the space industry. If you have experienced similar things, leave a comment. Other views and opinions are very welcome, too, as they may present a way forward. Please be kind to each other.


The author’s views are his own do not represent the views of Berlin Space Technologies.


[1] Success by 1000 improvements, Sinclair et al. – SSC-14-XII-1

[2] Stella Star Tracker, Würzburg University (02.10.2021)

[3] Ximea Subminiatur Camera shown on Vision 2011

[4] Development of the Pico Star Tracker ST-200 – Design Challenges and Road Ahead, Buhl et al. 11-IX-4

[5] Prototyping of a Star Tracker for Pico-Satellites

[6] Aros research project (23.10.2021)

[7] Aros final project report (page 4)

[8] Aros final project report (page 6)

“Da die Hardware-Entwicklung der Sternsensoren nicht das Hauptziel des Vorhabens ist, wurden Erfahrungen aus dem Vorhaben STELLA (FKZ: 50RM0901) verwertet.”

translation by author

The hardware development of the star trackers is not the main task of the project. Therefore the experiences from the Stella project have been used. 

[9] Prototyping of a Star Tracker for Pico- Satellites, T. Schwarz

[10] Introducing the ST-200 – A low cost star tracker dedicated to Cubesat missions, Segert et al. 2011 Cubesat Workshop Würzburg 

[11] Stella final report page page 9

“Die Firma BST mit Sitz in Berlin entwickelt derzeit einen Sternsensor mit dem Namen ST-200. Der Vorteil des Sensors liegt in seinen geringeren Abmessungen. Eine Qualifikation steht noch aus. Im Übrigen ist es äußerst fraglich, ob der Sternsensor von BST mit seiner sehr kleinen Optik seiner Funktion gerecht werden kann. Des Weiteren hebt sich STELLA mit seinen zusätzlichen Eigenschaften/Features (siehe Abschnitt 6) positiv gegenüber dem ST-200 ab. 

translation by Author:

Berlin Space Technologies is currently developing a star tracker called ST200. The advantage of this star tracker is his small size. It has not yet been qualified. In addition is it highly questionable whether the star tracker with its small optics can perform as intented. In addition Stella offers features that go beyond what the ST200 has.






Leave a Reply

Your email address will not be published. Required fields are marked *