Are We There Yet?

or

Three Mechanisms and the Long Road to the Roman Space Telescope

By

Johnathan W. Carson

Originally published September 2026

As I sit here writing this, the Roman Space Telescope is speeding its way to L2 after a successful launch from Launch Complex 39A at Kennedy Space Center just a few hours ago (that’s the same pad that Apollo 11 launched from, which I find strangely satisfying). I hate to admit it, but after putting several dozen of my designs into space, launches have become somewhat routine for me. Still, sometimes one does stand out. Roman is one of those launches. The potential significance of this observatory is part of it, but it’s mostly because of the sheer number of years that I’ve invested in the development of this instrument.

Some of what I said in the previous paragraph may sound rather blasé, so please allow me to elaborate. I fully understand how unique my job is, and I consider myself truly lucky to be paid to design hardware for spaceflight. Earlier in my career, I was extremely excited about launch day, but after a while, even extraordinary things become ordinary when you experience enough of them. It’s simply human nature. That's a slightly melancholy observation, but it's true well beyond aerospace.

Additionally, one of the strange things about designing spaceflight hardware is how much time passes between delivering your hardware and watching it launch. Launch isn't the culmination of the project for the engineer who designed the hardware. Unless you’re working the Integration and Test phase, a span of two or three years between hardware delivery and launch is fairly typical, sometimes more. Sometimes a lot more! By then I’m usually neck deep in the next project, if not the one after that, and the problems that once occupied nearly every waking hour have begun to feel surprisingly distant. A lot of work—and a lot of life—can happen in those few years. That emotional connection between finishing the work and watching it fly gets stretched very thin by time.

For several years, I was part of the mechanical team for the Coronagraph Instrument (CGI), one of the two instruments onboard the Roman Space Telescope. I was thoroughly involved with the development of three mechanisms that are fundamental to the operation of the Coronagraph: the Precision Alignment Mechanism (PAM), the Focus Control Mechanism (FCM), and the Fast Steering Mirror (FSM). The PAM and FCM were new designs for CGI. The FSM has a much longer history.

The first of these was what eventually became known as the Precision Alignment Mechanism (PAM). I developed the original concept, which was essentially a filter-wheel mechanism with a precision PZT-driven 2-axis fine alignment stage, and took it through an informal initial design review. Before things progressed much further, however, ownership of the mechanism was given to our German international partner, the Max Planck Institute for Astronomy (MPIA) in Heidelberg, Germany. At that point, my involvement with the device was significantly reduced, though I remained involved in its development as a reviewer and advisor. 

I was sorry to see this one go, but sometimes that’s just how these things work out.  I was really looking forward to the design challenges this little mechanism presented.  It would have been a lot of fun, but it was probably for the best, as I already had my hands full with the FSM and FCM development.  

MPIA took the design in a substantially different direction, developing an X-Y positioning plate of masks or filters driven by motors and leadscrews. Six nearly identical variations of that mechanism ultimately ended up in the flight instrument, carrying the various masks, filters, stops, and other optical elements required by the coronagraph.  

The second of the three, the Focus Control Mechanism, was probably the most challenging simply because of the ridiculously large dynamic range requirement. This device had to be able to move over 1 mm with a resolution of just a few nanometers. I spent several weeks working out the basic concept for this device, ultimately settling on a two-stage approach: a coarse stage driven by a gearmotor and leadscrew, and a fine stage driven by PZT stacks to provide 45 microns of fine travel, with flexures everywhere! The finished mechanism ultimately achieved 1.25 mm of total travel, with the fine stage capable of positioning on the order of just a few nanometers. For perspective, that’s molecular-scale—only a few molecules across. That gives the mechanism a dynamic range of several hundred thousand to one. 

I’ve touched on the FCM before, but I didn’t get into much detail.  I spent many weeks going through more than a half dozen different concepts before I finally settled on something that had real promise.  Many of those initial concepts could have been made to work, but they just didn’t “feel right.”  It’s hard to explain, but there was just something about them all that I just didn’t like.  I just knew there was a better way to get the job done, but it kept eluding me.  I’m glad this was only one of three mechanisms I was working on simultaneously.  That allowed me to shift my focus from one to another and give those ideas time to marinate.  

When what became the final design came into view, I knew immediately that it had real potential and I stopped trying to develop concepts.  I simply started to focus on developing this latest concept and soon made it the baseline design.  It evolved over the next several weeks as I began to run stress and frequency analyses on the design, but remarkably, it changed very little from that original concept. 

Of all the mechanisms I’ve designed over the years, the FCM may be the one I’m most proud of.  It’s not just because of the demanding requirements, but because it met those using a relatively simple, and dare I say elegant, approach.  It was a flexure designer’s dream (or nightmare, depending on your perspective).  The two stages worked together better than expected with virtually zero influence from one to the other.  I was also quite happy with the coarse stage travel limiter I developed.  It was inspired by a tone arm on a phonograph and it worked remarkably well.  

The last of these three mechanisms, the Fast Steering Mirror, or FSM, has the longest history by far!  The FSM was actually designed and developed over 15 years ago for the Space Interferometry Mission (SIM).  Unfortunately, SIM competed with the James Webb Space Telescope (JWST) for funding and SIM was ultimately cancelled while JWST survived. There were plenty of technical and scientific arguments for continuing SIM, and it retained strong support within the scientific community, but JWST had developed considerably stronger political support.  So, SIM ended after more than two decades of development, including prototype builds like the FSM. 

At that point, I had spent nearly my entire career supporting the development of SIM on and off.  I’ve worked on many missions that were cancelled.  It’s just part of the territory, but that one stung.  A lot!  As a team, we had worked for so long and moved the capability so far forward.  When SIM was first proposed, much of the capability required to make the instrument work simply wasn’t achievable at the time.  It took many years of development work and testing to move the state of the art far enough along to make SIM even remotely achievable.  When it was cancelled, we had demonstrated capability in nearly every technical area.  There were only a small number of hurdles left to negotiate, and we had very promising paths forward.  Many of us were simply in shock that this could have happened after getting so close!  

The Decadal Survey was a constant topic of conversation around the project at the time. Despite SIM's cancellation, the independent technical assessment performed for the 2010 survey concluded that SIM Lite was technically ready to enter Phase C and required relatively little additional technology development. The survey even cautioned that the technology developed for SIM should not be lost. 

Two FSM models were built for SIM.  They were tested extensively and used in many other optical testbeds.  They performed remarkably well.  Numerous scientists and engineers looked for many years to find missions where the design could be utilized.  The Roman Coronagraph Instrument was the first to actually fly.  

The FSM consisted of a 2-inch, flat mirror that rotated ±90 arcsec in two axes with a resolution of 15 milliarcsec.  The motion was provided by a set of three PZT actuators housed in a monolithic piece of titanium filled with flexures.  The mirror rotated about its reflective surface, which was also the location of the CG of the mirror cell.  It was fully reaction compensated to almost completely eliminate self-induced disturbance.  In fact, it was balanced so well that the disturbance input to the system proved extremely difficult to measure. 

I was the Cognizant Engineer (Cog E) responsible for the development and delivery of the FSM for SIM, but I had a lot of help.  Jim Moore (Don’s brother) did most of the heavy lifting when it came to the design details of this mechanism, but it was very much a collaborative effort.  Because of this, I was sought out to return to the role of Cog E for the FSM on Roman.  

Despite our best efforts to document as much of SIM as we could before shutting off the lights for good, not much remained of that little FSM.  There were many published papers on the performance of it, but very few technical details on how to build it.  Preserving evidence that a technology works isn't the same thing as preserving the knowledge needed to build it. Thankfully, after much searching, we found a solid model and were able to remake drawings.  

A few updates had to be made to account for some of the part and fabrication changes, along with new assembly procedures, that occurred over the 15 or so years since the last one was built, but the new build was relatively straightforward.  What didn’t survive were all the ground support equipment (GSE) and assembly tooling.  Also, a new mount needed to be designed that proved to be more challenging than anticipated.  

This new mount had to provide a first structural mode of 500 Hz or greater, allow  precision initial alignment in five degrees of freedom, and of course remain lightweight.  Oh, it also had to have a kinematic interface to the composite bench.  None of those are particularly challenging on their own, but this also had to fit in a rather crowded optical bench.  There was very little room around the FSM to provide this mounting structure.  That’s what caused all the problems.   

I spent the rest of my time on Roman shepherding the fabrication, assembly, and testing of the FSM, but I also stayed very involved in the FCM development as well.  The role of Cog E was given to a young engineer, but I was assigned to help guide that development while simultaneously training this engineer on optomechanics.  I worked very closely with the FCM team for about two years.  This approach worked very well and I found it very rewarding, as well as quite enjoyable.  Both the project and line management also noted the efficiency of that mentor/mentee approach, and on my very next job, Mars Sample Return, I was tasked with doing nearly the exact same thing, only for a few people instead of just one.  

Over the coming weeks, CGI will begin the long process of commissioning and calibration that will eventually allow it to do what we spent all those years designing it to do.  All of our efforts will be proven at First Light.  We will soon see if we got everything right!  

Lessons I Took With Me

• Give ideas time to develop.
  Not every design problem yields to more concentrated effort. Sometimes stepping away, working on something else, and giving an idea time to marinate is part of the design process.

• Preserve knowledge, not just documentation.
  Drawings, models, and test reports can show that something worked, but they don't necessarily preserve the knowledge required to build it again. The FSM was a very good demonstration of the difference. 

• Pass it along.
  As you gain experience, part of the job becomes passing along what you've learned. Guiding someone else through a design can be every bit as rewarding—and valuable—as doing the design yourself.

Explore Further

Nancy Grace Roman Space Telescope 

The Roman Coronagraph Instrument 

Space Interferometry Mission (SIM)

Precision Actuation in the Flight Design of the Roman-CGI Focus Control Mechanism (FCM)

An Introduction to Flexure Design

Related Conversations

Origins of the Back Porch

To Skin a Cat

Enjoyed this conversation?

If you'd like to know when the next engineering story is published, join the Mentee List. I'll send you a short note whenever there's another one waiting on the Back Porch.

After all, we should never stop being both mentors and mentees.