This is the eighteenth post in the LocalPapa Notes dev-log series. Last time we shipped the Gimbal Motor Sizing Calculator: plug in aerodynamic environment and stabilization bandwidth targets, and it works backward to the torque and current each axis motor needs. After it went live, we looked back at its input fields and noticed an unstated assumption — where do the numbers for stabilization bandwidth f_bw and maximum angular deviation target θ_max actually come from?
The calculator only handles half the problem: "given a known stability target, derive the motor spec." The other half — "once you've actually built the gimbal, what is its real LOS stability, and does it hit the target?" — is a measurement problem, and specifically an optical measurement problem, not something you can eyeball from a phone video. This post pulls together what we read while figuring out how to actually set up a LOS optical measurement rig.
What LOS is, and why "calculating it" isn't enough
LOS (Line of Sight) is the direction the sensor's optical axis actually points. Ideally the gimbal pins the LOS to the target no matter how the carrier shakes, how the motor bearings introduce friction, or how cables tug on the mechanism — the optical axis shouldn't move. In practice it does, and the angular magnitude of that motion is what the field calls LOS jitter.
Cornelius Dennehy at NASA Langley Research Center put it plainly in a systematic survey: LOS jitter sources are diverse — reaction wheel imbalance, cryocooler vibration, structural resonance, residual oscillation after solar-array deployment — and most of them can't be predicted precisely from analysis alone, because parameters like friction, cable drag, and resonant frequency only become accurate through actual measurement in practice (Dennehy, A Survey of the Spacecraft Line-of-Sight Jitter Problem, NASA NTRS 20200002690). A follow-on engineering practice guide built from the same survey makes the same point even more directly: LOS budgets need measurement-based verification baked in from early in the project, not simulation alone (NASA/TM-20210017871, Spacecraft Line-of-Sight Jitter Management and Mitigation, 2021).
The formulas our calculator uses — aerodynamic drag torque, inertial torque, friction margin, safety factor — all trace back to J. M. Hilkert's widely-cited review, essentially the textbook reference in this field (Hilkert, Inertially Stabilized Platform Technology: Concepts and Principles, IEEE Control Systems Magazine, vol. 28, no. 1, pp. 26–46, 2008). But that same paper is equally clear that these formulas give you the demand-side torque and bandwidth estimate — motor sizing is only half of the stabilization loop. Whether LOS actually holds the target depends on the loop's closed-loop disturbance rejection, and that has to be verified with a measured frequency response (Hilkert & Pautler, Reduced-Order Disturbance Observer Applied to Inertially Stabilized Line-of-Sight Control, Proc. SPIE 8052, 2011). In other words: the calculator tells you roughly how much torque you need; the measurement rig tells you whether what you built actually meets the target — you need both.
Three common optical measurement architectures
After a round of reading, LOS measurement approaches roughly converge into three tiers, decreasing in precision, cost, and setup difficulty.
1. Autocollimator + mirror (lab-grade)
This is the most common approach in aerospace and mil-spec EO/IR systems. A mirror is mounted on the gimbal's (dummy) payload; an autocollimator projects a collimated beam that reflects off the mirror back into the autocollimator's detector. The deflection angle of the mirror's normal relative to the autocollimator's optical axis is a direct measurement of LOS deviation, resolvable down to arcsecond levels. A paper on friction-induced jitter in electro-optic turrets describes this setup concretely: a calibrated commercial autocollimator (a UDT Model 1020, in that paper) aimed at a mirror on a dummy payload mounted to the gimbal, with the off-axis angle serving as the direct sightline-deflection measurement (Sightline Jitter Minimization and Shaping Using Nonlinear Friction Compensation, International Journal of Optomechatronics, 1(3), 259–283, 2007).
The upside is the highest precision and widest bandwidth (autocollimator detectors can sample at kilohertz-class rates); the downside is dedicated instrumentation and precise optical alignment, which raises the setup bar — better suited to projects with lab access.
2. Laser + position-sensitive detector (PSD / quadrant photodiode)
Conceptually similar to an autocollimator, but a collimated laser is aimed at a target point on the gimbal payload, and a position-sensitive detector (PSD) or quadrant photodiode is placed in the return or transmission path. These devices work by measuring the centroid position of the light spot on the sensing surface and converting it to an angular offset (RP Photonics Encyclopedia, Position-Sensitive Detectors). This architecture shows up throughout earlier tracking-system literature too, where quad-cell signals often double as both the low-bandwidth tracking error input and the LOS jitter measurement.
More accessible than an autocollimator (a laser diode + PSD module costs far less than precision autocollimator instrumentation), but still requires a stable optical path and a signal-processing chain, and the measurable range is limited by the PSD's linear dynamic range.
3. Target board + telephoto camera (accessible)
The first two are "optical-instrument-grade" approaches. For smaller gimbals — like the BaseCam-class systems our calculator targets, typically carrying payloads from a few to a few tens of kilograms — a more approachable alternative is to use the camera itself as the measurement instrument: mount a target board with a known pattern at a fixed distance, point the gimbal's camera at it, and use image-processing to track the pattern's pixel displacement frame by frame, converting it to angular jitter via the lens focal length and target distance. Small-UAV gimbal research takes this route, combining flight tests with image-based analysis of actual attitude stability (see the survey by Dhruv & Kaushal, A Review of Pointing Modules and Gimbal Systems for Free-Space Optical Communication in Non-Terrestrial Platforms, Photonics, 12(10), 1001, 2025, which systematically covers the pointing-accuracy, bandwidth, and inertia trade-offs across gimbal designs).
Less precise than the first two (limited by camera resolution and the sub-pixel accuracy of the tracking algorithm), but the cheapest to set up — a telephoto lens, a printed target board, and software capable of sub-pixel corner/blob tracking is enough for a small-to-mid project to build in-house.
Injecting disturbance and measuring bandwidth
Static detection alone isn't enough — what actually matters for a LOS stabilization system is its disturbance rejection: can the stabilization loop suppress LOS error when the carrier shakes, and how fast? The standard approach is to mount the gimbal on a rate table / shaker table, excite the base with a known angular-rate or acceleration profile, simultaneously record the LOS error using one of the optical methods above, and use system identification to derive the loop's closed-loop frequency response — this is precisely where the calculator's "stabilization bandwidth f_bw" input should, in principle, come from in the first place (Hilkert 2008; Hilkert & Pautler 2011). A simplified version uses handheld shaking with IMU-logged disturbance input — less precise than a rate table, but good enough for an order-of-magnitude bandwidth estimate, which is fine for verification-grade testing.
Analyzing the data: RMS and power spectral density
Once you have a time series of LOS error over some duration, two analysis methods are standard:
- RMS jitter: compute the root-mean-square error over a specific time window (seconds to tens of seconds), which maps directly onto the calculator's θ_max (maximum angular deviation target). The James Webb Space Telescope's pointing stability spec is defined exactly this way — for fixed targets, as the RMS error of guide-star position over a 15-second interval relative to the mean position over a 10,000-second observation, with a Noise Equivalent Angle (NEA) describing the sensor's own measurement precision floor (JWST User Documentation, JWST Fine Guide Stability).
- Power spectral density (PSD) analysis: transform the time-domain error into the frequency domain to find where the jitter energy concentrates — this is what tells you whether "the servo bandwidth isn't wide enough" or "a specific structural resonance got excited." During JWST's in-orbit commissioning, PSD analysis is exactly what surfaced a 0.3 Hz vibration-damper mode and a 0.045 Hz fuel-slosh signature — neither related to the reaction wheels or cryocooler. Without frequency-domain analysis, the RMS number alone would have given no clue as to the root cause (STScI, JWST Line-of-Sight Jitter Measurement during Commissioning, JWST-STScI-008271, 2022).
Even for gimbals far smaller than a space telescope, this "RMS tells you pass/fail, PSD tells you why" analysis logic carries over directly: check whether RMS jitter is under the θ_max target first; if it isn't, dig into the PSD to find which frequency band is the culprit — insufficient stabilization loop bandwidth (tune the controller or size up the motor) versus an excited structural resonance (stiffen the mechanism or add damping).
A minimal practical rig for small-to-mid gimbal builders
After going through the literature, here's the combination we'd recommend for a BaseCam-class project (not a satellite-class one), balancing cost against precision:
- Measurement side: default to "target board + telephoto camera"; upgrade to a laser + PSD module if budget allows (an order of magnitude more precise, still within hobbyist-affordable component cost). An autocollimator is usually only worth it if you genuinely need arcsecond-class precision.
- Disturbance side: start with handheld shaking + IMU logging to get a rough order-of-magnitude estimate; once the direction looks right, consider renting or building a simple shaker table if you need a precise bandwidth measurement.
- Analysis side: compute both RMS (compare against the calculator's θ_max) and PSD (find the jitter's root cause) from the recorded time series — do both, neither is optional.
- Feedback loop: if the measured RMS/bandwidth doesn't meet the target you originally set in the calculator, go back and adjust the safety factor SF or friction margin, then re-derive the motor spec — measurement and calculation should correct each other, not end once you've run the numbers once.
References
- Dennehy, C. J. (2020). A Survey of the Spacecraft Line-of-Sight Jitter Problem. NASA Langley Research Center, NASA NTRS 20200002690. PDF
- NASA (2021). Spacecraft Line-of-Sight Jitter Management and Mitigation — Lessons Learned and Engineering Best Practices. NASA/TM-20210017871. PDF
- Hilkert, J. M. (2008). Inertially Stabilized Platform Technology: Concepts and Principles. IEEE Control Systems Magazine, 28(1), 26–46.
- Hilkert, J. M., & Pautler, D. (2011). Reduced-Order Disturbance Observer Applied to Inertially Stabilized Line-of-Sight Control. Proc. SPIE 8052, Acquisition, Tracking, Pointing, and Laser Systems Technologies XXV.
- Sightline Jitter Minimization and Shaping Using Nonlinear Friction Compensation. (2007). International Journal of Optomechatronics, 1(3), 259–283. DOI: 10.1080/15599610701548837
- Dhruv, & Kaushal, H. (2025). A Review of Pointing Modules and Gimbal Systems for Free-Space Optical Communication in Non-Terrestrial Platforms. Photonics, 12(10), 1001. DOI: 10.3390/photonics12101001
- STScI (2022). JWST Line-of-Sight Jitter Measurement during Commissioning. JWST-STScI-008271. PDF
- JWST User Documentation. JWST Fine Guide Stability. link
- RP Photonics Encyclopedia. Position-Sensitive Detectors. link
Want to plug the formulas from this literature straight into motor sizing? That's exactly what the Gimbal Motor Sizing Calculator is for. Want to follow the rest of the series? Bookmark LocalPapa Notes.