Guides
Who this section is for. Engineers who know what they want to achieve and need the sequence of steps that achieves it. It assumes you have worked through Getting started and can look up flags in the Reference.
Reference pages describe one command exhaustively. Guides chain several into a workflow and, more importantly, tell you what to do when the first answer is wrong, which is the normal case.
Designing
State what you need, let the solver find a geometry, then verify it properly. The verification step is the one people skip.
Generate a Walker constellationSpecify the geometry directly, and check the phasing you chose is sound.
Import a real fleetAnalyze cataloged objects, with the epoch discipline that makes the result meaningful.
Propagating
From a constellation to a timeline and the exports that follow from it.
High-fidelity runsAdd drag, solar radiation pressure, and third bodies, and know what each one costs you in required inputs.
Analyzing
Measure what a design actually sees, and converge the answer rather than accepting the first number.
Pointed and pushbroom sensorsModel what the payload sees rather than what the horizon allows, and read the revisit gap in the light of how much was covered at all.
Access windows and pointingPer-pass azimuth, elevation, and range, and why the best pass is the one an azimuth-elevation mount cannot track.
Moving-platform accessConnectivity along an aircraft or ship route, and why substituting a fixed point agrees on the total while scrambling the assignment.
Closing a linkTake a design that fails its link budget and work out which constraint actually binds.
Measured MODCOD thresholdsMeasure the Es/N0 a scheme actually needs instead of assuming one, and see what the cheap default measurement costs in availability.
BER sweeps and coding gainTell a converged Monte Carlo point from one where the simulator ran out of budget, and compare codes on an axis that charges for bandwidth.
Doppler and orbital elementsSize a receiver’s frequency search from real geometry, and check a propagator’s secular rates against the closed form.
Inter-satellite linksRoute through the crosslink mesh, and find the terminal range at which a comfortable latency stops existing.
Eclipse and power sizingWhy the plane with the easiest power budget today has the hardest one in three months.
Operating
Filter the events that are not collisions, and see why collision probability is a statement about your covariance rather than about the encounter.
Station-keeping budgetsSize propellant for a fleet, and read the annual figure correctly once the tanks run dry.
Orbit determinationFit an orbit with batch least squares and an EKF, and find out why their uncertainties are not comparable numbers.
The habit these guides share
Every guide ends by checking its own answer. That is not padding.
The recurring failure in constellation analysis is not a wrong calculation, it is a right calculation of the wrong question: coverage quoted without an elevation mask, a design verified over a window too short to be representative, a link budget evaluated at zenith. Each produces a number that is internally correct and operationally useless.
Where a guide can show you that failure with real output, it does.
Where the answer changed under scrutiny
Each of these came out of running the software rather than reading it, and each is written up in the guide that found it.
| Guide | The first answer | What re-checking showed |
|---|---|---|
| Synthesize from requirements | “Feasible” | 95 percent availability against a 99.9 percent target |
| Coverage and revisit | 2160 s worst gap | 5760 s once the grid converged |
| Pointed sensors | 70.1 percent coverage | 12.3 percent for the payload actually flown |
| Access windows | 1.6 deg/s azimuth rate | 6.03 deg/s at a step fine enough to see it |
| Moving-platform access | Totals matched exactly | 24 of 60 satellites disagreed |
| Inter-satellite links | 32.9 ms | No route at all below a 3500 km terminal |
| Eclipse and power | One plane in permanent sunlight | The same plane among the worst three months later |
| Measured MODCOD thresholds | A 2.75 dB threshold | Still climbing at 200 frames, optimistic at every step |
| BER sweeps and coding gain | A BER twice the theoretical one | Six errors, so not a measurement at all |
| Screening for conjunctions | 1042 events under 10 km | Every one co-orbiting, none a collision |
| Station-keeping budgets | 120.00 kg/yr of propellant | Supply divided by horizon, not demand |
| Orbit determination | The EKF looked 12x worse | Process noise, not estimator quality |
In every case the first answer was optimistic, and nothing in the output marked it as provisional.
main (pre-release)