Skip to content

Overpass 0.2.0

Choose a tag to compare

@djsincla djsincla released this 25 Oct 17:49
· 164 commits to main since this release

The first week of running Overpass against a HAZ46 at the end of a pass turned up a few things worth fixing, and the object list got a lot more to say about what is up there.

Download

Overpass-0.2.0.dmg below, for macOS 14 or later, signed with my Developer ID and notarised by Apple. Open it and drag Overpass to Applications, replacing 0.1.0.

Tracking

  • Smooth tracking used to hold the planned speed constant for each second of the plan and step it at every sample, which near the top of a pass you could hear as a jolt once a second. The planned speeds now come from a cubic through the plan, so they change a little at every update.
  • How often Smooth tracking updates is a setting, from every 0.05 s to every 2 s. The correction gain and the planned speed adjust to the period, so the slow settings stay stable. Each update costs about 45 ms on a HAZ46, so 0.05 s is close to as fast as the mount answers. The setting only applies once the mount is on target and tracking; waiting at the start point and catching up run at four updates a second.
  • Catching up under rate control now runs at the full axis speed. It was limited to 2°/s, and one 216° catch-up took 110 s that should now take under 40 s.
  • The tracking trace keeps four minutes for export, enough for a whole pass. The chart still shows the last minute.

The mount's own slew

It turns out the HAZ46's :T0 slew treats +180° and −180° on the azimuth axis as two separate ends, and on a HAZ that is due north. Asked to cross it, the mount sets off the long way round, which on my mount was heading for the cables. It also accepts a slew to within a few degrees of ±180° and then quietly does nothing.

Native slews now stop 8° short of ±180° and cross north under rate control at full speed. Any native slew that starts heading away from its target is stopped within about a second and finished under rate control. Settings has a "Test the mount's slew" button that measures all of this on your own mount, which is worth running once on an HAE before trusting it, and the details are in docs/iOptron-Special-Mode-Protocol.md.

Objects

  • Satellites are listed in columns: number, international designator, type, owner, launch date, period, perigee, apogee, inclination, length by diameter, span, radar cross-section, a small, medium or large size class, and TLE age. Click a column header to sort that source by it, and again to reverse. Each source keeps its own sort, and columns that do not fit the window are dropped, least useful first.
  • Search compares attributes as well as names: launched>2020, period<100, perigee>500, inc>50, rcs>10, span>20, size=large, owner=PRC, type=debris, combined as you like.
  • The launch, orbit and radar cross-section come from CelesTrak's satellite catalogue. The physical dimensions come from Jonathan McDowell's General Catalog of Artificial Space Objects, under CC BY 4.0, which covers Tiangong, Starlink and most of what has no published radar cross-section. Neither needs an account. Together they are about 26 MB, downloaded in the background once a week and kept for offline use.
  • The Observe list's brightness sort falls back to radar cross-section and then physical size for satellites, since TLEs carry no magnitude.

Not yet tested on hardware

Crossing north under rate control during a slew, and the 0.05 s and 1 to 2 s update periods over a whole pass. Camera guiding is moving to its own branch while I work out what a cheap UVC webcam will and will not let me control, which so far is not the exposure.