Sharp optics do not guarantee a sharp exposure

A lunar crater is visible in the eyepiece. You touch the focuser and the crater shakes for several seconds. In a photograph, stars stretch into streaks even though focus is correct. A more expensive mirror or greater magnification does not remove motion between the tube, its support and the sky.

A mount must both carry the telescope and control its orientation accurately throughout an observation. Balance reduces unwanted load torque; stiffness limits deformation; the drive follows the sky. On an altazimuth mount, holding one star at the center does not by itself stop the rest of the field from rotating. The choice of support therefore follows the task: a brief look through an eyepiece, planetary video or a long exposure of a faint object.

We will examine axes, mechanical parts and control, then follow the same hypothetical stars under three tracking modes. Useful prerequisites are focus, eyepiece and camera and the construction of a Newtonian tube. Four visible mathematical branches are optional; the main explanation works without them.

Mount history is not one type replacing the next

Large nineteenth-century refractors needed controllable supports for long, heavy tubes. The Great Equatorial Telescope installed at Greenwich in 1893 used an equatorial support with an axis aligned with Earth’s rotation. Its mount was adapted from an earlier, smaller instrument, giving the new tube an unusual shape around the middle. This illustrates mechanics shaping a telescope, rather than dating the invention of the equatorial principle. The operating museum on its construction and history.

Another approach simplified and lowered the cost of the support while retaining a large aperture for visual observing. The Sidewalk Astronomers, founded in 1968, taught accessible telescope making; John Dobson popularized a low altazimuth mount with simple sliding bearings. Here “Dobsonian” describes the mechanical implementation, while the reflecting tube remains Newtonian. The organization’s own history, Stellafane’s original construction.

Compact catadioptric tubes also helped spread fork mounts: the tube is held on both sides, and the whole fork turns about a second axis. Celestron dates the popular orange C8 to the 1970s; the CPC manual documents a fork that can operate in altazimuth mode or equatorially on a wedge. Fork shape and axis orientation are separate characteristics. Celestron’s history, CPC modes and equatorial wedge, pp. 20 and 30.

Professional altazimuth telescopes are not a return to primitive supports. NTT achieved first light in March 1989 with a compact altazimuth structure; computer control and an instrument rotator solve problems that equatorial geometry partly solves through axis orientation itself. ESO: NTT.

Pointing, tracking and guiding are different actions

Pointing moves the tube to a selected part of the sky. It may be manual or automatic, GoTo. A successful GoTo command means the object enters the field; it does not yet prove that a long exposure will be sharp.

Tracking continuously changes orientation after pointing. Earth rotates, so the direction to a distant star changes in local coordinates. Keeping it centered requires periodic nudges on a manual mount or an appropriate motion law on a motorized one. The Moon, a planet or a satellite also has its own motion: one constant stellar rate is not exact for every target.

Guiding adds feedback from the star’s actually measured position. A camera or observer detects a departure and sends a correction. An axis encoder and a guide camera look at different things: the former measures a mechanical assembly’s angle, the latter the position of the light pattern. Flexure downstream of the encoder or camera motion relative to the tube may not appear in the encoder reading. Celestron: tracking and periodic-error correction, MMT: guiding and field rotation.

First define the horizon and celestial pole

An altazimuth mount turns about a vertical azimuth axis and tilts the tube about a horizontal altitude axis. These are the natural movements “left–right” and “higher–lower.” Tracking a star generally changes both angles, at rates that are not constant.

On an equatorial mount, the polar or hour axis is parallel to Earth’s rotation axis. The second, declination axis is perpendicular to it. For an idealized fixed distant star, after setting declination it is sufficient to track about the polar axis at the appropriate stellar rate. Real operation still needs corrections for mechanical errors, polar alignment and target motion.

The diagram specifies latitude 50° north. The horizontal direction to the right is north, upward is the local vertical, and east points out of the page. The north celestial pole is 50° above the northern horizon. In the Southern Hemisphere, the visible pole and drive settings differ. Polaris is close to the pole, but is not the pole itself. CPC: polar alignment and hemispheres, pp. 30–32.

Axes relative to the horizon and pole

Axes relative to the horizon and poleOur geometry in the local meridian plane at latitude 50° N. North is right, zenith is up, and east points out of the page ⊙. Azimuth axis AZ is vertical; polar axis RA is 50° above the northern horizon. Circles ALT/DEC show axes perpendicular to the page in one chosen orientation. This is a direction diagram, not a mount drawing. NCP means north celestial pole; Polaris is not exactly at that pole.AltazimuthNorthZenithEquatorialNorthZenithAZRA → NCPALT ⊙DEC ⊙50°

AZ: azimuth; ALT: altitude; RA: polar axis; DEC: declination. East: ⊙. Latitude φ=50° N.

Our geometry in the local meridian plane at latitude 50° N. North is right, zenith is up, and east points out of the page ⊙. Azimuth axis AZ is vertical; polar axis RA is 50° above the northern horizon. Circles ALT/DEC show axes perpendicular to the page in one chosen orientation. This is a direction diagram, not a mount drawing. NCP means north celestial pole; Polaris is not exactly at that pole. Source and limits.

Fork, German and Dobsonian describe construction

On a German equatorial mount, the tube sits on one side of the assembly and a counterweight on the other. Adjusting its position reduces load torque. This arrangement makes changing tubes convenient, but clearance between the tube, camera, counterweight and tripod must be checked. Crossing the meridian may require flipping the tube to the other side of the support; the procedure depends on the particular mount and controller.

A fork supports the tube with two arms. With its main axis vertical, the arrangement is altazimuth. Tilting the base on a suitable equatorial wedge and correctly aligning it with the pole turns the main axis into an hour axis. A fork is not a separate third way of describing the sky’s daily motion. CPC: the same fork system in different modes.

A Dobsonian is a low altazimuth support with large side bearings and a rotating base. For manual visual observing, its advantages are a simple, short load path and accessible construction. Tracking may be manual; some designs add motors or a special platform, but these are extra assemblies rather than an automatic property of every Dobsonian. Stellafane: altitude bearings.

Stiffness extends from the ground to the camera

The weight of the tube and accessories passes through the clamp, axes and bearing supports to the head, tripod or pier, and ground. A weak joint anywhere along the path changes the optical axis direction. Tripod bracing, short lever arms and sufficiently stiff connections therefore matter as much as the material of the head itself.

Not every bearing is a ball bearing. In the Stellafane design, altitude arcs are made from plywood: a circular disk is cut and finished first, then divided into two matching parts. Roundness and a square working edge matter, not merely nominal diameter. A laminated surface slides over PTFE pads. In the azimuth base, the pads sit near the three feet to shorten the bending load path; the central bolt establishes the pivot rather than clamping everything rigidly. Making the arcs, base and pads.

At another scale, the ELT design uses hydrostatic bearings with an oil film and large machined tracks. The documented structure requires track alignment to fractions of a millimeter over tens of meters; bearing geometry transmits forces to the foundation. These are requirements of a particular project, not a home recipe or a manufacturing result we claim to have achieved. A commercial head’s detailed factory process cannot be recovered from a promotional photograph. ESO: ELT bearings, tracks and load path.

Load path and two kinds of feedback

1 · Weight path to the ground

  1. Camera and tube
  2. Clamp and tube cradle
  3. Axes and bearing assemblies
  4. Head and tripod / pier
  5. Base and ground

2 · Motion command

  1. Target and clock → desired angles
  2. Controller → amplifier → motor
  3. Transmission, if present → axis
Functional diagram after Stellafane, Celestron and ESO documentation. The first chain transmits mechanical load. The second controls motion; direct drive may omit the transmission. An axis encoder and a guide-camera image measure different things: flexure downstream of an encoder need not appear in its readings. This is not a circuit diagram for a particular product. Source and limits.

Balance the tube with all its accessories

Mass alone is not enough: the distance of its center from the axis also matters. A small camera far from the assembly can produce appreciable torque. Adding a camera, changing an eyepiece or pulling a cable after balancing invalidates the earlier equilibrium. The motor then fights both the desired motion and a persistent extra load.

On a Dobsonian, the tube can slide relative to its altitude bearings; on a German mount, tube and counterweight positions are adjusted for the respective axes. Check the complete observing configuration. Hold the tube when releasing a clutch: even appropriately chosen masses do not prevent an unsecured assembly from falling. Stellafane: balance with the heaviest eyepiece and accessories, official Advanced VX manual, pp. 9–10.

The same support before and after adding a camera

The same support before and after adding a cameraOur horizontal lever: axis out of the page, gravity down, counterclockwise torque positive. Point masses: 5 kg at x=+0.18 m and 3 kg at −0.30 m. A camera adds 0.5 kg at +0.25 m. Blue gravity arrows are proportional to forces; CM marks the center of mass, dashed line the axis. After the addition τ=−1.22625 N·m, CM=+14.706 mm; support reaction is 83.385 N. Rebalancing moves the counterweight to −0.341667 m. This is not the exact spatial geometry of RA.Initial balance5 kg0.18 m3 kg-0.30 mCMτ = 0.00000 N·mCamera added5 kg0.18 m3 kg-0.30 m0.5 kg0.25 mCMτ = -1.22625 N·m

3 kg counterweight: −0.300 m → −0.341667 m to restore balance.

Our horizontal lever: axis out of the page, gravity down, counterclockwise torque positive. Point masses: 5 kg at x=+0.18 m and 3 kg at −0.30 m. A camera adds 0.5 kg at +0.25 m. Blue gravity arrows are proportional to forces; CM marks the center of mass, dashed line the axis. After the addition τ=−1.22625 N·m, CM=+14.706 mm; support reaction is 83.385 N. Rebalancing moves the counterweight to −0.341667 m. This is not the exact spatial geometry of RA. Source and limits.

Balance is not stiffness: an equilibrated but flexible tube still deflects in wind. Nor should advice about slight one-sided loading of a worm gear be transferred blindly to every direct drive. It is a setting for a particular mechanism, governed by its manual.

Branch · With equationsOptional detail; the main explanation continues below.

Torque and center of mass: a short calculation

Our example is a horizontal lever with its axis pointing out of the page; gravity acts down and lever arms are horizontal. Coordinate x is positive to the right of the axis, and counterclockwise torque is positive. For masses m_(i) and specified g = 9.81 m/(s)^(2):

\begin{aligned} \tau&=-g\sum_i m_i x_i,\\ x_\mathrm{CM}&=\frac{\sum_i m_i x_i}{\sum_i m_i}. \end{aligned}

A 5 kg mass at + 0.18 m and a 3 kg counterweight at − 0.30 m produce opposite torques of 8.829 N·m each. A 0.5 kg camera at + 0.25 m adds τ = − 1.22625 N m. The combined center of mass shifts right by 14.706 mm. Restoring equilibrium requires moving the same counterweight to x = − 0.341667 m.

The support reaction also balances total weight: zero torque does not mean an unloaded bearing. This is our own lever, not the exact spatial geometry of RA at latitude 50°; there, torque follows a vector cross product with the corresponding gravity projection. The “mass × lever arm” principle remains. Static equilibrium and torque.

Extra weight does not always make a system steadier

A gust of wind exerts a force on the tube and a torque about the support. Materials and joints deform elastically even without slipping or backlash. A changing torque can make the system oscillate around equilibrium. After touching the focuser, what we see is motion of the image rather than an instantaneous deterioration of the mirror polish.

Shorter extended tripod legs, a stiff connection and wind protection can help, but their effect must be checked on the actual structure. A heavy camera increases inertia and loads the focuser; mass without increased stiffness can lower the oscillation frequency. The exact settling time depends on damping and many assemblies, so it cannot honestly be guessed from tripod mass. Large telescope design considers structure, foundation and wind loads together. ESO: stiffness, mass and wind trade-offs.

Inertia changes frequency, not static deflection

Moment of inertia JStatic angle θFrequency f₀
2 kg·m²5 µrad15.915 Hz
8 kg·m²5 µrad7.958 Hz
Our single linear torsional mode: τ=0.1 N·m, kθ=20000 N·m/rad, f=1 m, p=5 µm. Only J changes. Both static angles are 5 µrad, giving about 1 px displacement. Undamped natural frequency does not determine settling time. These are not measurements of a real tripod. Source and limits.

Branch · With equationsOptional detail; the main explanation continues below.

A small deflection can already be one pixel

Take one hypothetical elastic rotational mode: angular stiffness k_(θ), moment of inertia J, and a slowly applied torque τ. In the linear model, static deflection and undamped natural frequency are:

θ = (τ)/(k_(θ)),; f_(0) = (1)/(2π)√((k_(θ))/(J)).

With our chosen τ = 0.1 N m, k_(θ) = 20000 N m/rad and J = 2 kg (m)^(2), we obtain θ = 5 × (10)^( − 6) rad and f_(0) = 15.915 Hz. For focal length f = 1 m and pixel pitch p = 5 μm, the displacement magnitude in central projection is:

N_(px) = (ftan⁡θ)/(p) ≈ 1.

At the same stiffness, increasing J to 8 kg (m)^(2) halves the natural frequency to 7.958 Hz without reducing static deflection under the same torque. These are not measured characteristics of any tripod, a wind-spectrum forecast or a settling-time calculation. The rotational analogue of the harmonic oscillator follows from J\ddot\theta=-k_\theta\theta. Oscillator, stiffness and inertia.

The drive needs to know what happened to the axis

In a worm drive, a motor turns a small screw-like element that slowly rotates the axis’s large wheel. Reduction provides low speed and the required torque. Imperfections in gear shape or alignment can repeat with its rotation. This is one mechanism behind periodic tracking error.

Backlash is a different problem: after a reversal, the motor first takes up a gap while the load does not yet move as expected. Excessively tightening the transmission can cause binding rather than precision. Software compensation shortens reversal delay but does not make the mechanical gap disappear. Celestron: periodic error, backlash and reversal.

Direct drive can rotate an axis without this toothed reduction stage. It requires appropriate motors, sensors and control; removing the worm does not remove wind, elasticity or angular measurement error. The published VLTI Auxiliary Telescope design has direct drives, tachometers and optical encoders on its Alt/Az axes. ELT’s design even separates position- and velocity-loop measurements in different parts of its large axis. ESO: AT drives, ELT loops.

In a closed loop, the controller compares required and measured motion and commands the drive. It must account for geometry, travel limits and cables. The mere presence of electronics does not establish that the whole mechanism is accurate enough for a chosen focal length and exposure.

Assembly and alignment follow a clear order

Here is the documented assembly logic for Advanced VX, without a requirement to buy that model. Its official manual starts by spreading and locking the legs, fitting the head and tightening the central attachment; the spreader tray braces the tripod. Next the counterweight shaft is installed and secured, the counterweight added, and the end safety screw replaced. The optical tube is secured to the dovetail saddle with the axes locked and counterweight already installed. This is one manufacturer’s sequence, not universal compatibility between arbitrary parts. Advanced VX manual, pp. 7–9.

After fitting observing accessories, RA and declination balance are checked while holding the tube as clutches are released. Check attachments, free motion, clearance to legs and enough cable slack. Polar-axis alignment uses the base’s altitude and azimuth adjustments; simply rotating the tube around RA/DEC does not change the polar axis’s direction. Advanced VX: balance and physical adjustments, pp. 10–11.

The stellar pointing model is adjusted separately: the controller relates axis positions to the sky. GoTo star alignment and polar alignment are not the same action. Moving the base after calibration can alter the relationship; consult the current instructions for that particular system. We have not physically assembled or tested the mount described here.

A star does not wait for a full solar day

Two nearby returns of the starry sky are separated by approximately 23 h 56 min 04 s, rather than 24 solar hours. Earth also moves along its orbit, so its orientation relative to the Sun and distant stars repeats over different intervals. A drive making “one turn per 86400 seconds” does not provide exact stellar tracking.

Direction and hemisphere also matter for practical control. In local coordinates, stars move around the celestial polar axis in the direction opposite to the Earth-fixed frame’s own rotation. Calling it “clockwise” without defining the viewpoint can confuse. Our diagrams and model therefore specify hemisphere, axes and hour-angle sign; a real controller needs the corresponding mode. USNO: sidereal time and UT1.

Polar alignment does not make every exposure indefinitely long: drive error, atmosphere, flexure and target motion remain. A proper test follows displacement over time and compares it with imaging requirements, rather than merely listening for a running motor.

Branch · With equationsOptional detail; the main explanation continues below.

A small rate error accumulates into pixels

Our short-interval model uses a rounded nominal period T = 86164.1 s on the UT1 scale, close to the period obtained from the Earth Rotation Angle coefficient. We treat the polar axis and distant stars as fixed, ignoring precession, nutation and refraction. This is sufficient for this example, but is not an exact ephemeris; the mean sidereal day measured from the moving equinox differs slightly from the ERA period. USNO Circular 179, §2.6.2, eq. 2.10–2.12.

ω = (2π)/(T) ≈ 15.041 (arcsec)/(s_(UT1)).

Suppose the drive mistakenly makes one turn per 86400 s on that same scale. At the celestial equator, after t = 300 s the angular difference is about 12.320 arcsec. For f = 1 m and p = 5 μm:

Δθ = (ω − ω_(drive))t,; N_(px) = (ftan⁡∣Δθ∣)/(p) ≈ 11.946.

At another declination, the small-circle geometry must be included; near the pole, the same hour-angle error gives less angular displacement across the sky. This number is our own equatorial calculation, not an Advanced VX measurement or the exact solar tracking rate, which comes from a model of the Sun’s motion. The correct rate makes this particular contribution zero, while other errors remain.

A fixed center does not imply a stationary field

An altazimuth support can follow the frame center very accurately with two axes. Its “up,” however, remains tied to the local vertical, while celestial north changes direction relative to that vertical. Neighboring stars therefore rotate around the center. A guide camera holding just one star does not by itself also determine the orientation of the whole field.

In the figure, the same hypothetical set of five stars is observed for 300 s, with the same focal length and pixel pitch. Without tracking the whole set moves; ideal altazimuth tracking holds the central star but leaves arcs from the others; ideal equatorial tracking preserves every position. We deliberately chose an example near zenith, where orientation changes rapidly, to make the effect visible. This is not a typical rate for the entire sky or a promise about a particular motor’s capabilities.

The same stars, time and scale: three modes

The same stars, time and scale: three modesOur ideal geometric model: 5 stars, 61 positions over 300 s, φ=50° N, δ=49°, H₀=−2°, f=1 m, p=5 µm. All three squares represent the same 8000×8000 px; scale is 0.03 SVG units per pixel. Open circle: start; solid dot: end; stationary points have no drawn arc. Virtual gnomonic plane: x right, y up, rather than a real camera’s optical parity. The altaz center remains fixed while the field turns +26.067° counterclockwise. The target is near zenith: 88.361°→88.889°, deliberately a rapid example, not a universal rate or motor capability. Refraction, precession, errors, distortion and derotation are omitted.No tracking08000 × 8000 px · 300 sAltazimuth08000 × 8000 px · 300 sEquatorial08000 × 8000 px · 300 s

Identical parameters and scale. Central star: 0; start ○, end ●.

Our ideal geometric model: 5 stars, 61 positions over 300 s, φ=50° N, δ=49°, H₀=−2°, f=1 m, p=5 µm. All three squares represent the same 8000×8000 px; scale is 0.03 SVG units per pixel. Open circle: start; solid dot: end; stationary points have no drawn arc. Virtual gnomonic plane: x right, y up, rather than a real camera’s optical parity. The altaz center remains fixed while the field turns +26.067° counterclockwise. The target is near zenith: 88.361°→88.889°, deliberately a rapid example, not a universal rate or motor capability. Refraction, precession, errors, distortion and derotation are omitted. Source and limits.

Long altazimuth observations rotate the camera or use appropriate optical derotation. This is a third controlled orientation with its own limits and calibration. At NTT, instrument rotators continuously maintain the required sky direction. Software alignment of short frames can correct orientation differences between them; it cannot recover information already smeared by rotation within a single long exposure. ESO: the rotator’s role, MMT: one guide star and residual rotation.

Branch · With equationsOptional detail; the main explanation continues below.

The parallactic angle checks field rotation

Let φ be geographic latitude, δ the star’s declination, and H its hour angle, positive west of the meridian. The angle between the star’s celestial meridian and the local vertical direction can be calculated with auxiliary quantities:

Q_(x) = tan⁡φcos⁡δ − sin⁡δcos⁡H,; Q_(y) = sin⁡H,; q = atan2⁡(Q_(y),Q_(x)).

In our virtual plane with x right and y up, the entire stellar pattern’s orientation change under altazimuth tracking is Δq counterclockwise. Another optical parity or coordinate direction changes the written sign; it cannot be transferred between instruments without checking. In our model this result is independently checked against spatial-vector rotation and projection onto the camera plane.

For the specified φ = (50)^(∘), δ = (49)^(∘), and initial H = − (2)^(∘), 300 s gives about 26.067°. Target altitude rises approximately from 88.361° to 88.889°. At zenith itself, azimuth and the vertical circle’s orientation become degenerate, so the model does not return a falsely defined angle. Real mounts also have speed and access limits near zenith. MMT: parallactic angle, rotation rate and zenith.

Similar trails can have different causes

A slow steady drift suggests checking tracking rate and polar alignment. A repeatable departure with the transmission period calls for examining periodic error and whether PEC is appropriate. A delay after reversal suggests possible backlash. Motion following a touch or under wind gusts calls for checking stiffness, joints and cables. If the center stays fixed while edges form arcs, check field rotation and derotation. These are diagnostic clues, not an unambiguous diagnosis from one photograph.

PEC replays a learned correction to the repeatable part of an error, indexed to the mechanism’s cycle. It cannot foresee a new gust. Guiding responds to measured departures, but overly aggressive or delayed corrections can worsen motion; mechanical flexure between guide and main cameras also limits the result. Identify the contribution first, then choose how to reduce it. The manufacturer on PEC, backlash and guiding.

Comparing repeated frames with time, orientation and configuration recorded is more useful than simultaneously changing several settings at random. Here we checked documentation, mathematical models and diagrams; the physical performance of particular equipment requires actual observing.

Choose the support to match the observation

For manual viewing, a wide field and comfortable, smooth movements may matter more than automatic pointing. Long-exposure photography needs agreement between angular error, focal length, pixel pitch, exposure and field orientation. A mass-capacity label alone does not describe tube length, camera lever arms, wind or required precision.

In space, the ground-based tripod disappears, but the direction-stabilization problem remains. Webb, for example, combines reaction wheels, star trackers, gyroscopic measurements, FGS and a fine steering mirror. This is another platform and set of loops, rather than an equatorial mount taken into orbit. NASA: pointing and stabilizing Webb.

Good optics determine the image that can be formed; the mount determines whether it stays in position for the required time. Greater magnification makes shaking more visible without removing its cause. For a concrete example, see the detailed construction of a Newtonian tube on a Dobsonian support: it shows how balance, bearings and mechanical clearances become part of a complete structure.