Why Autoguiding Won't Fix a Bad Polar Alignment

You guide all night with a clean RMS log, the guide star barely twitches, and the stack looks sharp in the center. But the stars near the edges are ovals stretching in the same direction, getting worse the farther they are from the center. This is not a guiding problem. This is field rotation, and it comes from polar alignment error that autoguiding cannot correct.

It is one of the most common misdiagnoses in deep-sky imaging. The imager sees elongated stars, checks the guiding log, finds it clean, and blames the optics, the seeing, or the mount. But autoguiding only corrects two axes of drift. It does not correct rotation around the guide star. If the polar axis is off, the whole field rotates slowly throughout the exposure. The corners show it first, and no amount of tuning the guide loop will fix it.

What autoguiding actually corrects

Autoguiding works by measuring the position of a single guide star and sending correction pulses to the mount when that star drifts in RA or DEC. The loop runs every one to three seconds, and it keeps that one star locked at the same sensor position all night.

This corrects two things: periodic error from the mount's worm gear, and steady drift caused by refraction, wind, or imperfect tracking. Both of these move the entire field together. Every star drifts at the same rate and in the same direction, so correcting the guide star corrects them all.

Field rotation is different. It does not move the whole field together. It rotates the field around the guide star. Stars closer to the guide star barely move. Stars farther from it trace longer arcs. On a modern camera sensor that is 15 to 20 millimeters across, those arcs become measurable at the edges long before they show up at the center.

The mechanism: why polar alignment creates rotation

When the mount's polar axis does not point at the celestial pole, the mount tracks around the wrong axis. The telescope still follows the sky's motion approximately, but the field of view slowly rotates relative to the true celestial coordinate grid. This is not a drift that guiding can chase, because it is not a translation of the whole field. It is a rotation, and the guide star is at the center of it.

Think of it this way: autoguiding keeps the hub of a wheel stationary. The rim still turns. On a small sensor at short focal length the rotation is negligible, which is why a wide-field setup can tolerate several arcminutes of polar error. On a longer focal length with a larger sensor, the rim of that wheel moves enough in a five-minute sub to turn round stars into ovals.

How much error is too much?

The relationship is linear and well documented. Hook's drift law gives 0.262 arcseconds of drift at the guide star per minute of exposure per arcminute of polar error, and Baudat's extension of that work translates the rotation into a linear blur at the corners of the frame.

In practical terms, for a typical deep-sky rig at 600 mm focal length with a four-thirds sensor, a polar alignment error under 2 arcminutes keeps field rotation below one pixel for a five-minute sub. At 3 to 4 arcminutes, which is what many electronic polar alignment routines deliver on a routine night, the corner stars begin to show elongation in subs longer than five minutes. At 6 to 8 arcminutes, which users have reported from ASIAIR's All-Sky Polar Align and from NINA's first rough alignment pass, the field rotation becomes visible even in three-minute subs on a moderately sized sensor.

For longer focal lengths the tolerance shrinks. At 1000 mm, aim for under 1 arcminute of error for clean five-minute subs. At 1500 mm and above, even 1 arcminute leaves detectable rotation at the edges, and an off-axis guider becomes necessary because picking the guide star from the same optical path eliminates the extra mechanical rotation that a separate guide scope introduces.

How to tell it is polar alignment and not something else

The symptom has a specific signature. Stars at the center of the frame are round. Stars become progressively more elongated toward the corners. The elongation points in the same direction across each corner, and the direction changes depending on where the guide star sits relative to the frame. If the guide star is at the center, the elongation is radial: each star stretches away from the center. If the guide star is off-center, the pattern shifts.

Compare this with a guiding problem. A guiding issue produces stars that are elongated the same way across the entire frame, because the mount is oscillating or drifting during the exposure. The guiding log will show a higher RMS or a periodic sawtooth pattern. With polar alignment field rotation, the guiding log can look perfect while the corners are oval.

Compare it also with optical issues. Coma produces elongation at the corners too, but it points away from the field center regardless of where the guide star is, and it is constant from sub to sub. Field rotation from polar error changes direction based on where the guide star sits and only appears during exposures long enough for the rotation to accumulate.

Diagnosing your own frames

Open a single uncalibrated sub and look at the corners. If the elongation is consistent across all corners and points radially toward or away from the center, open the same sub at full resolution and compare the corner stars with stars at the center. If the center is round and the corners are stretched, you are looking at field rotation.

Then check your guiding log. If RMS is below 1 arcsecond and the DEC graph does not show a sawtooth, the mount is tracking well and the guiding loop is happy. The problem is before the guide loop: the polar alignment.

If you use NINA or ASIAIR, check what your polar alignment routine reports. NINA's 3PPA gives a numerical error in arcminutes. ASIAIR's All-Sky Polar Align reports a combined error too. A reading of 3 arcminutes or more on a 600 mm scope is worth improving before your next session. One useful sign: if NINA reports "Azimuth error too large to compensate," the error is several arcminutes and well outside the range that a software-based correction can handle. In that case, the mount needs a mechanical adjustment to its altitude and azimuth before you can proceed.

Another quick test: take a short sub, 30 seconds, and a long sub, 5 minutes, of the same field. Compare the corners at the same brightness stretch. If the short sub shows round corners and the long sub shows elongated ones, the elongation is time-dependent, which rules out optical defects and points directly at field rotation.

What actually fixes it

In your own setup, the fix is mechanical: adjust the mount's altitude and azimuth until the polar alignment routine reports under 1 arcminute for long focal lengths, or under 2 to 3 arcminutes for a wide-field rig. Drift alignment is the gold standard, but a plate-solve based routine like NINA's 3PPA or ASIAIR's All-Sky Polar Align will get you close enough for most setups if you run it carefully and verify the result.

In a remote session, the polar alignment is done before anyone books the telescope. The mount is tripod-leveled, the polar axis is set to the site's latitude, and the alignment is verified with a plate-solve routine that achieves sub-arcminute accuracy. The guider is calibrated to that alignment and stays fixed across sessions. When a user connects, the first sub is already sharp corner to corner. No time spent aligning, no risk of a 6-arcminute error eating into your integration.

Autoguiding is essential. But it cannot polish a bad polar alignment. Understanding the boundary between what guiding fixes and what it does not is the difference between chasing the wrong problem all season and fixing it before the next sub.