You drove ninety minutes to dark sky. You polar aligned properly. PHD2 settled at 0.4 arcseconds and stayed there all night. And in the morning you opened the subs and half of them had eggs for stars.
If that has happened to you, the worst part isn't the lost data. It's not knowing what you did wrong — and quietly wondering whether the answer is another two thousand dollars of mount.
It usually isn't. In most cases the gear is fine. Something moved your image without moving your guide star, and your guider had no way to know.
That's a solvable problem, and you don't need to be an optical engineer to solve it. You need to know which three things your guider is blind to, and which piece of kit sees them. That's what this is.
What an autoguider can and can't see
An autoguider watches one star, measures how far it drifted, and tells the mount to push back. That's the whole idea. Which means — and this is the bit that catches people — it can only fix errors it can see.
If something moves your image without moving your guide star, the guider is blind to it. It'll report a lovely graph while your subs quietly rot. Jerry Lodriguss, writing specifically about mirror movement, puts it better than we can:
"The autoguider software will think everything is fine, and you won't get any hint in the autoguider display or autoguider logs that anything is wrong. But the stars in the image will be elongated or trailed."
His point generalises to everything in this section. There are three ways it happens.
Differential flexure
Your guide scope and imaging scope are two rigid objects bolted to a third. Anything that lets them move relative to each other — a flexing ring, a settling dovetail, a drooping focuser, a stiff USB cable tugging as the mount rotates — goes straight into the image and never shows up on the guide graph.
The tolerance is tighter than most people picture. At 1000 mm with 3.76 µm pixels, half a pixel is about 0.39 arcseconds. Across a 200 mm mounting baseline, that's a relative movement of 0.38 micrometres — a few hundredths of the width of a human hair, held steady through every temperature change and every meridian side.
Nobody's rings are bad. That number is just genuinely hard.
Worth knowing: flexure produces an angular error, so it trails your star by the same amount whatever your focal length. It doesn't get worse as you go longer. You just start noticing.
Mirror flop on SCTs and Maksutovs
Schmidt-Cassegrains and Maksutovs focus by sliding the primary mirror along an internal baffle tube. Celestron are refreshingly straight about this in their own knowledgebase: that's how the design works, and on bigger apertures the mirror can shift under gravity as the telescope tracks — "especially in SCTs with apertures 11 in and larger," and particularly when moving through the meridian.
Here's why it's so frustrating: that movement happens downstream of where a guide scope is looking. Your guide star sits perfectly still. Your image walks away. You can chase settings all night and change nothing.
On magnitude, we'll be honest rather than dramatic. Rack an SCT through focus at high power and owners typically report the image jumping 10 to 20 arcseconds — half a Jupiter or so. But that's focus-induced shift, which isn't quite the same thing as flop during tracking with the mirror at rest. Plenty of individual scopes show almost none, and we couldn't find a published instrument measurement we'd stand behind. So treat the number as unknown for your particular tube. What isn't in doubt is the direction of the problem: whatever the magnitude, a guide scope cannot see it.
Mount error arriving faster than corrections
Which brings us to the genuinely new part.
If you've just bought a harmonic mount, read this bit
Strain-wave mounts — the ZWO AM5 and AM5N, iOptron's HAE and HEM range, Sky-Watcher's Wave series, the Pegasus NYX-101 — are the biggest shift in amateur mounts in a decade. Serious payload without counterweights, packs into a backpack, and long focal lengths suddenly within reach of people who were never going to carry a traditional German equatorial up a hill.
They're also the most common source of "my guiding got worse when I upgraded" that we hear on the phone. Not because they're bad — because the settings that served you well on a worm-gear mount will actively work against you here, and nobody mentions that at the point of sale.
Why harmonic mounts need faster corrections
You'll see periodic error figures quoted for these mounts and next to a good worm drive they look alarming. Don't panic at the number on its own — peak-to-peak amplitude doesn't tell you how hard a mount is to guide. A field test published on ZWO's blog puts it well:
"What really matters for the guiding accuracy is not the absolute scale of the PE, but the slope of it, or how fast the location of the stars change."
Slope is roughly amplitude divided by period. The period on a strain wave is comparable to a worm's — Pegasus measure 430 seconds on the NYX-101, against a typical worm's 480 — but the error packed into that cycle is a lot bigger. ZWO specify the AM5 within ±20 arcseconds. A good worm with PEC running is down in the single digits. Same cycle length, much more movement, so the star travels faster and your corrections have to arrive more often.
And it doesn't repeat identically. Different gear teeth engage on each revolution, so you get a shape that's broadly periodic but never quite the same twice.
Don't skip PPEC — this one gets misreported constantly
You can't store a fixed PE curve for a strain wave the way you can for a worm. Pegasus explain why: "Strain waves can change the PE magnitude based on the load and direction of the telescope. Therefore, it is hard to apply a PE profile in each gear."
That gets repeated online as "PEC doesn't work on harmonic mounts." It's an argument against a static, factory-baked curve — not against PEC as a concept. Their actual recommendation is to compute it live: enable PHD2's Predictive PEC on the RA axis, set the period to 430 with auto-adjust, and they report 0.3 arcseconds of RA tracking error doing exactly that. iOptron and Rainbow Astro take a different route again, using encoders to absorb the error before the guider ever sees it.
If you own one of these mounts and you've left PPEC off because a forum told you not to bother, that's probably the highest-value ten minutes available to you tonight.
PHD2 settings for a harmonic mount
- Guide exposures of 0.5 to 1 second, not the 2 to 3 you'd use on a worm. Pegasus specify 1 second for the NYX-101; the ZWO field test found 0.5 or 1 second worked best on an AM5.
- Treat the Guiding Assistant's minimum move as a starting point, not gospel. Pegasus report it suggests values "2 to 3 times higher than the ideal value" on these mounts, and recommend trying a half to a third of what it proposes.
- You can guide declination in both directions. Strain-wave gearing keeps its teeth engaged, so there's effectively no mechanical backlash and the uni-directional dec trick worm owners rely on isn't needed. That's our inference from the mechanism rather than a manufacturer instruction, but it follows.
One honest caveat, because "zero backlash" gets oversold: no gear lash doesn't mean dec needs no attention. Owners do report the Guiding Assistant flagging dec backlash on these mounts — what it's actually measuring there is reversal delay, which is a different thing with similar symptoms. If the number looks odd and compensation makes your results worse, turn the compensation off.
The counterweight thing, which surprised us
These mounts are sold on not needing counterweights, so you'd assume running one balanced must guide better. In a same-gear, same-pointing test published on ZWO's blog, removing the counterweights left the measured periodic error unchanged but improved guiding RMS from 0.9 to 0.7 arcseconds. Less rotating inertia on the RA axis, faster response to short guide pulses.
That's one test by one person and we haven't seen it contradicted, but treat it as a lead worth trying rather than a law. The trade-off is a higher centre of gravity, so whatever you saved on counterweights is best spent on a stiffer tripod.
Why "calibrate near declination 0, close to the meridian" actually matters
This is the instruction everyone reads, most of us have ignored at some point, and almost nobody explains. It's worth two minutes, because getting it wrong quietly degrades every frame afterwards.
When PHD2 calibrates, it's working out two things: which way the camera is rotated relative to the mount's axes, and how far the star actually moves per unit of guide command. Get either wrong and every correction afterwards is scaled wrong.
The problem is that RA doesn't behave the same everywhere in the sky. Move in right ascension near the celestial pole and the star barely shifts; move the same amount near the equator and it travels the full distance. The relationship is a cosine of declination — and at declination 60°, a given RA pulse moves the star only half as far as the same pulse at declination 0.
So calibrate up at declination 70, slew to a target near the equator, and PHD2's idea of "how much does one pulse move things" is off by nearly a factor of three. It'll under- or over-correct on every single frame, all night, while the graph looks perfectly plausible. Calibrate between roughly −20° and +20° declination and the cosine term is within about 6% of 1, which is close enough to ignore.
The "near the meridian" half is about everything else that bends starlight. Down at low altitude you've got more atmosphere, more refraction, more differential flexure as the tube hangs at an awkward angle, and worse seeing. Calibrating there measures your mount plus a pile of noise.
Our guide scope suitability calculator works out your calibration step size too — the number of milliseconds per calibration pulse. PHD2's own rule is roughly 12 steps to move the star 25 pixels, and the right value depends on your guide camera's pixel scale and your mount's guide rate, so it's different for every setup. Feed in your gear and it'll give you the figure rather than leaving you on the default.
Image circles, in plain English
Your telescope projects a disc of usable light at the focal plane. That's the image circle. Your sensor has to fit inside it — measured corner to corner — or the edges go dark and the stars go bad.
Two numbers worth memorising: APS-C is about 28 mm across the diagonal (26.8 mm if it's a Canon), full frame is 43.3 mm.
And two separate failure modes, which get conflated constantly. Illumination is whether light reaches the corner at all. Correction is whether the stars out there are still round. Plenty of optics illuminate further than they correct — which is exactly why "corrected image circle" and "illuminated field" are different numbers, and why some manufacturers are vague about which one they're quoting.
Real image circle figures
| Optic | Corrected image circle |
|---|---|
| Celestron EdgeHD, native | 42 mm (all four apertures) |
| Celestron 0.7× Reducer for EdgeHD | 42 mm — unchanged, but 43% wider angular field |
| Starizona SCT Corrector LF | 42 mm — full frame |
| Starizona SCT Corrector IV 0.63× | 27 mm |
| Starizona Night Owl 0.4× | 16 mm |
| Takahashi FSQ-106EDX4, prime focus | 88 mm |
| Same scope, 645 reducer 0.72× | 60 mm |
| Same scope, CR reducer 0.73× | 35 mm |
Look at those last three rows. Same telescope, two reducers with almost identical reduction factors, and the corrected circle lands at 60 mm or 35 mm depending on which one you bought. Meanwhile Celestron redesigned the EdgeHD reducer specifically to hold 42 mm while widening the field.
So the rule isn't "reducers shrink your image circle." It's check the number for your specific reducer, because it's a design choice and the range is enormous.
Off-axis guiders and image circles
An off-axis guider puts a small prism in the telescope's own light path, just ahead of the camera, picking off a star from outside the sensor's footprint and bouncing it sideways to a guide camera.
Work through APS-C. The prism has to clear the sensor's half-height — about 7.9 mm — plus a few millimetres of margin, so its centre lands around 13 mm off axis and the guide star needs roughly a 27 mm circle. That's comfortably inside the 28 mm the sensor's own diagonal already demands. On APS-C, the OAG asks for essentially nothing your sensor wasn't asking for anyway.
Full frame is where it gets awkward — but not for the reason usually given. That 43.3 mm diagonal already exceeds a 42 mm corrected circle before you fit anything. Add a prism that has to clear the sensor and the camera hardware, and you're asking for light and correction from a zone the corrector was never designed to deliver. It's the sensor that's the problem; the OAG just removes your margin for error.
The reducer trap that catches everyone using an OAG
Here's the one that costs people whole nights, because it runs backwards from intuition.
Adding a reducer makes your off-axis guide star worse, not better. Twice over.
First, the aberration. The prism sits at a fixed physical distance from the optical axis. A reducer shortens the focal length, so that same physical spot now corresponds to a larger angle on the sky — and off-axis aberration scales with angle. Fit a 0.63× reducer and your guide star lands where a star half again as far off axis used to be. Same prism, same position, worse star.
Second, the light cone. The prism sits in a converging beam, and the width of that beam where it crosses the prism is just the distance to focus divided by the focal ratio. A faster cone is a wider cone. Put an 8×8 mm prism roughly 54 mm from focus behind a filter wheel and a cooled camera: at f/7 the cone there is 7.7 mm across and the prism catches essentially all of it. At f/4 it's 13.5 mm across and the prism catches under half. The reducer that made your imaging faster just dimmed your guide star.
None of which means don't use a reducer with an OAG — loads of people do, successfully. It means check it rather than assume, and don't be startled when the guide star that was fine at native focal length turns faint and comet-shaped at f/6.3.
While we're here: a comet-shaped guide star on an SCT is normal and not a fault. An SCT carries noticeably more off-axis coma than a classical Cassegrain of the same focal ratio, and at the radius an OAG prism lives at, several arcseconds of flare is routine. Guiding software centroids it perfectly happily. It's only worth worrying about if it changes.
Guide scope vs off-axis guider: how to actually decide
The usual answer is "OAG above about 1000 mm, because pixel scale." That lands on roughly the right conclusion for slightly the wrong reason. Modern guiding software locates a star far more precisely than the old rules assume, and raw pixel scale is rarely what limits people.
Here's the version we'd give you on the phone.
Buy a guide scope when
You're at short to medium focal length, on a refractor, with a rigid mounting arrangement. Easier to set up, easier to focus, wide field full of stars, and it will do the job well. If you're imaging under about 1000 mm with a refractor: get a decent guide scope with proper rings and stop thinking about it. You have better things to spend money on.
Buy an off-axis guider when
- You're on a moving-mirror SCT or Maksutov. This is the categorical one. A guide scope cannot see mirror flop, ever, at any price. An OAG picks its star off after the mirror, so any flop moves the guide star too and gets corrected like anything else.
- You're at long focal length, where flexure tolerances get down to sub-micron and no ring-and-dovetail arrangement holds them reliably night after night.
- You've eliminated everything else and still have trailed stars alongside a clean guide graph. That combination is very nearly diagnostic.
Think twice about an OAG when
You're at f/4 or faster with a small prism, or running full frame on a 42 mm corrected circle. Both are workable. Neither is comfortable, and you deserve to know that before you order.
Your next three steps
If you're staring at a folder of trailed subs right now, do these in order.
- Rule out the free fixes first. Check your calibration was done near declination 0 and near the meridian. On a harmonic mount, turn on Predictive PEC and drop the guide exposure to a second. Tighten every ring and dress your cables. None of this costs anything, and it resolves a real share of the cases that come to us.
- Find out what's actually limiting you. Run your gear through the Guide Scope & Off-Axis Guider Suitability Calculator. It names which of four things is the constraint — measurement precision, sampling, guide star availability, or signal — so you stop guessing.
- Then, and only then, buy something. And buy the thing the numbers pointed at, not the most expensive thing in the category.
Run the numbers for your own gear
Everything above is general. Your rig is specific, and the numbers move a lot with focal length, sensor size, prism dimensions and how dark your sky is.
We built a free Guide Scope & Off-Axis Guider Suitability Calculator that does it properly. Real diffraction and seeing to work out your guide star's true size, the CCD equation to check it's actually bright enough, and Cramér-Rao centroid theory for how precisely it can be measured. Pick your telescope from over 330 refractors and 79 Newtonians plus the SCT, Maksutov, Ritchey-Chrétien, CDK and RASA ranges, pick your cameras, and it tells you which of four things is limiting you, what to change, and where to start in PHD2 — including a minimum-move value and calibration step derived from your own equipment rather than a rule of thumb.
It runs in your browser and nothing is uploaded.
And if you'd rather just talk to a human, that's genuinely what we're here for. Tell us your telescope, camera and mount and we'll tell you honestly whether you need to spend anything at all. Sometimes the answer is a better set of rings and a shorter cable, and we'll happily say so.
Frequently asked questions about guiding
What focal length guide scope do I need?
Less than the old rules suggest, and for a different reason than you'd expect. What usually sets the answer isn't your imaging scope at all — it's getting the guide star onto a sensible number of guide pixels, which depends on your guide camera's pixel size and on diffraction in the guide scope. Both are independent of what you're imaging with. Our calculator gives you the shortest focal length that works and the one we'd call comfortable, scored against all four failure modes.
Is the "1/3 rule" for guide scope focal length still true?
It's a conservative shorthand that predates sub-pixel centroiding, built on the assumption that measurement precision scales with pixel scale. In angular terms it doesn't, until the star drops below about one pixel across. The rule isn't harmful — it just points at the wrong variable, and it'll send you to a longer guide scope than you need while saying nothing about flexure or mirror flop.
Why is my guiding RMS good but my stars are still elongated?
Three usual suspects, roughly in order: differential flexure, mirror flop, and mount problems guiding can't fix. The first two are invisible to the guide graph by definition — the guider is holding the guide star perfectly still while the imaging field moves independently. Two useful tells: if the elongation reverses direction after a meridian flip, suspect flexure; if you're on a moving-mirror SCT or Maksutov with a guide scope, suspect mirror flop first.
Can I use an off-axis guider on a fast astrograph?
You can, but prism illumination gets difficult — and in the opposite direction to most people's intuition. At a fast focal ratio the light cone at the prism is wider, so a small prism intercepts less of it. The fixes are a larger prism or a shorter distance from prism to focal plane: a thinner body, or dropping a filter wheel out of the train. Below about f/4 with a small prism, a guide scope is often more practical — and at those short focal lengths flexure matters far less anyway.
How much back focus does an off-axis guider use up?
Between about 13 and 30 mm depending on the model, out of the 55 mm most reducers and flatteners require. Extra-thin bodies come in around 13 mm; a ZWO OAG is 16.5 mm; a Celestron Deluxe is nearer 29 mm. Add a filter wheel at roughly 20 mm and a camera at 17.5 mm and the budget disappears fast. The calculator adds your specific train up and tells you whether it fits.
Mono or colour guide camera — does it matter?
Mono is worth roughly 0.6 of a magnitude. A colour sensor's Bayer filter throws away most of the light hitting any given pixel, and you gain nothing from the colour because the guider only wants a position. A colour planetary camera you already own will guide perfectly well — it just won't reach as faint, which matters most on an off-axis guider where stars are scarce.
What guide exposure should I use?
Two to three seconds suits most worm-gear mounts; 0.5 to 1 second for an encoderless harmonic mount. Shorter doesn't mean better tracking — below about a second on a worm you start chasing atmospheric seeing, which is random and uncorrectable, and chasing it actively degrades your stars. Longer exposures reach fainter, but temper expectations: guiding is normally limited by sky background and the star's own photon noise rather than read noise, so going from 2 seconds to 8 buys about three quarters of a magnitude — roughly twice the available stars, not four times.
Should I bin my guide camera?
If your guide star spans more than about six pixels, yes. PHD2 measures the centroid inside a fixed seven-pixel radius with its background annulus from seven to twelve pixels, so a star much wider than that is being clipped by its own measuring aperture and contaminating its own background estimate. Binning brings it back into range and costs nothing. This bites most often on an OAG at long focal length with a small-pixel camera, where a star can easily span fifteen or twenty pixels.
Will an off-axis guider fix mirror flop on my SCT?
Yes, and it's the strongest single argument for one. On a moving-mirror SCT or Maksutov the primary can shift slightly as the telescope tracks. A guide scope can't see it, because it happens downstream of where the guide scope looks — the guide star stays put while the image drifts. An OAG picks its star off after the mirror, so the flop moves the guide star too and gets corrected like any other error.
Why does my off-axis guide star look like a comet?
Because it's off axis, and your telescope has off-axis coma. On a standard SCT, several arcseconds of flare at a typical prism radius is completely normal and not a fault. It's also mostly harmless — guiding software centroids a comatic star happily. Two things to know though: the measured centre sits offset from the true position, so a German-equatorial meridian flip will shift your framing by roughly twice that offset (plate-solve and re-centre afterwards), and the extra blur does cost you some signal.
Do I need to guide at all with a harmonic mount?
In practice, yes, for anything beyond very short subs. The periodic error on an encoderless strain-wave mount is large enough that unguided exposures get short quickly — owners at around 1 arcsecond per pixel typically report needing to guide beyond roughly 10 seconds. Encoder models are a different story, since the encoders absorb most of the error before it reaches your image. The upside is that these mounts guide well once set up properly: 0.4 to 0.7 arcseconds RMS is routinely reported, which is competitive with far heavier traditional mounts.
Guide scopes · Guide cameras · Off-axis guiders · All our free astronomy calculators
Reviewed July 2026 by Ontario Telescope & Accessories. Periodic error, image circle and guiding figures are quoted from manufacturer documentation and published field testing, linked where possible. Mirror-shift magnitudes are experienced-observer estimates rather than instrument measurements and are flagged as such in the text. Specifications change — check current documentation before buying on the strength of any single number.