Galvo Scanner Selection Guide
This guide covers the selection of 2D galvo scanners only. Front-focusing 3D galvo scanners and rear-focusing 3D galvo scanners are a different dimension of the problem and will be covered in dedicated follow-up articles.
1. What Is a Galvo Scanner?
In One Sentence: A Galvo Scanner Is Two Galvanometers
In English, a galvo scanner is called a Galvo Scanner, where “Galvo” is short for Galvanometer. What is a galvanometer? It is the ammeter you saw in middle school physics class — pass a current through it and the needle deflects; the larger the current, the larger the deflection.
A galvo scanner works on exactly the same principle: attach a reflective mirror to the needle of a galvanometer. When the needle deflects, the mirror deflects with it, and the reflection angle of the incoming light changes — the needle’s deflection is used to control the direction in which light is reflected.
- X-axis galvo = one galvanometer with a mirror attached to its needle. When powered, the mirror deflects and controls the horizontal position of the laser beam
- Y-axis galvo = another galvanometer working on the same principle, controlling the vertical position of the laser beam
With one mirror oriented horizontally and the other vertically, the laser beam is reflected by the X mirror onto the Y mirror and then out from the Y mirror, so it can be aimed at any single point on the focal plane.
It is that simple. Apply a voltage to the X axis and it rotates to one angle; apply a voltage to the Y axis and it rotates to another angle. The combination of the two angles places the spot at a particular coordinate on the focal plane. Vary the voltages continuously, and the spot draws lines, text, and patterns — this is laser marking.
How It Differs from an XY Translation Stage
A traditional XY motion stage relies on motors to physically move the entire processing head or workpiece, marking wherever it travels. The problem is high inertia and low speed — moving an entire stage takes time to accelerate and decelerate.
A galvo scanner is completely different: it only rotates mirrors; it never moves the load. A mirror weighs just a few grams, so rotating it is virtually instantaneous. That is why a galvo scanner is tens or even hundreds of times faster than an XY stage. A 20W fiber marking machine can mark a string of characters in seconds precisely because of the galvo scanner’s ultra-fast deflection.
But the galvo scanner has its price: the deflection angle is limited, and the marking work area is constrained by the F-Theta lens focal length — unlike an XY stage, which can travel as far as you want. So galvo scanners suit “small area, high speed” applications, while XY stages suit “large area, low speed” ones.
The Three Layers of a Galvo Scanner
A complete galvo scanning system can be broken down, from the outside in, into three layers:
| Layer | Components | Function |
|---|---|---|
| Optical layer | X mirror + Y mirror + coatings | Reflects the laser; the coating determines which wavelengths can be reflected |
| Drive layer | Two galvanometer motors + servo driver board | Converts voltage signals into mirror deflection angles |
| Control layer | Control card + XY2-100 signal cable | Translates graphics commands from the marking software into X/Y voltage signals |
Every parameter you care about during selection — aperture, wavelength, speed, accuracy — is essentially a manifestation of the capabilities of these three layers. We will break them down one by one in the following sections.
An Easily Overlooked Key Point: The Mirrors Cannot Rotate Freely
Beginners often assume the mirrors can rotate by any amount they like. In reality, the deflection angle of a galvo scanner is limited:
- Typical scan angle: ±0.35 rad (optical angle), approximately ±20°
- Note: the optical angle is not the same as the mirror deflection angle. Because the angle of incidence equals the angle of reflection, the beam deflects 2° for every 1° the mirror rotates. An optical angle of ±20° therefore means the mirror actually rotates only ±10°
- This angle determines the marking work area once an F-Theta lens is added — the longer the lens focal length, the larger the work area covered by the same deflection angle
- But a longer focal length also means a larger focused spot (calculated in detail in Chapter 4)
So galvo scan angle × F-Theta focal length = your marking work area. These three quantities are bound together and must be considered together during selection — they cannot be separated.
2. Core Galvo Scanner Parameters — Reading the Spec Sheet Without Confusion
2.1 Mirror Selection: Not Every Mirror Can Survive the Laser
A galvo scanner mirror looks unremarkable, but it directly determines whether your system “survives”. Choose the wrong mirror and at best the coating ablates and service life plummets; at worst the mirror snaps off and the whole machine is written off.
Galvo scanner mirror substrates come in five types: silicon (Si), silicon carbide (SiC), fused silica, beryllium (Be), and aluminum nitride (AlN). Which one to choose depends on your top priority.
Core Parameter Comparison of the Five Substrates
| Silicon (Si) | Silicon Carbide (SiC) | Fused Silica | Beryllium (Be) | Aluminum Nitride (AlN) | |
|---|---|---|---|---|---|
| Density | 2.33 g/cm³ | Actually <2 g/cm³ (honeycomb structure, lighter than silicon) | 2.65 g/cm³ | 1.85 g/cm³ (lightest metal) | 3.26 g/cm³ |
| Strength | 70-250 MPa | 300-600 MPa | 1500-1900 MPa | 1670 MPa | 300-500 MPa (flexural) |
| Melting point | 1,414°C | 2,830°C | 1,690°C | 1,286°C | ~2,200°C (decomposes/sublimes) |
| Price | 💰 Cheap | 💰💰💰 Very expensive | 💰💰 Relatively expensive | 💰💰💰💰 Extremely expensive, even pricier than silicon carbide | 💰💰💰 Very expensive |
| Thermal expansion coefficient | 2.6 × 10⁻⁶ /K | 2.2 × 10⁻⁶ /K | 0.55 × 10⁻⁶ /K (lowest) | 11.3 × 10⁻⁶ /K (relatively high) | 4.5 × 10⁻⁶ /K |
| Thermal conductivity | ~150 W/(m·K) | 120-200 W/(m·K) | 1.4 W/(m·K) | ~200 W/(m·K) | 170-320 W/(m·K) (extremely high) |
| Laser damage threshold | Baseline | Relatively high | Highest | Relatively high | Relatively high |
| Toxicity | None | None | None | ⚠️ Toxic, carcinogenic | None |
💡 Why is silicon carbide’s density lower than silicon’s? Because the silicon carbide used for galvo scanner mirrors is not solid — the base layer is made honeycomb-shaped, and only the reflective surface layer is dense. The honeycomb structure cuts weight dramatically while retaining silicon carbide’s high strength and high melting point.
⚠️ Why is beryllium rarely used? Beryllium is the lightest metal, with a density of only 1.85 g/cm³, and its strength of 1670 MPa is on par with fused silica — an ideal mirror material in theory. But beryllium is toxic and carcinogenic, posing health risks during both machining and use. It is therefore generally not used in industrial galvo scanners; today only military or special projects are likely to use it.
💡 Aluminum nitride (AlN) — the rising star of the next generation? Aluminum nitride’s biggest trump card is thermal conductivity — 170-320 W/(m·K), more than 100 times that of fused silica and 1.5-2 times that of silicon carbide. This means that at high power the mirror dissipates heat extremely fast and thermal lensing is minimal. Its melting point of 2200°C is sufficient, its strength of 300-500 MPa is on par with silicon carbide, and it is non-toxic. But aluminum nitride currently has two shortcomings: ① a density of 3.26 g/cm³, the heaviest of all substrates, giving mirror blanks high inertia so speed cannot go up; ② high price and process maturity inferior to silicon carbide and fused silica. Aluminum nitride remains at the “rising star” stage in the galvo mirror field — not yet in large-scale commercial use, but worth watching.
The Four Things Everyone Wants from a Galvo Mirror, Ranked by Priority
- High speed — the lighter the mirror, the faster the motor can rotate it
- Power handling — survives high power and high peak power without burning
- Reasonable price — save wherever possible
- No snapping — the mirror must not fly off at high rotational speed
Choosing a Substrate by Application
Ordinary laser marking machines (low-to-medium power marking) → choose silicon (Si)
Most marking machines are not high powered, and everyone pursues speed. Silicon’s low density (light mirror, fast rotation) and low price make it the king of value for marking machines. For pulsed fiber marking machines up to 100W, silicon mirrors are entirely sufficient.
Pulsed marking above 100W / continuous galvo welding above 1000W → choose fused silica
Once power goes up, silicon cannot cope. Fused silica has the lowest thermal expansion coefficient, the least thermal lensing, and the highest laser damage threshold. Under prolonged high-power continuous irradiation, fused silica’s deformation is only about 1/5 that of silicon. Silicon carbide can also cope, but it is too expensive — customers in this power range generally will not pay that price.
Solar wafer scribing → choose silicon carbide (SiC)
Solar scribing customers demand stability and durability plus speed, and care relatively little about price. Silicon carbide is a perfect match: the honeycomb structure makes the mirror lighter than silicon (no speed penalty), the 2830°C melting point (durability that crushes the competition), and high strength (fewer failures). It is expensive, but one day of downtime on a scribing line costs far more than the mirror price difference.
Large-spot galvo welding → choose fused silica
Large-spot galvo scanners have big, inherently heavy mirrors with high rotational inertia. Silicon carbide is light thanks to its honeycomb structure, but its strength is only 300-600 MPa, so large mirrors risk snapping at high rotational speed. Fused silica’s strength reaches 1500-1900 MPa — 3-5 times that of silicon carbide — so only fused silica is safe for large mirrors. Many people do not know this: silicon carbide is not a panacea — the larger the nominal spot, the larger the mirror, and the higher the snapping risk.
Quick Reference in One Table
| Your scenario | Recommended substrate | Reason |
|---|---|---|
| Pulsed marking ≤100W | Silicon | Cheap, light, good enough |
| Pulsed above 100W / continuous above 1000W | Fused silica | Handles power, low thermal deformation |
| Solar scribing | Silicon carbide | Light + heat resistant + price is no object |
| Large-spot galvo welding | Fused silica | Highest strength; large apertures will not snap |
| Military / special projects | Beryllium | Lightest + high strength, but toxic and carcinogenic; not recommended for industry |
| High-power thermal bottleneck (future direction) | Aluminum nitride | Extremely high thermal conductivity, but heavy, expensive, and not yet in large-scale commercial use |
⚠️ Get This Straight First: “10 mm Spot Size” ≠ the Mirror Can Only Accept a 10 mm Beam
Many people assume a galvo scanner with a nominal 10 mm spot size can only catch a beam 10 mm in diameter. In reality it can handle far more. Look at the actual mirror dimensions (YAG/1064nm as an example):
| Nominal spot size | X-axis mirror dimensions | Y-axis mirror dimensions |
|---|---|---|
| 5mm | 10×8mm | 14.5×8.5mm |
| 7mm | Φ12×10mm | Φ16×11mm |
| 10mm | 18.3×13mm | 23.2×16mm |
| 12mm | Φ21×16.8mm | Φ30×19mm |
| 16mm | Φ28×23mm | Φ39×23mm |
| 20mm | Φ37×24mm | Φ49×31mm |
| 25mm | Φ45.2×30.1mm | Φ56.5×38.9mm |
The Y-axis mirror for a 10 mm spot size galvo is as wide as 16mm and can actually catch beams up to 12-13mm — galvo mirrors carry a genuine margin. The manufacturer nevertheless rates it at 10mm because the beam must remain unclipped across the full deflection range — as the mirror rotates, the spot moves across the mirror face, and at extreme angles the usable reflective area shrinks. Hence the conservative rating.
What does this mean? If your actual beam is 11mm, a galvo scanner rated at 10 mm spot size is perfectly fine at rest and still mostly usable at large deflection angles — because the mirror itself carries margin. So the matching rule is simple: beam diameter ≤ galvo aperture is all you need; there is no need to deliberately reserve an 80% margin.
Why Expand the Beam and Then Use a Large-Aperture Galvo Scanner?
Suppose a 100W fiber laser outputs a beam 6.5mm in diameter feeding directly into a 10 mm spot size galvo scanner. The beam occupies only 65% of the nominal aperture, which looks like plenty of margin. But the power density on the mirror face is:
$$P_{density} = \frac{P}{\pi r^2} = \frac{100W}{\pi \times 3.25^2} \approx 3.0 \text{ W/mm}^2$$
Add a 2X beam expander and the beam diameter becomes 13mm, requiring a 14 mm spot size galvo scanner, at which point the power density drops to:
$$P_{density} = \frac{100W}{\pi \times 6.5^2} \approx 0.75 \text{ W/mm}^2$$
The power density drops 4-fold! The safety margin of both the mirror and its coating improves dramatically.
The effect is even more pronounced for pulsed lasers: peak power density likewise drops 4-fold, which can easily be the difference between a coating that lasts one month and one that lasts three years.
Large-Aperture Galvo Scanners: Benefits and Costs
| Benefit | Cost | |
|---|---|---|
| Power tolerance | Lower power density on the mirror face; lower damage risk | — |
| Spot quality | With beam expansion, the focused spot shrinks; finer characters print more clearly | — |
| Speed | — | Larger, heavier mirrors mean higher rotational inertia and lower speed |
| Cost | — | More expensive mirrors, more expensive galvo scanners, more expensive complete systems |
| Size | — | Large galvo scanners demand more installation space |
Within the same series, a 10 mm spot size galvo always marks faster than a 14 mm one, and a 14 mm one faster than a 20 mm one. Each step up in aperture costs roughly 20%-30% in speed. So the principle for choosing aperture is: just enough is enough — do not oversize. Only step up to a large aperture when necessary (high power, or beam expansion needed to shrink the spot).
2.2 Speed Parameters: The Most Misunderstood Set of Numbers
Open a galvo scanner spec sheet and you will see a pile of speed parameters: marking speed, positioning speed, step response, tracking error, writing speed… Each governs a different thing, and mixing them up leads to the wrong choice.
Angular Velocity vs Linear Velocity — Why Does Marking Speed Change with F-Theta Focal Length?
This is a key point many people do not know: the galvo motor outputs angular velocity (how fast the mirror rotates), but the spec sheet states linear velocity (how fast the spot travels across the workpiece). The relationship between the two is:
$$v_{linear} = \omega_{angular} \times f_{focal}$$
The marking speed stated by the manufacturer is, by default, measured with an F160 F-Theta lens. For example, the SCANLAB basiCube 10 is rated at a marking speed of 2.5 m/s — that is the value at F160.
Change the focal length and the linear velocity changes with it:
| F-Theta focal length | Linear velocity (same galvo scanner) | Multiple relative to F160 |
|---|---|---|
| F160 | 2.5 m/s | 1.0× |
| F254 | 4.0 m/s | 1.6× |
| F330 | 5.2 m/s | 2.1× |
| F420 | 6.6 m/s | 2.6× |
Double the focal length and the linear velocity doubles too — the galvo scanner rotates just as fast, but with a longer F-Theta focal length, the same angular sweep covers a longer distance.
Why does this matter? Because you cannot judge by the spec sheet number alone. The “2.5 m/s” you see is at F160; if you use an F330 lens, the actual marking speed is 5.2 m/s — more than double the nominal figure. Conversely, if you pick a galvo scanner rated at 5 m/s assuming that is the F330 value when it is actually the F160 value, then at F330 it becomes 10 m/s — which is good news, but you will have paid extra for performance you never needed.
Remember: whenever you look at a speed parameter, always ask “at what focal length was this measured?”
Five Speed Parameters — What Does Each One Govern?
① Marking Speed — “how fast can it draw lines”
The speed at which the mirror carries the spot along the marking path. Unit: m/s. It affects the efficiency of marking lines and contours.
- Where is it critical? Large-area pattern marking, large-format graphics filling, on-the-fly marking on production lines
- Where does it matter little? Marking a small QR code or a string of date codes — short paths, speed is not the bottleneck
② Positioning Speed (Jump Speed) — “how fast can it move without drawing”
The speed at which the spot jumps from one marking point to the next without drawing anything in between (laser off). Unit: m/s. It affects the efficiency of switching between marking points.
- Where is it critical? Distributed marking — multiple widely separated marking positions on one workpiece with lots of empty travel in between; multi-part layouts, where the spot finishes one part and jumps to the next
- Where does it matter little? Continuous line patterns — almost no empty travel, one line drawn from start to finish
③ Step Response Time — “when told to stop, how quickly does it settle”
Give the galvo scanner a step command (say, “deflect 5° immediately”) and measure the time it takes the mirror to go from starting motion to settling within ±1% of the target position. Unit: ms. It reflects the galvo scanner’s acceleration and braking capability.
- Where is fast required? Micro-hole array machining — every hole is a point; after drilling one, the spot must jump to the next instantly and settle immediately, or the hole position drifts; precision marking — for small text and small graphics, stroke starts and ends must be clean
- Where does it matter little? Large-format line cutting, continuous curved welding — smooth motion, rarely any sudden stops
④ Tracking Error — “the time lag between command and motion”
Tracking error is not a position error; it is a time lag: how much later the galvo scanner actually reaches a given position than the moment the control command was issued. Unit: ms. Think of it as the galvo scanner’s “reaction sluggishness”.
For example: the control card says “the spot should reach point A in 0.1ms”, but the galvo scanner actually arrives at point A after 0.24ms — the tracking error is 0.14ms. The spot does eventually reach point A; it is just a little bit late.
What impact does this have?
In static marking, virtually none — you ask it to draw a circle and it draws the circle 0.14ms late, but the circle still comes out right, because the delay is identical on the X and Y axes.
In on-the-fly marking, however, it causes real trouble: the workpiece moves at constant speed on a conveyor, and the galvo scanner must compensate for that displacement. A 0.14ms tracking error means the compensation falls short by the displacement corresponding to 0.14ms, and the result is a systematic offset of the entire marking position. The faster the line, the bigger the offset.
- Where must it be watched? On-the-fly marking, high-speed production lines, precision contour tracking
- Where can you ignore it? Static marking, large-format low-speed processing
⑤ Writing Speed (cps) — “how fast can it type”
cps = characters per second — how many characters can be marked per second. This is a composite metric given by the manufacturer; it is not determined by any single parameter but is the combined result of marking speed + positioning speed + acceleration/deceleration + laser on/off delay.
Marking one character goes like this: jump to the start position → draw the stroke → jump to the next stroke → draw the stroke → … → finish all strokes → jump to the next character. Every step involves acceleration, constant velocity, deceleration, and settling, plus waiting for the laser’s on/off response. So cps reflects the galvo scanner’s overall efficiency in a real typing scenario and tracks actual throughput better than marking speed alone.
- Where is it critical? Production-line date codes, batch serial numbers, high-speed QR code arrays, multi-character on-the-fly marking — all “large volumes of short text” scenarios; 10 extra characters per second means hundreds more products per hour
- Where does it matter little? Large patterns, long lines — few characters, cps is not the bottleneck
cps also carries a hidden meaning: it determines whether you can run in “high quality mode” or “good quality mode”. SCANLAB spec sheets usually list two cps values — for example, the basiCube 10 is rated at 570 cps (high quality) and 800 cps (good quality). High-quality mode moves the mirrors more smoothly but slower; good-quality mode is faster but strokes may show slight jitter. If your application demands extremely crisp characters (e.g., micro medical device coding), you may be limited to 570 cps; if you are only marking production dates, 800 cps is entirely sufficient.
Speed Parameter Quick Reference
| Parameter | What it describes | Unit | Demanding scenarios |
|---|---|---|---|
| Marking speed | How fast it draws | m/s | Large-format patterns, on-the-fly marking |
| Positioning speed | How fast it jumps | m/s | Distributed marking points, multi-part layouts |
| Step response | How fast it stops | ms | Micro-hole arrays, precision small text |
| Tracking error | Reaction lag | ms | On-the-fly marking, high-speed contours |
| Writing speed | How fast it types | cps | Batch coding, high-speed QR codes |
2.3 Accuracy Parameters: Don’t Be Intimidated by the Numbers — Understand What They Describe First
Accuracy parameters are the most headache-inducing part of a spec sheet — μrad, bit, mrad/deg, ppm/K… all quantities too small to see. But they determine whether your marking is actually accurate.
Repeatability — Why an Angle? And Why “Repeat”?
The spec sheet states μrad (microradians); 1 μrad = 0.000057°, an angle utterly imperceptible to the naked eye. Why not millimeters? Because a galvo scanner’s essence is rotating a mirror — what it outputs is an angle. How many millimeters that angle corresponds to on the workpiece depends on the F-Theta focal length:
| F-Theta focal length | Linear displacement caused by a 10 μrad angular error |
|---|---|
| F160 | 1.6 μm |
| F254 | 2.5 μm |
| F330 | 3.3 μm |
| F420 | 4.2 μm |
The same 10 μrad galvo scanner paired with an F160 lens yields an error of only 1.6μm, but 4.2μm with an F420. That is why manufacturers can only quote an angle and let you convert it according to your focal length. Linear displacement = angular error × focal length. It is that simple.
So why is it called “repeat” positioning accuracy (repeatability) rather than positioning accuracy (accuracy)? These are two entirely different concepts:
- Repeatability: send the galvo scanner to the same target position 100 times and measure how much the 100 actual positions differ from one another. It is like walking blindfolded to a point: each attempt takes a different route, but you stop in roughly the same spot every time — that means your “repeatability” is good.
- Accuracy: send the galvo scanner to a target position and measure how far the actual arrival point is from the target. It is like walking blindfolded toward a mark 10 meters away and stopping at 9.9 meters every time — repeatability is excellent (9.9 meters every time), but accuracy carries a 0.1-meter bias.
The key difference: a galvo scanner with good repeatability is not necessarily accurate — it may deviate in the same direction by the same amount every time, but very consistently. That stable bias can be compensated through software correction. A galvo scanner with poor repeatability deviates differently in direction and magnitude each time, and compensation cannot help because you never know how much to correct.
That is why the galvo scanner industry quotes only repeatability — it is the core metric. Systematic accuracy bias can be corrected, but the random scatter of poor repeatability cannot.
Resolution — the “Pixels” of a Galvo Scanner
Resolution is the smallest angular step the galvo scanner can distinguish. Spec sheets usually quote it in bits (16-bit, 18-bit, 20-bit), meaning the control signal divides the entire scan range into that many steps:
| Resolution | Steps | Angle per step (±0.5rad scan range) | Linear displacement per step at F160 |
|---|---|---|---|
| 16-bit | 65,536 | 15 μrad | 2.4 μm |
| 18-bit | 262,144 | 3.8 μrad | 0.6 μm |
| 20-bit | 1,048,576 | 0.95 μrad | 0.15 μm |
The higher the resolution, the smoother the curves the galvo scanner can draw and the finer its positioning. It is like a phone screen — at the same screen size, 4K packs in denser pixels and renders a finer picture than 1080p.
But resolution ≠ accuracy! A 20-bit resolution means that in theory 0.15μm steps can be distinguished, but if repeatability is 10 μrad (1.6μm), a 0.15μm step has no practical meaning — the galvo scanner simply cannot position that precisely. Resolution must exceed accuracy to be meaningful, but exceeding it by too much is waste. In practice 16-bit is enough for the vast majority of marking applications; 18-bit and above are mainly for precision micromachining.
Linearity — How “Skewed” the Command-to-Motion Relationship Is
Ideally, the input signal you send to the galvo scanner and the mirror’s deflection angle should be perfectly linear: double the input, double the angle. Reality always deviates somewhat — like a spring scale that reads 1kg with a 1kg load, 1.98kg with 2kg, and 3.05kg with 3kg. Not perfectly linear.
Linearity quantifies that deviation: the maximum deviation between the actual deflection angle and the ideal linear value, expressed in mrad/deg or percent.
Practical impact: a galvo scanner with poor linearity may draw lines with a slight bow, circles that become ellipses, squares that become trapezoids. Fortunately, the linearity error of modern galvo scanners can be compensated by the control card’s correction table — deviations are measured point by point at the factory, written into a lookup table, and corrected automatically at runtime. So the linearity figure itself is not the most critical thing; what matters is whether the control card supports correction and how dense the correction points are.
Thermal Drift — the Invisible Killer of Long Operating Hours
After a galvo scanner powers up, its zero position slowly shifts as operating time accumulates — this is thermal drift. Its causes are not singular but a combination of mechanical and electronic effects:
Mechanically: the motor heats up → the mirror mount expands → the mirror’s physical zero position shifts. Easy to understand — thermal expansion and contraction.
Electronically: components on the servo driver board — resistors, capacitors, op-amps — change parameters with temperature. For example:
– Resistor temperature coefficients: ordinary metal-film resistors run roughly 50-100 ppm/°C; even precision resistors are 1-10 ppm/°C
– Capacitance varies with temperature: ceramic capacitors can deviate ±15% or more across temperature ranges
– Op-amp offset voltage and bias current drift with temperature
– The DAC (digital-to-analog converter) reference voltage varies with temperature
These component parameter drifts cause errors in the servo driver board’s output — you think you are sending the galvo scanner the same voltage, but because the resistance and the reference voltage have changed, the actual output is no longer the same. In many galvo scanners, electronic components contribute more to thermal drift than mechanical expansion does.
This is why high-end galvo scanners (such as the intelliSCAN series) adopt:
– Precision resistors and capacitors with low temperature coefficients
– High-precision temperature-compensated voltage references
– Digital encoders in place of analog position sensors (digital signals are immune to analog drift)
– On-board temperature sensors + software compensation algorithms
Spec sheets usually quote two values:
– Angle drift: μrad/K — how much the zero position shifts per 1 degree of temperature change
– Gain drift: ppm/K — how many parts per million the gain (sensitivity) changes per 1 degree of temperature change
The frightening thing about thermal drift is that it accumulates, and always in one direction. Repeatability errors are random and cancel out; thermal drift is systematic and keeps shifting the same way, growing worse over time.
Different applications have vastly different thermal drift requirements:
| Scenario | Thermal drift requirement | Reason |
|---|---|---|
| Static marking | ★☆☆ Relaxed | Marking one workpiece takes seconds; the job is done before the galvo scanner even warms up, so drift has no time to accumulate |
| Laser cleaning | ★☆☆ Relaxed | Cleaning itself has low positional accuracy demands; a shift of a few tenths of a millimeter is invisible |
| On-the-fly marking | ★★★ Strict | Production lines run for hours without stopping; drift makes every mark shift in the same direction — one inspection turns up a whole batch of rejects |
| 3D printing | ★★★ Extremely strict | One part takes thousands of layers over a dozen-plus hours; a little drift per layer accumulates until the part is simply scrapped |
Four countermeasures against thermal drift: ① warm up for 5-10 minutes before processing ② periodically return to the reference point for recalibration during operation ③ choose low-drift models (invar shafts and other low-expansion materials cut drift 5-10 fold) ④ a temperature-controlled, air-conditioned workshop beats one with large temperature swings by far
2.4 Electrical and Interface: How the Control Card “Talks” to the Galvo Scanner
A galvo scanner does not start rotating the moment it is plugged in — it needs the control card to tell it what angle to rotate to. The “language” between the control card and the galvo scanner is the control protocol. Pick the wrong protocol and the control card and the galvo scanner simply cannot communicate.
Three Major Categories of Control Protocols
| Type | Signal form | Typical interface | Characteristics |
|---|---|---|---|
| Analog protocol | ±10V / ±5V analog voltage | Wiring terminals | Simplest, but susceptible to interference |
| Digital protocol | Differential digital signal | DB25 / D-SUB | Mainstream; strong interference immunity, high accuracy |
| Proprietary protocol | Vendor-defined | Varies | High performance ceiling, poor compatibility |
① Analog protocol — what “analog galvo scanners” use
The driver board of an analog galvo scanner directly receives a ±10V (or ±5V) analog voltage signal, where the voltage corresponds to the target deflection angle: +10V = full positive deflection, -10V = full negative deflection, 0V = zero position. The DAC (digital-to-analog converter) in the control card converts digital commands into analog voltage for the driver board.
Advantages: simple, cheap, mature.
Disadvantages:
– Sensitive to electromagnetic interference — the moment variable-frequency drives and servo drives in the workshop switch on, noise piles onto the analog signal and the marking position jitters
– Short transmission distance — signal attenuation becomes obvious once the cable exceeds 3 meters
– Limited accuracy — constrained by DAC resolution and noise; actual accuracy falls far short of digital protocols
– No feedback — the control card has no idea where the galvo scanner actually rotated to, making in-process monitoring impossible
Applicability: analog galvo scanners are now essentially obsolete; they may still exist on laser equipment seven or more years old, but new equipment basically no longer uses them. Understanding the principle nonetheless helps explain why digital protocols replaced them.
② Digital protocol — the mainstream industrial choice
Digital protocols transmit position commands over differential signals, with far stronger interference immunity than analog. Three are currently mainstream:
XY2-100 — the industry standard, supported by nearly every brand
Originally developed by Cambridge Technology and popularized by SCANLAB into the de facto standard. 16-bit resolution (65536 steps), DB25 interface, differential signal transmission. It is an open protocol with the best compatibility:
| Brand | XY2-100 support |
|---|---|
| SCANLAB | ✅ Native support |
| Raylase | ✅ Supported (as XY2-100-Enhanced) |
| Han’s Laser galvo scanners | ✅ Supported |
| Cambridge Technology | ✅ Supported |
| Most Chinese domestic brands | ✅ Supported |
Choosing XY2-100 means your control card can drive the vast majority of galvo scanners on the market — swap the galvo scanner without swapping the control card.
XY2-100-E (Enhanced) is the enhanced version of XY2-100: the position signal is extended to 18-bit (262144 steps), additional feedback data channels are added, and status monitoring of digital galvo scanners is supported.
SL2-100 — developed by Nutfield, mainly used by Raylase
Developed by Nutfield Technology of the USA, with 20-bit resolution (about 1 million steps) — 4 times the accuracy of XY2-100. But it is a semi-closed protocol, currently used mainly on Nutfield’s own galvo scanners and Raylase’s SUPERSCAN series. SL2-100 outperforms XY2-100 in high-speed response and interference immunity, but compatibility is poor — your control card must support SL2-100 to use it.
HANS-100 — released by Han’s Laser in 2022
Han’s Laser judged that XY2-100 was not reliable enough for “lights-out factory” scenarios (its data verification mechanism is not strong enough), while SL2-100 was licensed only to Raylase — so it developed the HANS-100 protocol itself. The protocol has been published and is licensed free of charge to all laser companies, on the condition that the “HANS-100” name is retained in use. Currently Han’s Laser’s own galvo scanners support it; adoption by other brands is limited.
③ Proprietary protocols
Some vendors also have their own proprietary interfaces, such as Cambridge Technology’s ScanPack algorithm (24-bit, running over its own ScanMaster controller via Ethernet), and HSSI and RTFE-D15D supported by Polaris control cards. These protocols have high performance ceilings, but they lock you into that vendor’s ecosystem — you can change neither the control card nor the galvo scanner.
Power Supply: Not Just ±15V
| Brand/model | Supply voltage | Notes |
|---|---|---|
| Han’s Laser galvo scanners (most models) | ±15VDC, ≥3A | Most common |
| SCANLAB basiCube | ±15VDC | Standard supply |
| Cambridge Technology DC3000 | ±15V ~ ±28VDC | Wide selectable range |
| Raylase SUPERSCAN IIE | ±15V ~ ±18VDC | Small-format model |
| Raylase SUPERSCAN IV/V | +30V or +48VDC | Large-format, high-power models |
Raylase’s 48V supply deserves a separate mention. The SUPERSCAN IV processes at 30 rad/s on a 30V supply; switch to 48V and it jumps straight to 50 rad/s — a 67% speed increase! At the same drive current, higher voltage means more power, so the motor can deliver more torque, with both acceleration and speed rising. The cost is that you need an additional 48V power supply, and power consumption is higher.
Simple rule of thumb: ±15V is the standard; large-aperture / high-power / high-speed models may need higher voltages. Always check the supply requirements during selection — do not discover a power mismatch only after the purchase arrives.
Analog Servo vs Digital Servo — the Core Difference Between Driver Boards
First, a point of common confusion: “analog servo” ≠ “analog galvo scanner”.
- Analog galvo scanner (obsolete): an old-style galvo scanner taking ±10V analog signals directly; found only on equipment seven or more years old and basically no longer used in new machines
- Analog servo board: communicates over digital interfaces such as XY2-100, but its internal servo control uses an analog PID circuit — the basiCube 10 is exactly this type
- Digital servo board: has a built-in DSP processor with fully digital servo algorithms — SCANcube IV, intelliSCAN, and the like
So although the basiCube 10’s servo board is “analog”, it uses the XY2-100 digital interface — it is not an old-style ±10V analog galvo scanner. Do not confuse the two.
| Analog servo board | Digital servo board | |
|---|---|---|
| Communication interface | XY2-100 digital interface | XY2-100/SL2-100 digital interfaces |
| Control algorithm | Fixed PID parameters | Model-adaptive, real-time optimization |
| Tuning modes | One (hard-wired) | Multiple switchable (vector/step/micromachining) |
| Process monitoring | No feedback | Real-time position/temperature/current feedback |
| Remote diagnostics | Not supported | Remote status readout supported |
| Thermal management | None | Digital PWM output stage; low power draw and heat |
| Price | Cheap | 1.5-3 times more expensive |
| Typical models | basiCube 10 | SCANcube IV, intelliSCAN |
The digital servo board’s biggest advantage is not raw accuracy but switchable tuning modes: vector tuning for marking (fast), step tuning for micromachining (stable), switched at the press of a button without re-tuning parameters. An analog servo board can only compromise on a single tuning — either fast or accurate, never both.
Selection advice: if budget allows, go straight for the digital servo board. If your application is a single scenario (say, only marking or only welding), the analog servo board offers better value; if you need to cover multiple processing modes, the digital servo board is far less hassle.
Position Sensors: the “Eyes” of Galvo Scanner Accuracy
The servo board determines “how it is controlled”; the position sensor determines “how precisely it sees”. The accuracy ceiling of a galvo scanner’s closed-loop control depends on how small an angular change the position sensor can resolve. Three mainstream technologies exist today:
| Photoelectric sensor | Capacitive sensor | Digital encoder | |
|---|---|---|---|
| Principle | LED source + shutter vane + silicon photovoltaic cells; the vane rotates with the rotor, the illuminated cell area changes → differential current output | Differential capacitance; rotor rotation changes plate spacing/area, and the capacitance change reflects the angle | Grating disk (de) or interferometric (se); optical diffraction/interference measures the digital angle directly |
| Output | Analog signal | Analog signal | Digital signal |
| Repeatability | <2-5 μrad | <1-2 μrad | <0.5-1 μrad |
| Thermal drift | Relatively pronounced | Relatively small | Minimal |
| EMI immunity | Average | Fairly good | Strongest |
| Cost | Low | Medium | High |
| Typical examples | SCANLAB dynAXIS (basiCube/SCANcube/intelliSCAN III) | Thorlabs galvo scanners, Century Sunny (世纪桑尼) SS9310D | SCANLAB dynAXISde/se (intelliSCANde/se/IV), Han’s Laser grating-encoder galvo scanners |
How to choose?
- Ordinary marking: a photoelectric sensor is enough — cheap and fast enough
- On-the-fly marking / industrial production lines: photoelectric or capacitive both work; the keys are the servo board and tracking error
- Precision micromachining / 3D printing / long-duration stable operation: a digital encoder is a must — highest accuracy, lowest thermal drift, strongest interference immunity
A note on the SS9310D’s “moving-magnet capacitive sensor”: capacitive sensors offer better accuracy and stability than traditional photoelectric sensors, but their output is still analog — it must subsequently pass through an ADC to become digital, and that step also introduces some noise and thermal drift. “Truly digital” means the position detection is digital from the source (grating encoder), with no analog→digital conversion stage.
3. The Five-Step Selection Method
3.1 Step One: Determine the Aperture — the Laser’s Power Decides
Aperture is the first step in selection and also the easiest to get wrong. Too small, and the beam is clipped or the mirror burned out; too large, and you waste speed and spend money for nothing. The core principle: choose the aperture according to the laser type and its power.
Pulsed Fiber Lasers — Power Tiers, Simple and Direct
| Laser power | Recommended aperture | Typical applications | Notes |
|---|---|---|---|
| ≤150W | 10 mm spot size | Marking, engraving, light cleaning | Fiber lasers output a 6-7mm beam, which fits a 10 mm spot size with room to spare — no beam expander needed |
| 200-440W | 14 or 15 mm spot size | Deep engraving, cleaning, welding | At this power you must expand the beam to cut power density; with a 2X expander the beam is 13mm, so you must move to a 14/15 mm spot size |
| 500W | 20 mm spot size | Cleaning, welding, cladding | High power plus beam expansion makes the beam diameter larger; start at a 20 mm spot size |
One exception: for pulsed lasers within 150W, if you prioritize precision over speed (say, precision micromachining), you can also choose a 14 mm spot size with a 2X beam expander — the spot shrinks by half and small characters are crisper, at the cost of about 30% lower speed.
CW Fiber Lasers — You Must Calculate the Beam Diameter
CW lasers cannot be handled with simple power tiers the way pulsed lasers can, because the beam diameter of a CW laser after the collimator depends on the collimator’s focal length and the fiber’s NA:
$$\Phi_{beam} = 2 \times f_{collim} \times NA$$
For example, with fiber NA=0.06 and a 75mm collimator focal length, the beam diameter = 2 × 75 × 0.06 = 9mm. Swap in a 100mm collimator and the beam diameter = 12mm. Change the collimator and the beam diameter changes, so the aperture choice must follow.
| Laser power | Recommended aperture | Precondition |
|---|---|---|
| ≤1500W | 20 mm spot size | Beam diameter after collimation ≤20mm (must not exceed the aperture) |
| ≤1500W but beam >16mm | 30 mm spot size | Long collimator focal length or high NA pushes the beam beyond the 20 mm aperture |
| 2000-6000W | 30 mm spot size | High power demands a large aperture for heat dissipation |
After calculating the beam diameter, you must also check the power. The beam may fit into a 20 mm spot size, but at 2000W the mirror power density is too high, so you still have to go to a 30 mm spot size. Both conditions must hold simultaneously: the beam is not clipped + the power density is within the safe range.
⚠️ Special Attention: Back-Reflection Spot — the Hidden Killer That Keeps Destroying Galvo Mirrors
In laser marking machines or galvo welding machines, one situation is very common: the galvo mirrors keep getting burned out — you replace one and it fails again — while the F-Theta lens looks perfectly intact. The customer’s first reaction is “your galvo mirror coating is no good”. What makes it even more awkward: swapping in another brand’s galvo scanner may actually stop the failures — leaving the galvo manufacturer holding the bag.
But the real cause is often not coating quality — it is that the F-Theta lens’s back-reflection spot lands right on the galvo mirror.
What is the back-reflection spot?
Look at the last surface of the F-Theta lens (the one facing the galvo scanner): it is a concave mirror. In theory 100% of the laser passes through the F-Theta lens, but actual transmission is between 98% and 99.5%, meaning 0.5%-2% of the laser is reflected back by that concave surface.
Strictly speaking, the F-Theta lens has more than one reflective surface — a multi-element F-Theta lens has multiple lens elements and multiple air-glass interfaces, each with Fresnel reflection (reflection caused by abrupt refractive-index changes). But most surfaces are convex or flat, so the reflected light diverges and does no harm. Only the final concave surface acts like a satellite dish, focusing the reflected light onto a single point — that focus is the back-reflection spot. When this spot happens to land on the galvo scanner’s M2 mirror, the laser is effectively focused onto the galvo mirror, and of course the coating burns out easily.
Why might another brand’s galvo scanner stop the failures?
Because the distance from the M2 mirror to the F-Theta lens varies slightly between manufacturers. Swap in a different galvo scanner and the M2 position changes, so the back-reflection spot no longer lands on the mirror — and naturally it stops failing. But that is dodging the problem by luck, not solving it at its root.
Solution: add a spacer ring and push the F-Theta lens out by 3mm
The simplest fix is to raise the F-Theta adapter ring by about 3mm, or insert a washer so the lens cannot be screwed in all the way. The F-Theta lens then sits farther from the M2 mirror, the focus of the reflected light no longer falls on M2, and the problem is solved.
The right practice: check the PDF parameters as soon as you receive the F-Theta lens
Every F-Theta lens set should ship with a PDF document containing the back-reflection spot position parameter. With that number, you can precisely calculate the adapter ring height during design and ensure the back-reflection spot never lands on a galvo mirror.
Why was this rare before but increasingly common now?
Optical fiber marking machines used to be 20W and 30W; the light reflected back by the F-Theta lens’s last surface was only about 0.2-0.3W, and even if it happened to focus on a galvo mirror, that little power could not damage the coating. But now:
– 2000-6000W CW laser welding — a 1% reflection is 20-60W! Focused onto a galvo mirror, the coating burns through in seconds
– 200-300W pulsed laser cleaning — even higher peak power
– Picosecond/femtosecond lasers — extremely high peak power density
The higher the power, the deadlier this problem. When galvo mirrors keep failing, do not blame the coating first — check the back-reflection spot first!
3.2 Step Two: Determine the Wavelength — Mirror Material and Coating Decide
Wavelength determines what material and coating the galvo mirrors need. Get it wrong and transmission collapses off a cliff, and the mirrors burn through quickly.
Mirror Configurations for Mainstream Wavelengths
| Wavelength | Laser type | Mirror substrate | Coating type | Notes |
|---|---|---|---|---|
| 1064nm | Fiber/solid-state pulsed | Silicon or fused silica | Dielectric | Most mainstream; silicon for low power, fused silica for high power |
| 1064nm | Fiber/solid-state CW (high power) | Fused silica | Dielectric | CW high power requires fused silica |
| 10600nm (10.6μm) | CO₂ laser | Metal substrate (copper/aluminum) | Gold or enhanced metal coating | CO₂ cannot use ordinary glass substrates (opaque at 10.6μm) |
| 532nm | Green solid-state | Silicon or fused silica | Dielectric | Green light is demanding on coatings; metal coatings lose too much |
| 355nm | UV solid-state | Fused silica | UV dielectric | UV requires fused silica (BK7 fluoresces) |
| 1064+532nm | Dual wavelength | Fused silica | Dual-band dielectric | One mirror set compatible with two wavelengths |
| 1064+532+355nm | Triple wavelength | Fused silica | Triple-band dielectric | Rare and expensive |
💡 The F-Theta lens material is also strongly tied to wavelength — do not fixate on the galvo scanner and forget the F-Theta lens:
– 1064nm marking/welding: F-Theta lenses typically use Corning 7980 or 7979 fused silica (UV-grade fused quartz), high transmission and low thermal expansion
– 355nm UV: F-Theta lenses must use UV-grade fused silica such as Corning 7980 or 8655; ordinary optical glass sees transmission fall off a cliff in the UV
– 10600nm (CO₂): F-Theta lenses must be zinc selenide (ZnSe) or gallium arsenide (GaAs); ordinary glass is completely opaque at 10.6μmConsider wavelength for the galvo mirrors and the F-Theta lens together — do not select the galvo scanner correctly and then miss on the F-Theta lens.
Power × Wavelength = the Combination Logic of Substrate + Coating
It is not just wavelength that decides the coating — power and peak power also determine what substrate and coating you should use:
- Low-power CW (<50W) + 1064nm: silicon + dielectric coating — sufficient and cheap
- High-power CW (>50W) + 1064nm: fused silica + dielectric coating, to resist the thermal lens effect
- High-peak-power pulsed + 1064nm: fused silica + high-damage-threshold dielectric coating, to prevent coating burn-through
- CO₂ + high power: copper substrate + gold coating; gold reflectivity at 10.6μm is >99.5%
What About Special Wavelengths?
Some lasers’ wavelengths fall outside the mainstream range — 808nm, 915nm, 1550nm, and so on. Dedicated dielectric-coated galvo mirrors for these wavelengths are very hard to buy — volumes are small and coating vendors are unwilling to open a coating batch.
Practical alternative: use metal-coated mirrors.
Metal coatings (aluminum, silver, gold) are broadband: they offer decent reflectivity across a very wide wavelength range, unlike dielectric coatings that only hit high reflectivity at specific wavelengths. The cost is reflectivity 1-2 percentage points lower than dedicated dielectric coatings (say, 98% vs 99.5%), and metal coatings also oxidize and degrade faster over long service. But for niche-wavelength applications, this is the fastest solution — far better than finding no mirror at all.
| Metal coating type | High-reflectivity band | Reflectivity | Characteristics |
|---|---|---|---|
| Aluminum + protective layer | 400nm – 10μm | >95% | Most versatile; covers nearly all common wavelengths |
| Silver + protective layer | 500nm – 2μm | >97% | Highest reflectivity in visible to near-IR |
| Gold | 700nm – 10μm | >98% | Best from near-IR to far-IR |
Caution: metal coatings are unsuitable for high-peak-power pulsed lasers — single-pulse peak power density is too high and the metal coating punctures easily. For pulsed lasers, still try to find a dielectric coating matched to the wavelength.
3.3 Step Three: Determine the Speed — the Application Scenario Decides
Static Marking — a Standard Galvo Scanner Is Enough
Static marking is the most common application: the workpiece sits still on the table and the galvo scanner finishes one then moves to the next. Speed demands are not severe, and the vast majority of 10 mm spot size galvo scanners on the market can handle it.
| Brand | Model | Writing speed (F160) | Positioning speed (F160) | Characteristics |
|---|---|---|---|---|
| SCANLAB | basiCube 10 | 570/800 cps | 12.8 m/s | Classic entry model, great value |
| SCANLAB | SCANcube IV 10 | 700/950 cps | 20.8 m/s | New generation, faster, digital servo board |
| Raylase | MINISCAN III-10 | 800/1280 cps(@48V) | 16 m/s(@48V) | 48V supply, extremely high cps |
| Century Sunny (世纪桑尼) | SS9310D | 800/950 cps | 17 m/s | Moving-magnet capacitive sensor, purpose-built for high speed, ±15V supply |
| JCZ (金海创) | SG7210 | 559 cps | 15 m/s | Economy model, XY2-100 interface |
| Han’s Laser | ExtraScan II 10 | — | 10 m/s | Domestic mainstay |
Selection advice: with ample budget choose the SCANcube IV 10 (950 cps is a strong performer); on a tight budget choose a domestic 10 mm spot size model (JCZ 7210 / Century Sunny 9310D / Bairuitong 8910D / Han’s ExtraScan) — fully adequate for everyday marking.
On-the-Fly Marking — Speed and Tracking Error Are the Keys
In on-the-fly marking the workpiece moves at constant speed on a conveyor, and the galvo scanner must “mark while chasing”. This imposes two hard requirements on the galvo scanner:
- Small tracking error: the conveyor is moving, so if the galvo scanner is 0.1ms late the mark drifts by a distance — the faster the line, the more it drifts
- Fast marking speed: all content must be marked within the limited window
Signal path: encoder → marking card → galvo scanner (the encoder does NOT connect to the galvo scanner!)
The encoder reads the conveyor speed; the signal enters the marking card (e.g., SCANLAB RTC5/RTC6); the marking card computes the compensation and sends corrected position commands to the galvo scanner over XY2-100.
| Brand | Model | Tracking error | Writing speed (F160) | Positioning speed (F160) | Characteristics |
|---|---|---|---|---|---|
| SCANLAB | SCANcube III 10 | 0.14 ms | 600/900 cps | 15 m/s | Classic on-the-fly marking model |
| SCANLAB | SCANcube IV 10 | 0.12 ms | 700/950 cps | 20.8 m/s | New generation, faster and more accurate |
| Raylase | MINISCAN III-10 | 0.13 ms | 800/1280 cps | 16 m/s | Highest cps on 48V supply |
| Century Sunny (世纪桑尼) | SS9310D | 0.2 ms | 800/950 cps | 17 m/s | Purpose-built for high-speed/on-the-fly, ±15V supply |
| Century Sunny (世纪桑尼) | TSH8310D | 0.15 ms | 480/650 cps | 11 m/s | Economy on-the-fly marking |
| JCZ (金海创) | SG7210 | 0.144 ms | 559 cps | 15 m/s | Good value for money |
| Bairuitong (百瑞通) | YZD-8910D | 0.2 ms | 800/950 cps | 16 m/s | XY2-100 + analog interfaces, ±15V supply, domestic on-the-fly marking |
Selection advice: for slow lines (<30m/min), domestic 9310D/8910D/7210 will do; for medium-high speeds (30-80m/min), prefer the SCANcube IV 10 or MINISCAN III-10 (tracking error 0.12-0.13ms, high cps).
High-Speed Lines (>80m/min) — No Universal Solution; Needs System-Level Matching
Above 80m/min, a single galvo scanner struggles to cope alone. The core bottleneck is not line speed — lengthen the F-Theta focal length and line speed rises — but rather insufficient cps and a strained compensation range. As analyzed earlier, a basiCube 10 with an F330 F-Theta lens needs 930cps for 180 characters of 1.3mm, but it tops out at 800cps.
Viable options:
| Option | Principle | Advantages | Disadvantages |
|---|---|---|---|
| excelliSCAN | SCANLAB’s high-end series; SCANahead control eliminates tracking error | Tracking error = 0; highest cps at 1310 | Expensive; requires the RTC6 control card |
| intelliSCAN IV 10 | Digital encoder + SCANahead, 1310cps | Highest cps + zero tracking error + ultra-low drift | Most expensive; 10 mm spot size handles only 150W |
excelliSCAN is today’s speed ceiling for 2D galvo scanners: the core idea of SCANahead control is “compute the trajectory ahead of time” — the control card precomputes the entire trajectory before the galvo scanner actually moves, so the galvo scanner merely executes it strictly and tracking error is theoretically zero. This solves the most vexing offset problem in on-the-fly marking.
Fine Machining — Accuracy First, Speed Can Wait
The key to fine machining (micro-holes, micro-structures, precision etching) is not speed but precision: fast step response (settles immediately on arrival), high repeatability, low thermal drift.
Recommended: intelliSCAN III 14
- Digital servo board with up to 3 switchable tuning modes
- Analog photoelectric position sensor (dynAXIS 3), repeatability <2μrad
- Thermal drift Offset <15μrad/K, Gain <25ppm/K — half of the basiCube’s
- 8-hour drift <100μrad, twice as good as the basiCube’s <200μrad
- The 14 mm spot size aperture can pair with a 2X beam expander to shrink the spot — ideal for precision micromachining
If you need still higher accuracy (<1μrad repeatability, extremely low drift), step up to the intelliSCANde or se (digital encoder versions), but they cost considerably more.
Why a 14 mm spot size and not 10? Fine machining usually needs a small spot, which requires a beam expander. A 10 mm spot size with a 2X expander gives a 13mm beam that gets clipped, so a 14 mm spot size is mandatory. The 14 is a bit slower than the 10, but fine machining is not chasing speed anyway.
Batch QR Codes — cps Is the Lifeline
Marking QR/data matrix codes in bulk means every code is a dense little array of modules, placing extreme demands on cps. A 2mm×2mm DataMatrix code has roughly 14×14=196 modules, and outlining + filling requires numerous jumps and short line segments — exactly where cps matters.
| Scenario | Recommended galvo scanner | Rationale |
|---|---|---|
| Ordinary lines (static / slow on-the-fly) | SCANcube IV 10 (950 cps) | Adequate, good value |
| High-speed line on-the-fly coding | excelliSCAN 14 (1310 cps) | Highest cps + zero tracking error |
| Budget sensitive | MINISCAN III-10 (1280 cps@48V) | cps close to the excelliSCAN at a much lower price |
An easily overlooked point: batch QR codes are usually small (2-5mm per code) but numerous and closely spaced. That means positioning speed matters more than marking speed — the galvo scanner keeps jumping between adjacent codes, and a high jump speed is what truly raises throughput.
3.4 Step Four: Determine the Accuracy — How Precise Do You Need It?
The logic of accuracy selection runs opposite to speed selection: for speed, “good enough is fine”; for accuracy, “better to have it and not need it”. Fall short on speed and you merely work slower; fall short on accuracy and you produce scrap directly.
Three Accuracy Tiers, Simply Put
General grade: marking, coding, ordinary engraving — as long as deviations are invisible to the naked eye. An analog servo board or a domestic digital servo board is enough. Repeatability <10μrad; somewhat higher thermal drift does not matter, because each marking job takes only a few seconds — done before the galvo scanner even warms up.
Industrial grade: precision marking, on-the-fly marking, volume production — the 1000th mark must be as accurate as the first. A digital servo board is mandatory, repeatability <5μrad, thermal drift <15μrad/K. On-the-fly jobs run for hours; with high drift every mark shifts together, and one inspection turns up a batch of rejects.
Precision grade: micro-hole arrays, 3D printing, semiconductor processing — a few μm off means failure. Digital servo board + digital encoder, repeatability <2μrad, thermal drift <10μrad/K, 8-hour drift <50μrad. Expensive, but there is no alternative.
A trap many fall into: looking only at repeatability while ignoring thermal drift and 8-hour drift. Repeatability is “how much you can deviate each time you go to the same position” — random error; thermal drift is “how far the zero position shifts over time” — systematic error. Random errors cancel out positive and negative; systematic error only grows. A galvo scanner with 2μrad repeatability but 200μrad of 8-hour drift is less trustworthy after 4 hours than one with 5μrad repeatability but 50μrad of drift.
What Configuration Does Each Accuracy Tier Correspond To?
| Accuracy tier | Servo board | Protocol | Typical models |
|---|---|---|---|
| General grade | Analog servo board | XY2-100 | basiCube 10, domestic 10 mm spot size galvo scanners |
| Industrial grade | Digital servo board | XY2-100-E / SL2-100 | SCANcube IV 10, intelliSCAN III 14 |
| Precision grade | Digital servo board + digital encoder | SL2-100 (20-bit) | intelliSCAN IV 10 (encoder version), excelliSCAN |
3.5 Step Five: Determine the Configuration — Budget Decides
With aperture, wavelength, speed, and accuracy settled, the last word belongs to your wallet. Choose the most cost-effective combination that still meets the performance requirements.
Three Configuration Tiers
Entry level (galvo + card: CNY 3,000-8,000): domestic 10 mm spot size galvo scanner + domestic marking card + XY2-100. Suitable for small-batch marking and startup budgets. Usable, but do not expect long-term stability — thermal drift and 8-hour drift are considerably worse than imported units. Note: at this price point there may be old-style ±10V analog galvo scanners; for new equipment choose at minimum a galvo scanner with a digital XY2-100 interface.
Industrial level (galvo + card: CNY 20,000-50,000): SCANcube IV 10 or MINISCAN III-10 + RTC5 control card + digital servo board. Suitable for volume marking and on-the-fly marking. This is the most cost-effective tier — performance sufficient for most applications at an acceptable price.
High-end level (galvo + card: CNY 50,000-150,000+): intelliSCAN IV or excelliSCAN + RTC6 control card + SL2-100 (20-bit) + SCANahead. Suitable for precision machining, high-speed lines, 3D printing. The extra money mainly buys “flexibility” — multiple tuning modes, SCANahead zero tracking error, ultra-low drift from the digital encoder. If your application is a single scenario (say, only metal marking), these features go unused and you need not go high-end.
4. Aperture × Beam Expansion × F-Theta Lens — the Error-Prone Trio
These three constrain one another: change one and you must check the other two.
4.1 The Beam Must Not Exceed the Aperture — the Matching Rule: Beam ≤ Galvo Aperture
A larger aperture is not automatically better; the key is the ratio of beam diameter to aperture. Because galvo mirrors carry built-in margin (a 10 mm spot size mirror actually accepts 12-13mm), it is enough that the beam diameter does not exceed the galvo aperture — you do not have to deliberately hold to 80%. But if you pursue the finest spot quality or do precision machining, leaving some headroom on the beam yields better results.
Reference for beam-clipping loss at different ratios:
| Beam diameter / aperture | Clipping loss | Notes |
|---|---|---|
| ≤50% | ≈0.03% | Nearly zero loss; recommended for precision machining |
| ≤75% | ≈3% | Industrial standard; sufficient for marking/welding |
| 100% | ≈5% | Acceptable in most cases; the mirror margin can cover it |
| >100% | ≈15%+ | ❌ Spot distortion; barely usable only when mirror margin is generous |
Rules of thumb:
– Beam ≤ aperture — the basic principle for everyday selection; the mirror margin ensures most scenarios are fine
– Beam ≤ 75% of aperture — the recommended industrial standard; 3% loss has no impact on the vast majority of applications
– Beam ≤ 50% of aperture — recommended for precision machining (micro-holes, chip packaging); nearly zero loss, but it requires a larger-aperture galvo scanner and raises cost
A few examples make it clear:
- Beam 6.5mm + 10 mm spot size aperture = 65% ratio, ✅ loss <1%, no problem at all
- Beam 9mm + 10 mm spot size aperture = 90% ratio, ✅ within the aperture; the mirror margin covers it
- Beam 10mm + 10 mm spot size aperture = 100% ratio, ⚠️ exactly equal to the aperture; the mirror margin can absorb it, but not recommended for precision machining
- Beam 13mm (after 2X expansion) + 10 mm spot size aperture = 130% ratio, ❌❌ beyond the aperture, severe clipping
- Beam 13mm + 14 mm spot size aperture = 93% ratio, ✅ within the aperture, usable
- Beam 13mm + 15 mm spot size aperture = 87% ratio, ✅✅ more generous headroom
Simple memory aid: no expansion (6.5mm beam) → 10 mm spot size is fine; 2X expansion (13mm beam) → 14 or 15 mm spot size; 3X expansion (19.5mm beam) → 20 or 25 mm spot size. Just keep the beam within the aperture.
4.2 The Real Purpose of Beam Expansion — Shrinking the Spot
Many people think beam expansion means “making the light fatter”. It is exactly the opposite — the purpose of expansion is to shrink the focused spot.
The principle is simple: the fatter the beam, the smaller the spot after the F-Theta lens focuses it. Like water pipes — a thick pipe delivers water at low pressure, a thin one at high pressure. Fatter beam → smaller focusing angle → smaller focal spot.
The core formula:
$$\Phi_0 = \frac{4 M^2 \lambda f}{\pi \Phi}$$
Ignore the formula details; just remember one conclusion: double the beam diameter and the focused spot halves.
A few practical calculations (1064nm, M²=1.0):
| F-Theta lens | Beam 6.5mm (no expansion) | Spot | Beam 13mm (2X expansion) | Spot |
|---|---|---|---|---|
| F160 | 6.5mm | 33μm | 13mm | 17μm |
| F254 | 6.5mm | 53μm | 13mm | 26μm |
| F330 | 6.5mm | 69μm | 13mm | 34μm |
| F430 | 6.5mm | 90μm | 13mm | 45μm |
Longer focal length means a bigger spot, but 2X expansion halves the spot and makes up the difference.
How small must the spot be to mark small characters? Rule of thumb: for a 1mm-tall character, the spot diameter should not exceed 1/15 of the character height — i.e., within 67μm. For a 1.3mm character, the spot should not exceed 87μm. So:
- F160 + no expansion (33μm) → a 1mm character is 30 spots wide, very sharp
- F330 + no expansion (69μm) → a 1.3mm character is 19 spots wide, fine for simple characters; complex characters/Chinese text get fuzzy
- F430 + no expansion (90μm) → a 1.3mm character is 14 spots wide, blurry
- F330 + 2X expansion (34μm) → a 1.3mm character is 38 spots wide, sharp
- F430 + 2X expansion (45μm) → a 1.3mm character is 29 spots wide, acceptable; complex characters slightly fuzzy
Core conclusion: to mark characters under 1.3mm, keep the F-Theta lens at F254 or shorter (no expansion) or F330 (2X expansion). An F430 lens is simply blurry for small characters — no amount of expansion beats a shorter focal length.
4.3 How to Calculate the Field Size
Field size depends on the F-Theta focal length and the galvo scan angle:
$$field\ size = 2f \times \theta$$
Note: here the angle (in radians) is multiplied directly, not tan(θ). This follows from the F-Theta lens’s design property that image height = focal length × angle (h = f × θ) — which is exactly why it is called “f-theta”. With an ordinary lens, image height would be f × tan(θ) and the edges would distort.
No need to compute it yourself — just remember the common combinations:
| F-Theta lens | Typical field size |
|---|---|
| F160 | 110×110mm |
| F254 | 175×175mm |
| F330 | 220×220mm |
| F420/F430 | 300×300mm |
Note that Raylase galvo scanners have an optical scan angle of ±0.393rad, larger than SCANLAB’s ±0.35rad, so with the same F-Theta lens the field is slightly bigger. But a larger scan angle also means larger edge aberrations, demanding a better correction table.
4.4 Quick-Reference Table for the Trio — Look It Up, Don’t Calculate
| Application | Laser | Expansion | Aperture | F-Theta lens | Spot | Field size |
|---|---|---|---|---|---|---|
| Metal marking (small characters) | 20-50W pulsed | None | 10 | F160 | 33μm | 110×110 |
| Metal marking (large characters) | 50-100W pulsed | None | 10 | F254 | 53μm | 175×175 |
| Deep engraving | 100-200W pulsed | 2X | 14/15 | F254 | 26μm | 175×175 |
| Cleaning | 100-200W pulsed | None | 10 | F330 | 69μm | 220×220 |
| Welding (small parts) | 200-500W CW | None | 14/15 | F330 | 69μm | 220×220 |
| Welding (large parts) | 500-1500W CW | None | 20 | F420 | 56μm | 300×300 |
| Cleaning (large areas) | 1000-2000W CW | None | 20 | F330 | 69μm | 220×220 |
| Cladding / high-power welding | 2000-6000W CW | None | 30 | F420 | — | 300×300 |
5. Quick-Reference Selection for Common Application Scenarios
Look up your scenario in the tables directly — no need to calculate from scratch.
5.1 Laser Marking (Metal/Plastic)
| Laser | Recommended galvo scanner | F-Theta lens | Expansion | Field size |
|---|---|---|---|---|
| 20-50W fiber | Domestic 10 mm spot size or SCANcube IV 10 | F160 | None | 110×110 |
| 50-100W fiber | SCANcube IV 10 | F254 | None | 175×175 |
| 100-200W fiber | intelliSCAN III 14 | F254/F330 | 2X | 175-220 |
5.2 On-the-Fly Marking (Production Lines)
| Line speed | Recommended galvo scanner | F-Theta lens | Marking card |
|---|---|---|---|
| <30 m/min | JCZ 7210 / Century Sunny 9310D / Bairuitong 8910D | F160/F254 | Domestic |
| 30-80 m/min | SCANcube IV 10 / MINISCAN III-10 | F254/F330 | RTC5 |
| >80 m/min | excelliSCAN 14 | F330 | RTC6 |
5.3 Laser Cleaning
| Laser | Recommended galvo scanner | F-Theta lens | Notes |
|---|---|---|---|
| 100-200W pulsed | SCANcube IV 10 | F330/F420 | Cleaning has low accuracy demands; focus on field size |
| 500-1000W CW | intelliSCAN III 14/20 | F330 | High CW power; watch coating heat dissipation |
| 1000-2000W CW | hurrySCAN 20 | F330/F420 | Water-cooled galvo scanner; watch back-reflection |
| 2000-6000W CW | SUPERSCAN IV-30 | F420 | 48V + water cooling + fused silica mirrors |
5.4 Laser Welding
| Laser | Recommended galvo scanner | F-Theta lens | Notes |
|---|---|---|---|
| 200-500W CW | intelliSCAN III 14 | F330 | Can pair with a camera for process monitoring |
| 500-1500W CW | hurrySCAN 20 | F330/F420 | Water-cooled; watch the protective window |
| Above 1500W | SUPERSCAN IV-30 | F420 | 48V + water cooling; back-reflection must be assessed |
5.5 UV/Green Precision Machining
| Laser | Recommended galvo scanner | F-Theta lens | Notes |
|---|---|---|---|
| 5-20W UV (355nm) | intelliSCAN III 10 | F160 | Fused silica mirrors required; BK7 fluoresces |
| 5-20W green (532nm) | SCANcube IV 10 | F160 | Green light requires a dielectric coating; metal coatings lose too much |
| 20-50W UV | intelliSCAN III 14 | F160/F254 | 14 mm spot size + 2X expansion = smaller spot |
6. Brand and Model Comparison
6.1 SCANLAB (Germany) — the Industry Benchmark
| Series | Positioning | Aperture | cps (F160) | Tracking error | Servo board | Typical applications |
|---|---|---|---|---|---|---|
| basiCube 10/14 | Entry level | 10/14 | 570/800 | 0.14ms | Analog servo | Static marking |
| SCANcube IV | Industrial | 7/10/14 | 700-1840 | 0.12ms | Digital servo | Marking, on-the-fly, cleaning |
| intelliSCAN III | Precision | 10/14 | 480/680 | 0.15ms | Digital servo | Micromachining, precision marking |
| intelliSCAN IV | High end | 10/14/30 | 710-1310 | 0-0.08ms | Digital servo + encoder | 3D printing, precision + speed |
| excelliSCAN | Flagship | 14/20 | 1200/1310 | 0ms | Digital servo + SCANahead | High-speed production lines |
| hurrySCAN | Large aperture | 20/25 | 210/320 | 0.35ms | Analog servo | Welding, cleaning |
Naming pattern: basi = basic, SCANcube = compact, intelli = precision, excelli = flagship, hurry = large aperture. III = third generation, IV = fourth generation. The numeric suffix = nominal spot diameter in mm (note: the actual mirror is larger than the nominal value; see Section 2.1).
Suffix meanings:
– de = Digital Encoder — angle detection by a deflection grating disc with radial graduations; mature and stable, with extremely low drift and jitter and the highest linearity
– se = Interferometric encoder (SCANLAB patent) — angle measurement by the “optical pointer” interference principle; lower inertia at the rotor end, higher precision and faster speed; the fastest digital-encoder galvo scanner in the 10mm aperture class
In short: intelliSCANde uses a deflection-grating disc encoder (mature and stable), while intelliSCANse uses an interferometric encoder (more precise and faster). Both are digital-encoder galvo scanners, with far higher accuracy and long-term stability than traditional analog position sensors (photoelectric type).
6.2 Raylase (Germany) — the High-Speed Specialist
| Series | Positioning | Aperture | cps | Supply voltage | Typical applications |
|---|---|---|---|---|---|
| MINISCAN III | Compact | 10/14/20 | 800/1280 (@48V) | 30V/48V | Marking, on-the-fly, cleaning |
| SUPERSCAN IV | High power | 10/15/20/30 | 350-900 | 30V/48V | Welding, cleaning |
| SUPERSCAN V | Flagship | 7/10/14/20 | — | 30V/48V | Digital encoder, extremely low drift |
Raylase’s core selling point: 48V supply. The same galvo scanner runs 67% faster at 48V than at 30V; MINISCAN III-10 reaches 1280cps at 48V — the highest speed at its price point.
6.3 Cambridge Technology (USA, a Novanta company) — the Technology Pioneer
| Series | Positioning | Aperture | Highlights |
|---|---|---|---|
| 62xxH | High speed | 7/10/14 | Fastest motors in the industry |
| 83xxK | Precision | 10/14/20/25 | Large aperture; welding/AM |
| Lightning II | High-end scan head | 10-50 | 2-axis/3-axis; first choice for 3D printing |
Uses the standard XY2-100 interface; with its own ScanMaster controller + ScanPack algorithms it can reach 24-bit.
6.4 Domestic Galvo Scanners
| Brand | Representative model | Aperture | Positioning |
|---|---|---|---|
| Han’s Laser | ExtraScan II 10 | 10 | Economy; positioning speed 10m/s |
| Han’s Laser | UltraScan II | 7/10/14/20 | Industrial; <2μrad |
| JCZ | SG7210 | 10 | Economy; 559cps |
| Century Sunny | SS9310D | 10/12 | High speed; on-the-fly marking; 800/950cps |
| Bairuitong | YZD-8910D | 10 | On-the-fly marking; 800/950cps; XY2-100 + analog, ±15V |
Domestic galvo scanners typically cost 1/5 to 1/3 of imported ones. Their short-term specs are not bad, but they lag behind imports in long-term stability (thermal drift, unit-to-unit consistency, lifetime). They suit budget-sensitive, non-precision, delivery-sensitive scenarios. Not suitable for 24-hour production lines or precision micromachining.
💡 What is a grating galvo scanner? A new trend among domestic galvo scanners is the grating galvo scanner (e.g., some Han’s Laser models). Traditional galvo scanners detect angle with a photoelectric position sensor (silicon photocell + light-blocking vane) — hence “photoelectric galvo scanners”. A grating galvo scanner detects angle with a grating encoder — equidistant grating lines are etched onto glass or metal, and angle is measured via optical diffraction with resolution down to the nanometer level. Advantages: higher accuracy, stronger immunity to electromagnetic interference, lower thermal drift. Trade-off: higher cost. It follows the same technical route as SCANLAB’s intelliSCANde (digital encoder).
7. Common Selection Mistakes — the Pitfall Checklist
❌ Forcing 2X expansion onto a 10 mm spot size galvo scanner
After 2X expansion the beam is 13mm, exceeding the 10 mm spot size aperture, causing severe clipping at large deflection angles. Correct: 2X expansion → 14 or 15 mm spot size.
❌ Judging by marking speed while ignoring cps
A marking speed of 2.5m/s looks nice, but if you are marking 200 tiny 0.5mm characters — each requiring a jump, acceleration/deceleration, and settling — actual throughput is far lower than the marking speed implies. cps is the true character-marking efficiency.
❌ Wavelength and coating mismatch
Use a galvo scanner coated for 1064nm at 532nm and transmission drops from 99.5% to 60% — 40% of the power lands on the mirrors and the coating burns through quickly.
❌ Ignoring thermal drift
You buy a galvo scanner with 5μrad repeatability thinking it is very accurate, then it drifts 200μrad over 8 hours (32μm at F160). The line runs for 4 hours and drifts beyond 15μm — batches of rejects. Repeatability determines whether a single pass is accurate; thermal drift determines whether the system stays accurate over time.
❌ Not accounting for the compensation margin in on-the-fly marking
In on-the-fly marking the galvo scanner must “chase the workpiece”, so part of the deflection angle is consumed by compensation and the actual marking field is smaller than in static mode. Without calculating this in advance, the field may be too small to finish the job.
❌ Large aperture + large expansion = sharp speed drop
With a 20 mm spot size plus 3X expansion the spot does get smaller, but the mirrors are too heavy and marking speed may drop to 1/3 of a 10 mm spot size unit. Precision machining is about balance, not brute force.
❌ CW lasers without considering workpiece back-reflection
With 1000W on metal, surface reflectivity can reach 40% — 400W of reflected light returns along the original path. The galvo mirrors may burn out, and the laser’s output head can be damaged too. You must assess workpiece reflectivity; for highly reflective materials (copper, aluminum), add an optical isolator when necessary.
❌ BK7 mirrors with UV lasers
355nm UV striking BK7 glass causes fluorescence and transmission falls off a cliff. UV galvo scanners must use fused silica substrates — pricier, but using BK7 ruins the machine.
❌ Control card and galvo protocol mismatch
You buy an SL2-100 galvo scanner, but the control card only supports XY2-100 — SL2-100 is itself a 20-bit protocol, and only a control card and cable supporting SL2-100 can deliver its 20-bit precision. Connect it with an XY2-100 cable and it degrades to 16-bit XY2-100 mode; the SL2-100 galvo scanner’s advantage is wasted.
❌ Machining without warm-up
Right after power-up, a galvo scanner’s zero-point offset is large and the first few workpieces come out inaccurate. Build the habit: let it idle 5-10 minutes after power-up before formal production.
8. Selection Decision Workflow
Follow the steps in order and nothing gets missed.
Step 1: Determine the aperture
– Pulsed fiber ≤150W → 10 mm spot size
– Pulsed fiber 200-440W → 14/15 mm spot size
– Pulsed fiber 500W → 20 mm spot size
– CW fiber → first calculate the post-collimation beam diameter (Φ = 2 × f_collim × NA), then check the power
– CO₂/UV/green → choose aperture + coating by wavelength
Step 2: Determine the wavelength
– 1064nm low power → silicon + dielectric coating
– 1064nm high power → fused silica + dielectric coating
– 10600nm → metal substrate + gold coating
– 355nm → fused silica + UV dielectric coating
– 532nm → silicon/fused silica + green dielectric coating
– Special wavelengths → metal coating as a substitute (not recommended for pulsed lasers)
Step 3: Determine the speed class
– Static marking → standard galvo scanner
– On-the-fly marking → low tracking error + high cps
– High-speed lines → excelliSCAN
– Fine machining → high accuracy + low drift
– Bulk QR codes → high cps + high positioning speed
Step 4: Determine the accuracy class
– General → analog servo board
– Industrial → digital servo board
– Precision → digital servo board + encoder
Step 5: Determine the budget configuration
– Entry level → domestic galvo scanner + domestic card (CNY 3,000-8,000)
– Industrial → SCANcube/MINISCAN + RTC5 (CNY 20,000-50,000)
– High end → intelliSCAN/excelliSCAN + RTC6 (CNY 50,000-150,000+)
Step 6: Verify the trio
– Beam diameter ≤ galvo aperture?
– Focused spot meets the minimum-character requirement?
– Marking field large enough?
– Compensation margin sufficient for on-the-fly marking?
Only place the order when every box is checked.
This concludes the article. Selection of front-focus 3D galvo scanners and rear-focus 3D galvo scanners will be covered in dedicated follow-up articles. If you have any questions, feel free to contact us:
📞 Hotline: 159-2722-0467
🌐 Website: https://www.soar-laser.cn
🏷️ Brand: SOAR-LASER



Leave a Reply