What Backlash Actually Is — and How It Is Measured
In a precision planetary gearbox, backlash is the angular free play measurable at the output shaft when the input shaft is held stationary and the output is loaded alternately in positive and negative directions with a small test torque. It is the total angular dead-band that the output shaft sweeps through when load direction reverses — the gap between gear teeth in mesh, expressed as the angular equivalent at the output shaft.
The standard test method (per ISO 9283 and consistent with DIN EN 61800 servo equipment standards) applies a load equal to ±3% of the gearbox’s allowable output torque. This specific load level is chosen deliberately: it is large enough to fully take up any geometric clearance in the gear meshes, but small enough that torsional elastic deflection of the gearbox components is negligible — so what is measured is pure geometric backlash, not a mix of backlash and stiffness.
Gearboxes are rotational devices. Their inherent accuracy specification must be angular. Degrees are too coarse — a precision gearbox at 0.133° backlash sounds large, but that is only 8 arcmin, a very standard specification. Arcminutes provide the right resolution: 1 arcmin = 1/60th of a degree = approximately 0.0167°. The metric system equivalent for angular error is milliradians (mrad), but arcminutes dominate the planetary gearbox industry and all EP series datasheets are specified in arcmin.
Fix the gearbox input shaft rigidly. Attach a precision torque arm to the output shaft at a known radius. Apply a positive test torque equal to 3% of rated torque and read the angular position (encoder or dial gauge). Apply negative test torque of equal magnitude and read again. The total angular displacement between the two readings is the backlash value. Korea Ever-Power measures and certificates backlash for every EP series unit before shipment, with the measurement performed at the ±3% test load standard.
The Table Every Servo Gearbox Engineer Needs — Arcmin to Millimetre Linear Error at Five Load Radii
The following table converts every standard servo mjenjač backlash specification — from ultra-precision at 1 arcmin through standard-grade at 30 arcmin — into the actual linear positioning error at five practical load radii. All values are calculated using the exact formula E = R × tan(θ) where θ is the backlash angle in radians. For typical precision planetary gearbox backlash values below 30 arcmin, the small-angle approximation introduces less than 0.01% error.
The load radius is the distance from the gearbox output shaft centreline to the point where positioning accuracy is being measured or required — for example, the tip of a robot arm, the cutting tool of a CNC spindle, or the contact point of a conveyor drive roller.
| Negativna reakcija | Angle (°) | R = 50 mm | R = 100 mm | R = 200 mm | R = 500 mm | R = 1,000 mm | EP serija |
|---|---|---|---|---|---|---|---|
| <1 arcmin | 0.017° | 0.015 mm | 0.029 mm | 0.058 mm | 0.145 mm | 0.291 mm | Ultra-precision custom |
| <3 arcmin | 0.050° | 0.044 mm | 0.087 mm | 0.175 mm | 0.436 mm | 0.873 mm | High-precision CNC / laser |
| <5 arcmin | 0.083° | 0.073 mm | 0.145 mm | 0.291 mm | 0.727 mm | 1.454 mm | General servo positioning |
| <8 arcmin ★ | 0.133° | 0.116 mm | 0.233 mm | 0.465 mm | 1.164 mm | 2.327 mm | EP-ZDE / EP-ZDF (frames 60–160); EP-ZDS (all) |
| <12 arcmin | 0.200° | 0.175 mm | 0.349 mm | 0.698 mm | 1.745 mm | 3.491 mm | EP-ZDE-40; EP-ZDE 2-stage |
| <15 arcmin | 0.250° | 0.218 mm | 0.436 mm | 0.873 mm | 2.182 mm | 4.363 mm | EP-ZDE 3-stage; conveyors |
| <25 arcmin ▲ | 0.417° | 0.364 mm | 0.727 mm | 1.454 mm | 3.636 mm | 7.272 mm | EP-ZDWE / EP-ZDWF (80–160, 1-stage) |
| <30 arcmin ▲ | 0.500° | 0.436 mm | 0.873 mm | 1.745 mm | 4.363 mm | 8.727 mm | EP-ZDWE-60 (1-stage) |
★ = Standard precision class for EP-ZDE/ZDF/ZDS inline series. ▲ = Right-angle input series (ZDWE/ZDWF) — wider due to bevel gear stage contribution. Values calculated from E = R × tan(θ), where θ = backlash in radians.
A collaborative robot wrist joint with a 400 mm arm radius, using an EP-ZDWE-80 at <25 arcmin, will have a maximum backlash-induced positioning error at the end-effector of approximately 400 mm × tan(25/60 × π/180) = 2.91 mm. For a robot controlled by a servo drive in closed-loop position mode, this 2.91 mm is not a permanent error — it is the dead band at direction reversal. The servo controller compensates for this through position feedback from the motor encoder. However, any external disturbance during a hold position (after the encoder confirms position) can produce up to 2.91 mm of drift if the load torque causes the output shaft to move within the backlash dead band without the motor encoder detecting it.
Four Backlash Precision Classes — Matching Grade to Application Requirement
The standard industry precision class structure for precision planetary gearboxes maps backlash ranges to application categories. Choosing the right class is as important as not over-specifying: a <1 arcmin ultra-precision unit costs 3–5 times more than a <8 arcmin standard precision unit of the same frame size. If your application’s accuracy requirement is met by <8 arcmin, spending on a <1 arcmin unit adds no measurable performance benefit.
At 100 mm radius, <1 arcmin produces only 0.029 mm of backlash-induced dead band. Required for semiconductor wafer handling robots (silicon die positioning to ±0.01 mm), precision optical mounts, and research-grade direct-drive robotics where any dead band is unacceptable. Not typically available as a standard EP series product — requires contact with Korea Ever-Power application engineering for custom specification.
At 200 mm radius, <3 arcmin produces 0.175 mm maximum dead band. Appropriate for CNC feed axes where part dimensional tolerance is ±0.01–0.1 mm, laser cutting head positioning where kerf width is 0.2–0.5 mm, and multi-axis servo-driven positioning stages in Korean electronics assembly equipment. The servo position feedback loop readily compensates for backlash at this level in normal operation.
This is the specification range of the EP-ZDE, EP-ZDF, and EP-ZDS series (frames 60–190 at single stage). At 100 mm radius, <8 arcmin means 0.233 mm maximum dead band — completely adequate for industrial robot positioning, general automation indexing, and conveyor servo drives. The standard class represents the best value for the vast majority of Korean servo automation applications. For applications where cost matters and positioning requirements are moderate, this grade delivers consistent performance without the premium of tighter-tolerance alternatives.
The EP-ZDWE and EP-ZDWF right-angle input series fall into this range due to the bevel gear input stage adding angular clearance. The <25–30 arcmin specification is not a quality deficiency — it is an inherent characteristic of bevel-gear input designs across all manufacturers. For servo-controlled axes where the position loop compensates for gearbox backlash, this range is fully functional. Where it is not appropriate: open-loop stepper motor systems, where the backlash directly becomes a positioning error with no feedback compensation.
Backlash vs Torsional Stiffness — Two Different Causes of Positioning Error That Engineers Frequently Confuse
One of the most persistent misunderstandings in precision planetary gearbox specification is treating backlash and torsional stiffness as the same phenomenon. They are not. They affect positioning accuracy through completely different physical mechanisms, they are specified in the same units (arcminutes at the output shaft), and confusing them leads to incorrect gearbox selection. Buying a tighter-backlash unit does not solve a torsional stiffness problem, and vice versa.
Quantified Comparison: EP-ZDE-160 vs EP-ZDS-190 Elastic Deflection Under Variable Load
The following table uses the formula θ_elastic = T / Ct to show how the same applied torque creates very different elastic angular errors in the standard precision series vs the high-stiffness series. This is the actual data relevant for CNC rotary table and heavy robot joint specifications, where peak cutting or handling torques can reach 200–800 N·m.
| Applied Torque | EP-ZDE-160 Ct = 38 N·m/arcmin |
EP-ZDS-190 Ct = 130 N·m/arcmin |
Stiffness Ratio | ZDE-160 linear error at R=200mm |
ZDS-190 linear error at R=200mm |
|---|---|---|---|---|---|
| 50 N·m | 1.32 arcmin | 0.38 arcmin | 3.4× | 0.077 mm | 0.022 mm |
| 100 N·m | 2.63 arcmin | 0.77 arcmin | 3.4× | 0.153 mm | 0.045 mm |
| 200 N·m | 5.26 arcmin | 1.54 arcmin | 3.4× | 0.306 mm | 0.089 mm |
| 380 N·m (heavy CNC cut) |
10.00 arcmin | 2.92 arcmin | 3.4× | 0.582 mm | 0.170 mm |
| 800 N·m | 21.05 arcmin | 6.15 arcmin | 3.4× | 1.225 mm | 0.358 mm |
An engineer who specifies an EP-ZDE-160 with <8 arcmin backlash for a heavy CNC rotary table application has the backlash specification correct — but under 380 N·m peak cutting torque, the torsional elastic deflection adds another 10 arcmin. The total angular error at the output under load is 18 arcmin — more than twice the specified backlash. This is why heavy-load precision applications (large CNC rotary tables, heavy robot joints, servo press drives) require the EP-ZDS series with Ct = 130 N·m/arcmin, not merely a tighter-backlash EP-ZDE unit. The EP-ZDS-190 under the same 380 N·m load produces only 2.92 arcmin elastic deflection — a 3.4× improvement in dynamic accuracy.
How Backlash Grows Over the Gearbox Service Life — and What Accelerates It
A precision planetary gearbox does not maintain its initial backlash specification indefinitely. Angular dead-band increases over time as gear tooth flanks wear and planet carrier bearings accumulate running clearance. The rate of increase depends heavily on operating conditions — a correctly loaded, correctly lubricated gearbox running at recommended duty cycles will show only modest backlash increase over 20,000 hours. An overloaded or contaminated unit can double its backlash in under 5,000 hours.
| Service Hours | Approximate Backlash EP-ZDE-80, correctly loaded |
Linear Error at R = 300 mm | Notes |
|---|---|---|---|
| 0 h (new) | 7.5 arcmin | 0.654 mm | Factory-certified at ±3% rated torque test |
| 2,000 h | 8.0 arcmin | 0.698 mm | Normal run-in completed; initial surface conditioning |
| 5.000 sati | 8.8 arcmin | 0.768 mm | Steady-state wear rate; record baseline at 5,000 h inspection |
| 10.000 sati | 10.2 arcmin | 0.890 mm | Still within acceptable range for most standard applications |
| 15,000 h | 12.5 arcmin | 1.091 mm | Approaching replacement threshold for high-precision applications |
| 20,000 h (L10) | 15.1 arcmin | 1.318 mm | L10 rated life; schedule gearbox replacement |
Illustrative progression based on industry longitudinal data for correctly specified and loaded precision planetary reducers. Actual values depend on specific loading conditions, duty cycle, and ambient environment. The EP-ZDE/ZDF series lifetime lubrication significantly slows gear flank wear vs. improperly lubricated units.
Four Conditions That Accelerate Backlash Growth
Planet gear tooth flanks experience Hertzian contact stress above their designed surface fatigue limit. Pitting initiates and accelerates. Backlash can double within 3,000–5,000 hours rather than 20,000. This is the most common accelerant of backlash growth in Korean servo automation applications.
Water ingress (particularly in IP54 units subjected to direct washing) emulsifies the lifetime grease, reducing its film strength. Metal wear debris from early overload creates abrasive conditions. The resulting three-body abrasive wear acts on all gear mesh surfaces simultaneously, compounding the backlash growth rate.
Operating consistently above the recommended input speed (3,000 rpm for most EP series) increases planet gear centrifugal stress and generates heat that accelerates lubricant oxidation. Higher temperature reduces grease viscosity and film thickness, increasing metal-to-metal contact on gear tooth flanks.
Servo press main drives and robot collision-stop axes subject planet carrier bearings to repeated impact loads that exceed the steady-state fatigue design. Planet carrier bearing races develop micro-pitting, which adds to output shaft radial play — eventually contributing to measurable backlash increase beyond the gear tooth wear component.
EP Series Complete Backlash Specifications — All Frame Sizes and Stages
The following specifications are the factory-certified backlash values for all Korea Ever-Power EP series precision planetary gearboxes, measured at ±3% of rated output torque per standard test protocol. The wider backlash of the ZDWE/ZDWF series is a direct consequence of the bevel gear input stage — this is consistent with all right-angle input planetary gear reducers regardless of manufacturer.
| Serija | Frame Size | 1-Stage | 2-Stage | 3-Stage | Configuration |
|---|---|---|---|---|---|
| EP-ZDE | 40 mm | <12 arcmin | <15 arcmin | <18 arcmin | Inline, round flange |
| EP-ZDE | 60–160 mm | <8 arcmin | <12 arcmin | <15 arcmin | Inline, round flange — standard precision |
| EP-ZDF | 40–160 mm | <8–12 arcmin | <12–15 arcmin | <15–18 arcmin | Inline, square flange — identical to ZDE by frame |
| EP-ZDS | 115–190 mm | <8 arcmin | <12 arcmin | N/A | Inline, square flange, IP65 — same backlash as ZDE, higher Ct |
| EP-ZDWE | 60 mm | <30 arcmin | <35 arcmin | <40 arcmin | Right-angle, round flange — bevel stage adds clearance |
| EP-ZDWE | 80–160 mm | <25 arcmin | <30 arcmin | <35 arcmin | Right-angle, round flange — wider but servo-compensatable |
| EP-ZDWF | 60–160 mm | <25–30 | <30–35 | <35–40 | Right-angle, square flange — identical to ZDWE by frame |
When Backlash Does Not Affect Accuracy — The Unidirectional Exception
Angular dead-band only produces positioning error at direction reversal. If your application positions in one direction only — the load always approaches the target from the same angular direction, and the drive always maintains a positive torque in that direction during positioning — backlash contributes zero positioning error regardless of its magnitude.
- Solar tracker azimuth/elevation drives (always moving in the same sun-tracking direction within a half-day period)
- Single-direction conveyor drives
- Winding and unwinding spindles (unidirectional torque maintained)
- Gravity-loaded vertical axes where the load weight maintains positive tooth engagement
- Feed drives that always approach the workpiece from the same direction (with a one-sided approach strategy)
- CNC contouring axes (bidirectional movement within contour profiles)
- Robot joints (bidirectional by nature during path execution)
- Pick-and-place systems (approach and departure in opposite directions)
- Indexing tables (half the index movements are in positive direction, half in negative)
- Servo presses (ram descend and return are opposite directions)
Cost implication of this rule
A Korean solar tracker manufacturer who specifies <3 arcmin backlash for their azimuth drives — because “we need precision tracking” — is paying 2–3× the cost of a <8 arcmin unit for no accuracy benefit. The solar tracker always moves in the same azimuth direction (east to west through the day). Angular play only becomes relevant during overnight reset — a movement where ±5 mm positioning error at the panel face has no impact on energy yield. Specifying standard <8 arcmin EP-ZDE or EP-ZDS units and redirecting the budget to IP65 sealing (using EP-ZDS) for outdoor durability delivers more value than tight-backlash units exposed to the Korean coastal environment.
How to Measure Installed Backlash — Field Verification Procedure
Measuring backlash after installation establishes the system baseline — the reference against which future measurements are compared to detect wear-induced backlash growth. The procedure below uses servo drive diagnostics (no external instruments required for basic measurement) as well as the precision dial gauge method for definitive results.
Fix the input shaft (or engage the servo motor holding brake). Attach a precision dial gauge to the output shaft at a known radius R (measure to 0.01 mm resolution). Apply a test load of approximately 3% of rated output torque in the positive direction and zero the dial gauge. Apply the same test load in the negative direction and read the total displacement D. Backlash in arcmin = arctan(D/R) × (60/π × 180). This method directly measures the linear-equivalent value at your specific load radius — providing the most operationally meaningful measurement for your application.
Backlash Specification Decision Framework — Avoid Over-Specifying
The following decision questions will guide you to the correct backlash specification for your precision planetary gearbox without paying for tighter tolerances that provide no measurable benefit in your specific application.
Rule of thumb for Korean servo automation: <8 arcmin (EP-ZDE/ZDF inline, or EP-ZDS for heavy load/IP65) is the correct specification for approximately 80% of servo planetary gearbox applications in Korean industrial automation. The remaining 20% requiring tighter backlash are primarily semiconductor and precision optics applications, where it is worth paying the 3–5× cost premium. Right-angle input configurations (ZDWE/ZDWF) at <25–30 arcmin are appropriate whenever the space saving justifies the wider backlash — and in servo closed-loop systems, the backlash is typically fully compensated by the position feedback loop. For a complete five-step selection workflow including service factor and inertia matching, see the precision planetary gearbox selection guide.
Frequently Asked Questions on Planetary Gearbox Backlash
Korea Ever-Power’s application engineering team provides backlash-to-linear-error calculations and precision grade recommendations for your specific application — including load radius, accuracy requirement, and EP series product selection — in Korean and English. Provide your application parameters and receive a full specification recommendation before ordering.
Urednik: Cxm