{"id":1273,"date":"2026-04-27T06:47:54","date_gmt":"2026-04-27T06:47:54","guid":{"rendered":"https:\/\/worm-and-worm-wheel.com\/?p=1273"},"modified":"2026-04-27T06:47:54","modified_gmt":"2026-04-27T06:47:54","slug":"worm-gear-backlash-sources-measurement-and-control","status":"publish","type":"post","link":"https:\/\/worm-and-worm-wheel.com\/ko\/worm-gear-backlash-sources-measurement-and-control\/","title":{"rendered":"\uc6dc \uae30\uc5b4 \ubc31\ub798\uc2dc - \ubc1c\uc0dd \uc6d0\uc778, \uce21\uc815 \ubc0f \uc81c\uc5b4"},"content":{"rendered":"
\n

\uc6dc \uae30\uc5b4 \ubc31\ub798\uc2dc - \ubc1c\uc0dd \uc6d0\uc778, \uce21\uc815 \ubc0f \uc81c\uc5b4<\/h1>\n

A 0.05 mm backlash spec at the rim is not one number \u2014 it is the sum of five clearances. Decompose them, measure each, and the indexing accuracy you wanted is suddenly within reach.<\/p>\n

\uc5d4\uc9c0\ub2c8\uc5b4\uc640 \uc0c1\ub2f4\ud558\uc138\uc694 \u2192<\/a><\/p>\n<\/div>\n

\n
\ube60\ub978 \ub2f5\ubcc0<\/div>\n

Backlash on a worm and worm wheel pair is not a single quantity but the sum of five sources: keyway clearance, hub-to-shaft fit, output bearing radial play, tooth profile clearance, and thermal expansion mismatch. Total backlash measured at the wheel rim is typically 0.05 to 0.30 mm for general industrial drives and 0.02 to 0.05 mm for precision indexing. Reducing the total below 0.02 mm requires controlling every source individually, with duplex worm geometry handling the tooth-profile component down to near zero. Most “noisy reversing drive” complaints trace back to one or two dominant sources rather than a uniform increase across all five. Diagnosing which source dominates is the first step in any backlash reduction project.<\/p>\n<\/div>\n

Why backlash matters \u2014 the five-arcminute test<\/h2>\n

A new project lead at a Korean machine tool builder asked us last month for a 60:1 worm gear set with “industrial standard backlash.” The application turned out to be a 4-station rotary indexing table with a positioning tolerance of plus or minus five arcminutes. Industrial standard backlash on a typical worm gearbox is 30 to 60 arcminutes \u2014 six to twelve times the application tolerance. The mismatch was not the supplier’s fault and not the customer’s fault. It was the consequence of treating backlash as a single number rather than a system property assembled from five independent contributions.<\/p>\n

Every worm gear pair carries some lost motion between the worm thread and the worm wheel teeth. That lost motion is necessary to allow lubricant film, accommodate thermal expansion, and prevent jamming. The question is not whether to have backlash but how much to allow and how to control its sources. Articles that say “backlash is between 30 and 60 arcminutes” are repeating a catalogue number that may or may not match the application. Articles that talk about “anti-backlash worm gears” jump to the solution before identifying where the backlash is actually coming from. The right starting point is decomposition.<\/p>\n

The five sources of backlash \u2014 decomposition<\/h2>\n
\n
\n

Total backlash measured at the worm wheel rim is the sum of five components. Each component has its own physical mechanism, its own controllable range, and its own design action. The decomposition matters because you cannot reduce total backlash below the largest single component, no matter how hard you tighten the others.<\/p>\n

Most general industrial drives have one or two dominant components \u2014 typically tooth profile and bearing radial play \u2014 with the others contributing little. Precision indexing applications must control all five down to comparable levels.<\/p>\n<\/div>\n

\"\uc6dc<\/div>\n<\/div>\n
\n\n\n\n\n\n\n\n\n\n
Source<\/th>\nTypical contribution at rim<\/th>\nControllable range<\/th>\nPrimary design action<\/th>\n<\/tr>\n<\/thead>\n
1. Keyway clearance<\/strong><\/td>\n0.02 to 0.08 mm<\/td>\n0.005 to 0.10 mm<\/td>\nTighter key fit, retention set screw<\/td>\n<\/tr>\n
2. Hub-shaft fit<\/strong><\/td>\n0.005 to 0.04 mm<\/td>\n0.002 to 0.05 mm<\/td>\nShrink-fit or split-hub clamp<\/td>\n<\/tr>\n
3. Output bearing radial play<\/strong><\/td>\n0.01 to 0.05 mm<\/td>\n0.003 to 0.08 mm<\/td>\nPreloaded angular contact bearings<\/td>\n<\/tr>\n
4. Tooth profile clearance<\/strong><\/td>\n0.04 to 0.15 mm<\/td>\n0.000 to 0.20 mm<\/td>\nCenter distance, duplex worm, ground class<\/td>\n<\/tr>\n
5. Thermal expansion mismatch<\/strong><\/td>\n0.005 to 0.03 mm per 30\u00b0C<\/td>\ndepends on materials<\/td>\nMaterial pairing, housing temperature control<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Add the typical contributions and the picture becomes clear. A general industrial worm gear pair carries roughly 0.08 to 0.34 mm total backlash at the rim \u2014 which converts to 30 to 90 arcminutes on a 100 mm pitch radius. That range matches the catalogue numbers most articles quote without explanation. The decomposition reveals why those numbers are not destiny: each source can be reduced individually, and a 0.02 mm precision result is achievable when every component is held to the tight end of its range.<\/p>\n

How to measure backlash \u2014 the dial indicator method<\/h2>\n

Backlash measurement is straightforward but easy to get wrong on the first attempt. The procedure below works for any worm gear pair from miniature actuator to large industrial reducer. The key discipline is locking the worm shaft completely so that all measured motion at the wheel comes from the joint clearances, not from the worm rotating slightly under load.<\/p>\n

\n
    \n
  1. Lock the worm shaft against rotation. On a complete reducer, hold the input shaft against motion using a clamp or a flat across the keyway. On a bare gear set in a test fixture, clamp the worm shaft directly.<\/li>\n
  2. Mount a dial indicator on a stable reference surface adjacent to the worm wheel rim. Position the indicator tip against a flat surface on the rim, perpendicular to the wheel axis, at the largest available radius for measurement sensitivity.<\/li>\n
  3. Apply light tangential force to the wheel rim in one direction until the worm-wheel teeth fully engage on one flank. Zero the dial indicator.<\/li>\n
  4. Reverse the tangential force, applying equal magnitude in the opposite direction until the teeth re-engage on the opposite flank. Read the dial indicator displacement \u2014 this is the total backlash at the measurement radius.<\/li>\n
  5. Convert to angular backlash if needed: angular backlash (radians) = linear backlash (mm) divided by measurement radius (mm). Multiply by 3437.75 to convert radians to arcminutes.<\/li>\n
  6. Repeat at four positions around the wheel circumference (90 degrees apart). Backlash variation around the wheel reveals tooth-to-tooth spacing errors and runout effects that a single measurement misses.<\/li>\n
  7. Document the four readings, the average, and the variation range. The average is the working backlash; the variation is a quality signal for the wheel itself.<\/li>\n<\/ol>\n<\/div>\n

    Variation greater than 25 percent of the average usually indicates an out-of-round wheel or a tooth-to-tooth spacing error from a worn hob. If the variation is uniform around the wheel but the absolute number is too high, the dominant source is a fixed clearance (keyway, fit, bearing) and adjusting the wheel will not fix it.<\/p>\n

    \n
    \uc5d4\uc9c0\ub2c8\uc5b4\ub9c1 \ub370\uc2a4\ud06c \ub178\ud2b8<\/div>\n

    A measurement subtlety that catches first-time technicians: the dial indicator must read displacement of the wheel rim, not displacement of the dial indicator base relative to the wheel housing. If the indicator is mounted on the same housing that the wheel rotates inside, housing flex under tangential force shows up as fake backlash. Mount the indicator on an external rigid frame, not on the gearbox housing itself. The first time we ran a backlash audit on a Japanese customer’s indexing table, the apparent backlash dropped 40 percent the moment we moved the indicator base from the gearbox cover to a separate magnetic stand on the granite surface plate.<\/p>\n<\/div>\n

    Designing the backlash budget for an indexing application<\/h2>\n

    \"\"<\/p>\n

    Once the five sources are decomposed, the design exercise becomes straightforward. Allocate the total budget across the five components, recognising that the cheapest reductions come from the components that already have the largest controllable range, and the most expensive reductions come from components like tooth profile that need specialised geometry.<\/p>\n

    Consider a precision indexing rotary table for a Korean automotive parts welder. Index accuracy specification: plus or minus 30 arcseconds at the workpiece, located 250 mm from the wheel centre. That converts to plus or minus 0.036 mm linear at the workpiece radius, scaling to plus or minus 0.018 mm at a 125 mm wheel rim. Total bidirectional backlash budget: 0.036 mm at the rim. Allocating across the five sources at the tight end of each controllable range:<\/p>\n

    \n\n\n\n\n\n\n\n\n\n\n
    Source<\/th>\nAllocated budget (mm)<\/th>\nHow achieved<\/th>\n<\/tr>\n<\/thead>\n
    Keyway clearance<\/strong><\/td>\n0.005<\/td>\nHand-fitted parallel key + retention set screw<\/td>\n<\/tr>\n
    Hub-shaft fit<\/strong><\/td>\n0.002<\/td>\nH7\/p6 shrink fit interference<\/td>\n<\/tr>\n
    Output bearing radial play<\/strong><\/td>\n0.005<\/td>\nPreloaded angular contact pair, C2 fit<\/td>\n<\/tr>\n
    Tooth profile clearance<\/strong><\/td>\n0.020<\/td>\nDuplex worm with 0.02 mm\/mm axial adjustment<\/td>\n<\/tr>\n
    Thermal expansion mismatch<\/strong><\/td>\n0.004<\/td>\nSteel worm + bronze wheel, 20\u00b0C ambient swing<\/td>\n<\/tr>\n
    Total budget<\/strong><\/td>\n0.036<\/strong><\/td>\nMatches application requirement<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

    Notice that the tooth profile component takes more than half the total budget. That is typical \u2014 tooth profile clearance is structurally the largest source and needs the most aggressive reduction (duplex worm geometry) to fit within a precision budget. The other four components are easier to control individually and contribute proportionally less.<\/p>\n

    Duplex (anti-backlash) worm gear technology<\/h2>\n
    \n
    \"\"<\/div>\n
    \n

    A duplex worm carries a small intentional difference in thread pitch between the right flank and the left flank of every thread. The pitch difference creates a tooth thickness that varies along the worm length \u2014 thinner at one end, thicker at the other.<\/p>\n

    Sliding the worm axially relative to the wheel changes which axial position is in mesh, and therefore which tooth thickness contacts the wheel teeth. Move the worm toward the thicker end and the tooth profile clearance drops. Move it the other way and clearance opens up. The same gear pair adapts to a wide range of backlash settings without re-machining anything.<\/p>\n<\/div>\n<\/div>\n

    A typical duplex design changes backlash by 0.02 mm for every 1 mm of axial worm movement. With manufacturing tolerances on the wheel of plus or minus 0.045 mm, a 2 mm axial worm shift covers the full tolerance range from open clearance to zero clearance. Adjustment is done at assembly with a shim-and-lock-nut arrangement, and the setting holds for the life of the drive unless re-shimmed.<\/p>\n

    Two cautions on duplex geometry. First, zero backlash is rarely the right target \u2014 at zero clearance the lubricant film cannot establish, friction climbs, and wear accelerates. Most duplex applications target 0.02 to 0.04 mm tooth profile clearance, leaving room for oil film without giving up positioning accuracy. Second, duplex geometry is not retrofittable. The worm and wheel are matched as a pair from manufacture, and substituting a standard worm into a duplex wheel housing removes the adjustment capability entirely.<\/p>\n

    Backlash drift over service life<\/h2>\n
    \n
    \"\"<\/div>\n
    \n

    Backlash is not constant over the life of the drive. Each of the five sources drifts on its own time scale, and the total grows in a characteristic pattern that maintenance teams can monitor.<\/p>\n

    Tracking backlash through scheduled measurements is one of the cheapest condition-monitoring techniques available \u2014 a 5-minute dial-indicator check every quarter catches developing wear long before it becomes visible by other means.<\/p>\n<\/div>\n<\/div>\n

    Tooth profile clearance grows steadily with operating hours as the bronze wheel teeth wear. A typical industrial drive shows 0.003 to 0.008 mm of tooth-profile growth per 1,000 operating hours under nominal load, accelerating to 0.015 mm per 1,000 hours under chronic overload. Bearing radial play grows in steps when bearings wear past their fatigue threshold. Keyway clearance grows when the key fretts under reversing load. Hub-shaft fit and thermal expansion are essentially constant unless something fails catastrophically.<\/p>\n

    A maintenance team that records backlash quarterly and plots the trend can usually predict gearbox replacement six to twelve months in advance \u2014 well before the rising backlash starts to affect output positioning accuracy or trigger downstream alarms. For complete drive units, browse standard \uc6dc \uae30\uc5b4 \uac10\uc18d\uae30<\/a> options that include factory backlash specifications and field-adjustment provisions on most frame sizes.<\/p>\n

    Three real backlash control cases<\/h2>\n

    Case 1 \u2014 Korean machine tool indexing table<\/h3>\n

    A Korean automotive parts welder needed plus or minus 30 arcseconds index accuracy on a 4-station rotary table for door-frame welding fixtures. Initial specification: standard 50:1 worm gear reducer. Measured backlash on the first prototype was 35 arcminutes \u2014 70 times the application tolerance. Diagnosis: tooth profile clearance dominated at 0.12 mm at the rim, with the keyway adding another 0.04 mm. Solution: switch to duplex worm and wheel pair with 0.020 mm tooth profile target, hand-fitted parallel key reducing keyway clearance to 0.005 mm, preloaded angular contact bearings reducing radial play to 0.005 mm. Final measured backlash: 0.034 mm at the rim, equivalent to plus or minus 28 arcseconds \u2014 within the application tolerance with a small margin. Total cost premium over the standard reducer: roughly 2.4 times. Application required this premium because positioning error directly impacted weld quality.<\/p>\n

    Case 2 \u2014 Japanese semiconductor wafer stage<\/h3>\n

    A Japanese semiconductor equipment OEM needed sub-arcsecond positioning on a wafer-handling rotary stage. Backlash budget at the wheel rim: 0.005 mm \u2014 well below the practical limit of any worm gear technology. Diagnosis: worm gear was the wrong technology choice for this accuracy class. Solution: replace the worm gear concept entirely with a direct-drive torque motor and harmonic drive backup, abandoning the worm gear approach. Lesson: when the budget calculation shows that even the tightest control on every backlash source cannot meet the requirement, the answer is not better worm gear technology. The answer is a different gear technology. Worm gears with full duplex and tight bearings can reach roughly 0.02 mm at the rim; below that, harmonic drive or direct-drive becomes the right answer.<\/p>\n

    Case 3 \u2014 Vietnamese textile loom positioner<\/h3>\n

    A Vietnamese textile loom builder reported “noisy reversing” on a thread positioning drive after 4 months of operation. Initial assumption: worn bronze wheel needing replacement. Backlash measurement showed 0.42 mm at the rim, far above factory specification of 0.18 mm. Decomposition diagnosis: tooth profile had grown only modestly from 0.08 mm to 0.12 mm. The dominant new source was bearing radial play, which had grown from 0.02 mm at delivery to 0.18 mm \u2014 bearings were worn out, not the gear pair. Solution: replace bearings, retain original worm and wheel, restore backlash to 0.16 mm. Total cost: about 8 percent of a full gear pair replacement. Lesson: not every increased-backlash complaint means worn gears. Decomposition before replacement saves money on the parts that are still serviceable.<\/p>\n

    \uc790\uc8fc \ubb3b\ub294 \uc9c8\ubb38<\/h2>\n
    \n
    \nQ: Is zero backlash a realistic design target?<\/summary>\n

    Almost never. Zero backlash means the worm and wheel teeth are in continuous contact on both flanks simultaneously, which prevents lubricant film formation between the contacting surfaces. Friction climbs, heat generation increases, and wear accelerates dramatically. Practical “anti-backlash” designs target 0.01 to 0.04 mm of tooth profile clearance \u2014 small enough for precision positioning but large enough to maintain the oil film. True zero-backlash designs (spring-preloaded split worm) work but require careful lubricant selection and accept shorter service life as the trade-off.<\/p>\n<\/details>\n

    \nQ: How does backlash convert between linear and angular units?<\/summary>\n

    Linear backlash at radius R converts to angular backlash through the formula: angular backlash (radians) = linear (mm) divided by R (mm). Multiply by 3437.75 to convert radians to arcminutes, or by 206265 to convert to arcseconds. Example: 0.05 mm linear backlash measured at a 100 mm rim radius equals 0.0005 radians equals 1.72 arcminutes equals 103 arcseconds. The same 0.05 mm at a 25 mm rim radius gives 6.88 arcminutes, four times worse. Always specify the measurement radius alongside the linear value, or specify the angular value directly.<\/p>\n<\/details>\n

    \nQ: Can I reduce backlash on an existing gearbox without replacing parts?<\/summary>\n

    Sometimes \u2014 depends on which source dominates. If output bearing play is the dominant source, replacing bearings with a tighter clearance class often recovers 50 percent of the original backlash budget without touching the gears. If keyway clearance has grown from key wear, fitting a slightly oversized key restores the original spec. If tooth profile clearance is dominant, the fixed-geometry worm and wheel cannot be adjusted in place \u2014 replacement is the only path. Adjustable-centre-distance designs allow some tooth-profile recovery but only on housings designed for it. Diagnose the dominant source before deciding to replace gears.<\/p>\n<\/details>\n

    \nQ: What is the relationship between backlash and gear accuracy class?<\/summary>\n

    Accuracy class (DIN 5, 6, 7, 8) controls tooth-to-tooth profile error and total cumulative pitch error, not the average backlash. A DIN 5 ground worm gear pair has tighter tooth-flank geometry than a DIN 8 hobbed-only pair, but their average backlash can be set to similar values. Where they differ is backlash variation around the wheel \u2014 DIN 5 might show 0.005 mm variation while DIN 8 shows 0.030 mm. For applications where backlash variation matters (servo positioning, smooth motion control), accuracy class matters as much as average backlash. For applications that just need consistent reversing position, average backlash is the dominant spec.<\/p>\n<\/details>\n

    \nQ: How does temperature affect backlash on a phosphor bronze wheel and steel worm?<\/summary>\n

    Phosphor bronze has a thermal expansion coefficient of roughly 18 ppm per degree Celsius, while case-hardened steel is 11 ppm per degree Celsius. The bronze wheel grows faster than the steel worm and housing as temperature rises. For a 100 mm pitch diameter wheel, a 30\u00b0C temperature swing changes the wheel diameter by approximately 0.054 mm \u2014 most of which translates directly into reduced tooth profile clearance at the operating temperature. Cold-start backlash is therefore larger than hot-running backlash, and precision applications that operate across a wide temperature range need to design for the cold-start case (largest backlash) while ensuring the hot-running case never reaches zero clearance.<\/p>\n<\/details>\n

    \nQ: Should I specify backlash in arcminutes or millimetres in my drawing?<\/summary>\n

    Both. Korean and Japanese OEM specifications typically state the angular value as the primary specification (e.g. “12 arcminutes maximum bidirectional backlash”) with the equivalent linear value at a defined radius as a secondary reference (e.g. “equivalent to 0.07 mm at 100 mm pitch radius”). The dual specification eliminates ambiguity for the supplier and gives the inspection team a direct measurement target. Standalone linear values without specified radius are ambiguous; standalone angular values are precise but harder to measure on the bench. Both together make the spec unambiguous and inspectable.<\/p>\n<\/details>\n

    \nQ: How does backlash interact with worm gear ratio calculation and mounting choices?<\/summary>\n

    Higher ratios produce more backlash at the output for the same input motion, because the output rotates less per unit of input. A 100:1 ratio with 0.1 mm rim backlash shows 10 mm of input shaft travel before output engagement reverses \u2014 irritating but harmless on a conveyor, intolerable on a servo positioner. Mounting method also matters: split-hub clamping introduces zero joint backlash because the friction grip is uniform around the full bore circumference, while keyway mounting always carries the keyway clearance contribution. For high-ratio precision applications, both the ratio choice and the mounting choice need consideration alongside the gear-pair backlash spec.<\/p>\n<\/details>\n<\/div>\n

    Backlash on a worm gear pair is not a single number to negotiate down with the supplier. It is a budget assembled from five independent sources, each measurable, each controllable through specific design actions, each subject to drift over service life on its own time scale. Articles that quote “30 to 60 arcminutes typical” without explaining the decomposition leave the design engineer no path to a precision result. The engineer who decomposes the budget, allocates each component honestly, and measures the assembled drive against the budget reaches the application tolerance reliably the first time.<\/p>\n

    For Korean and Japanese OEM design teams developing precision indexing, machine tool, or servo positioning applications, our engineering desk runs a five-source backlash decomposition against your accuracy requirement and recommends the gear pair, mounting, bearing, and key arrangement that fits within budget. Standard catalogue precision and duplex worm gear sets<\/a> cover the full range from general industrial to indexing-grade applications. Custom geometries are made to drawing on 6 to 8 week lead times \u2014 request a backlash budget review<\/a> with your accuracy specification and our team will return a five-source allocation within one Korean working day.<\/p>\n

    \n

    Designing a precision indexing or positioning drive?<\/h3>\n

    Send the accuracy specification (in arcseconds or millimetres at the workpiece radius) and the operating temperature range. We will decompose the backlash budget across the five sources and recommend the gear pair, mounting, and bearing combination that fits within tolerance.<\/p>\n

    Request a backlash audit \u2192<\/a><\/p>\n<\/div>\n

    \ud3b8\uc9d1\uc790: Cxm<\/p>","protected":false},"excerpt":{"rendered":"

    Worm Gear Backlash \u2014 Sources, Measurement, and Control A 0.05 mm backlash spec at the rim is not one number \u2014 it is the sum of five clearances. Decompose them, measure each, and the indexing accuracy you wanted is suddenly within reach. Talk to an engineer \u2192 Quick Answer Backlash on a worm and worm […]<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[2821],"tags":[30,33],"class_list":["post-1273","post","type-post","status-publish","format-standard","hentry","category-worm-and-worm-wheel","tag-worm-gear","tag-worm-gear-worm"],"_links":{"self":[{"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/posts\/1273","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/comments?post=1273"}],"version-history":[{"count":2,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/posts\/1273\/revisions"}],"predecessor-version":[{"id":1275,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/posts\/1273\/revisions\/1275"}],"wp:attachment":[{"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/media?parent=1273"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/categories?post=1273"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/worm-and-worm-wheel.com\/ko\/wp-json\/wp\/v2\/tags?post=1273"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}