How To Design With Precision Ballscrews
How To Design With Precision Ballscrews
How To Design With Precision Ballscrews
to
design
PRECISION
vide ability to carry heavy loads with iow friction and high efficiency.
As co&e?*te~*sof rota,ry to linea:r motion, the2 are P*eplacing the acme pokes screw in many application,s. Heq*e are their load alzd speed chmacteristics, meth.ods for preventing ?,eversibility and, . , over-travel, and a simplified lo-step design pyocedwe.
JOHN
Beaver
A HOPE, v;ce.presiden)
Precision
How
a ball
screw
operates
I~(~l~ents: rl
&xv
the screw or nut cansesthe rollinq balls to move along the p:lth. The lxalls trar,el at approximately half the ,IXLXI #)f the ~;ICC~ and esit at the trailing end of the nut. 1L)I\c~) the balls rolling continuously in the system, a re(llilj trll)c Jcflects ;~11,1 recirct~i;ltCs thC 1X111\ to tllL* lC.ltlillS
i.ll(l tji the nnt.
li.l<i~: oi)cratioll of ;* ball screw is illustrated in I;ip 2: l !A! Rotarv motion applied to the screw (manually or 1-r~ 2 po\vcr q0IIrc.c) drives the nut linearlv along the CCKW .I:\I\. conlinuec!. next paw
PRODUCT ENGlNEERlNG
l
REPRINTED-Circle
59
I-DIMENSIONS
,* .r -
AND
RATINGS
OF
8Al.L
SCREWS
SW
HUTDATA
~&I&II ,,poted Max lhrwds kilt Turns Mia screw CjperotlRg~ static Pw of 0.0. .-ma, kr. @lC lb Lead, ihin. El? I IkJlls In. i,;-,-,. f-. s$ ; -: %[I*. T, l/2:%48 ,q;*;.g .~:z yg : ;g l/16 -- 7 T/2 0.54 .-.7)16 .i,300 - 7:oow..s 5/t%: 8.,,1/2Q.G4 -ID - ,t,m -to,ooo 8 s/64 8 .I/2 0.70 2,604 14,OOD 7 .3J#$ 8 v-p.- 0.07 SD 3/4 :-3,300:. 18,DQb .6 8-lfi?.l.D3
LOAD
VS
LIFE
.?
iY: . 10. .66 ., 98
Rated bad,
We, miltion inches trovsl ,. 1ooo.O ,- ., 6$..0 . . 8.0 ..s 2.4 1 .o. 0.512 0.296 0:186 D.l,%
l (Bj Rotary motion applied to the nut (manuallv or tx a transmission drive such as gears) drives the screw l&~~rl~~.
c, ! _
Because of the low coefficient of friction, both rotx!.-tolinear functions arc reversible; that is, a linear motion of either the screw or nut can cause the other mcmbcr to rotate. Thrcacls on most types are machined or rolled. \\,l~cn machined, they are rough cut, hardened and then ground to obtain highest positioning accuracy, cfficicncy. load capacity and Iifc. Rolled threads, on the other hand. arc cheaper to produce and offer considerable cost s;nings if quantities are sufficiently high. _
BALL-SCREW CHARACTERISTICS
(A)
Load-life characteristics of ball screlvs arc similar to those of ball bearings. Because most of the loading is in thrust, ball scrc\vs operate similarly to a series of an;nlarcontact ball bearings. The sides of the ball grooves extend almost to the midpoint of the ball.
Static loading 2. BASIC OPERATIONS: Rotary motion to (A) screw, 01 (B) nut, causes other member to tra\,el linearly or provide linear force.
* (,.
.\ ball can carry a much greater load in a closely couforming racen.ay than on a flat surface. In fact, \\,hen the radius of the raceway or groove is only slightly larger than that of the ball, the load-carrying ability is o\.er three times greater than that of a ball rolling on a flat plate. In ball scrflvs. conformity of raceway to balls is closer than generaily rcalized-groove radii of 51 to 52% of the ball diamctcr .ire normal. (Cloqer conformity will result in too much scrubbing action during rolling.1 St,ltic load-carrling ahility of a ball on ;I surface varies with the fourth bower of the Rriucll hardness number of that surface. This esplniuz whv rxeways with a Rc 63 h,lrtllle<s are desired for lx111 bearings. Ball scrcxs and nuts ;lrc u~.lclc to Rc 6%hut this I-,I~IICis somctinlcc :cduccd bccausc of hi+ structural-strength rcquircments.
Dynamic loading
llle dt n,~ni~c load-c,irr!ing c,ipncity (or rated opcratin;: lr,ad, l.il)lc I) is usually of grc,ltcr ililpulI.kllcc tll.111 the
60
static load capacity. Since fatigue is the primary CI~\C ,I: failure, the load-life rating of a ball screw is based on the number of times a ball rolls over a particuklr poiut in the raceway, and the load on that point as exh ball p<l\ycj 01x3 it. Life varies inversely as the cube of the !~klil: cycles to failure decreases rapidly as the lox1 per Ix111 increases. Ball screws are rated at 1 million inchrs of travel. 1.1l~lc II. lkublilrg the rated Iodd cuts life to one-cishth (12 i.000 in.): reducing the load in half increases life eight times (to S,OOO,OOO in.). About 90% of any !argc group of identical units may be expected to bc equal to or c~ccd the I,lting. Iliis rated ~lluc is only abalit oric-fif!ll the .lctu i! GIIcr;lgc lift. Illus if a Ixlll scrcw is rlllllliilg ~ri~ootlilv at
PRODUCT ENGINEERING
l
3. TO
REDUCE
AXIAL
PLAY,
preloaded circulating
system
the time it reaches its rated life, it will probably continue operation again as long as a new unit.
Groove hardness
. similar to a clccp-groove, single-row ball bearing. If both nut and ccrew grooves have the same radius, the axial freedom is a function of four times the difference between the groove and ball radii. Axial freedom also varies ivith the angle at which the ball contacts the groove. The contact an@ is measured between the normal to the screw axis and a line through the ball center at 90 to the rolling axis of the ball. For typical ball screws with a radius 52% of the ball diameter and with the balls making contact at 45 angle, axial play equals 5.6% of the ball diameter. Decreasing the groove radius reduces the end play. Also, if a groo\e is ground several ten-thousands of an inch shallower, it has a reduced contact angle and, as a result, less axial freedom. However, a shallower groove increases the ball pressure dnd reduces life (for a given external load) faster than the axial freedom is decreased. Ogival (or gothic-arch) groove forms are often requested in ball scrams to reduce axial freedom without greatly increasing the ball pressure. Vaiiations in groove depth do not change the contact angle as rapidly as with a round groove. However, the ogival form reduces ball conformity and its load-carrying ability. Axial play can be avoided by preloading. The ball screw in Fig 3(A) uses a double nut for this purpose. The method is similar to that used for face-to-face or back-toback angular-contact ball bearings. There is little sacrifice of efficiency at full load because the preload is alleviated as the external load builds up. Axial play can also be removed by a method analogous to that used in double-row ball bearings, Fig 3(B)-but at a sacrifice in ease of adjustment. Numerically controlled production machines frequently utilize the double-nut type, shown in (A), for high system stiffness. Positioning accuracy directly from the tape has been held to 0.000,15 in., and repeatability accuracy to 0.000,050 in.
DESIGN Axial load CONSIDERATIONS
life of ball screws. Reducing the recommended hardness from Rc 60 to Rc 50 cuts the expected load capacity in ( calf. A further reduction to Rc 45 lowers life approxi\ mately 30%.
Race hardness does not affect the spring rate. but races with greater hardness are able to u-ithstand more deflection safely.
Efficiency
Ball screws have efficiencies consistently in. escess of 90%. Best techniques for determining ball screw efficiencies involve measuring only the losses due to friction. Methods based on measurement of total output thrust and the total input torque often give confusing results because errors of determination can easilv be greater than the losses in an efficient 1~111 scre\v. \i.i& ball scrt~vs of conventional proportions. the efficiencies in com.erting torque to thrust 3nd thrust to torque are not signilicantly different.
End play and preload
.4sial load-attachment points on the nut, such as a flange or V-thread, should be located with reference to the load-attachment points on the screw (on the end of the nut away from the load end of the screw) so that there are like stress signs (tension or compression) in the screw and nut body. Avoid thin-walled nuts with trunnions at the midpoints for bvo reasons. Trunnion deflection forces the trunnion base on the compression side to move inward and causes highly localized ball loading. Also. on one end of the nut, the stress sign for carrying the load uill be opposite to that in the screw.
Overturning and side loads
Ball screws possess a certain amount of illlicrcllt ;lui;ll freedom or end play. In this rcspcct, the hall screw is
PRODUCT ENGINEERING - APRIL 24, 1961
\Vhere\-er possible, the load direction in ball scrcn5 should coincide with the axis.. This avoids overturnin: or side loads. An overturning loading increnscs the tot;)1 load on- the balls in the end turns of the nut; side load, increase the stresses on a limited number of ball5 and a limited portion of the groove. If present, side loading or overturning moments must be allowed for in the same Iuanncr xs the combination of asial and radial loatl.~ in Ml-bcariug applications. Normally, it is casicr to 2vuid
61 .
60
Effectire column length, in,bosed on moximum screw I D. LOAD CAPACITIES 4. COLUMN
100
200
4 5 6
1000
4 5000
hollow
side loads in design than it is to make the ball screw large enough to withstand them.
Column loading
Ball screws are frequcntlv employed as pin-ended columns. The curves in Fig 4 &e safe compression loads for simply supported hollow screws of various lengths and diameters. These are minimum values because varying degrees of end tixity will increase the load capacities considerably. The left end of each curve approaches the
maximum static load for the corresponding size; the right
components under load, there is -also a very substanti:lI amount in the anti-friction bearings. It is inadvisable to fasten the wheel of a worm-gear ict directly on a ball screw or nut unless the thrust bearings are preloaded. Otherwise, deflection of the ball bearing? under ball-screw loads will allow the worm set to move out of alignment, with resulting loss in efficiency and life of the gear set.
Ball-screw enclosure
end stops at a point where column length divided by the radius of gyration equals 150.
Critical speeds .
Ever]; shaft has several natural frequencies at which it will vibrate se\.erely. Resonant frequencies will also occur when the screw is fixed and the nut rotated, if the relati1.e rotational speed of the nut is equivalent to the natural frequehcv of the screw. It i, belt to ctav at least critical speed in ball SC&S. 20% below the Critical speed is a function of the end mounting (ends can be clamped, merely supported or entirely free), length and screw diameter. Of all the variables, the unsupported length has most effect because the critical speed is inverseI! proportional to the square of the length. A shaft 100 in. long will have a critical speed of one-fourth that of a shaft 50 in. long. Critical speed is directly proportional to the diameter-doubling the diameter doubles the critical speed. Resonance problems can be avoided by increasing the screw lead to lower the rotating speed, or by employing hollow screws to increase the critical speed. For difficult problems, retractable fingers can be provided to support the midppint or several points along the screw length. The curves in Fig 5 show the lowest, or first order, critical speeds for solid shafts with simply supported ends (such as radial or angular-cont4ct bearings). If the shaft can be clamped at one end, multiply the rpm by 1.6. If it can be rigidly held at both ends, multiply the pm by 2.3.
iojvest
It is best to completely enclose ball SCICMS to c\c:~lc dirt. This also preserves the lubricant. Telescoping shirltl~ with dirt-tight glands give good protection, and bellow+ tvpc enclosures serve well where the screw requires protc.ction from dirt only. Where complete enclosure is impossible, use nipcr 5, I:. on the ball nuts to prevent entrance of metal chips and 11: but the finest dirt. Where the screw rotates, the Rut c.111 be sealed within an extension tube connected between :!lc ball nut and the load-but this requires a longer o\er-.I!? assembly length than that of the trunnion type.
Preventing reversibility
Because these devices are highly efficient in converting thrust to torque, a heavy thrust load can cause the SCKN to reverse. Some designs rely solely on the losses in hi;h ratio gear trains to prevent reversal, but frequentlv m>re
positive means are necessary.
Axial
deflections
Spring-loaded, electrically reIeascd brakes are someti;::ct used with electric drives, or with a hydraulically rele.ascd counterpart where hvdraulic motors proGde the drive. Spragging locks, Fig 6(A) with a means of unlockIn; from the driving side, can prevent runback, but have 3 tendency to chatter when the load attempts to run ahead of the drive. These devices do depend on rotation to de\-elop braking torque, therefore they require some method of releasing the spragging member when the drive torque is reapplied. Load-sensitive no-back devices, Fig 6(B) used directly on the screw are very effective. \Vhen power is applied through the scar to drive the load to the right, the overrunning clutch or mtchct tu;ns
PRODUCT ENGINEERING
l
10 STEPQDESIQNINQ
Step 1: Prepare n gcomttry of tile linkage;. This establi~l~e* :hc rtwkc 2nd. wth relntcd dapa, linear speed. Stc-p 2: Dct,:rlllinc !he Io;I~I:I~ on the ball screw. Colunw aAm may ucxr if +hc ~~~~!,ing is in compression un 91c 11nk which bccnrnc~ the l!n~xr acluator. If the column is short, skip to Step 4. Step ?: Determine the necessary screw diameter from Fig 4 to avoid rdunu~ action. For solid steel shafts USC cthe formula:
WITH
BALL
SCREWS
-.; -.
f : : !I,
-. .?
I = load or force, lh = lead of thread, in. screw efficiency (more than 90/with normal F = Loads) If the cornptited tilue is too high, another s&w &e, or nr.e with a Spe&l lead, may & .requi!ed. Step 9: Check critical speed of the rotating screw, I from Fig. 5. This shouhi be done even if the nut is rotating- %,.. .._ and the screw translating. A Step 10: Check the systerri stiI?n& In nudleri&lT$ A =I cross-sectional art+ sq in. coritrolled machines or missile controls, ,the system stiff- ( L = length of column (pin to pini, in. p = Radius of gSlRtion of rro&+aeetioF, in. ness (in axial direction) requirements w d+rmine the ,m diameter, Sti&ess is best giveri in ~+TITIIS of : ,-_ In deterrniriing A xnd p, the aie~ge %tween .&tttal spiug.ratc. (linear de&ctjon or elongation of the,bar h .: OD and root gives the equivaknt OD of the sxew. (ruin .z,;.?: unit load), Ii thk Joad is constant, d&&o~ \%ere the screw is only part of the pctuating, link, and an extensioti tube from nrtt to pin at theother eird~~&-~ . &ffness) occurs at midpoint of the a&r@ when !he.- j ;: p&es half or more, the average area i&l: radius of gym; :.-> screw is stretched slightly between,. $hr%$ *in@. ,P!{.,.r,&+ tion for the full length can be det~nnined by the fol@e screw is supported at one ed ohfy * s&m >p, _ !.. : T -sketch (A), the spring rate of the screw $:. * .; s-. :.: -. lowing appro?rimatiOn:
.,., ._ / ,.*:. ; _
. K,-+
I
and
Step 4: Find the peak +&tic load that &I be,imp&d :-on the screw link. -This vjrlut is- thi %x&&p, s&& .,d load column of Table I (.p 60). It also establishesI& _ of the screw if column loading does not exist. Step 5: *Find the permissible average operating load from the second colpn of Table I. These figures are based on a life of IO* in. of ball-screw travel. By knowing length of the full stroke vou can determine min life of the system in terms of operating cycles (one cycle equals twice the stroke). .4 jacking load fnising and lowering) will give half the life of a friction load (pushing and pulling). With a f&i& load the ball load moves from one sidi: of the groove to the other as the direction of the nut reverses; with a jacking load, the ball load is always on same side of the groove. 03) Step 6: From the table, find the lead of the thread. .; * ., / .., threads per io. (tpi). . .prn of the screw. Step 7: With this value, vpute Stifhess is ah i&uttbd M d&&on of the b& * Rpm equals linear speed, in./min., divided by the lead, in the ball SCT~W.For a ICIU& @@te, spring ties of in inches. Lead is the reciprocal of tpi; for example, preloaded ball nuts can.* from lq tip.to 5(r b/in. In 5 tpi equals 0.2~in. lead. This establishes gear-reduction some IX.& rates of 100 have been obtained inytiy large . . ntio between the motor (or crank) drive and ball screw. dia screws with numerous Media balls. The lead yu be varied to suit an existing gear chive, St&r systcrns will be attained wi* larger quantities although generally it is more economical to vary the of balls because the deflection of balls ie. directly pro. gearing. poitional to the f power of the Gd and inversely proStep 8: Compute the required torque input. This is . portional. to the f power of the ,$xll ,diameter.
(Al
frecl\.. \\Iien the power is relcascd, the load trying to rcwrw the scrw acts through the friction disk and attclnpts to turn the overxnnin, 0 clutch in its locking dircction, and the Iodd is h&l. \\hen tile power is applied to drive the load to the left, it olercomcs the hr.iking
PRODUCT ENGINEERING
l
action of the friction disk-e\.en though the overrunning clutch remains locked. arrangement. In Fig 6(C) is shonn a bidirectional \Vhen the load is applied to the left, the friction disk and overrunning clutch arrangement on the Icft is effectively 63
6.
NO-BACK
DEVICES
prevent
reversal
under load.
. out of the system. The reverse is true when the load is applied to the right. Such devices are limited by their a liity to dissipate heat. The bidirectional arrangement is b used for wing-flap actuators of aircraft. Servo devices also prevent reversibility. One type employs interlehved friction plates, n.hich are locked together by the load and released by the drive motor. A modification of this device has a spring to lock the plates together against backward rotatioh. Its system of bidirectional ramps permits the drive mechanism to rclicve spring pressure and drive the screw against the load or hold it with the load. All of these servo devices require the first :ncrement of drive motion to effect the release. They do not have to dissipate any substantial amount of motor power and have greater thermal storage than the load-sensitive types. On the other hand, the seno devices are generally more com\ plicated in design. r.
Travel-limiting devicei
and the stored energy helps release the ball screw for rcturn travel.
Materials
Most ball screws are either qrburized, or surface-hardened by induction or flame heating. The heat-treatable and precipitation-hardening stainieis steels are now finding widespread application in corrosive environments. .4ll-aluminum units are becoming pop&r in servo svstems \vhere a low mass moment of inertia is desirable and where loads are light.
lubrication
Oil is best hrbricant for ball screws, but a good grade of light greaseis frequently satisfactory. In either case, onh enough lubricant should be applied to maintain a continuous film to assure long life at the rated load. .4void greases containing graphite because an excess builds up and eliminates internal clearances in the unit.
For REPRINT of above article, just check Reader Service cards found in this issue. EDITORS NOTE: to Ball 601 on one of the
Travel of the nut should be limited, if for no other reason than to prevent accidental disassemblv. When travel stops must withstand very high impact. loads, resilient materials, such as nylon or rubber, or springs, can be employed. Helical springs or Bellcvillc-spring .packs sometimes flatten under the impact loads, creating a nonrelcasing condition. Ring springs are better-they dissipate much more energy than other spring devices within the same space, and still release properly. Timed-dog stops, Fig 6(D), often used by designers, operate by missing each other one revolution (before impact), and then making contact entirely in torsion one re\~olution later. Torsional resilience absorbs the impact,
64
linear
devices
see :
basic types oPlinear slides, load caiacities and best mounting methods. For the complete text ol eight ear&r articles presenting
design considerations in the selection and application of ball and roller bearings, order a copy oP the new Wpage combined reprint Rolling Element Bearings from Reader Service Dept. Product Engineering. 330 W (2nd St. NY 36. Price $2.
-Nicholas Chironis
PRODUCT ENGINEERING
FLAP SLAT
RECIRCULATI
ULTIMATE DESIGN LOAD (Ibl) (CURRENT APPLICATION) AIRCRAFT
T6NSll.E
COMPRESSIVE
1CTRIDENT 1E 2E 38 TRIDENT BELFAST BELFAST CARAVELLE BAC l-11 BAC l-11 BAC l-11 BAC l-11 BAC l-11 BAC l-11 TRIDENT lE2E3B 1 I 18
494
501 502 503 518 517 522
I I
I I I
I I I
I
2,940
I
I
563 564 565 562 566 567 568 571 572 577 600 601 v I z 3
NG-BALL
SCREW
JACKS
SHEET
1 OF 2
OVERALL
DIMENSIONS Ins.
DRY WEIGHT
lb.
1.25 1.125 2.00 2.00 1.50 1.00 1.00 1.00 1.00 1.00 1.00 0.875 0.875 1.00
0.16 0.16 0.2450 0.2857 0.0582 0.1765 0.1765 0.1765 0.1765 0.1765 0.1765 0.0926 0.0926 0.1765 0.1765 0.1765 0.1765 0.1785 0.1765 0.1765 0.1765 0.1765 0.1765 0.1765 0.1765 0.1765 0.1765
34.13 8.84 65.62 87.76 27.72 35.85 35.85 35.85 35.85 35.85 35.85 17.35 14.32 35.85 10.04 35.85 35.85 35.85 35.85 35.85 35.85 35.85 35.85 35.85 35.85 35.85 35.85
3.75 x 4.62 x 36.60 3.75 x 4.62 x 12.80 68.06 x 5.00 x 4.72 90.20 x 5.00 x 4.72 30.22 x 10.35 x 6.20 50.25 x 6.00 x 5.75 50.25 x 6.00 x 5.75 48.10 x 6.50 x 6.00 50.25 x 6.00 x 5.75 48.30 x 6.10 x 6.00 48.30 x 6.50 x 6.00 19.95 x 5.10 x 4.76 17.40 x 5.10 x 4.76 48.30 x 6.50 x 6.00 15.55 x 4.70 x 3.38 48.30 x 6.50 x 6.00 48.30 x 6.10 x 8.00 48.30 x 6.10 x 6.00 50.40 x 8.00 x 5.75 48.30 x 6.50 x 6.00 48.30 x 6.10 x 6.00 50.40 x 6.00 x 5.75 48.30 x 6.10 x 6.00 48.30 x 6.10 x 6.00 50.40 x 6.50 x 6.00 50.18 x 6.10 x 5.75 50.18 x 6.10 X 5.75
18 13.25 33.75 40.43 32.75 19.10 22 19.70 19.70 21.20 21.20 7.43 7.12 20.50 10.20 20.20 21.20 19.70 19.70 21.20 19.70 22.00 21.20 21.20 22.20 21.60 21.60
1.oo
1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
m m
FLAP SLAT
RECIRCULATI
ULTIMATE DESIGN LOAD (Ibl) (CURRENT APPLICATION) AIRCRAFT
TENSILE
COMPRESSIVE
603 604 624 654 655 656 657 658 677 700 711 712 713 714 715 716
24.29 24.29 18.10 17.64 17.64 14.26 14.26 30.54 19.90 52.52 52.40 47.83 43.19 39.74 35.32
r-M 9N LU
!i!g sm nz az 1
NG-BALL
SCREW
JACKS
SHEET 2 OF 2
OVERALL
DIMENSIONS Ins.
35.85 1.00 1.00 0.875 0.875 0.875 0.875 0.875 1.00 I 0.1765 0.1765 0.1050 35.85 35.85 34.02 33.05 iii025 0.0826 0.0826 26.74 26.74 38.14
50.18 x 6.10 x 5.75 50.18 48.30 x 6.10 X 5.75 x 6.10 x 5.75 x 4.18 X 4.10 ', \\
21.85 21.85 22.56 9.87 9.70 9.70 9.15 9.15 15.67 28.02 35.78 76.61 55.65 43.26
I
1
37.20
36.23 x 4.18 x 4.10 36.23 x 4.18 X 4.10 29.92 x 4.18 x 4.10 29.92 x 4.18 x 4.10
0.1250
I
I
74.10 x 6.50 x 6.25 76.90 x 6.25 x 6.80 69.14 x 6.50 x 6.25 64.50 x 6.50 x 6.25