instr'1ang

background image

Zbigniew Wasiak, Tadeusz Wojciechowski

Ć

w i c z e n i e nr 1

CHECKING MACHINE TOOL GEOMETRIC ACCURACY BASED ON LATHES

One of the conditions to obtain high accuracy and dimensions-shape repeatability of items made by ma-

chining is to preserve relevant geometric accuracy of machine tool. The purpose of this exercise is to learn

the methods and ways of checking geometric accuracy of machine tool, in practical part selected meas-

urements will be carried out for universal central lathe.

1. Introduction

The accuracy of each machine, and hence the machine tools for metals, is determined by:

- Geometric accuracy which is meant a dimension-shape mistakes and mistakes of relative positions of

components and machine assemblies.

- Kinematic accuracy, determined by the accuracy of kinematic couplings (joints),

- Accuracy of the adjustment, the specified precision mechanisms for traversing the dimensional setting

machine

- Precision machining.

This manual is devoted to only check the geometric accuracy of technological machine from a group of

machine tools. This check is covered by the Polish Standard PN-93/M-55580/01 title: " Dokładność geo-

metryczna obrabiarek pracujących bez obciążenia lub w warunkach obróbki wykańczającej " based on

ISO/DIS-230-1.

Checking geometric concerns the size, shape and position of the components of the machine and their

mutual displacement (surface flatness, compliance and intersecting axes, parallel and perpendicular of

straight lines, flat surfaces with respect to the surface plane or any other). It relates only to the dimensions,

shapes and relative movements which can affect the accuracy of the machine and machining. Values of

checked points must be between fix bounds, depending on the class of machine accuracy, as the elements

included in the machine tool components and assemblies are made of dimensionally-shaping tolerance

specified by the constructor. It requires that used gauges were at least a class more accurate than the

measured values.

Sketch nr. 1 shows basic quantities referring to measurements during geometrical precision validation pro-

cess of shaping machine.

background image

On vertical plane

perpendicularism

Shape deviation

On horizontal plane

surface flatness

Between two planes

parallelism

two axes or concentrici-

ty of two axes

Validation

of

Plane and axis

Coordinative
deviation

Two planes

perpendicularism

Two axes

Axis and plane

Radial

Spin imbalance

axial

Frontal plane

Distance to plane

displacement

perpendicular

Distance to axis

Distance Between

curves

Rys.1. Basic quantities concern-

ing measurements during the
process of geometrical preci-
sion validation of shaping ma-
chine.

Plane and curve

parallel

Axis and curve

Distance between

curves

2. Preliminary checking operations

- Before executing shaping machine examination, it is necesarry to place it on proper foundation and lev-

el it. By doing that machine gets its stability, which is helpful during examination,

- examination should be executed on machine Fuldy prepared to work. Disassembly of particular parts is

allowed Only special circumstances and With thorough respekt to producer instructions.,

- during geometrical examination each part (eg. Spindles) should be in the state of regular work parame-

ters (temperature should be as referred by producer instruction) ,

background image

- geometrical examination should be exercised on idle run Or clogged machine.

3. Methods of examination

During geometrical precision examination several metods are allowed. A few of them are described be-

low on sketch nr. 1.

3.1. Line measurement process.

It refers to shaping tool's devetal ways, on which tables, sleighes or supports are being moved while

work is on. Work piece or tool are stiffly constrained with these supports, so line measurement deviations

of devetal ways are strictly related to dimensions-shape repeatability of work pieces.

Sprawdzanie to jest tematem odrębnego ćwiczenia i nie będzie tu bliżej omawiane.

3.2. Surface flatness validation.

Surface is considered as flat, when each point resides between two planes parallel to the surface and dis-

tance between planes is within flatness tolerance.

Flatness validation may be done by: estimation plate, estimation plate and indicator, ruler and size

block, ruler and strict level and indicator, strict level and also optical methodes (with autocollimator, opti-

cal angle bar and laser). Such validation is not a part of today's exercise.

3.3. Validation of planes' and lines' parallelism deviations.

Two lines are parallel to each other, when the distance between them is within parallelism tolerance on

any given normal plane.

Two planes are considered paralell, when distances betw themeen measured in at least two planes per-

pendicular to each other and to planes are equal (within parallelism tolerance).

Max. parallelism deviation is different between maximal and minimal taken result during the examina-

tion.

Differences are measured on given planes' (vertical, horizontal, perpendicular to examined plane, cut-

ting examined axis etc.) distance (e.g. on the sector of 300mm or whole plane).

During axes paralellism validation, these axes should be represented by oval planes with high accuratici-

ty of shape, relevant smoothness and sufficient lenght. If it is an interior plane, on which sensor can't be

placed, it should be taken from extra cyrindrical surface as a test tongue.

Mount and centering of test tongue is being done on its tapered end, in cylindrical or tapered hole sub-

jected to setting tool and other tools. Usually it is not possible to set-back of tongue in machining spindle

axis in a way, so it represents pivotal axis. While machining spindle rotates tongue axis describe hyperbo-

loid or tapered surface. On measured plane we can observe two extreme tongue positions B-B’ (sketch nr.

2).

Parallelism validation (when using tongue) may be done in

any given

position of machining spindle, but measure has to be repeated

after

ro-

tating machining spindle by 180

o

. Parallelism inaccuracy on

given

B'

B

A

Rys.2. Setting a tongue in an

average A position
sketch.

background image

Wspornik

Plaszczyzna odniesienia

prowadzacy

Linial

Rys.3. Test of parallelism between

two axes using ruler and indica-
tor.

plane is being set as artmetic average of both results.

Tongue may be set to middle position A. In this case, validation should be exercised only in given posi-

tion. „average whipping position” of tongue is being set by observing gauge of mesurement device, while

machining spindle is rotating slowly and there is contact between contact of indicator end with oval plane

referring to pivot axis on measured plane. Machining spindle is in middle position, if indicator’s pointer is

in middle position between its two extreme positions.

3.3.1. Validation of two planes' parallelism.

It should be performed on two planes perpendicular to each other and to given planes. There are two

methods of validation:

- with ruler and indicator

- with a strict level

In the first one dial indicator (sketch nr. 3) is constrained to console on a flat base. Is being moved tan-

gentially to ruler and moved along plane through given distance, while its test end is moving along the sec-

ond plane. In strict level method,a level is set on console connecting two comparable planes. Console is

to be moved along planes with level bound to it and get results from level. Maximal difference of results

(angle) determines angular parallelism deviation, reading (angle) multiplied by length "l" determines linear

parallelism

deviation

(sketch nr.

4).

3.3.2. Vali-

dation

of

two

axes

parallelism.

It is

done by

three methods:

- on plane by both axes

- on relative plane parallel to plane by both axes

- with a level set on horizontal plane and extra tools

3.3.2.1. In parallelism examination on plane by both axes measurement

tool

is

constrained to console allowing to move along oval shape representing

one

of

the axes test end is moving along oval shape representing the other axis.

Plytka

prowadzaca

Wspornik

l

Rys.4. Test of parallelism between two axes

using a strict level.

S. 5. Validation of

axis parallelism
on plane by both
axes.

background image

Tool has to be leaned slightly in the direction perpendicular to both axes (sketch nr. 5).

3.3.2.2. Validation may be done with extra relative plane. as much parallel to plane going through both ax-

es as possible. Each surface parallelism has to be taken into consideration separately and set it in rate to

this surface in a way described in 3.3.4.

3.3.2.3. Validation with a level set on horizontal plane and extra tools is being done in case given planes

are not both horizontal. When dip angles are narrow, extra block is used (sketch nr. 6), but when they are

wider constant angle bar or that one you can change is used (sketch nr. 7). Block or angle bar is set on two

ovals representing axes . Air bubble of a level has to be set on 0 by rotating correct handwheel.

A level with a block angle bar is moved along axis by given distance and reading of results are done.

przesuwa się Measure is related to distance between two axes. If, for instance, distance is 300mm, and

odczyt

poziom-

nicy

shows

0,06

mm/1000

mm, then

parallel-

ism devi-

ation

equals 0,018mm.

3.3.3. Validation of two axes parallelism to relative planes.

Measurements of parallelism are in fact parallelism measurements (3.3.4). parallelism referres to dis-

tance between axes and relative plane. Parallelism occurs, when plane containing axes is parallel to relative

plane. At first, measurement of both axes' parallelism to plane has to be done. Then, using the same dial

sensor

as

during

measure-

ments

of

ovals rep-

resenting

both axes,

the

dis-

tance between these axes and plane is measured. (Sketch nr. 8).

Pokretlo

Klocek

Sketch nr. 6.

Katownik

Pokretlo

Sketch nr. 7. Validation of two

axes parallelism with con-
stant angle bar.


Rys.8. Test of parallelism of

A and B axis with
reference surface
sketch.

A

B

A

B

background image

In case ovals are not identical, radial variation has to be considered.

3.3.3. Validation two axes' concentricity

3.3.3.1. Definition.

Two lines or axes are considered concentricity (which are covered by each other), when distance be-

tween them measured in several spots on given test part does not exceed particular value, referring to con-

centricity tolerance. Distance may be measured between lines themselves or between their prolongation.

Centricity inaccuracy of one axis in relation to the other is to be subjected to length of measured part. Un-

der special circumstances, extra information has to be added.

3.3.3.2. Measurement method.

Measure device is constrained to an arm and rotates full circles around axis „1”. End of indicator of

measure device moves along given A cut along oval ambit representing second axis (sketch nr. 9). Be-

cause of the

fact, that in

measured

cut,

both

axes

may

interfere

with

each

other, measurement has to be repeated in second B cut. Dis-

tance

between cuts should be as long as possible. If inaccuracy has

to

be

examined in both given planes (e.g. H and V on sketch nr. 9),

devia-

tions have to be written down separately.

In case one of the axes is pivotal axis, which handles meas-

ure de-

vice imbed on tongue reprezenting pivotal axis.

If it is necessary for measure device to rotate about fixed

oval, it

should be imbed on a ring rotating with the smallest possible

back-

lash (sketch nr. 10).

If both axes are pivotal, measured oval may be reduced to

its

whipping average position on measured plane. (3.3).

3.3.4. Validation of parallelism axis and plane.

Measure device is constrained on console with flat base and

moves

along

plane by

given vec-



Rys.9. Sprawdzanie

współosiowości
dwóch osi 1 i 2

H

V

A

B

1

2

Rys.10. Validation with device

rotating about fixed axis

Rys.11. Validation of parallelism

plane and axis.

A

B

A

B

Rys.12. Validation of

parallelism of rotat-
ing axis and plane.

background image

tor of test end is moved along oval representing axis. (sketch nr. 11).

In each measurement spot the smallest distance is measured by moving measure device along vector

perpendicular to axis.

In case axis rotates, only setting the oval representing axis to middle position is required. Measure has to

be taken in two extreme positions. „A” and „B” (sketch nr.12).

3.4. Validation of perpendicular deviations.

Two planes, two straight lines or straight line and plane are perpendicular to each other when perpen-

dicular deviation of one in relation to the standard angle bar (representing right angle),set on the other does

not exceed maximum allowed perpendicular tolerance deviation value. Validation of perpendicular devia-

tion comes to measurement of parallelism.

3.4.1. Validation of two planes' perpendicularism.

Cylindrical angle bar is set on one of the planes „2”

(sketch nr.13). Dial sensor is moved along the other plane,

get-

ting readings in equal parts of time. Then, cylindrical angle

bar

rotates half circle and the second set of reading is made.

From this two sets of readings are emerges average im-

age

of deviations.

3.4.2. Verification of perpendicularism of two axes.

Two cases are considered:

- both axes are fixed,

- one of the axes is pivotal.

3.4.2.1. both axes are fixed.

On oval representing one of the axes e.g. „1” an-

gle

bar with special base plate is set (sketch nr. 14). Parallel-

ism

of free arm of this angle bar and the other of the axes

„2”

is validated with methods of parallelism validation. (3.3.4).

3.4.2.2. One of the axes is pivotal.

Dial sensor is set on arm constrained to the tongue representing pivotal axis , its end touches the oval

representing the other axis on two spots A and B (sketch nr. 15). Reading differences are expressed in rela-

tion to distance AB.

1

Katownik
wzorcowy

2

sketch nr.13. Validation of planes'

perpendicularism with cylin-
drical angle bar and sensor

1

Katownik

2

Rys.14. Validation of perpendicularism

of two fixed axes „1” i „2” with
an angle bar and sensor.

background image

If

the

other axis

is pivotal

as

well,

the

oval

represent-

ing it is

set to an

average

position of

whippng on measurement plane, using method described in parallelism validation paragraph (3.3).

3.4.3. Validation of perpendicularism of axis and plane.

Two cases are examined:

- axis is fixed,

- axis is pivotal.

3.4.3.1. Fixed axis

An angle bar with special arm of base plate is set to touch the oval representing axis. (sketch nr.16).

Parallelism of free arm of an angle bar and plane is examined in two directions perpendicular to each other

with the use of method described in paragraph 3.3.1.

3.4.3.2. Axis is pivotal.

An arm which holds a dial sensor is constrained to

machin-

ing spindle „1” (sketch nr.17) and end of indicator is set

as paral-

lel to its pivotal axis. During machining spindle rotation,

sensor

makes a circle, which plane is perpendicular to pivotal

axis.

Validation of perpendicularism of axis and plane „2”

with us-

age of end of dial sensor, comes to examining parallel

deviation

between ambit plane, which is drawn by end of indicator,

and giv-

en plane.

Such deviation is clear as a ratio to diameter of circle

drawn by

indicator during its rotation.

If measurement plane is not given, dial sensor rotates

full cir-

cle and registers the biggest variation of indicator’s read-

ing.

If measurement planes are given (e.g. planes I i II), the variation for two positions of sensor placed be-

tweeen 180 degree angle is registered for each one of the planes.

Rys.15 Validation of perpen-

dicularism of two ax-
es, when at least one
is pivotal.


B

A

Rys.16. Validation of perpen-

dicularism fixed axis
and plane.

Sketch nr.17. ScValidation perpen-

dicular deviation of pivotal
axis and plane.

Plaszczyzna I

180°

Plaszczyzna II

18

1

2

background image

This method of measurement is used for validation of perpendicularism of axis of machining spindle of

vertical mill to suraface of it’s table.

3.5. Validation of whipping

There are three types of whipping:

- radial whipping,

- axial whipping,

- front surface whipping.

3.5.1. Radial whipping

In case geometrical axis does not covered with

pivotal

axis (concentricity mistakes), (distance between axes

on plane

normal to pivotal axis on given spot is called radial

whipping

of axis (sketch nr.18).

If ovality is not considered, we get doubled value of radial whipping of axes on given sector is simply

called whipping (sketch nr. 18).

Generally measured whipping is an outcome of:

- radial whipping of axis

- ovality of diameter,

- inaccurateness of bearings.

3.5.1.1. Whipping of exterior surface

End of dial sensor is set to touch rotating plane, which is being under examination and reading of meas-

ure device during slow work of machining spindle during rotation by 360 deg. are being observed (sketch

nr. 19a). Variation of extreme positions of sensor during this process is a quantity of whipping.

On tapered surface the end of sensor is set perpendicularly to forming line. If during rotation of machin-

ing spindle

any

axial

movement

ocurrs, di-

ameter of

examined

circle will

change. Therefore, whipping on tapered surface is possible to be measured only when concurrence of cone

is not very significant. Each time before examination, axial translation of machining spindle should be

measured (sketch nr.19b), and then an influence of it on measurement should be calculated according to

concurrence angle.

3.5.1.2. Whipping of intrinsic surface

"O" - os geometryczna

O

"w" - os obrotu

w

promieniowe

Bicie

Srodek
obrotu

Bicie

sketch nr.18. Definition of „whipping”

and „radial whipping”

a

b



Rys.19. Measurement of whip-

ping of exterior surface a)
cylindrical, b) tapered

background image

If in the cylindrical or tapered hole it is not possible to measure sensor directly, a measure tongue is

used.

If measurements are being undertaken in only one cut of a tongue, position of just one measure circle in

relation

to

axis is taken

under con-

sideration.

Because

tongue axis

may

inter-

fere with pivotal axis on measured plane, examination should be done in two separate cuts A i B along giv-

en stand-off (sketch nr. 20).

For expamle one measure should be takan near housing of a tongue (A), and the second in specyfic

lenght from (B). To eliminate mistake of set-back of a tongue in hole, particularly by tepered holes,

measures should be repeated 4 times, rotating tongue by quarter a circle each time in relation to machining

spindle. As an outcome we apply arthmetic average of these measures on both planes. Each time whippnig

should be measured vertically (position C

1

), and then horizontally (position C

2

on sketch nr. 20).

3.5.2. Axial whipping

It is a range of plane and reflexive movement along rotating full-circle axis of given part after erasing

axial backlash by adding axial force in particular direction (rys. 21). Minimal axial backlash is the smalles

value of possible translation of rotating part, measured during idle time on several spots in relation to piv-

otal axis. (rys. 21).

For purpose of eliminating the influence of resistance bearings backlash, we have to put a small value

force

to

machining

spindle

along

measure-

ment

vec-

tor. End of

dial sensor should be touching the middle of front surface and set as precise as possible along pivotal axis.

Readings are done during constant rotation of machining spindle with small speed, constantly maintaining

required clamp.

If machining spindle is cored, a short tongue should be entrench in it, with front surface normal to axis,

against


Rys.20. Validation of whipping of

intrnsic surface with a
measure tongue set in the
hole.

C1

A

C1

C1

B

C2

J

P

J

j

d

j

P


Sketch nr. 21. Description of axial

backlash by measurement of
axial whipping.

J – maximal axial backlash,
j – minimal axial backlach,
d – periodic axial movement

a

b

Rys.22. Measure of axial whipping of

cored machining spindle with: a)
short tongue and indicator, b) cir-
cle end tongue and flat indicator

background image

which circle end of dial sensor may be based (sketch nr. 22a). A tongue with globe end may be used as

an alternative, end of dial sensor would be flat then. (sketch nr. 22b).

If machining spindle has a center hole,steel globe which has contact with flat end of dial sensor, may be

placed inside it (sketch nr. 23).

Axial whipping is determined as boundery axial translation during its full circle slow rotation with

clamp of small axial force.

3.5.3. Front surface whipping

Flat surface whipping rotating around axis is axial

whipping

of this surface. That whipping is a disadventage of flat

surface,

which does not remain normal to axis during pivot.

Whipping

is determined by distance between two planes normal to

axis,

which are set as boundaries for moving spots during its

rotation.

It is a result of several surface disadvantage and pivotal

axis:

- surface is not flat,

- surface and pivotal axis are not perpendicular,

- periodic axial movement occurs.

Because front whipping has a tendency to gain magnitude during going away from pivotal axis, meas-

urements should be exercised on ambit related to exterior spots.

During front whipping validation dial sensor

should

be set in particular distance A to center and perpen-

dicular-

ly to front surface (sketch nr. 24), then make machin-

ing

spindle rotate slowly and write down extreme indica-

tor’s

reading. Variation of these results is a quantity of

front

surface whipping. During rotation small siłę osiową

should

be put to eliminate influence of resistance bearings

back-

lash.

3.6. Validation of translation parallelism.

Translation parallelism refers to position of path of rotating machine part in relation to:

- plane (support or devetail ways),

- straight line (axis, edge of surface cross),

- trajectory of other moving machine part point.

Methods of examination are identical to these used for examination of line and plane parallelism

Measure device, if it is to be moved, should always be constrained to moving part, which should be

powered in natural way, to reveal real influence of backlash and iraccurateness of devetail ways.

Rys.23. Examination of axial

whipping using marble
placed in center hole

czujnika zegarowego

Rozne polozenia

A

1

A

osiowa

A

2

Sila

A

3

A

4

Rys.24. Validation of front whipping

of flat surface, rotating about
axis

background image

3.6.1. Validation of parallelism between trajectory and

plane.

Two cases are considered:

- plane is placed on mobile part,

- plane is placed on fixed part.

In case the first dial sensor is constrained to fixed part and its

end

lies on measured surface under right angle. Mobile part should

be

translated along given vector (sketch nr. 25), and then variation

of re-

sults indicator should be read on the lenght of this sector.

This method is used mainly for milling machine and grind-

ers,

where work piece is constrained to machining table. Dial

sensor

is set on fixed end of machining spindle as shown on

sketch

nr. 25, and table is an moving object. This is the way of

exam-

ination whether table plane is parallel to vector of its

movement.

In case the plane lies on fixed part (e.g. table of radial

drill-

ing machine) measured piece is constrained to its mobile

part

and moved with it by given vector. End of indicator

touch-

es the plane perpendicularly and moves across it. (sketch

nr.

26).

3.6.2. Validation of parallelism trajectory and axis.

Przyrząd pomiarowy is constrained to części

ruchomej and moves along given vector. End of indica-

tor

moves along a tongue representing axis (sketch nr. 27).

In

a

place where pivotal axis is, a tongue which represents it

should

be set in „place of average whipping” (3.3 - sketch nr.

2). If

each plane is the same to all other, measurement should

be

done on two planes perpendicular to each other.

3.6.3. Validation of parallelism between two trajecto-

ries.

maszyny

Stol

Listwa

Rys.25. Validation of parallel-

ism of table to trajectory
of it’s move

Rys.26. Validation of parallelism of

trajectory of movement of
radial drill spindle totable
axis.

Rys.27. Validation of parallelism tra-

jectory and axis.

background image

Dial sensor is set on one of mobile parts,

so its end touches the second mo-

bile

part in given spot. Both parts are

being

moved together, along same vec-

tor,

reading of measure device are be-

ing

observed during process (sketch

nr.

28). In case when parallelism of

trajec-

tories on two given planes is sig-

nifi-

cant (e.g. horizontal and vertical), measures should be done on both planes.

3.7. Validation of translation perpendicularism

Translation perpendicularism for machining tools refers to subsequent positions of particular spot of

mobile part of machine in relation to:

- plane (base or devetail ways),

- straight line (axis or edge of two planes' crossing),

- path of other mobile part spot.

In this process, previously used methods subjected to translaton parallelism measurements are applied.

4. Progress of the task

Before executing this exercise student should:

- check, whether machining tool is in heat-set state,

- plug it out of electric network,

- get know with machining tool handling procedure required to execute task.

- get know with measure devices required to execute task.

- clean surface of contact between machining tool and measure device.

Execute practice part of the task, examining deviations of quantities assigned to protocol.

5. Compilation of results.

Results calculated during task should be written into protocol of geometrical acuteness. Upon calculated

results, form wrap-up, which should consist of:

- picking observed geometrical deviations exceeding admissible deviations,

- Sum up its influence on dimention-shape acuteness of made object.

Rys.28. Validation of

parallelism be-
tween two trajec-
tories.

background image

Wroclaw University of Technology

Name, Surname:

Institute of Machine Technology and Automation Study:

Year: Group

Exercise data:


Exercise 1

CONTROL OF MACHINE TOOL GEOMETRY

background image

6. Measurement protokol

LATHE TYPE . . . . . . . . . . . . Span between centers (DC) . . . . . . . . . . . . . . Maximal diameter over the bed D

a

-. . . . . . . . . .Factory number . . . . . . . . . . . Production date

………………….

No

Sketch

Measured value

Measure tools

Deviations [mm]

Measurement description

max.

measured


1

Spindle

center

pulsation.

Sensor with
sensitivity
0.001 mm

0.015

Dead center mount inside the spindle. End of sensor
mount perpendicular to cone surface. Rotate spindle by
hand with little speed and read maximal and minimal
values. Measured deviation divide by cosα. α is a half
of cone center angle value.


2

Spindle face pul-
sation.

Sensor with
sensitivity
0.001 mm

0.02

Put end of sensor to spindle face in perpendicular di-
rection. Rotate spindle by hand with low speed without
stop. Read indications.


3

End of spindle
outside

center

surface

pulsa-

tion.

Sensor with
sensitivity
0.001 mm

0.01

Put end of sensor to end of spindle outside center sur-
face perpendicular. Rotate the spindle constantly with
little rotation speed. Read maximal end minimal val-
ues.


4

Spindle

inside

surface

radial

pulsation.
a) near a spindle
face
b) in a distance
D

a

/2

but

not

more

than

300mm

Measure
mandrel,
sensor with
sensitivity
0.001 mm

a) 0.01
b) 0.02

Put measure mandrel inside the spindle cone. Put end
of sensor to mandrel surface perpendicular. Rotate
spindle by 2

π

angle at least and read maximal and

minimal values. Measurement should be done in two
crosses places: a) near the spindle face and b) at the
end. Measurement replace four times with spindle po-
sition changing on

π

/2 angle. Final deviation is a mean

value of four measured deviations in two crosses.
Measurement do in two perpendicular directions (ver-
tical and horizontal).

No

Sketch

Measured value

Measure

tools

Deviations [mm]

Measurement description1

max.

measu-

red.

background image


5

b)

a)

Control of parallel
spindle axe to par-
allel

cross-way

move a) vertical
and b) horizontal.

Measurement
mandrel, dial
sensor 0.001
mm sensitivi-
ty.

a)
0.015/300
in vertical
direction
b) 0.02/300
in parallel
direction

Fix the mandrel into spindle cone. Fix the dial sensor in
stand on the cross-way. Put end of sensor to mandrel
surface in vertical direction. Move cross-way parallel to
spindle axis and observe the dial sensor indications.
Rotate the spindle about

π

angle and repeat measure-

ment. Describe parallel deviation as mean value of two
measured deviations. Repeat measurement in horizontal
direction.


6

Height of axes
Difference
Sprawdzenie
różnicy wysoko-
ś

ci osi wrzecio-

na i osi konika

Measure
mandrel,
sensor with
sensitivity
0.001 mm

0.04 horse
axe higher
than spin-
dle axe

Tail stock and his bush blocked as in regular work.
Mount mandrel inside the tail stock bush. Mount sensor
to cross way. Put end of sensor to mandrel surface and
read measured values on the mandrel ends.


7. Conclusions.

background image

Ad.4

Vertical surface

Horizontal surface

A

B

A

B

0

min

Deviations

min Deviations

min Deviations

min Deviations

max

max

max

max

2

π

min

Deviations

min Deviations

min Deviations

min Deviations

max

max

max

max

π

min

Deviations

min Deviations

min Deviations

min Deviations

max

max

max

max

π

2

3

min

Deviations

min Deviations

min deviations

min Deviations

max

max

max

max

background image

4

Ad 5.

Vertical surface

Horizontal surface

0

A Deviations



………………….

A Deviations



……………….

B

B

π

A Deviations



……………………

A Deviations



………………….

B

B

Average deviation
…………………….


Wyszukiwarka

Podobne podstrony:
INSTR KLASYF DLUZNE

więcej podobnych podstron