Siemens SINUMERIK Backlash Measurement and MD32450 Setting: Step-by-Step Axis Backlash Compensation

This page explains, step by step, how to measure mechanical axis backlash with a dial indicator and how to enter the compensation value in MD32450 on Siemens SINUMERIK controls. It draws on two official Siemens sources: the “Extended Functions” function manual (valid for SINUMERIK 840D sl / 840DE sl / 828D, CNC software 4.4 generation) and the “Machine Data and Interface Signals” parameter manual (valid for SINUMERIK 802D sl V1.4 SP7). The parameter value ranges and data types therefore come from the 802D sl manual, while the functional description comes from the 840D sl / 828D manual; small differences between the two generations are possible. SINUMERIK 840Di sl is not in the validity list of this generation and appears only in older documentation. For SINUMERIK ONE, refer to the manual of your own version. Menu names, parameter access levels and function availability also differ between machine builders, NCU types and software versions.

What This Page Covers — and What It Does Not

This page focuses on one job only: measuring the mechanical play in an axis and compensating it in the SINUMERIK control through MD32450 $MA_BACKLASH.

The following topics are outside this scope:

• Friction compensation and quadrant error (MD32490, MD32500, MD32510, MD32520, MD32540)
• Dynamic feedforward and following error setup (MD32620, MD32630, MD32810)
• Neural QEC learning
• Temperature and sag compensation

These functions can produce the same machining symptoms as backlash, but they are different functions and are commissioned in a different way. Increasing MD32450 simply because a circular contour shows bumps is a common and damaging mistake. Measure the real mechanical play with a dial indicator first.

What Backlash Is and How It Shows Up on the Part

Backlash is the total mechanical play in the motor, coupling, gearbox, ball screw nut, rack and pinion and bearings. When the axis reverses, the motor and encoder start to move, but the table or slide stays still until that play is taken up.

The distinguishing characteristic is this: the error appears only on direction reversal and it is repeatable. Approach the same dimension always from the same side and the part is correct; approach it from the opposite side and it shifts by a constant amount. This single test is the most practical way to separate backlash from friction and servo-related errors.

With an indirect measuring system (motor encoder) the control cannot see the play at all, which is why compensation is needed. With a direct measuring system (linear scale) position is read from the scale, but any play between the scale mounting and the drive still affects the position loop.

Before You Start Measuring

Equipment

• A dial indicator with 0.001 mm graduation. A 0.01 mm dial gauge cannot reliably resolve play below 0.01 mm.
• A rigid magnetic base with a short arm
• A clean, flat contact surface for the indicator tip
• The machine builder’s service or commissioning access level
• A form or notebook to record every reading

Machine condition

1. Warm the machine up by running it for 20–30 minutes. Play measured on a cold machine can differ from play at operating temperature.
2. Slideway and ball screw lubrication must be working.
3. There must be no loose coupling, damaged bearing or obvious mechanical noise on the axis.
4. Reference all axes. Backlash compensation becomes active in the operating modes only after reference point approach.
5. Remove the tool and workpiece so no collision is possible.
6. Move the axis under test to a safe position near mid-stroke.

Backup

Before touching any parameter, archive the NC machine data, PLC data, drive data, tool and work offset tables and user programs. In addition, write down the current MD32450 value separately. If the archive ever fails to restore, that note is the only way back.

Critical Point: Existing Compensation Before Measuring

If a backlash value is already entered on the machine, what you read on the indicator is not the real mechanical play — it is the residual error left after compensation. Miss this distinction and the parameter will be entered wrongly.

Pick one of the two methods below and stay with it from start to finish:

• Method A – measure with compensation zeroed: record the old value, temporarily set MD32450 to 0, activate and re-reference, then measure. This gives the total mechanical play directly and is the preferred method in a commissioning situation.
• Method B – correct via residual error: keep the existing value and add or subtract the measured residual error. This is used for a quick correction between production runs and forces you to work in small increments.

Mixing the two is what goes wrong in practice: if 0.010 mm is already entered and you write the measured 0.004 mm residual straight into the parameter, you have reduced the compensation and made the axis worse.

Step-by-Step Backlash Measurement on the X Axis

1. Mount the indicator

Clamp the magnetic base to a fixed part of the machine and let the indicator tip touch the table, slide or a clamped gauge block that moves with the X axis.

Three details matter:

• The indicator spindle must be as parallel as possible to the measured axis motion. An angled setup introduces cosine error and reports the play smaller than it is.
• Keep the magnetic base arm short. A long arm adds its own deflection to the very play you are trying to measure.
• Preload the indicator so the tip sits compressed near the middle of its range.

2. Approach the measuring point from one direction only

The final approach must be unidirectional so that the play is taken up in that direction.

Move away in X-
Approach the measuring point in X+   (last motion ends in this direction)

3. Zero the indicator and record the position

Set the indicator dial to zero and note the actual axis position shown on the control. From here on, the screen position and the indicator reading are evaluated together.

4. Command a known move in the opposite direction

Move the axis a small, known distance in the reverse direction in incremental mode. 0.100 mm is a practical figure: large enough to take up the play, small enough to stay controlled.

Mode  : JOG INC or incremental move in MDA
Move  : X axis, 0.100 mm in the negative direction
Speed : low jog feedrate

5. Read the difference and calculate

If the control reports 0.100 mm of travel while the indicator shows less real movement, the difference is the mechanical play.

Travel shown by the control : 0.100 mm
Travel shown by the indicator: 0.087 mm

Backlash = 0.100 - 0.087 = 0.013 mm

If the difference is close to zero there is no significant play on that axis, and the error on the part almost certainly comes from somewhere else.

6. Repeat and average

Never enter a parameter from a single reading. Take at least five measurements at the same point with the same approach.

Reading 1 : 0.012 mm
Reading 2 : 0.013 mm
Reading 3 : 0.012 mm
Reading 4 : 0.014 mm
Reading 5 : 0.013 mm

Average = (0.012 + 0.013 + 0.012 + 0.014 + 0.013) / 5 = 0.0128 mm
Starting value ≈ 0.013 mm

If the spread between readings exceeds roughly a quarter of the average, the play is not stable. Go back to the mechanical inspection before entering anything.

7. Measure at several positions along the stroke

Repeat the same measurement near the start, the middle and the end of the travel. The aim is to see whether the play is constant along the axis.

The Most Common Measurement Mistakes

Entering the Value in MD32450 $MA_BACKLASH

The base machine data for backlash compensation is axis/spindle specific and is entered separately for each axis:

MD32450  $MA_BACKLASH

The data type is DOUBLE and the factory default is 0.0. With a value of 0 no compensation is applied. The unit is mm on linear axes (or inch in an inch system) and degrees on rotary axes and spindles; the parameter is interpreted in whatever input unit is valid for that axis.

Positive or negative?

Siemens defines the sign by the position of the measuring system relative to the machine part:

• Positive backlash (the normal case): the encoder leads the machine part, so the axis actually travels less than commanded. Enter a positive value. This is the typical situation on axes with an indirect measuring system.
• Negative backlash: the encoder lags behind the machine part. Enter a negative value.

Do not guess the sign. A wrong sign does not reduce the error — it doubles it. If you are unsure, try half the measured value first and repeat the measurement: if the error shrinks, the sign is right.

Sequence on the operator panel

The flow below is generic for SINUMERIK Operate and HMI Advanced. On a machine builder’s custom screens the headings may be named differently.

1. Switch to the service / commissioning access level (builder or service password).
2. Open Startup / Commissioning → Machine Data → Axis Machine Data.
3. Select the correct axis. X, Y, Z and A/B/C are entered individually; skip the axis selection and the value lands on the wrong axis.
4. Find the parameter by typing 32450 or $MA_BACKLASH into the search field.
5. Read the current value and write it down; take a screenshot if possible.
6. Enter the new value. For a measurement of 0.013 mm with a positive sign: MD32450 = 0.013
7. Activate the change and re-reference the axis.

Activating the change

The parameter manual lists the activation of MD32450 as NEW CONF, meaning the new configuration must be taken over after the machine data is confirmed. Depending on the control version and the builder’s configuration, one or more of the following may be required:

• Confirming the machine data / pressing NEW CONF
• NCK Reset
• Power On Reset (restarting the control)

After activation the axis must be referenced again. Backlash compensation only takes effect once reference point approach is complete, so a verification measurement on an unreferenced axis is misleading. This is by far the most common reason a correct setting appears not to work.

If There Is a Second Measuring System: MD30200 $MA_NUM_ENCS

Some axes have two measuring systems, for example a motor encoder together with a linear scale. The number of measuring systems on an axis is read from:

MD30200  $MA_NUM_ENCS      ; number of measuring systems on the axis

With two measuring systems, the backlash value must be defined separately for each one. The two systems are attached at different points in the mechanical chain and therefore see different amounts of play: the motor encoder sees the whole drive train, while the linear scale only sees what lies beyond its mounting point.

When the measuring system is switched over, the corresponding compensation value becomes active automatically. So if only the first index is filled in and the second is left empty, the axis is suddenly left uncompensated at the changeover. Measure each system separately as well — never copy the value from one to the other.

When Is MD32452 $MA_BACKLASH_FACTOR Used?

MD32452  $MA_BACKLASH_FACTOR
Value range : 0.01 – 100.0
Default     : 1.0

This parameter weights the backlash defined in MD32450 according to the parameter set. The effective compensation is approximately MD32450 × MD32452.

MD32450 = 0.010 mm , MD32452 = 1.0  →  effective ≈ 0.010 mm
MD32450 = 0.010 mm , MD32452 = 1.2  →  effective ≈ 0.012 mm

The typical application is a spindle or rotary axis whose gear stages show different amounts of play, each stage running with its own parameter set.

On a standard X, Y or Z axis this parameter is left alone. Unless the machine builder has a specific application for it, keep it at 1.0. Adjusting through MD32450 alone is always easier to trace than moving two parameters at once. Do not assume this function is supported identically on every control variant.

Applying the Compensation Step by Step: MD36500

MD36500  $MA_ENC_CHANGE_TOL

The primary definition of this machine data is the maximum tolerance permitted when the actual position value is switched between measuring systems. In the backlash context it takes on a second role: it determines that the compensation value is applied over several cycles rather than in one go.

Applying the full amount of play in a single cycle at reversal produces an abrupt step in the axis setpoint. Applying it gradually smooths that out. The number of cycles is approximately:

Number of cycles ≈ MD32450 / MD36500

If MD36500 is larger than MD32450, the compensation is applied in a single cycle. An excessively small MD36500 spreads the application over too long a time and can trigger a standstill / zero speed monitoring alarm.

This parameter is not touched during an ordinary backlash correction. Because it is also the measuring-system changeover tolerance, arbitrarily raising or lowering it changes encoder difference monitoring behaviour as well. Any change should be evaluated together with the servo and monitoring settings by a Siemens commissioning engineer.

How to Calculate the New Value When One Already Exists

This is the point most often confused in the field. Writing the measured residual error into the parameter is wrong; the correct action is to add it to the existing value when the direction confirms it.

Existing value    : MD32450 = 0.010 mm
Measured residual : 0.004 mm  (same direction, i.e. still under-compensated)

WRONG   : MD32450 = 0.004
CORRECT : MD32450 = 0.010 + 0.004 = 0.014 mm

If the residual error is in the opposite direction — the axis is now over-compensated — the value is reduced by the same logic:

Existing value    : MD32450 = 0.020 mm
Measured residual : 0.005 mm in the opposite direction (over-compensation)

New value         : 0.020 - 0.005 = 0.015 mm

The sequence to follow:

1. Record the old MD32450 value.
2. Determine both the size and the direction of the remaining error.
3. Change in small increments; never enter a large value in one step.
4. After every change, activate, re-reference and measure again.
5. Stop when the improvement stops — what remains is probably not backlash.

Verification Checklist After the Adjustment

1. Re-measure with the indicator

Repeat the measurement with the same indicator setup and the same approach direction.

Play before adjustment    : 0.013 mm
Residual after adjustment : 0.001 – 0.003 mm  (expected)

2. Repeatability

Approach the same point from the same direction at least five times. If the results are not consistent, the problem is not constant play and compensation will not solve it.

3. Different positions

Check the start, middle and end of the stroke again. An axis that is correct in the middle but off at the ends is mechanically worn.

4. Low and high feedrate test

First watch the reversal at a low jog feedrate; the axis should change direction without sticking or jumping. Then, at a safe distance, try a reversal at a higher feedrate. If you get a hard knock, a sudden jump or an alarm, take the value back.

5. Machining test on a milling machine

• Mill a square pocket and compare opposing wall dimensions in both directions.
• Mill a circular pocket and an external circle; measure the diameter in two perpendicular directions.
• Mill a contour with reversing directions and check for reversal marks.

6. Machining test on a lathe

• Approach the same diameter in X from two different directions and compare.
• Measure a groove width.
• Check shoulder lengths in Z from both directions.
• Machine the same diameter repeatedly and observe stability.

If the Result Got Worse, What Should You Check?

When Not to Increase Backlash — Repair the Mechanics Instead

In the situations below, do not try to buy time by enlarging the electronic compensation. Compensation does not repair a mechanical fault; it only corrects small and repeatable positional differences electronically.

• Loose or cracked motor coupling
• Worn ball screw nut or a screw that has lost its preload
• Damaged fixed bearing or a bearing with axial play
• Loose motor mounting bolts
• Worn rack and pinion drive
• Misadjusted gib or a worn linear guide carriage
• Loose linear scale mounting or read head
• Play that keeps changing along the axis
• A different value at every measurement
• An audible metallic knock or impact at reversal

A practical limit: if the measured play is clearly above the figure recorded when the same machine was commissioned, the problem is wear, not setup.

Backlash or Friction? A Quick Distinction

Do not raise MD32450 the moment you see four marks on a circle. Measure the mechanical play with an indicator first; if there is no play, the subject is friction compensation and servo optimisation.

Variant-Dependent Support on the 802D sl

In the machine data tables of the SINUMERIK 802D sl parameter manual, some rows carry system variant abbreviations. The presence of a parameter in the manual does not mean it is active in the same way on every 802D sl variant.

Entries such as 0/0 or 1/1 in those tables indicate different defaults or limits per variant. Before entering a parameter, check whether your own NCU, software version and function package actually support it.

Parameter Summary

Summary Procedure

1. Archive the parameters and the system; note the current MD32450 value separately.
2. Bring the machine to operating temperature and check the mechanical condition.
3. Reference the axis.
4. Decide on the measurement method (zeroed compensation or residual error).
5. Measure the play with a dial indicator: five repeats, three positions.
6. Enter the new value on the correct axis and the correct measuring system index.
7. Carry out the required confirmation / reset.
8. Reference the axis again.
9. Re-measure with the indicator and record the residual error.
10. Test at low and high feedrate.
11. Verify with a machining test and measuring equipment, and document the result.

Safety Warning

MD32450 intervenes directly in the actual position value of the axis. A wrong sign or an excessive value can cause axis jumps, dimensional errors, contour degradation, servo monitoring alarms and tool or workpiece collisions.

• Always take a backup and record the old value before making a change.
• Remove the tool and workpiece from the working area.
• Run the first tests at low feedrate and low rapid override.
• Work in small increments; never enter a large value in one step.
• Unauthorised users must not modify service-level parameters.

Disclaimer

This content was prepared to explain how backlash measurement and MD32450 adjustment work on Siemens SINUMERIK systems. All numerical values shown are examples, not ready-made figures to be entered on any machine. Machine data, encoder and compensation settings must not be changed without proper backups and without reviewing the machine builder’s procedure. The same parameters can behave differently on different SINUMERIK models and software versions; the official manuals of the machine builder and of Siemens always take precedence. Mentor CNC cannot be held responsible for material damage, dimensional errors or workplace accidents resulting from this application.