5
Troubleshooting Techniques
The time is gone when machines, controls, and drives were ‘‘add-ons,’’ as
shown in Figure 1.
Early machines used add-on drives. These drives had one thing in
common: There was no interaction among the control, drive, and machine.
To a large extent these machines were open loop.
Today’s numerically controlled machines are tightly integrated
machine systems. There is a great deal of interaction between the machine,
drive, and control. It is this interaction that results in problems with
stability, surface finish on the work part, and accuracy. The actual block
diagram for today’s machine systems looks more like the block diagram
shown in Figure 2.
5.1 TECHNIQUES BY DRIVE
The primary difference between the machine drives of yesteryear and those
of today lies in the number and type of feedback signals that occur with
modern control systems. It is true that early drives had load forces fed back
to the drive, and these forces could cause a droop in the feed rate with load
(called load regulation). Nonlinearities (stiction, lost motion, etc.) would not
affect the stability because they were outside the servo loop, but they could
cause inaccuracies. By and large, drives were ‘‘hooked’’ onto a machine and
in spite of resonances in the mechanical drive, stick-slip, lost motion, etc.,
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
the machines performed with less than acceptable accuracy by today’s
standards. With the ever-increasing demand for more accurate machines,
more and more of the machine was included in the control/drive feedback
loops. Today the machine, control, and drive are an integrated machining
system. The existence of resonances between the drive motor and machine
slide can make the drive unstable. A stick-slip condition on the machine
slide can cause a null hunt (discussed in the next section) in the drive. Poor
surface finish can result where load forces are fed back to a drive that has
poor servo stiffness. A contouring drive with poor resolution also
demonstrates poor low feed characteristics affecting the surface finish.
Electrical noise feedback into transducer cables, as the result of poor
shielding and isolation, can cause stability and accuracy problems.
With just about everything in the machine system being a variable that
can affect machine accuracy and stability, where does the technician start to
look for the cause of a machine that will not perform properly? It is virtually
impossible to diagnose the trouble of a malfunctioning machine with all the
control loops closed. Therefore, the control and machine should be
Fig. 1 Machine control system block diagram.
Fig. 2 Machine control system block diagram with nonlinearities.
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
separated. With a hydraulic drive the separation should be at the servo valve
connector. With a DC electric drive the separation should be at the input to
the electric motor. With brushless DC drives the battery box should be
connected to the drive amplifier input. The troubleshooting should now
proceed in the following sequence.
Closed Velocity Loop
The next logical step in the ‘‘setting up’’ or troubleshooting of the servo
drive is to close the tachometer or velocity loop and check the performance.
Hydraulic Drive
The velocity loop performance can be checked by closing the servo loop
external to the control. A battery box connected to the velocity drive input
can be used to operate the servo drive independently of the control. By
operating the velocity drive back and forth and putting step changes in input
voltage to the drive, the servo performance can be observed for stability.
Any unacceptable performance must be corrected before proceeding further.
Electric Drive
Setup instructions for the electric drive are provided by the drive supplier.
This type of drive can also be controlled by a separate variable DC voltage
applied to the drive amplifier input terminals. Likewise, by moving the
machine slide back and forth and putting step changes in input voltage to
the drive, the servo performance can be observed for stability.
Unacceptable performance must be corrected before proceeding to
close the position loop.
In either the hydraulic or electric drive the performance of the velocity
loop should be checked with an external DC voltage source. The numerical
control voltage sources such as the feed-forward voltage should not be used
for trouble shooting. The most obvious test to make with the velocity loop is
to make sure the drive can be controlled from standstill to the traverse rate.
Whether the feed rates are smooth or jerky, especially at the lower feed
rates, should be observed. With either electric or hydraulic drives the
velocity loop gain should be set at its highest possible setting. The velocity
loop performance is directly related to the gain. With too low a gain the
drive will be sluggish and exhibit a speed droop with applied load. One of
the prime reasons for poor surface finish with numerically controlled
machines is a velocity loop with too low a gain (thus also poor performance
and bandwidth). There are engineering evaluation tests that can be used to
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
ascertain the servo loop performance. However, these tests should only be
made by a trained analyst. For general use the gain setting can be made by
raising the velocity loop gain until the drive goes unstable and oscillates. The
oscillation is usually observed as an audible rumble often called ‘‘tachometer
growl.’’ When this condition occurs the gain should be reduced until the
‘‘growl’’ ceases. The gain should then be reduced slightly further to allow
some margin in reliability of the gain setting. With electric drives the gain
setting is often a component part of the setup instructions.
Hydraulic Drives
Machine piping should be checked to assure that all return lines run directly
to the ‘‘tank.’’ The drain lines should have a loop at the motor to be higher
than the motor. Any restriction in the drain line can blow out a motor seal.
A blown seal is obvious with a puddle of oil on the floor. Hydraulic pressure
and return line pressure can be checked at the servo valve as a rapid way to
determine if a piping problem exists. Return line pressure at the motor
should be less than 100 psi when the motor is rotating.
Electric Drive
Wiring for electric drives does not pose the same problems as hydraulic
drives. However, motor armature cables carry currents with very fast rise
times from pulse-width modulation (PWM) DC drives and brushless DC
drives. These cables should be isolated as much as possible from tachometer
feedback cables and position transducer (resolver or Inductosyn scales)
cables.
Open Loop
Battery boxes can be used for both hydraulic and electric drives.
Hydraulic Drive
The battery box connected to the servo valve can be used to move the
machine slide back and forth. A normal machine slide will move (break
away) with about þ0.004 A or less. This current may not necessarily be
symmetrical around zero. The servo valve mechanical null adjustment may
cause a slight servo shift if it is not centered.
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
Electric Drive
For DC electric motors, the breakaway current should be approximately
less than one-half the rated current. Excessive breakaway currents in either
drive indicate a sticking or binding machine slide. Vanes or pistons in a
hydraulic motor and armature winding slots in electric motors will cause a
cogging effect when these motors are operated as an open loop. The cogging
is noticeable at low speeds. It is normal and should not be a cause for alarm.
A properly operating tachometer loop will smooth the open-loop cogging
action. Machine slide stick-slip can be observed with a battery box in cases
of friction slides. With antifriction slides (hydrostatic or roller bearing)
stick-slip should not be a problem.
In the case of friction slides a breakaway current two or more times the
run current can cause a position loop null hunt if the stick-slip is inside the
position loop.
Hydraulic drives usually have some amount of internal leakage. This
leakage is a form of damping for the hydraulic resonance in the drive. It will
be of some benefit in damping and will permit higher gains to be used. When
the leakage becomes excessive and variable, the low feed rates may become
jerky. This is the same resultant effect obtained from varying loads on a low-
gain velocity servo loop. The cure is the same for both situations. The
velocity loop gain (and thus the bandwidth) should be set to its highest
feasible setting. With some hydraulic motors the leakage has been found to
be excessive and the motor should be replaced. Some control suppliers make
an unnecessary practice of adding cross-port leakage to all hydraulic
motors. Cross-port leakage in the form of a drilled set-screw plug in the
servo valve manifold should only be used if needed. Cross-port leakage
reduces the sensitivity of a drive and if required should be kept as small as
possible. The diameter of the drilled plug should be about 0.001 in. With the
larger hydraulic motors, experience has shown that a 0.0025- to 0.003-in.
diameter orifice will suffice.
In general the setup and troubleshooting of electric drives has been
found to be simple and reliable. The numerous variables associated with
hydraulic drives have proven to be challenging at times.
Closed Position Loop
Each control manufacturer will probably have a different technique for
closing the position loop. The one thing in common to all controls is that the
position loop gain must be set at a predetermined value. The techniques vary
for setting this gain. It is quite common to set the gain by measuring the
position lag (command position7actual position, called the following error)
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
at some given speed. In other cases the output position responses to a step
input position may be measured and recorded on a chart recorder. The
position-loop gain is then adjusted until the desired response is observed on
the chart recorder. Whatever the techniques used to set the position-loop
gain, the control manufacturer will have a step-by-step procedure for these
adjustments.
Once the position-loop gain is set to the prescribed value, the servo-
driven axis may not perform properly for a variety of reasons. These are
some symptoms of faulty position loops:
1. The servo drive may be unstable or it may null hunt at a low
frequency.
2. The feed rates may not be smooth.
3. The drive may not be stiff enough.
4. The drive may not position accurately.
To relate all the possible faults to the symptoms would result in endless
discussion. The next section attempts to relate different kinds of problems to
their possible causes.
5.2 PROBLEMS: THEIR CAUSES AND CURES
Table 1 illustrates the causes of problems.
Poor Surface Finish
Causes: Other than tooling problems, a poor surface finish can result from a
poor low-feed contouring control. If the low-feed-rate contouring control
does not operate smoothly it can be defined as a drive with poor drive
resolution (the difference between breakaway error and run error is too
large). The smaller the control signal required to cause the drive to move,
the smoother is the drive and the better the surface finish. The surface finish
is also related to the bandwidth of the velocity loop and the position loop.
Poor surface finish is very often the end result of a low gain (and thus low
bandwidth) tachometer loop.
Stiction and high-friction machine ways can also contribute to poor
surface finish. A machine way is the surface area where a machine slide
makes contact with the structure of the machine. Nonlinearities such as
nonuniform leakage, as found in hydraulic motors, can definitely cause
surface finish problems.
Cures: It is recommended that all feed drives use the highest
tachometer loop bandwidth (highest gain) possible. A wide bandwidth
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
Table 1 Causes of Problems
Problem
Possible causes
Poor surface
finish
Unstable position
loop
Null
hunt
Tach loop
growl
Feed rate
not smooth
Low
stiffness
Insufficient
accuracy
1. Mechanical backlash p X X p X p
2. Stick-slip p X p p
3. Low spring rate p X p p p
4. Undamped hydraulic resonance p X p p
5. Too much hydraulic leakage X p X X X
6. Too much friction p p p
7. Tach loop gain too low X X X X
8. Tach loop gain too high p X p
9. Position loop gain too high p X p
10. Position loop gain too low X X X X
Note: X, Definite cause; p, possible cause.
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
tachometer loop will minimize the nonlinear effects of hydraulic motor
leakage. The machine slide should have no lost motion (backlash) in the
mechanical drive train, and the machine ways should be antifriction (roller
bearings, hydrostatic, Rulon way liners, etc.).
Unstable Position Loop
Causes: There are several causes for an unstable position loop, with the most
obvious being too high a position-loop gain for a given system. When
resolver feedback is used at the drive motor, the nonlinear mechanical
problems such as lost motion, stiction, and low spring rates will not affect
the stability of the drive. However, when direct slide feedback is used (using
Inductosyn scales) these mechanical nonlinearities are inside the position
loop.
Cures: For contouring servo drives the posit ion-loop gain should be in
the ‘‘soft servo’’ range (0–2 ipm/mil). The machine dynamic problems can be
minimized by using a low position-loop gain when direct machine slide
position feedback is used. Some control manufacturers use 0.6 ipm/mil
position-loop gain on all axes. In general, most large machines with direct
feedback should have position-loop gains of about 1 ipm/mil. It is very
important that lost motion (backlash) be absent from drives with machine
slide feedback. If a wound-up gear train should lose its windup it will
probably result in an unstable position loop.
Null Hunt
Causes: A null hunt is a form of instability. A null hunt is a very-low-
frequency oscillation sometimes called a limit cyle oscillation. Lost motion
(backlash) can cause a null-hunt condition when combined with large
friction forces. Usually a backlash condition inside a position loop results in
an unstable drive. The fact that the period of oscillation may be fast or slow
is not significant. Stick-slip on a machine slide can definitely cause a null
hunt. In general, a null hunt is possible when the breakaway friction is at
least twice the running friction. The null hunt associated with stick-slip can
also be aggravated by a spring (unstiff drive screw) inside the position loop.
With hydraulic drives, excess leakage will cause an unstable position loop,
which will oscillate at a low frequency and appear as a null hunt.
Cures: For most cases of null hunt caused by backlash, antibacklash
gearing (wound-up gearboxes) should be used. When the windup is lost in a
wound-up gearbox the position loop is almost guaranteed to be unstable.
Friction-type slides should be avoided. For machine ways using roller
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
bearings, hydrostatic ways, or Rulon way liners, null hunt from stick-slip
should not be a problem. Excessive cross-port leakage on hydraulic servo
drives will cause a null hunt, and care should be taken that if an orifice is
required for hydraulic damping the orifice is in place. There have been cases
where the orifice was omitted. The orifice is located in the servo valve
manifold. Hartman motors have enough internal leakage and do not require
additional cross-port leakage.
Tachometer Loop Growl
Cause: The instability of a velocity loop will cause an oscillation at a
frequency high enough to be audible, thus the term ‘‘tach loop growl.’’ The
most significant cause is trying to set the loop gain too high for the existing
parameters.
Cures: The most obvious cure for an unstable tachometer loop is to
lower the loop gain. However, if the loop gain must be lowered the total
contouring drive will have reduced performance.
Feed Rate Not Smooth
Causes: There are numerous causes for an unsmooth feed rate. Excessive
and varying hydraulic leakage can also cause a varying low feed rate in
hydraulic drives. Other things, such as the drive nonlinearities of stick-slip,
lost motion, low mechanical resonances can affect the drive smoothness.
Cures: A prime consideration is to maintain as high a tachometer loop
gain (thus bandwidth) as possible assuming other restrictions, such as
underdamped hydraulic resonances and mechanical resonances, have been
dealt with.
Low Stiffness
Cause: Stiffness as referred to here is the servo drive loop stiffness. It is a
measure of how many lb–in. of torque is required to rotate the motor shaft a
given number of degrees. Too much leakage will reduce the stiffness of a
hydraulic drive.
On a machine, the stiffness measured at the motor increases by the
square of the drive ratio. The stiffness can also be measured at the machine
slide where the stiffness increases by the square of the ratio/lead. Drive
stiffness is discussed in depth in Part II.
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
Cures: To maintain adequate stiffness the design of the drive must
include the proper sizing. After the drive is assembled the only variables left
to adjust are the loop gains.
Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved
. 5 Troubleshooting Techniques The time is gone when machines, controls, and drives were ‘‘add-ons,’’ as shown. the drive amplifier input. The troubleshooting should now proceed in the following sequence. Closed Velocity Loop The next logical step in the ‘‘setting up’’ or troubleshooting of the servo drive. block diagram for today’s machine systems looks more like the block diagram shown in Figure 2. 5.1 TECHNIQUES BY DRIVE The primary difference between the machine drives of yesteryear and those of