TMR SpeedTronic™ Mark V LVDT Calibration Basics

This document attempts to explain how the TMR SpeedTronic Mark V

calculates position feedback from LVDTs (this includes RVDTs, also) and
how this can cause the servo output currents to appear to be mismatched
and/or out of tolerance. This explanation also attempts to define when it is
necessary to adjust Null Bias current values, and when it is not.

This document is primarily directed at heavy-duty gas turbine applications,

though small- and medium steam turbines which employ only two LVDTs
per control valve, etc., do experience the same kind of problems.

(Steam turbine applications which use three LVDTs per device use a
unique method of downloading individual LVDT calibration information to
each control processor which does not result in the problems experienced
when one I/O Configuration is downloaded to all three control
processors—which is the case in the overwhelming majority of TMR Mark
V applications.)

But first, some basics of both LVDTs and three-coil electrohydraulic servo
valves are necessary. Then, a basic explanation of how feedback relates
to servo output current is given. Then a description of how the TMR
SpeedTronic Mark V converts LVDT feedback voltages into positions is
given. Finally, a description of when—and when it is not—necessary to
adjust servo valve Null Bias currents is given.

LVDT Basics
For most every heavy duty gas turbine application GE specifies to its
vendors that the voltage output of LVDTs is to be linear between
approximately 0.700 VAC RMS and 3.500 VAC RMS over the specified
range of travel of the LVDT. (The output voltage can be less than or
greater than the limits specified, but the voltage with respect to position
must be linear between the limits specified. The limits of 0.700 – 3.500
VAC RMS are almost always the standard for LVDTs; RVDTs sometimes
have higher or lower minimums, and usually have lower maximums, as
well.)

This is not to say that every device with an LVDT will have an output of
exactly 0.700 VAC RMS to 3.500 VAC RMS. Quite the contrary, most
devices will have an output of 0.700 VAC RMS at zero stroke, and
something less than 3.500 VAC RMS at maximum stroke (maximum
actuator travel)—but almost never more than 3.500 VAC RMS at
maximum actuator travel.

Further, GE specifies the range of travel of the actuators of devices to be

equipped with LVDTs should not exceed the linear range of travel of the
LVDT. In other words, when properly adjusted the output of an LVDT will

Practical Turbine Controls Page 1 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
be between 0.700 VAC RMS and 3.500 VAC RMS over the range of travel
of the device actuator and that output will be proportional to stroke (linear)
over the range of travel of the device actuator. ‘Properly adjusted’ means
that the output of the LVDT is adjusted such that when the device is at it’s
fully closed mechanical stop the LVDT output voltage is approximately
0.700 VAC RMS, +/- 0.020 VAC RMS and does not usually exceed 3.500
VAC RMS at maximum actuator travel.

Servo Valve Basics

The three-coil electrohydraulic servo valves which GE uses to control the
position of hydraulic actuators have springs in them which, in the absence
of any electrical current being applied to the coils (all three coils) will
cause the device to move to shut off the flow of fuel or air or steam to the
turbine. This spring is called the fail-safe spring because it causes the
servo-valve to port the flow of high-pressure hydraulic oil to the actuator of
the device to move the device to a “safe” position in the event of a loss of
all electrical signals (“safe” meaning reducing the flow of air or fuel or
steam, etc., to the unit).

The servos which GE uses require negative current to cause the device to
increase the flow of fuel or air or steam to the turbine; positive servo
current decreases the flow of fuel or air or steam to the turbine. For this
reason, it is necessary to verify that the polarity of the servo current being
applied to each coil is correct before beginning any calibration procedure.
This document assumes that has been done and so the effects of
incorrect polarity will not affect the descriptions in this document.

To cause the servo valve to position the device at some mid-stroke,

‘steady-state’ position it is necessary to apply an electrical current to at
least one of the coils to overcome the spring tension and keep the device
at the mid-stroke position. This current is called the Null Bias current. The
servos which GE uses require a total of –0.8 mA, +/- 0.4 mA, in order to
overcome the fail-safe spring tension and keep the device at some mid-
stroke, ‘steady-state’ position.

Since there are three coils in the servo, the total can be divided by three to
determine the Null Bias current that each processor would supply under
ideal conditions to keep the device at some mid-stroke position. -0.8 mA
divided by three equals -0.267 mA per processor. Since the Mark V scales
0-10 mA as 0-100% for servo output current, this translates into
-2.67% servo current per processor under ideal conditions. (NOTE: The
I/O Configurator automatically converts the value entered into the Null
Bias Current field to a negative value when a ‘Save’ or a ‘Save and Exit’ is
performed, saving the value in IOCFG_Q.DAT as a negative value.
Therefore, it is NOT necessary—except under one circumstance, for the
Gas Transfer Valve of a DLN-I combustor-equipped gas turbine—to enter
a negative value of Null Bias current into the I/O Configurator.)

Practical Turbine Controls Page 2 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
Note that the specified range of total Null Bias current required is -0.8 mA,
+/- 0.4 mA, meaning that the total can fall anywhere between -0.4 mA and
-1.2 mA. This means that the servo current output of any processor can
range between -0.133 mA and -0.400 mA and still fall within the
specification, or between -1.33 mA and -4.00 mA.

The fail-safe spring tension is adjusted by the servo valve manufacturer

and is not intended to be field-adjusted. In fact, most field attempts to
adjust servo valve fail-safe spring tension result in replacement of the
servo valve because it is just not possible to properly adjust the spring
tension without the use of calibrated pressure gauges and specified
hydraulic pressures, etc.

Further, most field attempts to adjust fail-safe spring tension are driven by
a desire to bring a single, or sometimes two, servo currents ‘into’
specification. Since there is only one spring, adjusting the spring tension
affects the servo current of all three coils. Resist the temptation to field
adjust a servo valve’s fail-safe spring—particularly on a new unit
installation or the installation of a new servo valve—and do so only if a
spare is readily available—again, since most field adjustment attempts
result in replacement of the servo anyway.

Mark V Regulators and Servo Output Currents

The regulators used in the SpeedTronic Mark V subtract the position
feedback from a position reference in what’s called a summing junction.
The difference is used to increase or decrease the amount of current
applied to the servo valve coils. Again, this is done in each of the three
control processors. If the feedback were equal to the reference, the
difference would be zero. If zero current were output to the servo valve,
the valve would not move to the required position because of the current
required to overcome the fail-safe spring. So, the value of current
specified as the Null Bias Current in the I/O Configurator is added to the
output of the regulator summing junction and that current is always
“added” to the output of the summing junction. In other words, the output
of the summing junction is biased by the addition of the Null Bias current.

Under ideal conditions—that is, when the calibrated position feedback for
all three processors is exactly identical—the total of the three servo
current outputs will be between -4.00% and -12.00%, and the individual
servo currents will be nearly equal. In fact, if the proper hydraulic pressure
is available and the correct regulator gain has been input to the I/O
Configurator and downloaded to the control processors, the individual
servo currents will be very close to -2.67%, if not nearly exactly -2.67%.

However, ideal conditions are not attainable in the TMR SpeedTronic

Mark V turbine control panel except: 1) immediately after an
AutoCalibrate has been performed, or, 2) if a special method is employed

Practical Turbine Controls Page 3 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
to download individual 0%-stroke and 100%-stroke voltages to the three
control processors. (This special method is employed on most large
steam turbine applications where devices are equipped with three LVDTs.)

The I/O Configuration which is downloaded to all three control processors

includes the 0%-stroke and 100%-stroke voltages which the control
processors use to convert the LVDT feedback voltages into position
feedbacks. It also includes the Null Bias current value. Note, that all three
control processors get the same 0%-stroke and 100%-stroke voltage
values and the same Null Bias current value.

The TMR SpeedTronic Mark V panel typically has all LVDT feedback
voltage signals connected to a single I/O terminal board (usually the
TBQC card on <R> processors), and those voltages are “fanned out” to
three ribbon cables which are connected to individual cards (usually
TCQA cards) in <R>, <S>, and <T>. The analog-to-digital conversions
take place on the individual TCQA cards. Because of things like ribbon
cable lengths not being exactly equal, varying resistances of connectors
and printed-circuit card traces and other component tolerance values all
adding up, the voltages that the analog-to-digital converter on each
individual TCQA card reads can be—and, in fact, are—different.

(Resistors and capacitors and other electronic components are almost

never precision components—precision components are costly—and
always fall within a certain tolerance range: 5%, or 10%, or 1%. Cost and
the precision required by the application dictate the tolerance, or range, of
deviation from the desired value that can be acceptable when building
electronic equipment. This is true of all electronic equipment that employs
analog components. This is one of the reason many analog circuits
employ potentiometers, or “trimming pots”.)

If one manually reads the LVDT feedback voltages at the TBQC card
while performing a calibration of the LVDT feedback and then inputs these
values into the I/O Configurator for the 0%-stroke and 100%-stroke
voltages, then downloads these values to the three individual processors,
the slight mismatches in the actual voltages being read by the analog-to-
digital converters on the three TCQA cards are not accounted for and the
LVDT feedback for each control processor will be slightly different from the
other control processors—not usually different by a great amount, but
different enough to cause the servo output currents of the three control
processors to vary. This is because the slight differences will cause each
of the three control processors to determine the device to be at a slightly
different position, and therefore the output of the three summing junctions
will be slightly different—meaning the servo output currents of the three
control processors will be slightly different.

Similarly, if one uses AutoCalibrate to determine the 0%-stroke and 100%-

stroke voltages to put into the I/O Configurator, one has to take the

Practical Turbine Controls Page 4 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
average of the three values since each control processor’s 0%-stroke and
100%-stroke voltage values are calculated and displayed and only one
value can be entered into the I/O Configurator and downloaded to all three
control processors. The act of averaging the values before entering them
in the I/O Configurator and downloading them to the three control
processors will cause each control processor to display a slightly different
feedback value from the other two processors—just as downloading the
voltage values read manually at the TBQC will cause.

The slight differences (usually just tenths of a percent) in themselves don’t

cause a problem—except that each control processor looks at the position
reference for the device (a single, voted value from sequencing) and then
compares the reference to the position feedback it believes the device to
be at and adjusts the servo output current accordingly to try to make the
feedback equal to the reference. With this happening in all three control
processors, the servo output currents of all three control processors can
be very different from each other.

For example, if <R> thought the IGVs were at 78.9 DGA, and <S> thought
they were at 78.6 DGA, and <T> thought they were at 79.1 DGA, and the
reference was 79.0 DGA, <R>’s servo output current would be
approximately -0.279 mA, or -2.79%, <S>’s servo output current might be
approximately -0.344 mA, or -3.44%, and <T>’s servo output current might
be approximately -0.121 mA, or -1.21%.

On first glance, this might appear to be a problem: the servo currents are
not approximately -0.267 mA, or -2.67%, and they appear to be
unbalanced. But, given the differences in position feedback seen by the
three processors it’s not a problem at all.

If one performs an AutoCalibration of LVDT feedback and then

immediately uses the Manual feature of AutoCalibrate to position the
device to some mid-stroke position, one can see the three position
feedback values for the three control processors would be nearly identical
(within hundredths of a percent, usually)—and the three servo currents
would most likely be nearly identical, and usually very close to the desired
-2.67% value, +/- 1.33%.

This is because immediately after an AutoCalibrate (before an I/O Config.

download has been performed and the processors have been re-booted)
the 0%-stroke and 100%-stroke voltages in the TCQA RAM are the exact
values they need to be for that TCQA and processor to make it’s
calculation of position feedback from the LVDT voltages it reads nearly
identical to the other two TCQA cards and processors.

(The values in TCQA RAM are put there by AutoCalibrate after an

AutoCalibrate has been performed, and are written over by the values
from the I/O Config. download when the processor is re-booted.)

Practical Turbine Controls Page 5 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
It should be clear that mismatches in servo currents can be attributed to
the mismatches in calculated position feedback done by the individual
control processors. This occurs when a single, or average, set of 0%-
stroke and 100%-stroke voltages are downloaded to the processors when,
in fact, each processor really sees slightly different voltages and needs
slightly different 0%-stroke and 100%-stroke voltages in order to calculate
position feedbacks identical to the other control processors.

Sometimes, one TCQA card will calculate a much higher or lower position
feedback than the other TCQA cards. This can cause concern, and could
even be cause for replacement of the TCQA card which has the abnormal
reading after all other possible reasons for the difference have been
investigated (such as poor ribbon cable connections; corrosion on the
ribbon cable connect pins; a chafed or worn ribbon cable with an exposed
conductor which might be in contact with a ground source, etc.).

When to Adjust the Null Bias Current

Now, the question which should be on the reader’s mind is: When is it
necessary to adjust servo Null Bias current values like the Control Spec
says should be done? The answer to that is: When the calibrated position
feedback differs from the reference by more than approximately 0.5%.

Assume that LVDT feedback has been recently calibrated, including the
downloading of the 0%-stroke and 100%-stroke voltages to the three
control processors and the re-booting of the processors. Then, when
verifying the accuracy of the calibration it is noted that the feedback differs
from the reference by more than 0.5%, it is acceptable to make small
changes in the Null Bias current value, download, and re-boot, and re-
check the accuracy of the positioning of the device.

For example, while checking the accuracy of LVDT feedback calibration

the device is commanded to go to the 80.0% stroke position. The device
moves to a position and stops moving, and the observed position
feedback is 79.2%, which is unacceptable. Assume another LVDT
feedback calibration is performed, and when the device is commanded to
go to the 80% stroke position, the feedback is 79.3%. Assume another
calibration is performed and the feedback is still 0.6% below commanded
position. Looking at the servo-currents, one can see that they are all fairly
well-balanced and within the specification. In this case, it would be
acceptable to increase the Null Bias current to, say, 3.00, download, re-
boot, and repeat the verification check. If the difference between the
actual position feedback and the commanded position was less than
0.5%, no further adjustment of the Null Bias current value is required.

However, assume that during the verification of the accuracy of LVDT

feedback calibration the total of the servo output currents is within the
desirable range of -4% -to -12% but the individual currents themselves are
not well balanced. Assume also that the actual position feedback for each

Practical Turbine Controls Page 6 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics
of the three processors is within 0.5% of the commanded position, as is
the “voted” value of position feedback for the device. In this case, it is not
required nor recommended to change the value of Null Bias current.

Remember when changing the value of Null Bias current in the I/O
Configurator, that single value gets downloaded to all three control
processors. If one is trying to get a single processor’s servo output current
‘into’ range, changing the Null Bias current will most likely make that
processor’s value only worse, and will affect the other two processors, as
well. If a single processor’s servo output current is well out of range, the
problem is most likely that processor’s calculated position feedback is also
out of range of the other two and that is a different problem which cannot
be solved by adjusting the Null Bias current.

Practical Turbine Controls Page 7 of 7 Rev. 3

Mk V LVDT Feedback Calibration Basics