LVDT Calibration

B

Thread Starter

B K MEENA

I want to know the correct procedure for calibration of LVDT of GCV/SRV in Mark_VI Machine.

And how to download the .m6b file to controller after changing the null bias value.

Any body please suggest
 
ProcessValue,

Can you explain specifically your section on null bias current? I'm not sure what the intent is nor how to accomplish what is to be done.
 
P

Process Value

Servo null bias configuration.

Here comes the long answer once again :)

Servo valve operation : to explain in detail about the servo null bias one should know how a 2 way servo valve actually works. i am uploading a small pic here in which i have explained in detail the working of a flapper/nozzle servo which is used by moog. the pics that you will see represent a very similar design used by moog and by all servo manufacturers.

a. first part - the main parts of a servo is explained.

b. operation of a servo valve to move the actuator to the right is explained ( the reverse operation is vice versa)

c. the resetting of the servo valve after the required movement of the actuator is explained

http://www.2shared.com/photo/_LkyHPfy/servo_expain.html

please read through the above so that what i am going to say below can be understood in greater detail.

In the pic above the centering spring is present on both sides of the servo spool piece. This centering spring is responsible to bring back the servo spool

to the initial position so that there is no control oil flow into the actuator chamber. in our case there is only one spring installed in the system. this spring is called the failsafe spring/ compression spring . in the absence of any electrical commands this pushes the spool piece to open a very small port which causes the actuator rod to retract (open/close the vavle depending on the design). During normal operation , such a thing should not happen , and thus the current required to over come the spring force and return the servo to null (zero control flow) position is called the servo null bias.The servo null bias is often expressed as a percentage of the full scale servo current. in moog design its 10 ma full scale and for a null bias current of 0.2 ma means a null bais of 2%.

i am uploading another pic here , in that you will see the servo chara for a typical moog valve ( control flow vs. input currnet ) in the same i have marked the servo null current for better understanding.

http://www.2shared.com/photo/4CNdVGGl/moog.html


now coming to the null bias setting in mark Vi.

in mark vi TMR systems the servo is a three coil one , with each of the three processors supplying the current to the servo. in GE designs the servo feedback

is accomplised by two methods.
a. LVDT - IGV / Gas based designs

b. pulse rate inputs - fuel valve in distillates

the servo null bias has a range of about -0.133 ma to -0.4 ma. the required null current will be somewhere between this value. this is the usual norm but it may not be true for all cases. i have seen a site with a null current of -0.75 ma in IGV conrol servo working perfectly. please see the servo data sheet to get the actual value.

for an example let us say that we have fixed the null value at -0.15 in the servo configuration (-1.5%) . now given that the processors are online and you are in the servo calibration page use the manual command to give a stroke of 50%. if you have a LVDT then you can see the value easily other wise you need a dial guage to measure the valve stroke. let us say for the stroke of 50% we get the following values.

R - 47.5
s - 47.8
t - 47.7
dial guage - 48.

now this means that the control oil ported is not enough to attain the correct stroke and you need to increase the null bias value , theoretically a 2% increase in the null bias ie from 1.5% to 3.5% must get you to 50% value (assuming that the dial gauge reading is most accurate), but it rarely happens; with three processors and three coils and three feedback inputs attaining a perfect 50% is well quite difficult , anything in the 49-51% range is acceptable (at least for me). further calibration is quite tedious and frankly a waste of time.

now enter the new null value (3.5%) in the servo calibration page and do the stroking once again , you will get a value close to 50%. if you still want to have some fun you can change the resolution of the servo null bias and try to get the perfect fifty ( upto 5 decimal places are accepted if i remember correctly).

most probably you will see that in GE machine the null bias value to be at -2.667 %. this is the median value of the range specified by moog , and yes this will work most of the cases. it is quite rare that you need to change this value.

this is to the best of my knowledge how a servo works and this is the method to calibrate the servo null current. hope CSA will be satisfied by the explanation and can give more pointers in the area :) .
 
P

Process Value

Chris the link is working well , only problem is that they have disguised the download button very well. in the bottom of the page you will have a small button which will say "save file to computer" you need to click it to down load the doc. till now 20 downloads have been done so i think it has worked for that many people.
 
P

Process Value

Sigh ... well the links are indeed working, as i have said in the earlier post at the bottom right side of the page there is a posting like this

Save file to your PC: click here

click the link and you will able to download the file.

from the last time i saw the screen it is now downloaded 44 times :) hurray :).
 
i dunno, but the message says "The file link that you requested is not valid. Please contact link publisher or try to make a search. "

Please upload to another file sharing, or send email to cikenkari(at)gmx.com

Thanks ya
 
Seems the link can't find the file, so using search to find "sevo_calibration",

thanks Process Value, very nice procedure
 
P

Process Value

Good to hear that you did download the file , and as far as the invalid link i do not know what the problem is i will upload to a different site and post here.

and yes glad to be of help :)
 
Well, I'm going to have to take exception to some of the material presented here. While the basic premise is correct, the drawings used to explain the concept along with the disclaimers about the drawings, and the math doesn't seem to work out.

If there is a centering spring that returns the servo spool piece to a condition that blocks oil and prevents opening or closing of the servo-operated actuator, isn't that contrary to the "fail-safe" concept? In other words, in the absence of any servo current, the servo valve is supposed to port oil to the actuator to shut off the flow of fuel or air or steam. If there were no current applied to the coil in the drawing, the centering spring would just "block" the actuator in its current position, which could keep fuel or air or steam flowing in a "loss of control" situation (no servo current). So, this doesn't seem to make sense.

ProcessValue then says, oh, forget about the centering spring, because the servos at his site only have a spring at one end. A spring at only one end would cause a properly applied servo to shut off the flow of fuel or air or steam in the event of loss of servo current to the coils--which meets the definition of "fail-safe".

So, we're given some drawings with explanations that we're told to read and study, and then we're told, "Oh, that's okay, but that's not what really happens."

The Speedtronic servo regulators used by GE put out 0.00 mA when the actual position is equal to the reference (the error is zero), and then the null bias current value is added to that to overcome the null bias spring tension in order to keep the actuator/device at the desired position.

GE specifications for servos used for heavy duty gas turbines state that the null bias current (total) should be -0.8 mA, +/- 0.4 mA. Take that value and divide it by three for a three-coil servo and you get -0.267 mA per coil, for a total of -0.8 mA.

Presuming that no one has adjusted the spring tension of the single fail-safe spring of the servo (which should never be necessary except under extremely unusual circumstances), an individual null bias current of between -0.133 mA to -0.400 mA should be sufficient to make the measured position match the indicated position (from the LVDT feedback).

ProcessValue makes a very important and excellent point in the explanation: <b>The null bias current adjustment should only be used to make the actual position (and, hence, indicated position) equal to the reference</b>. It's not to be used to make the indicated position (feedback value displayed on the screen) equal to the reference if the actual position isn't being measured!

In other words, the calibrated feedback value needs to be very nearly equal to the actual physical position as measured (using a dial indicator or vernier caliper, or machinists' protractor, as appropriate) and <b>then and only then</b> should the null bias current be adjusted to make the physical position (feedback) equal to the reference value. So, if the reference is 50% and the actual position is 48.9 and the indicated position is 49.0, then one can adjust the null bias current to make the actual position equal to the reference value.

And we're speaking about averages here for a three coil servo. If one processor's indicated feedback is radically different from the actual position (and hence the indicated positions of the other two processors) then adjusting the null bias current value <b>is not</b> the proper method for correcting this problem.

And, lastly there's this statement about changing the null bias current by 2% to counter a 2% error between the actual physical position and the reference. I've never witnessed a one-to-one relationship between current and position reduce the error between actual and reference positions. It's not that simple, and a trial and error method is usually required. (And why was the original null bias value fixed at -1.5% to begin with? Why wasn't it -2.67%? The error would have been less had the original value been -2.67%, which is what the starting value should always be.)

Again, I want to stress that the most valid and important points in ProcessValue's explanation are these:

(1) The null bias current is to be adjusted to make the actual, measured physical position match the reference <b>after the calibrated feedback has been proven to match the actual physical position by comparing the measured physical position to the calibrated feedback</b>. If the actual, measured physical position is not very nearly equal to the calibrated feedback value on the display, then the feedback <b>IS NOT</b> calibrated properly, and adjusting the null bias current value isn't going to fix this error.

(2) The normal range of null bias current values should be between -0.133 and -0.400 mA per processor, or, -1.33% to -4.00% (remember, in most Speedtronic systems the negative sign is not used; the positive null bias value on the display is "inverted" in the Speedtronic).

Most sites <b>never</b> measure the valve or IGV position and just adjust the null bias current to try to make the feedback value on the display equal to the reference--and that's just <b>WRONG!</b> Or, they try to adjust the null bias value when one processor's indicated value is significantly different from the other processors' indicated value.

By the way, the feedback spring in the drawings provided by ProcessValue would usually be the "fail-safe" element in a servo with a centering spring. Null bias current would be required to overcome the force that the feedback spring would apply to the coil armature in the loss of servo current to cause oil pressure to be ported to the actuator to shut off the flow of fuel or steam or air. Presuming the application required a fail-safe servo.
 
P

Process Value

> Well, I'm going to have to take exception to some of the material presented here. While the basic premise is correct, the drawings used to explain the concept along with the disclaimers about the drawings, and the math doesn't seem to work out. <

well here are the answers to the "exceptions" found by CSA :) . we all love a healthy argument don't we :)

exception - 1

"If there is a centering spring that returns the servo spool piece to a condition that blocks oil and prevents opening or closing of the servo-operated actuator, isn't that contrary to the "fail-safe" concept? In other words, in the absence of any servo current, the servo valve is supposed to port oil to the actuator to shut off the flow of fuel or air or steam. If there were no current applied to the coil in the drawing, the centering spring would just "block" the actuator in its current position, which could keep fuel or air or steam flowing in a "loss of control" situation (no servo current). So, this doesn't seem to make sense."

All servo valves manufactured need not have a "fail safe" concept , and fail safe concept need not necessarily mean a full open or a full close condition. an example of a servo system which does not have a "fail safe concept" of full open or full close is the damper control in an FD fan. these i have typically seen in boilers , and in some furnaces. most boilers follow a air rich or fuel rich control mode ,in both the concept the air flow is maintained by these dampers. in case we need to increase in the boiler load the damper opens first to let in more air and only if there is a air increase the fuel is increased to the boiler. Here the inlet air needs to be maintained at 3-4% excess of the stotiometric air. in-case of absence of electrical signals the servo in the damper control just holds the damper in the last position. you cannot close the damper as the air inlet will be reduced , leading to black smoke then eventual tripping of the boiler , you cannot fully open the damper as you will let in excess air which leads to white smoke in the chimney , very inefficient operation , slight load reduction and lowering of stack temperature , lowering of temperature gradient in the boiler zone and a whole lot assorted problems. thus the servo just holds the damper in the last position. this servo "does have" a centering spring.

Servos are hard to explain without any diagrams , hope you will agree with me on that one. i got the pic from the net , i modified it to a large extent to suit my explanation. I was trying to tell people how a servo actually works so that one can get a grasp of what the null current is. I do not think that anyone who does not know how a servo works will understand the concept of null bias current. i provided the explanation specifically for that.

exception - 2

"ProcessValue then says, oh, forget about the centering spring, because the servos at his site only have a spring at one end. A spring at only one end would cause a properly applied servo to shut off the flow of fuel or air or steam in the event of loss of servo current to the coils--which meets the definition of "fail-safe".

So, we're given some drawings with explanations that we're told to read and study, and then we're told, "Oh, that's okay, but that's not what really happens." "

ha ha ha hurray for the tongue and cheek once again but i would like to add that my "Parent site" has all kinds of servos one with centering spring ( boiler FD fan in boilers ) , one with the fail safe spring at one end ( GT fuel recirculation moog valve for one ) , servos with no compression springs ( desalter rotary stirrer in CDU unit). In my opinion the explanation given in the diagram with the "centering spring" servos is correct , and i don't remember saying that is not happens (it happens at a lot of places as i explained) , i just said that is not what happens in GT fuel control.

exception - 3

" And, lastly there's this statement about changing the null bias current by 2% to counter a 2% error between the actual physical position and the reference. I've never witnessed a one-to-one relationship between current and position reduce the error between actual and reference positions. It's not that simple, and a trial and error method is usually required. (And why was the original null bias value fixed at -1.5% to begin with? Why wasn't it -2.67%? The error would have been less had the original value been -2.67%, which is what the starting value should always be.) "

i am copy pasting my exact wordings " theoretically a 2% increase in the null bias ie from 1.5% to 3.5% must get you to 50% value (assuming that the dial gauge reading is most accurate), but it rarely happens; with three processors and three coils and three feedback inputs attaining a perfect 50% is well quite difficult , anything in the 49-51% range is acceptable (at least for me) " . You just repeated what i have told ; it rarely happens , meaning there in never a perfect lenier relationship. I was giving an example by quoting that you begin with -1.5%., i have also mentioned in the post that a null bias value of -2.67% will work out in most cases and it is rare that you need to change them. and yes i always start with -2.67 % .

so i hope that there are no points of singularity in my explanation , and the exceptions regarding the material , disclaimers ??!! and math are all resolved. further discussions , criticisms ;) from CSA are always welcome.
 
We were, and aren't, talking about servos used for FD fans or rotary stirrers. The thread was about:

>What is the correct procedure for calibration of >LVDT of GCV/SRV in Mark-VI machine.

In the case of servos used for GE-design heavy duty gas turbines (as this thread regards), fail-safe operation is required. I don't see posts from ProcessValue in any threads other than GE-design heavy duty gas turbine-related threads (with the exception of speed/frequency control threads), so it's odd that FD fans and rotary stirrers are introduced into this thread.

You say you've read and learned a lot from previous GE-design heavy duty gas turbine-related threads on control.com, which were kept as "real world" as possible. The driving concern has always been to present information that is relevant to the discussion, and most closely matches real world circumstances that the originator, or a subsequent poster, would encounter.

Nothing more; nothing less. (Okay; an occasional editorial comment, meant to reinforce the information.)

Diagrams of servos used on GE-design heavy duty gas turbines can usually be found in the Service/Instruction Manuals provided with the units; not always, but quite often (at least, they used to be included in years past in the Control System section). I agree, that it's difficult to explain servos without diagrams, but not impossible. And, some people have difficulty understanding any drawings, even with notes and circles and arrows and paragraphs; even engineers! (Sad, but true.)

Your treatment of the null bias current adjustment issue in your write-up was lacking and superficial, and presumes that changes to the null bias current value are always necessary. And the explanation wasn't that much better.

There are <b>thousands</b> of GE-design heavy duty gas turbines with TMR Speedtronic turbine control panels running around the world at this very instant (or any instant, for that matter) using null bias current values of -0.267 mA per coil, and they run just fine with no perceptible problems whatsoever.

At the same instant in time, there are likely tens, if not more, GE-design heavy duty gas turbines running around the world with wildly incorrect null bias current values, and they, too, run just fine with no perceptible problems whatsoever (at least that people want to pay attention to; remember, Diagnostic Alarms are just nuisances, and as long as the unit doesn't trip, most people ignore them, and even if the unit trips, they ignore them still).

I can probably count on one hand the number of sites that I've visited in nearly three decades that actually use a dial indicator (or machinist's protractor for the IGVs) when calibrating LVDT feedback, or that even use the scales provided at the valves/IGVs for checking the accuracy of the calibration. They just "click" and "click" and "click" and ass-u-me that the calibration is correct, without anyone ever observing the device, or without ever measuring the physical position.

And, again, there are literally thousands of GE-design heavy duty gas turbines running without any perceptible problems around the world, but with inaccurate LVDT feedback calibrations (for a variety of reasons, not just improper null bias current values). In some cases, very inaccurate LVDT feedback calibrations, by the way.

And, until some time in the last 10 years there was virtually NO null bias current adjustment ever made or even attempted during most LVDT feedback calibrations. It's really only since the advent of DLN combustion systems and the need for very accurate position feedback calibration in order to use fuel splits to control emissions very precisely did this issue ever arise. And, it has arisen with virtually no proper documentation or instructions from GE or its packagers whatsoever.

Bad documentation is worse than no documentation. It causes people not to trust, and even not to read, documentation going forward. It causes people to put stock in myths, legends, and wives' tales, and worse, to perpetuate them over time. That's something which should be ended.

To quote a GE T.A. I worked with many years ago, "This stuff ain't rocket science!" No matter how people try to make it out to be rocket science. Providing concise and correct information with real world application and examples relevant to the questions being asked is the goal.

The only thing I'm going to argue is whether or not correct and relevant information is being provided. This is trying to keep the information presented on track, relevant, and correct. It seems that, as with the incorrect usage of the word "doubt" to express a question or questions, some want to use the word "argue" to mean discussion. When I see incorrect and irrelevant information being provided, I'm going to call attention to it.

Just as others have chosen to do.
 
P

Process Value

I still do not understand what is irrelevant or wrong about the information i have given. I did not write about Boiler internals in a servo related thread.

The simple thing is that the picture presented there provided a clear idea about a servo working. i have MOOG drawings , data sheets etc , but they have cross sectional diagrams , and it did not have the actuator part in it as in the one i have presented. I do not know how a 10 point explanation about the working can be called superficial. If you have a better explanation please post it in here i would be glad to read it. i have gone through many control.com posts but i have not seen anyone post a picture or an accompanying explanation.

It is not that i reply to gas turbine related posts alone, i have also replied to steam turbine and boiler related posts also.It just so happens that not many people here ask about boilers. My career started with the commissioning of a co-generation steam turbine power plant. and i have been maintenance coordinator for boiler and HRSG shutdowns.

" Your treatment of the null bias current adjustment issue in your write-up was lacking and superficial, and presumes that changes to the null bias current value are always necessary. And the explanation wasn't that much better."

i have never said in my post that null bias changes are always necessary , i have said quite in the contrary , the value of -2.67 % almost never needs to be changed ; please go through my earlier posts. Explanations are liked by some , disliked by some , if you did not like it , tough luck ... but the explanations are the best i can do and for the given diagram they are correct , and i am pretty sure a lot of people out there would have gained "correct" and knowledgeable information from it.you probably do not need them , as you have seen them for a long time. but let me assure you control.com is visited by a lot of new and young engineers , i was one of them a couple of years back. Information and knowledge that you take for granted is not the same for all people , i have been there first hand. And providing a additional bit of information is not going to hurt them or anybody. This is my way of giving back to the community from which i have learned.

" There are thousands of GE-design heavy duty gas turbines with TMR Speedtronic turbine control panels running around the world at this very instant (or any instant, for that matter) using null bias current values of -0.267 mA per coil, and they run just fine with no perceptible problems whatsoever."

The question was about LVDT calibration , yes it is true that thousands of machine are running with the standard value but what was i supposed to say. Just fix up -2.675 it will work fine. i just uploaded the document which had the procedure. then you asked to explain more in detail about null bias current , which is why i posted the second thread with the explanation about the servo and all.

i agree that most of the units do not use the dial gauge , I do , the people who taught me to do it did the same ; so you can add one more to your count on that.

" To quote a GE T.A. I worked with many years ago, "This stuff ain't rocket science!" No matter how people try to make it out to be rocket science. Providing concise and correct information with real world application and examples relevant to the questions being asked is the goal. "

Actually Moog is more famous for using its servo valves in flight controls , missile controls and yes Space craft controls. Servos form a integral part of flight control and missile control systems , so they do form a small part of rocket science.

And lastly , you would probably know this but i am not getting anything out of posting servo workings in control.com , The explanation that you deem superficial took me a whole 2 hours ; to hunt a good picture , modify and present. The only thing that i am getting out of posting the irrelevant servo information is the satisfaction that i am giving back to the community which had helped me out. you have every right to raise an attention when you see a irrelevant material , but try to appreciate the fact that i am taking the time to help out young engineers and readers with information which is to the best of my knowledge correct.

This will be my last posting in this thread.
 
N

Not the real CSA

please CSA can you help me

1 how to calibrate SRV with TSM; telnet on MARKVI

2 i have some problem in MS-5002D, when there is trip turbine SRV =-13%. so every startup i calibrate SRV; when the turbine is at 100% i take look on LVDT, so min lvdt is 1.02 and max is 3.7 so
what is the solution?
 
The problem will not be solved with TSM.

If you have a combined SRV/GCV assembly, the problem is detailed in this thread:

http://control.com/thread/1294809227#1295331639

The SRV and GCV are virtually the same as far as operation and construction. The valve stem does indeed go below 0% (it will go negative in many cases), and the there is a Control Constant that is preventing the start when the value is too negative.

If the problem happens after a trip I would say the problem is that the bar to which the LVDT core is attached is not pinned to the valve stem, and when the valve slams shut the bar is sliding down more and more each time.

There should be a roll pin that is placed in a hole in the bar and shaft which keeps the bar from sliding down when the valve is slammed shut by the closing spring during a trip. It's possible the roll pin is missing, or it has sheared off.

Please write back to let us know what you find.

Lastly, the zero stroke voltage should be set for 0.700 VAC RMS (it's CRITICAL to use a TRUE AC RMS voltmeter when making the adjustment), plus-or-minus 0.020 VAC RMS. That will make the 100% stroke voltage less than 3.700 VAC RMS which is at the upper limit of the linearity of the LVDT. (Also, the SRV should not be at 100% stroke when the turbine is operating--unless the gas fuel supply pressure is lower than it should be.)

Again, please write back to let us know what you find.
 
Again, I'm presuming the turbine has a combined SRV/GCV assembly (two valves in a single casting). There is supposed to be a gap between the actuator rod and the valve stem, of approximately 0.030-0.050 inches, which, when the LVDT is properly calibrated, <b>WILL RESULT</b> in a slight negative value when the valve stem and actuator rod are fully down. (The actuator rod and valve stem move UP to open the valve.) But, if the gap is per specification, the LVDT feedback is almost always not less then -5.0%, sometimes very close to it, or slightly less than -5.0%, but not much more than -6.0%. If the gap is greater than it should be then the feedback will be more negative--if the LVDTs are properly calibrated.

The most common specification for LVDTs used on GE-design heavy duty gas turbines is that the output of the LVDT is to be linear between 0.700 VAC RMS and 3.500 VAC RMS, plus-or-minus 0.020 VAC RMS (at both ends of the range). That <b>does NOT</b> mean that when properly calibrated the LVDT feedback will be exactly equal to 0.700 VAC RMS at 0% and 3.500 VAC RMS at 100%. Because the same LVDT may be used on a valve with a physical stroke (range of travel) of 1.500 inches, or a valve with 1.625 inches of travel, or a valve with 1.250 inches of travel. The output at 100% will vary with the range of travel of the LVDT--but the zero stroke voltage should always be 0.700 VAC RMS, plus-or-minus 0.020 VAC RMS. As long as the zero stroke voltage is 0.700 VAC RMS, plus-or-minus 0.020 VAC RMS, the LVDT output should not exceed 3.500 VAC RMS at 100% stroke--but as long as the voltage is less than 3.500 VAC RMS the the feedback from zero stroke to 100% stroke will be linear--and that's the important factor, linear output over the range of travel, not the exact values at <b>BOTH</b> ends of travel. As long as the feedback at one end of travel (by default, the zero-stroke end) is on the linear output range, the LVDT feedback should be linear over the entire range of travel and the output at 100% should not exceed 3.500 VAC RMS (plus-or-minus 0.020 VAC RMS).

As long as the feedback over the range of travel is linear, the exact values at each end of travel are not critical (though, again, by default the zero stroke voltage is usually set to 0.700 VAC RMS just to be sure the LVDT is in a position to output a linear voltage with stroke). If the zero-stroke voltage were set to 1.013 VAC RMS and the output voltage were found to be 3.419 VAC RMS at 100% stroke, this would be fine--because the output from 1.013 to 3.419 VAC RMS would be linear. The Speedtronic doesn't care what the voltages are--as long as they are linear over the range of travel.

But, again--if every time the turbine trips the SRV LVDT feedback is going negative and you are having to "recalibrate" the LVDTs then something is causing the physical stroke of the valve to change--or the bar that is supposed to be firmly clamped and pinned to the valve stem is not (firmly clamped or pinned).

TSM isn't going to fix this problem. From the information provided, the problem is mechanical. Either the bar is sliding down every time the valve is closed, or something else is mechanically wrong with the valve, or, the gap between the actuator rod and the valve stem is not correct.

Please measure (with a set of feeler gauges) the gap between the actuator rod and the valve stem (when there is NO hydraulic pressure) and write back to let us know what the gap is. It will be necessary to use something to push the valve stem and actuator rod down (a piece of lumber used to pry against the LVDT bar that's clamped to the valve stem usually works well--just DON'T push too hard!), and then use the same method to push the valve stem up to see the gap and insert the feeler gauges to take the measurement.
 
csa, (interesting choice of moniker)

Diagnostic Alarms can be seen for various cards (the VSVO in this case) using Toolbox. The feedback can be seen from the LVDTs using Toolbox (it's not always easy to find, but it's there).

If you are referring to the problem of the SRV LVDT feedback changing (going more negative) when the turbine trips, TSM/telnet isn't going to fix that problem. The problem is NOT with the Mark VI. It's EITHER:

1) The way the LVDT is being, or has been, calibrated;

OR

2) There is a mechanical problem with the device, the actuator and/or the LVDT.

Please answer the following questions:

A) Is the SRV contained in the same cast steel assembly as the GCV?

B) What is the zero stroke voltage when you perform the LVDT "calibration"?

If the SRV and GCV are "combined" in the same cast steel assembly, then before a calibration is performed it is necessary to do something to ensure that the calibration starts at zero valve stroke and ends at zero valve stroke. (That's what AutoCalibrate does--it moves the device to the "minimum" (usually zero) position and measures the LVDT output (feedback) voltage, then moves the valve to the "maximum" open position (which may or may not be 100%--depending on the device) and records the LVDT output (feedback) voltage, and then moves the device back to the "minimum" position and checks to see that the LVDT output (feedback) voltage is nearly identical to that obtain in the first measurement. Then it calculates the necessary gains and offsets required to properly scale the feedback (per the AutoCalibrate configuration values).

For a normal SRV (or GCV) in the combined assembly there is a 0.030-0.050 inches "gap" between the top of the valve stem (which is <b>NOT</b> physically attached to the valve plug) and the bottom of the valve plug. (The stem RISES to open the valve by pushing the valve plug off the valve seat.) To measure this gap, one needs to have the hydraulic system off and at zero pressure, and use some sort of 'pry-piece' or piece of lumber to gently push down on the bar that is attached to the valve stem to ensure the actuator piston/rod is fully depressed, <b>THEN</b> use that same pry-piece to gently pry upwards on the same bar to push the valve stem up until it contacts the bottom of the valve plug. There should be a small gap visible between the top of the actuator rod and the bottom of the valve stem--and it should (normally) measure 0.030-0.050 inches. (You may have to apply continual upward pressure to be able to insert the feeler gauge to measure the gap; this is normal--but it shouldn't take much upward pressure to do so.)

Using feeler gauges (the ones used to measure the gap are fine) between the top of the actuator rod and the bottom of the valve stem to fill that gap--and LEAVE THEM IN PLACE DURING THE CALIBRATION PROCEDURE. This is the TRUE zero stroke position as measured by the LVDTs--when the top of the valve stem is in contact with the bottom of the valve plug.

At this point you need to measure the LVDT zero stroke voltage--and you can do that in the junction box closest to the SRV LVTs with a TRUE AC RMS voltmeter. Adjust the LVDT core until the LVDT output voltage until it is approximately 0.700 VAC RMS, plus-or-minus (+/-) 0.020 VAC RMS--so anywhere between 0.680 VAC RMS and 0.720 VAC RMS is acceptable. And then lock the LVDT core with the jam nut and check to see that the LVDT output voltage has remained between 0.580 VAC RMS and 0.720 VAC RMS. Do this for BOTH LVDTs (if so equipped).

Check Toolbox to see that the null bias current value for the SRV is set for approximately 2.67%. It shouldn't be less than 1.44% nor more than 4.00%; if it is, either something is very wrong with the servo-valve <b>>>>OR<<</b> (more likely) there is something very wrong with the way the servo-valve null bias has been adjusted in the past. If it's not 2.67%, the best thing to do to get a good starting point is to set it to 2.67% and start with that to try to get a good LVDT calibration.

While in Toolbox, check to see that the AutoCalibration "bounds" are set for 0% and 100% (typical for an SRV).

Then do what's necessary to establish hydraulic pressure and get it to the SRV servo-valve and then perform an AutoCalibrate routine. When the AutoCalibrate routine is finished, you should then verify the accuracy of the calibration using the Manual positioning function of AutoCalibrate, while actually measuring the valve position. If the calibration was done for 0% to 100% from minimum to maximum, you will need to use the Manual position to move the valve to a reference position of, say, 120 % (which it can't achieve--but you are just trying to drive the valve to the maximum opening), and measure the stroke (either using a dial indicator, zeroed at the zero stroke position, or a dial caliper). If the measurement of stroke from minimum to maximum is, say, 37.5 mm, when you input a reference position of 25%, you should measure approximately 9.375 mm; for a reference position of 75% you should measure approximately 28.125 mm. If the reference is 25% and you measure 8.846 mm and the feedback is, say, 23.56%, the likely problem is the null bias current needs to be a little more than it currently is. ADJUSTING THE NULL BIAS CURRENT TO MAKE THE ACTUAL POSITION (AND THE FEEDBACK) EQUAL TO THE REFERENCE IS NOT EASILY LENT TO A CALCULATION--RATHER IT'S KIND OF A TRIAL-AND-ERROR THING. Try increasing the null bias, in this example, to say, 3.00% and then verifying the calibration again (NOTE: It's NOT necessary to recalibrate the LVDT feedback if you change the null bias!!!)

When you are done with the calibration/verification procedure remove the feeler gauge--but be sure to note the actual gap. NOTE THAT WHEN YOU REMOVE THE FEELER GAUGE THE VALVE STEM MAY DROP DOWN WHICH WILL MAKE THE LVDT FEEDBACK GO SLIGHTLY NEGATIVE. <B>THIS IS NORMAL--AND EXPECTED!!!</b> If the gap was measured at between 0.030-0.050 inches, the feedback should not go much less than -5.00%, but sometimes it does. If that happens, we can fix that--just let us know.

<b>Based on the information provided,</b> it seems that the zero stroke voltage is MUCH less than the typical minimum value of 0.700 VAC RMS. This means that the feedback is NOT linear at low stroke--which could be part of the problem.

The above should get the LVDT calibration to a known good point to start from. Run the turbine. If after a trip or shutdown the LVDT feedback goes to -13% again, then it's a safe bet that the bar which has the LVDT cores bolted to it is NOT pinned to the valve stem. Or, it's not a combined SRV/GCV assembly. While it's not likely, the valve plug and valve seat may be improperly assembled or so damaged that the valve doesn't close properly each time--which should be evident at some point by a leaking ("passing") SRV. But, nothing should change the calibration, or the LVDT output voltage (presuming they are good LVDTs, they are properly mounted, and the core jam nuts are properly locking the cores in place). Nothing should be affecting the actual stroke of the valve--if the valve is in good shape.

<b>Since we don't know what kind of SRV the unit has</b>, it's possible that it has a rotary cam vee-ball type of SRV (usually made by Fisher). If this is the case--and it's an older type of rotary cam vee-ball valve it may have an RVDT (Rotary Variable Differential Transducer)--which has a different linear range of travel and output voltage than the typical (linear) LVDT. Newer rotary cam vee-ball SRVs were equipped with LVDTs which made calibration and operation better than the RVDT-equipped valves. But, if the valve actually has an RVDT instead of an LVDT, then the voltage ranges you reported are more typical of the linear output range of the RVDT.

There is one other possibility--but, again, it's not a Mark VI issue. And, that is that there is something wrong with the actuator (oil or dirt or a binding actuator piston/rod) that is not allowing the actuator rod to fully descend every time the valve closes. This has happened on a number of machines. Also, sometimes the valve stem, itself, where it passes through the seal, gets hung up and won't drop down fully every time the valve closes--but the closing spring is so forceful (unless it's broken!!) that it should closed the valve fully every time.

But, again, there is most likely <b>NOT</b> a Mark VI issue that TSM/telnet is going to help with finding out what's wrong.

Finally, some turbine are equipped with IS (Intrinsically-Safe) barriers on the LVDT circuits. And, these barriers have been known to fail and cause intermittent problems--especially the powered ("active") barriers; but also the "passive" (non-powered) barriers.

<b>PLEASE</b> write back with the answers to the questions, perform a good LVDT calibration, and report those results (measured gap; verification results; as-found and as-left null bias current values), and the results of initial operation.

(It's been a VERY long time since I've used TSM/Telnet, and I don't have access to my notes at this writing. I do know that based on the information provided it's not going to help in solving the SRV LVDT feedback problem. It's only going to consume more time and generate more questions.)
 
Top