sg24 shunt calibration

[NSWC Ben]

Hi,

I apologize in advance for the long post. 

I am using an SG-24 in conjunction with a pressure gage (P900 Schaevitz - spec's attached) but my measured pressures are about ~13% less than what my precision calibrated pressure transducer is measuring.  I have confirmed that my precision instrument is working correctly, so I am stuck with the reality that some aspect of my setup or shunt-calibration of the SG-24 is providing the error. 

Due to my confusion over how to perform the shunt calibration, I am guessing this is the source of my error.  I am attempting to determine the correct calibration factor to multiply my voltage measurements by.  Following the instructions on pages 18 and 19 of the SCC-SG User Guide, I am doing the following steps: 

-Calculate the expected voltage (Vex).  This is based off of the resistance values of the pressure gage, and of the shunt calibration resistor. 
-Null the circuit using the nulling pot's (get voltage reading to be 0.0 at ambient pressure)
-Determine the actual voltage (Vmeas) by using the SG11 to shunt the SG24.  The result is (Vmeas).  The ratio of the (Vex/Vmeas) gives the calibration factor, although the gain of the SG-24 must also be accounted for by dividing the result by 100. 

Again, this is all outlined in detail on pages 18 and 19 of the SCC-SG User Guide. 

I understand this process, however where I am unclear is which resistance values to use in the calculation of (Vex) in the first step.  The User Guide says:

(Rscal) is the value of the shunt-calibration resistor
(R4) is the nominal active strain gage element value
(R3) is either the value of a completion resistor or a second active strain gage element nominal value

It is clear that (Rscal) is 301kΩ since this is what is provided by NI with the SG-24.  However, I am not clear on what (R4) and (R3) should be.  I am not using any sort of completion resistor to my knowledge, so to my understanding (R4) and (R3) are the values of the resistors in the Wheatstone Bridge inside of my pressure gage (P900 Schaevitz - spec's attached).  The User Guide gives an example for a quarter-bridge configuration, however my pressure gage is a full-bridge configuration with different resistance values.  I have measured these values with my multimeter and two of them are 874Ω, the other two are 742Ω, and across any pair measures 1005Ω.  So, I'm not sure which value is (R4) and which value is (R3). 

I have followed the User Guide's example (Figure 18) of shunting (Vex+) and (AI+) on the pressure gage.  This results in having one of the 874Ω resistors shunted.  But, in order to get a calibration factor that resulted in sensical pressures, I substituted into the (Vmeas) equation an Rscal of 301kΩ, an (R3) of 1005Ω, and an (R4) of 742Ω.  This results in an overall calibration factor (Fc/100) of ~0.8, but like I said the measurements are about 13% lower than they should be. 

Some questions: 

-If I only shunt one resistor, I assume this resistor would be considered (R4), so then what is (R3)?  Do I just pick a random value of the other resistance values that works?  Because I've tried this and still haven't found a combination of using 874Ω, 742Ω, or 1005Ω that results in creating a calibration factor that provides the correct pressure reading. 
-Should I shunt two resistors?  If so, will any two work?  Then I'm assuming (R4) and (R3) would be the two resistors that I shunt, but which one I chose for (R3) and which one I chose for (R4) will dramatically affect the calibration factor.  So which one is which? 

I'm just generally confused over this.  Can someone please help?

Thanks,

Ben

[Neal M]

Hello Ben,

If I am understanding correctly, what you are trying to do is a shunt calibration on a pressure transducer with a built in full bridge completion.  The problem comes in when trying to calculate the correction factor, more specifically the delta V Expected.

The way that I figure out what is going on with problems like this is to step way back and look at this from a circuits standpoint.  What we are doing with the shunt calibration is to take our bridge and skew it a known value.  We then calculate what we expect to see across the bridge, and measure it.  The ratio of these two values will give use our correction factor (what we are expecting over what we see). So all that first equation (on page 18) is doing is calculating the voltage difference across the bridge.  It is the difference between two voltage dividers, one for the side with the shunt in parallel to leg 4 and then leg 3 minus the other for the 2 equal resistors (bridge legs 1 and 2, which is just V_ex/2).

This means that  from the sounds of things, you need to recalculate these voltage dividers since the legs of your bridge are not equal (?).  This could be causing your error. If you do this instead of using the voltage equation on the documentation, it does not really matter which leg you shunt across (but you should only shunt across 1).  Usually, it is the same side of the bridge as the transducer to "simulate" pressure.

So, after all that, here is a summary:

1. You need to shunt across one of the legs on the side with the transducer.  If using the equation given, it needs to be R4.
2. The equation on page 18 is just a difference of voltage dividers (with simulated strain).  It assumes equal legs. If your legs are not equal, you need to calculate this yourself.
• V_exp =
  
• V_exp = V_BetwLegs3$4 - V_BetwLegs1&2
• V_exp = V_divider1 - V_divider2

[Neal M]

<CORRECTION>
the first bullet is suppost to say V_exp = <attached image>

I just didn't edit the post to put in the image fast enough...

[NSWC Ben]

Thanks for your response Neal.  I actually stepped back a little bit, and realized that I was consistently having an error of about 13% when my correction factor was 0.87 and an error of about 21% when my correction factor was 79%.  So, I decided to totally eliminate my correct factor and now my values are correct.  I'm going to verify this at much higher pressures later today, but it seems to at least be true from about 0 - 30 psig and on two seperate gages.  So this leads me to a new, much simpler question: 

When using the SG-24 in conjunction with the SG-11, is it necessary to go through the shunt calibration process and calculate the correction factor at all?  I had assumed this was true all along, but it's now appearing to be the case that if the SG-11 is installed in the chassis with the SG-24, one of the modules or the chassis, or the LabView software or something takes care of all this math for you. 

All I am doing is multiplying by a v/psi gage sensitivity value to get pressure in psi and my answers appear to be dead on.  The gain of 100 and everything appears to be taken care of somewhere within the hardware or software.  In fact, to make sure I wasn't going crazy I took out the SG-11 and compared the values of when it was installed in the chassis and the values appear to change real time from when the SG-11 is not installed which further supports my theory that somehow the SG-11 in conjunction with the SG-24 is doing all the work of calculating the correction factor, accounting for the SG-24's gain of 100, etc. 

This, however, is not spelled out in the manual, at least not obviously, so I'm still a little confused.  It appears to be working and that's great, and like I said I'm going to go confirm this holds for much higher pressures as well (up to probably 600 psi) but I would like to know why it works so that I'm not surprised later. 

Does the the SG-11, when used in conjunction with the SG-24, take care of the correction factor calculation and the gain and everything, so that these correction factor calculations are only necessary when shunting without the SG-11?

Thanks Again,

Ben

[Neal M]

Hello again Ben,

I am glad you got your setup working.  From what I have found, the gain is taken care of in software, but there is no interaction between the two modules and the correction factor must still be taken into account manually.  I found this Knowledgebase that has step by step instructions on how to use the SG11.  It indicates that the Fc must be implemented either by a custom scale or programatically (so you can do it every time you run the program).

This makes me wonder what you are seeing when you say "the values appear to change real time from when the SG-11 is not installed".  I would like to find out more about what you are doing and what you are seeing.  If there is an interaction between the two modules, I would like to clarify the documentation.  If you could describe exactly what you are doing and seeing, I would appreciate it.  I will also look into the issue more here.

[NSWC Ben]

Neal,

Thanks for the link, but unfortunately the process that's outlined in the link is the same process I have been following.  Here's a more detailed description of what I am doing:

I have an SC-2345 chassis with the SG-24 and SG-11 modules installed.  I have tried using both the PWR-01 and -02 options, both yielding the same results that I describe below.  I have a Schaevitz P903 full-bridge pressure gage installed and wired according to the SG-24 full bridge documentation.  All four of the legs on this gage are 1000Ω.  So, R3 = R4 = 1000Ω.  Rscal = 301,000Ω.  Vex = 10V.  So, when calculating the expected voltage I get the following results: 

Vexpected = (R3*Vex*(Rscal+R4)) / (R4*Rscal+R3(Rscal+R4))-(Vex/2)
= (1000*10*(301,000+1000)) / (1000*301,000+1000*(301,000+1000)) - (10 / 2)
= 0.0083

My VI simply has a raw voltage measurement, and then a psi measurement which is achieved by multiplying the voltage by 0.0000092 V/psig.  After nulling the SG-24 using the trim pots to near zero volts (~0.000001V which corresponds to ~0.10 psi on my gage), I shunt the gage by activating the relay on the SG-11 through one of the digitial channels on the SCC-2345.  This results in measurements of ~0.0095 V, which corresponds to ~1044 psi.  So, with using a Vmeasured ~= 0.0095 V:

Fc = Vexpected/Vmeasured ~= 0.87. 

So, I then would add a block into my VI where I multiply all the resulting data by 0.87 according to the SCC-SG literature.  However, when I was confirming the accuracy of my setup in our calibration lab all of my readings measured at 50, 100, 200, 400, and 600 psi were ~13% off.  It took me a while to put the pieces of the puzzle together, but I eventually realized that if my correction factor so 1, so that Fc = 1, then all of my measurements would be spot on.  This trouble-shooting process I've described above was done over a matter of weeks, but I have since gone back and confirmed that from 0 - 600 psi in the aforementioned steps this setup, with Fc = 1, is within 1% of the calibrated psig value (which is pretty darn good). 

So, again I am left with the question of do I really need to do the shunt calibration or is this math and process somehow done for me?  It really appears to be taken care of somehow, whether it be in LabView behind the scenes, in the hardware of the SG-11 and/or SG-24, in the SC-2345 or somewhere else.  I would very much like to know how this is being done so that I am not surprised later by not fully understanding the system. 

Also noteworthy, as I said in a previous post, removal of the SG-11 appears to change the resulting voltage measurements.  Here is an example:

I first null the SG-24 with the trim pots to ~0 V and ~0 psig with the SG-11 installed in the chassis.  I then turn off the power to the SC-2345 and take the SG-11 out of the chassis.  If I now take measurements, my channel reads ~0.0085 V (934 psi).  If I then turn the chassis power off again and put the SG-11 back in the chassis the measurements return to their original readings of ~0V and ~0 psig.  So, obviously, somehow the SG-11 is playing some active or passive role in the calculation or measurement of the pressure gage data.  I have tested this with both the PWR-01 and PWR-02 options both having the results described above. 

Another note is that this change does not occur if I remove the SG-11 module with power going to the chassis.  So, if I null the SG-24 and then remove the SG-11 with power still being delivered to the chassis, there is no dramatic change in the measured voltage of the SG-24, even if I stop and restart my VI with the SG-11 removed.  I can't make sense of this. 

So, finally I am still left with the same question of whether or not I really need to do the shunt calibration or is this math and process somehow being done for me behind the scenes?

Your help is greatly appreciated.

Ben

[Neal M]

Ben,

The short answer to your question is that no, you do not need to do the shunt calibration.  The reason for this is not, however, that it is done automatically.  What the shunt calibration is doing is exactly what you are doing to check the correction factor:  Seeing if, when under a known load, what you are measuring is what it you expect.  The shunt calibration does this by changing the value of one of the legs in your bridge and doing the math, where you are actually doing it with pressure.

I still am unsure what is going on with the math though.  I thought that it could be that the excitation value may not be exactly correct (not exactly 10V), but after measuring my device (I checked out the SG24 and SG11 to check the behaivior you are seeing) and doing the math, it is not off by enough to change the Fc significantly.

I solved the equation for Vexpected in reverse (solve for R4=R3=x, Vexpected=0.0095, Rscal=301k) and found that in order for your bridge to have the voltage we expect, the legs need to have a resistance of 1145.98Ω.  Because they do not (you say they are 1000Ω), I think that maybe more than one leg of your bridge may be active.  This would mean that >1 leg changes with pressure, skewing the bridge differently than our equation expects.

As far as the behaivior you are seeing with removing the SG11,  I am pretty sure that there is no communication between the two modules.  The fact that you must configure the SCC chassis yourself in Measurement and Automation Explorer suggests that there is no way for the driver to detect if the module is actually there.  I tried to see the change in voltage when the SG11 is removed, but on my SCC-68, there was no difference.  I think what may be going on is that the power supply is somehow changing when the module is removed.

You should not hot swap the modules (it is not reccommended, but I personally have not seen one blow because of it), so I don't know what is going on when you do that...

So again:  What you are doing to test the sensor is equal to or better than shunt calibration,  but I do not know why the math is so off.