Pressure Sensor Behavior 1

[tlerunner]

I recently observed some strange pressure sensor behavior and am looking for some troubleshooting assistance. We have a number of 0-10 barg, 4-20 mA output pressure sensors. They are installed in an air and coolant supply system. The air compressor motor in the system is a brushless DC motor. When it is running, a small amount of electrical noise is transmitted to the frame. This noise has a peak to peak amplitude of about 220 mV.

If I connect one of the pressure sensors to its electrical harness and just let it hang in air, I get 4 mA out of the sensor which is read by my controller as 1 VDC, since the current is run through a 250 ohm resistor. This is expected. Now if I apply a 60 psig pressure source to the sensor I read about 2 VDC or 4.2 barg, which is as expected. However, if I take a wire and briefly connect the sensor housing to the frame, our controller now measures 4 VDC, or about 8 barg. If I remove the wire, the
controller still reads 4 VDC. The only way to restore the sensor to its correct output is to unplug it from its electrical connector and plug it back in.

This only occurs when our air compressor motor is running.

I recently read that it is not a good idea to install a metal housed pressure sensor into a metal pipe. Is this true?

 

[danw2]

Your 220mV noise reading is likely seen as normal mode noise. Your setup likely has excessive common mode voltage.

Make two DC voltage measurements, one from one side of the 250 ohm resistor to ground, the other from the other side of the 250 ohm resistor to ground, when the motor is running and when it isn't. Let us know the results.

Is this a 2 wire loop powered transmitter with a power supply in series with the 2 signal wires? Or 3 wire (shared power supply ground with signal common?) or 4 wire (2 power wires, 2 signal wires)

What kind of power supply is being used?

The power supply output might be grounded, try floating the p/s output if it's grounded.

The input of whatever you are reading the signal with could be 'not isolated' and have its negative terminal grounded, which is serving as a ground path. What are you reading the voltage with? (Brand/model)

I am unaware of advice to not mount metal housed pressure transmitters in metal pipes. I've seen an awful lot of them mounted that way.

Dan

 

[tlerunner]

Hi Dan,

Thanks for the response. The setup is at another facility but I will be traveling back there again soon. I will take the measurements you have suggested.

The 24 VDC power supply is floating, its negative is not connected to the frame of the system.

The transmitter is a two wire, 24 VDC, current loop. The power supply is an Excelsys, Xgen series, switching power supply.

We are measuring the sensor output across a 250 ohm resistor with a National Instruments NI9205 analog input module mounted in a cRIO chassis.

I am currently leaning toward a ground loop problem. I suspect that the negative of the 24 VDC power supply and the frame are at different potentials. Here are two possible solutions:

1. Isolate the metal housing of the sensor from the pipe it is installed in. We have a couple other sensors that happen to be installed this way and they working fine.

2. Connect the negative of the 24 VDC power supply to the frame.

Tim

 

[itsmoked]

You need to also measure the AC differences between the frame and the sensor body. It could be 0VDC but plenty enough AC to cause latchup on your sensor or the NI9205.

The correct solution is to make sure your frame is grounded to the same ground as the NI9205's ground not isolation.

The motor power cable doesn't sound like it is very well grounded either. It should probably be sheilded and grounded on both ends. That would be a better place for those ground currents to circulate.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[itsmoked]

BTW tlerunner: I gave you a star for the clear and concise details you've provided. A rare treat.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[tlerunner]

Keith,

Thanks for the additional ideas and the kind words.

What do you mean by "latchup?"

Tim

 

[danw2]

I suspect Keith says 'latchup' like I say 'saturate', and if that's true, then either refers to your situation where the transmitter's reading 'locks up' when you connect the wire from the frame to the xmtr housing and the reading doesn't back to its pressure reading, it gets 'stuck'.

I echo Keith's commendation of your initial well-worded 'situation report' and your reply with specifics. You might not be aware of how unusual that is on public fora. Unusual enough to warrant a 'star'.

Dan

 

[itsmoked]

Yes danw2 has it.

Saturation is generally non harmful and immediately reversible. Latchup is not reversible and often results in destruction of semiconductors.

As you can imagine inside the ICs used in your systems there are internal voltages and currents. There are also some strict level requirements. If they are exceeded or levels reversed the device's outputs are entirely non-specified. It often causes the internal structures to turn ON in a manner that they cannot recover from.

In some rare cases like yours the device seems to tolerate this and is not damaged. In many others it causes over current and immediate thermal damage.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[tlerunner]

Thanks again, I appreciate it. It sounds like the pressure sensor is experiencing saturation resulting from a ground loop situation. I wonder if the sensor has been damaged making it more susceptible?

I am currently putting all your ideas into a test procedure that I can run on my next trip to the facility.

Please keep any other suggestions coming, I will add them to the test procedure. Also, I will provide an update after I run the test.

Tim

 

[tlerunner]

Hi, just wanted to provide an update. It appears that what is happening is that the DC brush-less motor and controller are radiating a significant amount of noise. Some of my sensors are installed in stainless steel pipes that have no electrical connection to the frame of the power plant. The radiated noise is causing a significant voltage potential between the pipes with the sensors and the frame of the power plant, up to 328 V peak to peak. See attached picture of the o-scope. This high of voltage is being discharged through the sensor either causing it to go into saturation or latch up. The solution appears to be ensuring that all piping with sensors installed is tied to the frame of the power plant.

Does this make sense?

Thanks,

Tim

 

[itsmoked]

That would be the correct solution. All the framework should grounded. For noise and safety.

Keith Cress
kcress -

http://www.flaminsystems.com