Pressure Sensor for water level measurement

[premkumar9]

OK, so the sensor is sitting at the bottom or really close to the bottom of a 3 meter vertical column of water that's capped at the bottom, right?

Yes. Close to the bottom.

If this was a test was it inside where line frequency noise would be present?

Yes

Did you use twisted pair shielded cable with the shield connected at one end only or something else?

Shielded, but not twisted pair. Shield connected to circuit ground.

 

[KeepItSimpleStupid]

Try a piece of shielded twisted pair cable and connect the shield to ground/EARTH at ONE end only.
Shielding supresses RFI
Twisting supresses EMI (power line noise)

What model meter are you using?

You can also try to lower the input Z of the meter. Put, say a 10K resistor at the meter. That way you don't have a 10 M ohm long wire.

In the long run, you may want a 4-20 mA or 0-20 mA transmitter. Then all, that''s required at the receiving end is a resistor and a power supply.

 

[premkumar9]

Try a piece of shielded twisted pair cable and connect the shield to ground/EARTH at ONE end only.

The cable is coming along with the sensor. We cannot take connection directly from sensor. It is sealed.

What model meter are you using?

Fluke 15B

 

[KeepItSimpleStupid]

https://www.fluke.com/fluke/inen/di...-Purpose-Multimeters/Fluke-15B+.htm?PID=78680

Accuracy is 0.5% + 3 digits, resolution is 1 mV. Input Z is 10 M. Lowering the input Z, should help a lot. Actually connecting the shield to ground, should also help.

 

[alec_t]

Accuracy is 0.5% + 3 digits

That gives a worst-case error of ~8mV in the measurement, which would swamp any variation in the post #31 readings unless a large part of the error remained constant.
If the post #31 readings do involve a mainly constant error, then the consistent decrease of the water pressure readings over a 4hr period suggests that the water sensor has good stability, whereas the more random variation of the atmospheric pressure readings suggests the air pressure sensor is less stable so may need upgrading.

 

[premkumar9]

If the post #31 readings do involve a mainly constant error, then the consistent decrease of the water pressure readings over a 4hr period suggests that the water sensor has good stability, whereas the more random variation of the atmospheric pressure readings suggests the air pressure sensor is less stable so may need upgrading.

Thank you for your valuable observations. Here the pressure sensor used for atmospheric pressure sensing was not temperature compensated while the one used for water level sensing is said to be temperature compensated. Could it be the reason for the non relationship between these two readings? If so what type of pressure sensor you suggest for compensating the variation of the water level sensor? (I presume that you consider the drift is due to atmospheric pressure variation)

 

[Tony Stewart]

To do an error analysis, I would measure the sensitivity from different sources of variance such as, temperature, supply AC, Vref, noise, and drift in ADC. To make precise measurements of 1000 ppm , your error budget , your calibration instruments need to have an accuracy of < 500 ppm and a resolution of ~100ppm for each measurement.

Your DMM may not be adequate.

 

[alec_t]

Here the pressure sensor used for atmospheric pressure sensing was not temperature compensated while the one used for water level sensing is said to be temperature compensated. Could it be the reason for the non relationship between these two readings?

Could be. Perhaps repeat the measurements, logging temperature too?

 

[KeepItSimpleStupid]

Accuracy and repeatability are two different specs and repeatability didn't appear to be listed. It's not specified, but I would bet, that +-1 digit which is indeed the resolution.

Contrary to what I thought, you need absolute pressure, no no port to the atmosphere.

Lowering the input Z of the meter to 10K||10M (nearly anything will do) and grounding the shield are the two things that might help.

This thing is powered by something too and we don;t know what that is.

I think we need to go back to the beginning. We went from a DAM to a controlled experiment in a tube, The game of 20 questions is turning to 100+

List everything and spec sheets and models and describe the experiment. The power supply and the lengths of wire.

here https://www.wunderground.com/US/NY/Buffalo.html, I just picked a city. Note the graph of barometric pressure vs time. A reminder that this is normalized to sea level. It's not the actual pressure.

 

[Tony Stewart]

BTW 1000 ppm is 50 mm/50 m
50 ppm resolution on a 2V scale is 0.1mV

Your resolution is only 1 mV and a typical Fluke 80 Series V DMM has an accuracy of 1000 ppm +/-1 but the model 87V has a resolution of 10 uV and accuracy of 500 ppm or 0.05%, which is adequate.

 

[premkumar9]

the consistent decrease of the water pressure readings over a 4hr period suggests that the water sensor has good stability

Do you think that the drift seen must be due to atmospheric pressure change and not due to any problem with the sensor?

 

[Tony Stewart]

I think you need to follow my advice and determine ALL sources of influence with more accurate tools and rigourous stress measurements with correlation to output with statistical analysis on spreadsheet.

This includes RF noise measurements, that may get rectified as a DC offset.

 

[premkumar9]

I think you need to follow my advice and determine ALL sources of influence with more accurate tools and rigourous stress measurements with correlation to output with statistical analysis on spreadsheet.

Mine is not a full fledged development facility. I have limited tools with me. Since this is my first experience with this type of project, I was expecting practical guidance from people who faced similar problems and solved them. Since I am shifting my work place, I could not do the experiments for some time.

 

[Tony Stewart]

All you need is a good scope with precision reference offset or better voltmeter and good test records for mean, 3sigma from different variables on error budget to analyze problems.

Understand that bandgap references used everywhere from DMM to ADC are usually 100 ppm per degree C. , better or worse.

 

[premkumar9]

All you need is a scope and better voltmeter.

Have some difficulty in arranging a better voltmeter

 

[Tony Stewart]

keep temperature constant +-1 deg? If not possible then design thermal regulator for ADC to max temp using LDO , thermistor, pot transistor with foam as a temperature control for chip like 40'C +/-0.5 deg.

Any LM317 will work.

 

[KeepItSimpleStupid]

I looked at what eric provided and I also looked here: https://en.wikipedia.org/wiki/Pressure_sensor

It really looks like the parts provided by GE are engineered correctly. The sensor vents to atmosphere above the water and then transitions into a non-vented cable. They offer devices that will power the sensor with 4 mA and you can get a current output which can travel long distances without error. The Wiki article is unclear in places, but it does point out pitfalls with a certain type of sensor and corrections.

Sort of sound like, we bought this, make it work. That sort of scenereo happened when I was hired for my first real job.

 

[ericgibbs]

hi kiss,
My company designed and sold well over a hundred Tide and River gauges worldwide using the sensors I mentioned, they work very well.
They used either the 0 to 20mA, 4 to 20mA loop or 1 to 5Vdc output sensors, for 1Bar thru 2.5Bar depth ranges.

The dredging companies and harbour authorities are very fussy about accuracy as it effects the cost of dredging operations.

The only problems we had were due to cables damaged in service.

Sort of sound like, we bought this, make it work. That sort of scenereo happened when I was hired for my first real job.

I agree with your statement, seen it so many times.

E

 

[Tony Stewart]

Without more accurate data and monitoring temp, nothing indicates if is the sensor, cables, EMI, diff amp or the ADC at fault.

 

[KeepItSimpleStupid]

"We" know it's a handheld DVM that has accuracy issues. Without accuracy issues, it should be repeatable within the noise. It's drifting.

What from? Donno. The xtal ball is having some issues. I have my nose in the sand and I don't know how deep it is. This "thing" is supposed to tell me but it looks like a rock on a string and it didn't come with any instructions.