Measuring water level in a tank

Interesting.... Cannot envisage the sensor on a rod, however, what is the reading of the pressure from empty to full?
For example is it 0-0.75 bar or mbar etc. or even just a digital value ?
for example 0 - 2546
If you have a good range reading from the sensor for empty to full that you can scale into 0-100 then there is no problem.

parky said:

Interesting.... Cannot envisage the sensor on a rod, however, what is the reading of the pressure from empty to full?
For example is it 0-0.75 bar or mbar etc. or even just a digital value ?
for example 0 - 2546
If you have a good range reading from the sensor for empty to full that you can scale into 0-100 then there is no problem.

Click to expand...

Bad choice of words on my part. When I said vertical rod, I just meant that the sensor is attached to a rod (actually just a vertical piece of unistrut). I'll check the pressure readings at the PLC next week. Cheers again.

Mas01 said:

Bad choice of words on my part. When I said vertical rod, I just meant that the sensor is attached to a rod (actually just a vertical piece of unistrut). I'll check the pressure readings at the PLC next week. Cheers again.

Click to expand...

Is it absolute or relative pressure? Is is submerged? Does it have a little pipe (tube) in the cable?

Most level sensors have the sensor face built into the fitting so that the sensor pressure face is directly into the vessel via a boss and screws in so the face is almost flush with the inner surface of the vessel, this is more to do with hygiene than accuracy but a short stub is not a problem if the medium is water.

If the tank's shape is as drawn, then the 0 - 100% indication will be non-linear. So an indication of 50% full would only be approximately 33% of the tank's capacity.

i.e. it would be wrong to indicate 50% full on a tank that can hold 20,000 gallons, when in fact it only has about 6,000 gallons in it.



This is easily correctable with a dip or strapping table. It depends on what the OP needs.

I sort of agree Daba, but he only wants to display it as a rough level on an HMI the actual contents is not important I was going to suggest he could use the two portions of the tank and when it reached the cross over level change the scaling but seems pointless as I assume for the existing top up controls probably use the sensor so it kind of works already. The process it controls is secret to their company and he is not allowed to reveal it as such. so it is difficult to post real solutions and this is only one of many posts asking for ideas.

Yes, all I'm really after is an indication of the water level, ie a bit like a see-through panel on the side of a kettle. I'm only interested in water level, rather than volume of water.

Mas01 said:

Yes, all I'm really after is an indication of the water level, ie a bit like a see-through panel on the side of a kettle. I'm only interested in water level, rather than volume of water.

Click to expand...

Is it submerged as a whole? If it is, it should be a hydrostatic level sensor and not a pressure sensor.. (it is the same thing but not the same thing )
Look at the cable from the sensor, is it thick and with a tube in it...

m_turk said:

Is it submerged as a whole? If it is, it should be a hydrostatic level sensor and not a pressure sensor.. (it is the same thing but not the same thing )
Look at the cable from the sensor, is it thick and with a tube in it...

Click to expand...

Yes, it's fully submerged.

Mas01 said:

It's not below any low level on the tank.

Click to expand...

It will not be able to measure any level below the sensor.


I would not even bother with figuring out the pressure-to-level stuff, instead do a simple calibration:

• Whatever low_raw_value you get when the level is below it, that is 0%.
• Whatever hi_raw_value you get when the tank is full is 100%.

The level, in percent is then [raw_value * m + b], where


m = 100 / (hi_raw_value - lo_raw_value)


b = - 100 * low_raw_value / (hi_raw_vale - lo_raw_value)

Mas01 said:

Yes, it's fully submerged.

Click to expand...

I think this is a hydrostatic level sensor, check the tube... That would mean that your "bar" measurement is actually just meters divided by 10...

So you could just change the unit and multiply by 10, or do as drbitboy said...


edit: but it is very weird that someone would install one, with the rod-holder and then not show it as a level but as a pressure value..

Already posted some code for him but have included In_Low, In_High Out_Low Out_High as fixed values just in case he needs to re-use the code for other purposes. however, re-iterated the fact that those fixed values could be retentive variables so he could have an engineering page for calibration but that seems unlikely because of the tank shape. I also converted the raw value to a real so the scale would be a little more accurate should he need the formula code for other purposes.

parky said:

Already posted some code for him but have included In_Low, In_High Out_Low Out_High as fixed values just in case he needs to re-use the code for other purposes. however, re-iterated the fact that those fixed values could be retentive variables so he could have an engineering page for calibration but that seems unlikely because of the tank shape. I also converted the raw value to a real so the scale would be a little more accurate should he need the formula code for other purposes.

Click to expand...

You told everything right.. It is just that at first I thought he was talking about a pressure sensor.. Mounted on the "wall", pipe.. And then he mentioned the rod, what is a standard when using a hydrostatic level sensor so it doesn't bounce, and it has to be submerged... He should have a range of the sensor on it in depth not bars. It is neve used for pressure measurement, it is weird he has it as pressure..

Also, is submerged sealed pressure sensor (never seen that) would have to be absolute and not relative without a tube. That would not give him a proper reading.

From what I have seen on some of the code & some of the equipment that appears to be on this system it is not very well designed and probably to a cost rather than a spec.
For example they have used Thermocouples on a low temperature range instead of PT's it is well known these are not very accurate at low temperatures and they used multicores for interface between the main panel and the TC panel interface so in effect the cold junction appeared in the interface but the PT for cold junction was on the card so it caused an error of about 4 degrees, they compensated using a fixed value subtracted from the actual value in the PLC code. There were some other oddities where they changed certain parameters of the analogue card buffer memories but did not make a lot of sense.
So That is why I suggested checking the low & high level in the tank and the corresponding raw values so he could scale it in some form.

parky said:

For example they have used Thermocouples on a low temperature range instead of PT's it is well known these are not very accurate at low temperatures and they used multicores for interface between the main panel and the TC panel interface so in effect the cold junction appeared in the interface but the PT for cold junction was on the card so it caused an error of about 4 degrees, they compensated using a fixed value subtracted from the actual value in the PLC code. There were some other oddities where they changed certain parameters of the analogue card buffer memories but did not make a lot of sense.

Click to expand...

On this one, I decided to bypass the second terminal block (and multicore cable) and wired the TC directly to the terminal block inside the PLC cabinet. I connected the RTD back in its original place. But, yes, a PT would be better.