Help with electronics

[trundlenut]

I'm in the process of putting together a system to monitor water level using a differential pressure sensor and an arduino thingy.

 

On thing I'm not sure on is to do with cable lengths, voltage drops and signal problems.  Ideally the sensor would be about 30-40m from the arduino, 15m is probably the shortest distance that I want to get away with.  The sensor is a four wire jobbie which takes 10-16V to power and gives a mV output (nominally 47-53mV for the 0-5 psi range).  I have used similar sensors in the past with 10-100m cable lengths but these were propriety systems.

 

I am planning on using separate power supplies to power the sensor and the arduino.

 

Does anyone forsee any potential issues?

[wonkyewok]

As long as the arduino input is fairly high impedance it shouldn't draw much current, giving little voltage drop from the sensor. Will the cable be run through anywhere electrically noisy? Is it shielded? Does the sensor tie the 0V power line and 0V signal lines together? I think it could work fine , depending on circumstances.

[rustic]

I have been involved with the design of electronic weighing systems, which use strain gauges in a bridge formation.

The Bridge impeadance was 350 ohms per load cell. Systems could have 1, 4, 6 8 + loadcells.

They were driven with 10-15 Vdc and had an output of 2mV/V so max output was 20-30 mV, so the application is similar to yours.

Driving them 20+metres was not a problem, but we used high quality 6 core screened cable.

The systems were Weights and Measures approved, so would not be unduly affected with temperature or moisture or barometric changes.

The accuracy of the system was 1 part in 5000 or 0.02% However, to achieve this it was essential to sense the voltage at the load cells, hence the need for 6 wires. Additional circuitry in the electronics was required to monitor this voltage drop. Alternatively you can have a power supply built with remote sense, so the voltage is constant at the sensor. The High impeadance of the electronics, means that the signal line is not greatly affected by voltage drop or temperature changes etc.

So... you can see what can be achieved, not using sense will affect the accuracy of the system, but if you only want say 1 part in 200 or 0.5% then it should work reasonably well, but I would run a 6 core screened cable anyway, for later upgrade, and in the mean time double up on the both supply wires.

Screen needs to be grounded at the instrument, and unterminated at the sensor.

DO NOT run other cables bundled in the same conduit as this screened cable, unless those too are instrumentation cables carrying low voltages and low currents. Mains cables MUST be avoided at all costs.

The largest error with your method of level measurement will be changes in barometric pressure.

Using a differential pressure sensor should help to reduce this error.

Hope it helps,

best regards,

Richard

Edited 4 January, 2015 by rustic

[wonkyewok]

What he said      Definitely don't run the signal cables in the same conduit as mains cables etc.

My only other thought is that in "The art of electronics" (Horowitz and Hill) I think they suggest that it's better to ground the screen at the sensor, not the measuring instrument. (But it's a long time since I read that and my copy is at work).

 

It might be worth thinking about ground referencing too. Are you using differential or single-ended a/d at the Arduino end? The best (I think) would be differential measurement. But there may need to be a path to ground for input bias current from the a/d converter, depending on its design. Hopefully the manual discusses this. But... if you are using two separate power supplies, both of which are floating (ie their outputs have no connection to mains neutral), then even single-ended may be fine as you'll be defining the earth for the arduino board by the sensor 0V wire (I think).

 

Sorry if I didn't say this clearly. Having circuit diagrams really helps to clarify things.

[rustic]

Screening can be an art in it's self. The key point is never to use the screen to pass current, which might be the case if both ends are terminated.

From many years of design and RFI interference, I have found that electronic weighing equipment performs better against spurious RFI transmissions when the screen is connected at the instrumentation only.

Other instrumentation can behave differently, depending on the main cause of interference.

RFI could be from CB radio, mobile phones, PMR - 2 way radios, even wireless transmissions from car key fobs to remote weather stations and houshold alarms.

There can also be electrical interference from electric motors, flourescent lights, electric tools and kitchen appliances.

In extreme industrial circumstances, I have had to install twin screens to cables, sometimes the screen is connected at both ends to reduce the problem, or even fit additional supressor components.

It is a black art, but for this application, it looks fairly straight forward, and to be honest, I couldn't wait to get started, just to see how well it did perform.

So, I do hope you are able to post back how well you got on, and what accuracy and resolution and repeatability that you achieved overall.

I am still an engineer at heart... ( Retired but still messing around, more so, now we have a grandson lol).

We bought him one of those red plastic BMW pedal cars, from Toy's R Us, it must be the only one with orange flashing LED indicators, white LED head lights and Red LED tail lights.

Next Mod... as he is getting older, is a horn on the steering wheel.

Now.... how can I install a maxidot display and central locking? hmmm...

Well it is running on 12 volts, with a non spill rechargeable gel battery, and all correctly fused and wired up of course.

So anything is posssible, maybe turning fog lights.

Now what about a tow bar, and trailer lights..?lol..

[trundlenut]

Wow. A lot of information there thanks.

 

As long as the arduino input is fairly high impedance it shouldn't draw much current, giving little voltage drop from the sensor. Will the cable be run through anywhere electrically noisy? Is it shielded? Does the sensor tie the 0V power line and 0V signal lines together? I think it could work fine , depending on circumstances.

 

The sensor doesn't tie the power and signal lines together and I was planning on doing this.  I hadn't got that far thinking about the cable but I will be looking at shielded cable.

 

Is there a difference between shielded and screened cable, it seems like these terms are used interchangeably.

 

I have been involved with the design of electronic weighing systems, which use strain gauges in a bridge formation.
The Bridge impeadance was 350 ohms per load cell. Systems could have 1, 4, 6 8 + loadcells.
They were driven with 10-15 Vdc and had an output of 2mV/V so max output was 20-30 mV, so the application is similar to yours.

Driving them 20+metres was not a problem, but we used high quality 6 core screened cable.
The systems were Weights and Measures approved, so would not be unduly affected with temperature or moisture or barometric changes.
The accuracy of the system was 1 part in 5000 or 0.02% However, to achieve this it was essential to sense the voltage at the load cells, hence the need for 6 wires. Additional circuitry in the electronics was required to monitor this voltage drop. Alternatively you can have a power supply built with remote sense, so the voltage is constant at the sensor. The High impeadance of the electronics, means that the signal line is not greatly affected by voltage drop or temperature changes etc.
So... you can see what can be achieved, not using sense will affect the accuracy of the system, but if you only want say 1 part in 200 or 0.5% then it should work reasonably well, but I would run a 6 core screened cable anyway, for later upgrade, and in the mean time double up on the both supply wires.
Screen needs to be grounded at the instrument, and unterminated at the sensor.
DO NOT run other cables bundled in the same conduit as this screened cable, unless those too are instrumentation cables carrying low voltages and low currents. Mains cables MUST be avoided at all costs.
The largest error with your method of level measurement will be changes in barometric pressure.
Using a differential pressure sensor should help to reduce this error.

Hope it helps,
best regards,
Richard

 

I am definitely using a differential sensor vented to atmosphere, doing compensation on a non vented transducer just makes it more complicated. What I am looking to do it measure the stage (i.e. water depth) in a stream at the end of my garden, which has an annoying habit of flooding my garden.  The sensor I'm looking at has a 3.5m range, normal water depth is around 0.5m but this can reach in excess of 1.5m during a flood event.  Accuracy wise I'm looking at 1-2% of full range, the sensor will be in some kind of stilling tube but given the nature of the stream (especially during flood events) and the variations in level accuracy of circa 5cm is probably acceptable.  I am planning on taking multiple readings over say a 30 second period to give an average reading.

 

I'm planning on putting the arduino in the shed, or in the house and then running a 4 core wire down to the stream. So no mains power nearby.

 

What he said      Definitely don't run the signal cables in the same conduit as mains cables etc.

My only other thought is that in "The art of electronics" (Horowitz and Hill) I think they suggest that it's better to ground the screen at the sensor, not the measuring instrument. (But it's a long time since I read that and my copy is at work).

 

It might be worth thinking about ground referencing too. Are you using differential or single-ended a/d at the Arduino end? The best (I think) would be differential measurement. But there may need to be a path to ground for input bias current from the a/d converter, depending on its design. Hopefully the manual discusses this. But... if you are using two separate power supplies, both of which are floating (ie their outputs have no connection to mains neutral), then even single-ended may be fine as you'll be defining the earth for the arduino board by the sensor 0V wire (I think).

 

Sorry if I didn't say this clearly. Having circuit diagrams really helps to clarify things.

 

I'm afraid you've lost me on the second paragraph.

 

Where I used to work I just happened to manage a telemetry monitoring system, mainly doing water levels and flow rates, so I did have a telemetry logger installed in the stream with a GPRS modem so I had realtime monitoring. But since I changed jobs I had to hand back the kit.  The sensors we used were four wire mV output and thinking about it the cables were screened.  Also I have kind of answered my own question as the pressure transducers were sometimes installed 50-100m below ground level.

 

Unfortunately you are looking at probably £200 for a proper pressure transducer and cable, so I need to make something a little more Heath Robinson.

[VWD]

Looking at Heath Robinson ideas, first that springs to mind is some form of float with a magnet attached .At the warning point, you have a normally open reed switch . Alternatively - you could use a series of switches at differing levels to see the levels, via a set of LEDs. Add on to this using an increasing voltage level, (each switch increases the voltage) and a set of LED s fed via zener diodes to give a barograph display .

[trundlenut]

Looking at Heath Robinson ideas, first that springs to mind is some form of float with a magnet attached .At the warning point, you have a normally open reed switch . Alternatively - you could use a series of switches at differing levels to see the levels, via a set of LEDs. Add on to this using an increasing voltage level, (each switch increases the voltage) and a set of LED s fed via zener diodes to give a barograph display .

The problem with this is the fact it will be in a stream with a rather turbulent water flow and a fair bit of debris coming down stream.

Also the rate of change in water level is as important as the actual level.

Sent from my Nexus 4 using Tapatalk

[wonkyewok]

I interpret "shielded" and "screened" as the same - though there are different versions of these for specific applications (e.g. microphone cables). My off-the-cuff reaction is that you might get away with shielded 4 core cable (2 power, 2 signal), as long as the sensor doesn't draw too much current either, but Rustic's suggestion of 6-core so that you could measure the voltage received at the sensor is right if you want to be extra sure. (And especially if you have a power supply with "sense" inputs").

 

The detail may be important. If you want to post links to any of the documentation we can tell you anything that jumps out.

 

If you get the shielding / grounding wrong, at worst either the sensor reading will apparently drift or it will be noisy. But it should be fixable by changing which things are connected to which. You could make it up roughly to test first.