Help with Magnetic Flow Meter Project

[gradyh]

I have undertaken a project that has me a bit past my skill level in electronics and could use some help troubleshooting. I'll try to keep this as brief as possible. I am trying to build a proof-of-concept electromagnetic flow meter. If you aren't familiar, wikipedia has a good overview (https://en.wikipedia.org/wiki/Magnetic_flow_meter). The basic premise is that the a conductive fluid passing through a magnetic field will generate an EMF between two electrodes which is proportional to the fluid velocity. Here's some photos (http://imgur.com/a/9PNV5) and a description of my current setup:

• I'm using an arduino as a signal gen to supply a biphasic square wave to a dual-channel H-bridge
• The H-Bridge is running off a bench power supply (12v) and driving a pair of hand-wound electromagnet coils
• There are four electrodes in contact with the fluid flowing through the pipe - two are diametrically opposed, perpendicular to the electromagnets (signal), the other two are about 6 inches from the magnets on either side (shield)
• The two signal electrodes go into a seeed studio differential amplifier circuit with the two shield electrodes as the reference
• I'm measuring the signal from the amp using my o-scope

The problem I'm having is that the noise coming from the amp is really high - at least much higher than any actual EMF generated by the water passing through the magnetic field. I can see the initial voltage spike when the magnetic field turns on or off, but other than that, there doesn't appear to be any difference in the signal between when the magnets are on and off. Here's some more details if they help:

• I have verified that the signal from the H-bridge to the electromagnets is working correctly.
• The power to the differential amplifier circuit is provided by a separate 5VDC transformer (not the power supply running the magnets).
• If I touch the shield electrodes, the amplitude of the noise goes down.
• Adding a capacitor between the signal electrodes did nothing.
• I've found one other person who has done and posted about it, but she did not post any circuit diagrams or photos. I've emailed her but did not hear back (https://nehagirme.wordpress.com/2011/11/19/electromagnetic-flowmeter-design/)
• The pump is a 120VAC sump pump. I've calculated the fluid velocity at about 1 ft/s (0.3 m/s).
• The electromagnet coils are 600 turns of 30 AWG wire. I think about 35 ohms each.

Any ideas on how to reduce the noise or just generally get a usable signal from this demo are very much appreciated.

[retrolefty]

Correct me if I'm wrong but is not water a poor/bad media to try and develop a mag flow meter. Unless the water has salts or significant water ions, it will not generate the potential difference you are attempting to measure. Mag meters are used with conductive media.

[gradyh]

Correct me if I'm wrong but is not water a poor/bad media to try and develop a mag flow meter. Unless the water has salts or significant water ions, it will not generate the potential difference you are attempting to measure. Mag meters are used with conductive media.

Magnetic flow meters are used for water and even potable water (as opposed to raw water, wastewater, etc.) but you make a good point. Perhaps if I test out a more conductive liquid I would have a lower signal to noise ratio. Thanks.

[SeanB]

A quick way to reduce the coupling of the electrical noise is to use shielded cabling for all those long antennas you have on the system, connecting the shields together at the amplifier. Shield the reference electrodes as well. Use a metal box ( even a small cardboard box with aluminium foil glued to the outside and a lug crimped to the foil to make an electrical junction for the shields and connect those to ground) for the amplifier. As well the leads to the coils need to be twisted wire, so grab some network cable and replace the croc clip leads with some twisted pairs so you do not couple extra energy into the measurement system.

Then to further reduce noise take the coils and wrap with copper tape. leaving a small insulated overlap so you do not make a shorted turn. Shield of the coils connected with coax cable braid to the PSU for low impedance, following the twested pairs ( or put the twisted pair inside the braid as well) so you keep the switching noise away from your measurement.

Your major problem is poor layout, with large loop areas and long unshielded wiring with high impedances picking up both mains hum and any RF energy in the area. Reduce loop areas in wiring, and shield all high impedance wiring, and shield all RF noise sources from the measuring system and you should get a measurable signal out.

Once you get a measurable signal then you can look at synchronous detection to reduce the noise and improve your accuracy, but you need to reduce the overall noise first.

[Kleinstein]

The "noise" seen at the scope is mainly higher frequency. So a low pass filter might help to see more.
For later real reading synchronous detection is a way to suppress noise quite a lot. Shielding definitely helps. The magnetic field is not very strong. Especially during development a stronger magnetic field helps to make the signal visible.

[gradyh]

Thanks for the help guys. I think I'm going to head back to the drawing board with these thoughts in mind.

[coppice]

Correct me if I'm wrong but is not water a poor/bad media to try and develop a mag flow meter. Unless the water has salts or significant water ions, it will not generate the potential difference you are attempting to measure. Mag meters are used with conductive media.

Magnetic flow meters are used for water and even potable water (as opposed to raw water, wastewater, etc.) but you make a good point. Perhaps if I test out a more conductive liquid I would have a lower signal to noise ratio. Thanks.

Electromagnetic meter work very well for potable water. The water needs to be very pure before they fail. Purer than you will ever seen in a public water system, which has to reason to remove benign salts from the water. It can come as quite surprising when you first work with them.

[Ian.M]

It probably doesn't help that you are likely to have a lot of turbulence in the sensor pipe, which may be partially cancelling your signal.  I assume the water inlet is the end with the pressure gauge.  The small inlet fitting bore will be creating a jet down the center of the pipe with back-eddies at the sides, and then the upstream guard electrode will introduce further turbulence.   

A stack of drinking straws can be used to make the flow laminar - simply jam enough in the pipe end to hold themselves in place, with a couple of layers of disks of green pan scourer at the upstream end to further damp any turbulence + catch any debris before it can block any of the straws.  The inlet fitting should either be baffled or be at right angles (like the gauge) so its jet doesn't directly impinge on the face of the laminar flow stack.

The guard electrodes should probably be re-made out of strips of stainless shim formed into a slightly oversized ring so they lie flat against the pipe interior to minimise turbulence.   Bring a bent tag out through the pipe wall to connect to them.  Reinforce the pipe at that  point by gluing a split ring of pipe to the exterior to get a double thickness and tap the hole for the tag so the sealing caulk has a good grip on the pipe.  Connect to the electrode by punching a hole in the tab and bolting it to a small ring terminal, glued in place with more caulk for strain relief.

The potential electrodes should similarly be made of stainless shim.  They should each occupy 1/6 of the pipe circumference and 1/2 of the magnetic field length to maximise the area which will minimise the contact resistance while avoiding excessive eddy currents in them.   They will need at least two lugs each to secure them flush against the wall.

If you have problems getting lugs sealed, you could try stainless countersunk bolts, inserted through minimum clearance holes in the shim from the pipe interir outwards.  Use slotted heads and a right-angle screwdriver for assembly.  Drill the hole in the pipe oversize so the shim dimples into it and the head sits nearly flush. The glued on split ring of pipe should be drilled with exact size holes.  Caulk and fit two nuts as a jam nut, then a third nut on top + a spring washer to retain the ring terminal. 

If you can find a big enough scrap two pole shaded pole motor that you can pass the pipe through with the rotor removed, you could salvage the stator core and coil former.  Cut the copper or aluminium shading rings off the pole pieces, and use the core to concentrate the field where you need it.  Unless it was a low voltage motor, you will have to rewind the coil with thicker wire.

If you think it may be a conductivity issue, chuck a handful of baking soda in the tank and see if the results improve!

[Circlotron]

"In science, the term observer effect refers to changes that the act of observation will make on a phenomenon being observed. This is often the result of instruments that, by necessity, alter the state of what they measure in some manner. A commonplace example is checking the pressure in an automobile tire; this is difficult to do without letting out some of the air, thus changing the pressure. This effect can be observed in many domains of physics and can often be reduced to insignificance by using better instruments or observation techniques."

https://en.wikipedia.org/wiki/Observer_effect_%28physics%29

So put a cover over the eyes on the top of your scope and see if that changes anything. 

[Richard Head]

I like your scope eyes.