We have a customer with an underground storage tank. Installed in the tank is a magnetostrictive float based level sensor that measures the level of the brine solution between the inner and outer tank walls, thus providing leak indication if solution level should fall or if it should rise below a fixed point.

The problem is every time there is a lightning storm, we end up getting a false alarm on the panel. We've verified all system grounds and the cable between the sensor and building for proper shielding and that it is in a isolated conduit. Any body know what's up, would appreciate all feedback

TIA,
Kris

 

That certainly is a high-end means of level detection for an inner wall. As you are aware, the device works by sending a “voltage” signal on a fine wire down the probe. The magnetic lines of force (ring magnet in float) cuts this signal and gives an extremely accurate level position. The same principle is used in hydraulic cylinders for cnc machining and robotics etc.

Is it possible the vessel can carry a charge thereby exciting the sensitive signal wire? I am sure this is not a common problem with a magnetostrictive system.

The unit simply sends a proportional signal. You must have some sort of control box that converts that signal “at a point” to a high or low level. Is the box changing the state on a surge?

Maybe contact the supplier?

Thanks Bob Hogg
http://www.almegcontrols.com

 

Magnetostrictive technology involves an ultrasound reflection in a wave guide surrounded by a magnet where the magnetic field causes a very small premature reflection. I suppose it might possible for a near lightning strike to casue a premature reflection. I suggest you put a timer on the alarm to filter nuissance alarms, any real alarm will still exist at the end of the time period and be recoginzed as valid.

 

Hi Kris -
Like the other posts have said, the problem is caused in the sensor or the data signal. (Your data signal is usually either analog or PWM depending on the sensor, but SSI data is also a possibility.)

Your application doesn't appear to be high speed, so your best bet may be filtering the data where you're reading it. If you know that the data cannot change by more that "±X" each time you read it, then compute you're delata between readings and if it's more than ±X, ignore the reading. After a few seconds, your readings should return to normal.

Of course it's a little more complicated than that. For example, What do you do if the last reading that got through was actually bad and you're now basing your delta on a bad data point. But, the basic idea should work.

[email protected]

 

I had some further thoughts -

Instead of throwing out the data value if the delta is greater than ±X, limit your delta to ±X as a maximum. Therefore, if you have a catastrophic failure in the tank, (which means the delta is > ±X and sitll valid), your program will eventually sound a warning.