Bouncing 4-20 current loop with ultrasonic sensor

[bretberger]

Utah Integrator at wits end and about to comment ritual suicide with my multimeter probes. Please help a brother out.

Culinary water tank at the end of a one mile shielded 2-pair (untwisted I think) 18 gage wire. One pair goes to hatch intrusion switch, the other tank sensor. The sensor on the tank is Ametek Drexelbrook US-21 ultrasonic. PLC is SCADAPack 350.

Intermittently the loop starts bouncing. Voltage drops from 24 to 10 or so and the current mirrors this. Period of bounce is on the order of a few seconds and may directly correspond to the ultrasonic ranging sensor firing... not sure.

I have had several sensors installed at this location (same model) and each behaves a little differently but I always run into a time of year when they start bouncing. They may go for 6 months without an episode and then it will start. Summer seems to be the worst.

When I power the sensor locally from 3 nine volt alkaline cells it works perfectly.

Any thoughts? Am I at the margins of what this setup can source current-wise and when my ultrasonics fire off and pull a big slug of electrons it droops and then bounces? Could I modify this setup with caps at the sensor or some other approach? My distributor and manufacturer have not much help.

 

[George Graziano]

First, welcome to the forum. What power supply voltage can that transmitter handle? When it is bouncing, how close to 20mA are you?
George

 

[Honga]

If you attached the products manual it may help you/us.

 

[bretberger]

Ametek Drexelbrook USonic Datasheet

Here is datasheet for the transmitter. Is it kosher to upload 100MB pdf manuals?

Whoops, answer to my own question: No. 300kB limit.

 

[bretberger]

Usonic spec sheet

Relevant portion of attachment:

Input Power
19 to 30 VDC
19 VDC required @ 4 mA
minimum

 

[OkiePC]

... water tank at the end of a one mile shielded 2-pair (untwisted I think) 18 gage wire.

My first thought: "That's a long skinny wire for DC...no way is it going to be reliable"

How much signal lag time can you afford?

Last winter I saw a demo of a Banner Surecross wireless module that was quite impressive. It was very easy to configure and set up, and worked through three brick walls with the tiny stock antenna. I am pretty sure they have accessories to handle that distance.

IIRC, the latency was around 20-50 milliseconds. Can your system tolerate that signal delay?

Push AC power down the existing wire (caution: induction may affect other shielded pair), and put a small 24vdc power supply at the other end to drive the sensor and wireless transmitter. Put the receiver at the PLC end and it can generate the 4-20ma to the PLC input.

Post a link to the manual if you can find one on the www.

 

[bretberger]

More info

George and Honga, thanks for helping. Long time lurker, first time poster

George, the device seems to be more prone to bouncing as it goes higher scale.

 

[R2DS]

Ultrasonic sensors depend on the speed of sound through air. Is there any chance that a gas other than air is present? CO2 is one that causes me fits at times, also H2 and He will drive ultrasonic sensors crazy. Also, if they are being aimed down a pipe, resonances which may be temperature dependant can cause problems as well.

 

[bretberger]

My first thought: "That's a long skinny wire for DC...no way is it going to be reliable"

How much signal lag time can you afford?

Last winter I saw a demo of a Banner Surecross wireless module that was quite impressive. It was very easy to configure and set up, and worked through three brick walls with the tiny stock antenna. I am pretty sure they have accessories to handle that distance.

IIRC, the latency was around 20-50 milliseconds. Can your system tolerate that signal delay?

Push AC power down the existing wire (caution: induction may affect other shielded pair), and put a small 24vdc power supply at the other end to drive the sensor and wireless transmitter. Put the receiver at the PLC end and it can generate the 4-20ma to the PLC input.

Post a link to the manual if you can find one on the www.

Thanks OkiePC. The delay of a wireless link is no problem. The PLC at the the other end of the wire is connected to the HMI/Base station via licensed VHF (GE MDS) over a 10-mile radio link. The same radio could be installed at the tank but, alas, no budget for it.

 

[bretberger]

Ultrasonic sensors depend on the speed of sound through air. Is there any chance that a gas other than air is present? CO2 is one that causes me fits at times, also H2 and He will drive ultrasonic sensors crazy. Also, if they are being aimed down a pipe, resonances which may be temperature dependant can cause problems as well.

I have seen temperature stratification affect ultrasonics in a closed culinary water tank. That is not the case here since I had the tank open during todays testing and air was well mixed. This is spring water... no gases present. Straight shot to the water, no obstructions. Really smells like a transmission line issue to me.

 

[danw]

>Input Power: 19 to 30 VDC
19 VDC required @ 4 mA minimum

I suspect that this is a case for the dreaded Ohms Law (I can hear the groans now)

18g = 0.00751 ohms per foot * 10,000 ft = 75 ohms
20g = 0.0119 ohms per foot * 10,000 ft = 120 ohms

What is the input resistance on the analog input of the PLC? A typical 250 ohms?

If your power supply is nominally 24.0 Vdc, that allows a 5v drop before the transmitter is voltage starved below the minimum 19V required.

If the input resistance 75 ohms of wire and 250 ohms on the analog input, the total load resistance is 325 ohms.

At 0.020A (20mA) the IR drop is 0.020A * 325 ohms = 6.5 volts

At 15.38ma, the IR drop is 0.01538 * 325 ohms = 5.0 volts, the cutoff point.

15.4mA is about 71% of span.

If the wire is not a continuous unbroken run, but has junctions along the run, the junction points could add resistance from weathering and corrosion over time, increasing the IR drop and lowering the point at which the transmitter loses the minimum voltage needed. And that's if the wire really is 18g and not 20g or 22g

When the level increases above the voltage starvation point, the transmitter chokes. It might even cycle through initialization. It could drop, charge up, fire a ping, try to output the result, choke, oscillate, who knows what? The thing's designed to run at adequate voltage, not on the edge.

Boost the power supply to 30Vdc and see if the problem doesn't disappear.

 

[Mrslipmat]

Last winter I saw a demo of a Banner Surecross wireless module that was quite impressive. It was very easy to configure and set up, and worked through three brick walls with the tiny stock antenna. I am pretty sure they have accessories to handle that distance.

As quoted by OkiePC, the above units are great when it comes to 4-20ma, I use these in a factory full of welding equipment with no problems both for wireless I/O and current feed back, distance wise, with the correct anttena you should have no problems and would be able to get rid of the cable !!!

Regards

John

 

[Lancie1]

Culinary water tank at the end of a one mile shielded 2-pair (untwisted I think) 18 gage wire. One pair goes to hatch intrusion switch, the other tank sensor.

Bret, Put a padlock on the hatch, disconnect the hatch intrusion switch, then parallel the two sets of wire, so that your effective resistance for the tank sensor drops by 1/2.

Another alternative is to run a new, larger cable. This time get a 3-pair cable so that you have a spare pair.

 

[OkiePC]

> Boost the power supply to 30Vdc and see if the problem doesn't disappear.

I would do this first. This has got to be the simplest and least expensive fix. It might end up being a permanent one, too.

Kudos to DanW for doing the math.

 

[bretberger]

Thanks for the help

Thanks to everyone for their thoughtful comments. DanW, thank you for cranking through the numbers. In retrospect this should have been the first thing I did. Yes the Analog input resistance to ground is 250Ohms on the SCADAPack. Your math looks correct.

After viewing the trends from the HMI it was clear that the device always started to fail as the water was rising in the tank. At around 75% of full scale it started chattering.

Wire is 18 gauge. One continuous run. Not even close to line of sight back to vault where the PLC is located so 900MHz/2.4GHz radios would not work even if there was budget for them.

Tank hatch is already locked with 2 (!) locks and operator still wants hatch open detection.

The power supply is goose-able to 28.5 Volts (Phoenix Controls Quint-ps-100). Next time I am on site I will turn it up and report back.

thanks again!