Pressure Transducers Failing

[Edmhydraulics]

Hi everyone, I wonder if anyone can help me figure this out. I have a test system that I built for a customer. It has a Red Lion Graphite HMI with an analog input module and 2 - 0-10,000 psi pressure tranducers, 24Vdc input, 4-20ma output. In under a year I have had to replace 4 transducers already. The customer is not overpressurizing them because I have logic in the PLC that prevents this. The test media is normal tap water. There are also standard gauges in the line to give a backup reading to the transducers. The system is pressurized with an air-over pump.

What has been happening is that I have had to adjust the zero points of the transducers a few times before they fail completely, they seem to slowly float from 4ma to up to around 10ma and then get stuck at that point.

The transducers are Wika so its not like they are cheap. I have a switching 24Vdc power supply connected to the incoming power at 220VAC.

Any thoughts on why this might be happening?

Thanks in advance.

 

[g.mccormick]

Have you contacted Wika and sent back the bad ones?

I would be inclined to stop buying that transducer.

 

[gclshortt]

http://www.setra.com/blog/6-ways-to-prevent-prevent-transducer-failure

The above article may help you out. I would look at the power supply and have the source fed through a UPS to filter the power.

Regards,

 

[Edmhydraulics]

I have contacted Wika here in Edmonton and dropped off the first 2 that failed, but I never heard anything back from them.

Its generally a slow pressure increase to the setpoint, but the depressurization can be faily quick, taking maybe 1/4 second. I suppose that might be a bit of a problem, I guess I could try a snubber to restrict the rate of change somewhat. Anyone have any other recommendations for low cost general purpose transducers for this application?

 

[James Mcquade]

What you are describing is what is called shock failure.
this is when a load cell (in your case a pressure sensor) is subjected to the initial load(shock) and it can be as much as 6-10 times of the load when the initial impact wares off.
for an example a 25 lb load cell has to be beefed up to a 250 lb load cell. a 1000 lb load cell is increased to a 10k load cell.

I have had to replace several load cells with much higher load cells due to this.

hope this helps,
james

 

[Edmhydraulics]

Interesting theory James, generally the test fixture is prefilled with water, the pressure is slowly increased by increasing air pressure to the air-over pump, its a 10 to 1 pump and it slowly increases as the air pressure increases. Since the test fixture is prefilled, the pump rarely goes through a full cycle, so I would like to think that there are not any transient pressure spikes, and even if there are any, they are few and far between.

I have used these same Wika transducers in hydraulic applications that are much more demanding, lots of transient spikes, high temp, vibration, etc and they work fine. Thats why this has me stumped.

 

[goghie]

Just to mention that when you are measuring the liquid, tapping point should be at the side, near the bottom of the pipe. It's not good to measure the pressure from the top of the pipe (because of airlocks) or the bottom (because of dirt).
System should be evacuated from air.
Just to mention, I had problem once with hydrostatic level transmitter (E&H FMX 167), due to very hard water (mostly due to Mn), and at the end we had resolved it, according to advice, by using condom and cable tie!!! Pressure still existed, I needed to calibrate a little, but transmitter was well protected!
Maybe you should try to mount some protective gauge rubber cover - I think Wika and lot of others have it in their offer.

 

[Andybr]

I would also check the mechanical installation is exactly as required for the sensor. We have had problems before where sensors have been stressed due to things like parallel threads into tapered sockets and sensors bottoming out on shoulders at the bottom of sockets. The sensors often work initially but fail after they have been mechanically/thermally cycled a few times.

 

[kamenges]

I would bang on Wika a little harder to have them get you a failure analysis. They really should be able to do that for you.

That being said, it really does sound like failure due to repeated transients. Just because you aren't commanding something that would be responsible for the transients nor are you measuring them doesn't mean they aren't there. Do you have a hydraulic snubber between the transducer and the line you are measuring?

Keith

 

[Edmhydraulics]

Thanks Kamenges, I will push Wika some more on this, not very impressed that they didnt get back to me on the first 2 sensors that failed. I dont have a snubber in the system, as I have put in a lot of these systems over the years and havent had an issue like this before. But I will definitely be putting them in now.

Cheers.

 

[Peter Nachtwey]

Its generally a slow pressure increase to the setpoint, but the depressurization can be faily quick, taking maybe 1/4 second. I suppose that might be a bit of a problem, I guess I could try a snubber to restrict the rate of change somewhat. Anyone have any other recommendations for low cost general purpose transducers for this application?

1/4 is very slow compared to what we see on presses. The rate of change in pressure shouldn't be a problem. Since this is an air over water application it should be quite compliant. Have you looked at the pressure with a scope? PLCs are quite slow and can miss fast pressure transients.

You shouldn't need or want to use a snubber. If spikes really do exist there are better ways to handle them.

 

[kamenges]

Originally posted by Peter Nachtwey:

You shouldn't need or want to use a snubber. If spikes really do exist there are better ways to handle them.

Don't stop there.

 

[rootboy]

It doesn't sound like it's shock related, and Wika should be able to tell you if the sensor became de-bonded.

Another failure mode is moisture making its way up the interior of the cable, this has taken out more than one load cell.