Valves drop every 2 weeks

[ebucktron]

The problem we are experiencing right now is that we loose communications with the valve about every 2 to 3 weeks.
The first thing that happens is that we find that every instrument in the segment won’t show readings. When I look what instruments are present in the segment with the 375 communicator the only one that I can’t find is the valve, but every other instrument is present.
The way to fix it is that we only unplug the valve and every instrument in the segment starts reading right away. After we do this we can plug the valve back in and everything starts to work.
I don’t believe that the positioner is bad because it happens on every segment that has a flowserve valve. They can’t all be bad. It doesn’t happen on any segment that does not have a flowserve valve. We have other valve manufacturers and we don’t see this behavior. We got Fisher, Metso, and Mesoneilan.
What I am looking for is that if there is a parameter or setting that may be wrong and perhaps that is why the valve locks up.
It can’t be noise because it happens only on segments with flowserve valves.

The other thing that we found is that this valves come set up from the factory as link masters. We changed that setting to basic to see if this will help. Could this be the problem?

I am sure the valve checked ok on your end because it usually works for weeks.

We are using Allen Bradleys 1757-FFLD as our host.

Please advice!!!

[rezabejd]

Which Flowserve positioner are you using?

Have you inquired with Flowserve about this? What vintage are these devices? Were they delivered some time ago?

Are you sure the link master capability is disabled? It almost sounds like the positioner takes over - so the host doesn't get any messages. When you power down the positioner, the devices start talking to the host again.

If you have another BLAS capable device on the segment, you could try setting its address lower than the Flowserve device. If there is reason to invoke the BLAS, the lower-addressed device should take over first.

Have you run any physical layer diagnostics (voltages, noise, jitter, etc.) on the segment? When the segment stops communicating with the host, it would be very interesting to know if any communications are still going on.

When you hook up the 375, it "sees" the devices but the host does not?

[ebucktron]

I am using the PMV D3
Flowserve is a swedish manufacturer and everybody in Flowserve US that I talk to tell me they don't know much about fieldbus. Currently they are trying to get me a contact from the sweden factory.
Yes I checked it and the link master was disabled after I changed it and restarted the positioner, but I did that change just yesterday. I know some of them are a couple of years old. They were purchased about two years ago, but we are doing the start up of the plant just now.
All I have to do is power down the positioner and the rest of the instruments start talking with the host again. I don't even have to power down the instruments nor the host.
You just thought me something new. I didn't know the value of the address would dictatate who takes over first as the LAS.
I connected the 375 while the segment had the problem and it shows noise, but the isnteresting thing is that the noise goes away as soon as i power down the positioner.
When I hook up the 375 it sees all the devices exept the positioner. The host shows the same thing. Every device is present on the host exept the positioner. The only thing is that the host can't report the actual values. Basically the links between the AO blocks of every device and the host don't work.

[smr]

Did you verify the shield wiring terminations? I've seen devices that came with the shield wire grounded at the device. Since we were landing all the shields back at the marshaling enclosure these devices were double grounding the shields which in turn was creating a lot of noise on the segment. Once we cut the redundant ground at the device all the noise went away.

[ebucktron]

I will check for the double grounding. I heard the same from another person. I know the costruction company landed all the shields at both ends. I am not sure yet were they landed the shield on the flowserve because I didn't see a shield terminal in the manual. I will check for that on monday.

[MarceloDultra]

Is this an I.S. installation? What Power Conditioner and Junction Boxes (connection Blocks) are you using? Do they have power limitations?

[ebucktron]

We are using for power conditioner the Relcom multisegment redundant fieldbus power supplies F890-ps
The connection blocks are the Pepperl+Fusch F2-JBSC-_8.FF.CGB
The installation was done by an outside contractor. We have comissioned everything and we did the start up last week. This is the only problem we have seen with the Fieldbus.

[aagostin]

I suggest to check the physical layer with a MTL/Relcom FBT-6 fieldbus handheld tester. This will give you accurate readings of each device. You might want to post your readings here, so that others can take a look, too.
It will give you noise readings in different frequency bands, which can help to determine whether there is a problem with shielding (HF noise), supply and conditonier (LF noise), or field instrument (FF noise). It will also give you the communication signal amplitude per device. This will tell you whether your segment is properly terminated (all amplitudes around 850mV), or terminator missing (amplitudes too high, e.g. >1100mV), or 3 terminators (amplitudes too low, e.g. <700mV).
It also shows the LAS address; if your segment goes haywire, simply connect the FBT-6, and it will give you all readings immediately. If John is right, it would show the Valve Positioner as LAS.
It could also be that the Valve Positioner hardware or firmware (comms board) simply has problems in your particular installation. Such things happen. In the past, some device vendors could update the firmware, or replace the comms board. The Fieldbus Foundation constantly improves the interoperability tests, so that the device might have past the test some years back, but might not pass the tests of today. An update then can solve your issue.
However, from experience, 95% of the problems are installation related. Only the remaining 5% are related to hardware or software issues. As such, a check with the FBT-6 can give you confidence about these 95% (or highlight your problem quickly). If you made sure everything is ok, then it is time to look into device problems.

[MarceloDultra]

You can also do a couple of simple tests using a multimeter. Measuring the impedance between shield and ground (after disconnecting the segment shield from the ground in the control room side) will tell you how good is your insulation. It should be infinite or above 20Mohms.

If you can disconnect the other wires you can also check their insulation to the shield and to ground. (again above 20Mohms)

[ebucktron]

I bought a FBT-6 to check the media. I am attaching the report of one of the segments. Right now it seems like is operating properly, but the only error I get is retransmits from the devices listed below. My control system is not picking up on this, which from what I read is what happens. I am not getting any download errors when I download the schedule.

Adrress 14h is an rosemount analytical xmt-c/t
Adrress 1Ah is a Flowserve positioner PMV D3
Address 24h is a Rosemount 848T

This was done at the power conditioner.

[aagostin]

You measured in the control room, which gives you a high reading in the fieldbus power supply voltage. This will change when you walk to the field junction box. It would be good if you can repeat the measurement there and post the data again. Particularly the noise readings can be significantly different as well, compared to the measurement in the control room, since there is a long cable in between which attenuates the noise.

The data you provide show that the segment is properly terminated and has no shorts to shield. Noise levels are low, which is good. FBT-6 can not determine whether your shields are terminated at both ends, but if a measurement at the field side also give you low noise readings then it is not too much to worry about.

The retransmissions shown for address 36 are unusual and a bit worrying. Repeat the measurement with the FBT-6 directly at the device to get most accurate readings of noise etc. If the retranmission counter keeps on counting up, it suggests that there is a problem, likely on the device/device connection. If the counter is stable, there could be other reasons for the retransmission, e.g. host software. For info, the retransmission shown is a "probe node" message, which occurs if the device is not responding.

Since address 36 shows high retranmissions and you can also see it is dropping off and coming back on the bus ('Most recent add/drop'), you could try to remove the double shielding and check for improvement, or check with the supplier whether there is any known issue or try to get a latest model just to try it out.

[MarceloDultra]

An easy way to find if your shield is terminated at both ends is to measure the impedance between shield to ground at the control room. It is also important to guarantee the continuity of the shield to all spurs and devices in the segment. For instance, the impedance to ground for the trunk shield may be ok but the spur shield may not be properly connected at the junction box leaving the spur shield disconnected from ground.

You can measure the shield to ground impedance for all spurs at the junction box to check for continuity and/or double grounding as well. All shield to ground impedance should be infinite or above 20Mohms.

[ebucktron]

I should be back on the site tomorrow. I will check for noise at the instrument. Tomorrow we will actually run everything in the system, so I should be able to see if any power cables are causing noise because we will be running all the motors.

[marisg]

Let me give you some information about fieldbus valves. This may or may not apply to your situation:

A few years ago there were problems with valves. Initially it was thought that this was caused by wiring components or the fieldbus power supply. It turned out that the MAU chip used in the valve had a very aggersive fieldbus signal filter. While the filter did remove noise frequencies from the fieldbus signal, it introduced its own considerable signal distortion. The clock recovery circuit following the filter was a bit lame and could not tolerate a great deal of distortion. Consequently, the valve worked most of the time but occassionally could not hear the incomming signal. As it turned out, a number of valve manufacturers had used this same fieldbus signal receiver design.

I believe this problem has been corrected but you may still have a valve of the old design. Please let me know what you find. maris.graube@relcominc.com

[vermav]

One of our engineers here with an Emerson DeltaV host and a flowserve positioner also experienced a similar unexplained dropout every couple of weeks. He eventually traced it to a parameter setting after a lot of heartache and research.

You might want to contact him on email ambrose.hargan@csbp.com.au

He is willing to help if it is the same issue he went through.

[ebucktron]

We ended up checking everything in detail. We opened every junction box and instrument to look for signs of anything that might be an issue.
We found two segments with shorts between the (+) and the shield.
We also found some instruments were the shield was landed on the chasis ground. This are the major ones we found. I haven't heard of any issues for a couple of weeks.

[aagostin]

Good to hear that the FBT-6 helped you to identify the installation issues. It confirms again the experience that approx. 95% of the problems are installation related.
Since you have the F890 power supplies, you could install the F809F diagnostic module onto the baseplate, which gives you the ability to monitor all these details online and conveniently via your Asset Management station. This is very useful as a predictive maintenance tool. You can send me a personal email if you require a contact within MTL.