Ultrasonic Flow Meter noise

[AutomationGuru 0]

I have run into an issue that I can't explain.  We have several ultrasonic flow meters utilizing their pulse outputs to a 1756-HSC module to detect the pulse frequency (corresponding to an instantaneous flow rate).  The flow meters have a maximum output frequency of 5000Hz representing an instantaneous flow rate of 280,000 cu. ft/day (Details are for background only, not really important to the discussion at hand.)  It's been discovered that at normal flow rates (frequency in the range of 1500 Hz output of the ultrasonic meter), the HSC module is reading frequencies of up to 2000 Hz.  Rockwell tech support initially recommended that we replace the HSC modules.  This was done  and we saw the same results.  

The next step was to rule out any floating commons.  I had the technicians at the site do some wiring modifications to source 24VDC to the pulse output from our PLC panel.  (Initially we jumpered the 24VDC supply of the flowmeter to the pulse output and used the PLC panel common.  All of our commons should be grounded together, but I wanted to rule out the possibility that this was causing the issue.  Once again, no change. 

The next option was to verify that the flowmeter frequency output was actually what was expected.  To this point we had assumed that the frequency output was correct, but had not measured the frequency.  The technicians at site used a Fluke meter set to measure Hz and verified the frequency at the output terminals of the flow meter.  Next they went in to the PLC panel and checked the frequency at the input terminals of the HSC module.   In both cases the measured frequency was dead on at around 1530 Hz, however, Logix 5000 software showed the present value of the frequency as around 2000 Hz.  

As they continued testing, the technicians shut down two VFDs (VFDs are controlled by a different processor and have no interaction with the HSC modules) and the HSC module read the correct frequency in Logix5000.    They further tested by running the drives and removing the HSC signal cable from the cable tray and routing it along the ground to isolate it from the VFD motor cables.  With the drives running, the frequency still reads in the neighborhood of 2000 Hz and 1550 Hz when the drives are stopped.  

Rockwell tech support now has recommended installing a common mode choke on the output of the 753 VFDs to reduce the noise, claiming that the VFDs must be causing the noise and increased frequency reading.  The concern is that if that does not resolve the problem it will be add another filter, add some other hardware, on and on until things finally work and then no way to know if all the added components were needed or if only the last one solved the problem.  There are additonal meters to be added to this site in the future, so I'm sure this issue will pop up again.

Here are my concerns:

1.  If the Fluke meter is reading the correct frequency at the input terminals of the HSC module, how could the CPU/Logix software be displaying a higher frequency?  The only interaction between the VFDs and the HSC is in the cable tray where the VFD cables are in a separate tray section with barriers in between.  For the life of me I would assume that if the VFD was inducing noise on the signal, that they would see that increased frequency on the Fluke meter.

2. The HSC modules are mounted in a remote rack.  The remote rack resides in a control panel not more than 3 feet from the main CPU rack.  The remote rack communications is via an isolated network, while the VFD controls are hardwired from a Micrologix controller provided by a 3rd party.  We communicate to that controller over a plant wide network that is separate from the RIO network.

3.  We initially intended to use 4-20mA analog signals for the flow rates, not pulse outputs and only added the HSC modules after the fact.  Since this system resides in a Class 1, Div 2 area and conduit seals had been poured already, we made use of a second twisted shielded pair  to connect the pulse output back to the control room PLC panels.  Is there any reason to believe that cable could be causing the issue.  

4.  In the overall system, there are 4 ultrasonic flowmeters.  3 of them are Flexim clamp on flow meters, the 4th is a Daniels ultrasonic flowmeter.  The Daniels meter frequency reading is dead on regardless of the VFD, where the flexim meters all are seeing increased frequencies in the PLC.

Any suggestions or solutions will be greatly apppreciated.  I'm at a loss and i'm not sure whether or not to believe the tech support claim that the higher frequency is due to electrical noise, when the correct frequency is on the input terminals of the HSC card.

 

Thank you

[Joe E. 259]

Disclaimer: I'm by no means a grounding/bonding/noise expert...

Noise and grounding can be tough. Like you, I would expect the HSC to read the same frequency as the Fluke, so noise may be getting into the HSC module from somewhere else. I can also see there being some noise in the system that the HSC would see since it has a common ground that the Fluke may not since it's power supply is isolated.

You said that the frequency reading from the Daniels meter is dead on. Is it wired to an HSC module in the same chassis as the one that's acting weird? If so, have you tried to swap the leads at the chassis so the signal from the Daniels meter is wired to the HSC module that's acting up and the meter that's giving weird readings is connected to the HSC for the Daniels meter? That would help you isolate the problem to the field or to the chassis. If the weird reading stays with the HSC module, then the field wiring is ok. If it moves, however, you can rule out the chassis and investigate in detail the shielding and grounding of the field wiring. Not all field devices are the same, and your different flow meters may be bonded differently internally.

Is the ControlLogix chassis with the HSC module grounded properly?

Can you add or change the routing of the ground connection for the chassis frame to the system ground?

Look at the installation and user manuals for the ControlLogix system and the VFDs to find the right way to bond their frames. Both have a lot of details on it.

Do you have an oscilloscope you can use to look at the incoming power to the chassis power supply while the VFDs are stopped and running to compare readings? I think this would be less likely, but worth checking.

These are all very speculative on my part. We get machines built to European standards and they always come in with both line and load filters on VFDs, so tech support isn't completely out of line here. In the long run, it may be better to go ahead and put in the filters on the VFDs, to avoid future issues, even if they don't completely fix this issue.

 

[AutomationGuru 0]

Joe,

Thanks for the input.  The grounding is something that we are going to have the customer check out at the site.  We came up with the same thoughts.

In answer to some of the questions you posed:  The Daniels meter is not only on an HSC module in the same chassis as the Flexim meters that are having issues, but it is also on the same card as one of the affected meters.  

We are seeing the issue on two different HSC modules, and on 3 of the 4 available channels.  The fourth channel being the Daniels meter.  We have moved wiring around and still see the issue on the three channels that the Flexim meters are connected to, leading us to believe it is something in the field side.  But again, that does not explain why the Fluke frequency reading matches the flowmeter output, while the ControlLogix monitoring raw channel data shows a frequency boost of 10% of full scale or greater.  

Unfortunately, as of right now, no one has an oscilloscope to look at the waveforms of either the power or the pulse output of the meters.  

Our biggest issue is that these meters are being used for custody transfer, meaning they are relied on heavily for billing and payment, so it's not just the technical guys looking at this problem, but the accountants and executives are very much aware as well.  

Thanks again for the input, and if you have any other thoughts, please don't hesitate.

[Joe E. 259]

That does seem to isolate the problem to the field. It's possible that the noise exists on all of the flowmeter channels but that there's something unique about the Daniels meter and its internal isolation that cancels it out or keeps it from affecting the reading.

Is there a Flexim flowmeter near the Daniels flowmeter? I'd be strongly tempted to swap the cabling at the flowmeter end to see if the problem stays or moves.

Are the cable runs the same between the Flexim and Daniels meters? Any additional terminations in one or the other where a shield didn't make it through continuously?

[AutomationGuru 0]

Joe,

None of the meters are close enough to one another to do anything on the field termination side.  All of the cable runs are home runs from the device via conduit and cable tray (conduit only from device to tray, ,tray for the remainder of the run back to the control room.)  there are no junction boxes, termination points, etc between the meters and the PLC cabinets.

Once again, my frustration is that I can't for the life of me logically explain how I can get a signal in the software that is 10% of full scale  (as much as 50% of full scale higher when the pulse frequency is 5000Hz).  When the frequency measured at the module terminals is exactly the same as the frequency at the output of the flow meter, I can't figure out how noise is being introduced from a VFD that is completely isolated from the PLC system that the HSC modules are in.

I've actually been asked to go to site tomorrow to further investigate, so I may come back with some additional information.  I will update as appropriate.

[DanW 41]

Are the Flowmeters AC or DC powered?

Some frequency outputs are passive that need loop power to operate. If a 24Vdc power supply is in the frequency loop, is the 24V common grounded or floating?   Can you ground the DC supply if floating, or float it if it is grounded in order check alternative performance?

I have heard of some field instrument outputs picking up noise from switching DC power supplies, where the solution was to substitute a linear DC power supply.   Linear supplies are now uncommon, but it might be worth the effort to locate one for testing to see if it makes any difference.

Presumably the Fluke is battery powered and floating.    

Are the frequency test results done with the field wires disconnected from the input terminals of the AB input card, or with the field wires connected?

I wonder what an AC powered, grounded chassis oscilloscope would show for an ”as connected” test (in parallel with the connected field wiring)?

I’ve seen 4-20mA common mode issues disappear when the field wiring is disconnected and its signal checked with a floating battery powered meter where problem reasserts itself when the field wiring is re-connected, because common mode is ’common’ between two points - remove the commonality (wiring connected, not floating) and the common mode disappears.

Try the same frequency test with the (-) low side meter probe grounded to see if you get the same results.