Averaging Analog input from Flow meter

[Dennis g]

I have a Flow meter that returns a 4-20 signal and the problem I have is the pump is a diaphram pump so theres alot of pulsing causing my totalizer to be innaccurate. I am grabbing the data every one tenth of a second but still not accurate when checked against graduated cylinder. Any ideas would be great. Thanks

 

[ndzied1]

I think your success is going to ultimately depend on whether your 10Hz sample rate is fast enough to give you an accurate read on what is going on.

What is the frequency of the pump pulses?

 

[Dennis g]

It varies depending on a signal from a mass flow meter. will the ML 1100 grab the data faster than 1/10th? I did not build the system but am tasked with getting it running correctly.

 

[Tom Jenkins]

You could try putting an accumulator/attenuator upstream of the flow meter to dampen the pulsations.

 

[JHarbin]

It varies depending on a signal from a mass flow meter. will the ML 1100 grab the data faster than 1/10th? I did not build the system but am tasked with getting it running correctly.

How about installing a pulsation damper after the pump and before the flowmeter?

I can't imagine that the flowmeter is actually changing its flow output as quickly as 1/10th of a second so speeding that up probably won't help. Without knowing what its signal is, and if it has built-in dampening, I can't say for sure.

 

[bosko1]

try smoothing the 4-20 signal

I work in water/wastewater and we have this on chemical pumps. we use smoothing on the signal before it's converted to EGU.

 

[Dennis g]

I work in water/wastewater and we have this on chemical pumps. we use smoothing on the signal before it's converted to EGU.

Can you elaborate on this

 

[Bit_Bucket_07]

If dosing accuracy is critical, then maybe you should bite the bullet and consider buying a different pump that doesn't pulse.

If your error is consistent, then perhaps there is a calibration error with your flowmeter and/or you might be able to massage the signal with software to correct the error. (I know... That's cheating.)

I suspect that the error is created by the pump pulsations and that the error is not consistent. Am I right? The prior suggestions to damp the pulsations with an in-line accumulator sound logical.

 

[Dennis g]

when I try to adjust,it either gains or loses depending on which way I go with the scaling.

 

[Tom Jenkins]

I work in water/wastewater and we have this on chemical pumps. we use smoothing on the signal before it's converted to EGU.

Can you elaborate on this

I think the attached filtering algorithm is what he had in mind. You might still lose a little accuracy, though. A pulsation damper (what I called an accumulator) is probably a better way to go.

 

[JHarbin]

How about installing a pulsation damper after the pump and before the flowmeter?

I can't imagine that the flowmeter is actually changing its flow output as quickly as 1/10th of a second so speeding that up probably won't help. Without knowing what its signal is, and if it has built-in dampening, I can't say for sure.

Woops - that should say pulsation *dampener*.

In a former life I worked for a Metering Pump Manufacturer (i.e. Chemical Feed Pumps). We recommended and sold, at least I think we sold - remember it was a former life - pulsation dampeners when smooth flow was needed.

 

[Peter Nachtwey]

I think the attached filtering algorithm is what he had in mind. You might still lose a little accuracy, though. A pulsation damper (what I called an accumulator) is probably a better way to go.

Physical averaging using an accumulator is definitely the way to go.

The digital sampling would need to be very fast to get an accurate estimate of the flow. The digital sampling would need to be much faster than the highest pulse rate.

 

[Kev77]

What types of volumes are we talk about here? Is the graduated cylinder test checking the accuracy? Most flow meters come with a pulse option with adjustable pulse settings. Just totalize the pulses and multiply by the volume.

I also have an application doing the same thing and the accumulators work great.

 

[Peter Nachtwey]

What types of volumes are we talk about here? Is the graduated cylinder test checking the accuracy? Most flow meters come with a pulse option with adjustable pulse settings. Just totalize the pulses and multiply by the volume.

I also have an application doing the same thing and the accumulators work great.

That is actually a good idea but the PLC must be able sample fast enough to catch all the pulses.

 

[daba]

That is actually a good idea but the PLC must be able sample fast enough to catch all the pulses.

It's not about the PLC, it's about whether the flowmeter itself is giving good data.

There's no point in accurately capturing/recording/processing data that is not accurate, otherwise you chase a non-existent goal. Many flowmeters cannot cope too well with the sort of rapidly changing (i.e. pulsating) flowrates produced by diaphragm or peristaltic pumps.