Stable speed readout

Hello

I have a 2.745 inch pulse from a line shaft on a machine running at 150m/min, I want to display a m/min for the operators. Any basic count and * ladder I produce gives an output that is very unstable jumps up and down 20m/min, only other way could think to do this was a running average, but this is very slow, machine is stopped but display is still ramping down
Anybody have a ladder for a filtered time based pulse count, that will respond quickly.

I have used a 1756-M02AE in a flying cutoff machine for two years where line speed goes up to 120 meter/min and the speed readout from the plc is steady and in accordance with actual line speed.
I´m using motion instructions.
Hope this will help

Carples said:

Hello

I have a 2.745 inch pulse from a line shaft on a machine running at 150m/min, I want to display a m/min for the operators. Any basic count and * ladder I produce gives an output that is very unstable jumps up and down 20m/min, only other way could think to do this was a running average, but this is very slow, machine is stopped but display is still ramping down
Anybody have a ladder for a filtered time based pulse count, that will respond quickly.

Click to expand...

I'm afraid that quickly and filtered are oxymorons. Some machines vary greatly in speed while runnning just fine. The question is what you want to display, do you want an instantaneous speed display or an average over a unit in time...

With what hardware are you measuring this speed (PLC, encoder resolution, scan time etc...?) This determines more than anything what can be done to improve your displayed speed.

I have a Mitsubishi Q series PLC, no motion instructions on this unit. I want to display the speed of a printing press, speed is fairly constant.
PLC = Q01
Scan time = 1-2ms
Pulse = 2.754 inches from prox and cam lobe

I have an old process meter on the machine using the same sensor and it displays an accurate stable output in m/min

I have tried
Counting pulse per sec then mul by 60 and convert to meters
Counting .01 sec per pulse then div 60 by result and convert to meters
An average of both of the above.

At the moment I have a running average of 5 values produced from a running average of 5 secs of pps, gives a nice stable result that take 10secs to catch up with the machine.

Both process meter and PLC values match, most of the time.
May be I'm just being picky, but it gets my goat that the process meter gives an almost instant response using same inputs and the PLC can't.

Carples said:

Hello

I have a 2.745 inch pulse from a line shaft on a machine running at 150m/min, I want to display a m/min for the operators. Any basic count and * ladder I produce gives an output that is very unstable jumps up and down 20m/min, only other way could think to do this was a running average, but this is very slow, machine is stopped but display is still ramping down
Anybody have a ladder for a filtered time based pulse count, that will respond quickly.

Click to expand...

Okay, you are getting a pulse every 2.745 inches, and the product flow rate is meters per minute?

So @ 150 meters per minute you're getting (150,000mm / 25.4 inches)

That equates to 2151.37 pulses per minute, or 35 pulses per second. That is your resolution per second, which is important to consider in your expected accuracy.

So the input logic connected to the pulses, needs to be PDQ (~.014 seconds) to get all of them in the first place, and the ensuing math needs to be timed quite repeatably too.

I would probably want to take samples about 4 times a second, of the pulse totalizer, and average the last three or four most recent samples. That will put the displayed value up to one second behind, plus comms delays (if any).

Of course, I have no idea what the root cause of your inaccuracy might be, so this is just sort of a WAG.

And, I may have mis-interpreted your units.

BTW what PLC are you using?

You should be able to construct a quick and effective filter, but also must consider communication delays to the display which can add to the problem you describe of seeing the display still ramping down after the mchine is stopped.

There are several good posts here on signal filter mathematics.

I personally like to trend the raw data first, to see what kind of "noise" I want to filter out, then take it from there.

Paul

Carples said:

May be I'm just being picky, but it gets my goat that the process meter gives an almost instant response using same inputs and the PLC can't.

Click to expand...

What is the process meter?
Why can't you be happy with the process meter's data?
Does the process meter use the same pulse every 2.745 inches?
Does the process meter use the same time base as the PLC? No.
I bet the process meter has access to the control signal from the drive to the motor.
Obviously there are ways of predicting the conveyor speed more accurately than just using a simple low pass filter. These other techniques they require more math and the control signal.

Think I would put a tach on it and see if the speed is varying or is constant. Maybe your system is telling the truth.

Dan Bentler

Thinking outside the box,maybe you could simply use a 4-20mA retransmit from the process display into the PLC and use that? Or as Peter Natchway has said above maybe there is an Analog from the Drive to the Process Display you could use. Is the intention to replace the process display?

Cheers