View Full Version : Pulse Count - not exactly what flowmeter says it should be (Logix 500)
October 2nd, 2007, 12:55 PM
I have a flowmeter with a pulse module. The flowmeter generates one pulse per gallon. I am reading approx. 5GPM. The pulse width is approx 2-sec.
I am sending 24v through the meter's pulse module, every pulse is 'seen' by the controller and through a one-shot, recorded. I am using an ADD block because it's count resolution is 9,999,999
I created a counter (see above) with help from this thread yesterday that works perfectly. I set my totalizer to match the meter and for 10 minutes the count was absolutely perfect, however, I let the controller count away overnight, when I showed up this morning, my gallon reading on the flowmeter was off from my plc version by almost 5% (flowmeter was ahead).
My initial thoughts were perhaps I need to execute the pulse read code more often than the rest of my ladder code? The manufacturer (GWF) swears that the mechanics and pulse module are dead-on.
My program is about 90 lines of ladder, no subroutines, two PID loops (0.30 sec rate).
If anyone can shed light on my dilemma, please do! Any help appreciated.
October 2nd, 2007, 01:20 PM
Quick update, I studied the flowmeter intensely, it appears that the wheel which is marked: 0 1 2 3 4 5 6 7 8 9 is 'slipping'. Approx every 5 revolutions, I notice it sticking for a fraction of a second. I think this might be the issue. Funny, both flowmeters are doing this to a degree.
Anyone with similar experience, please chime in, I'd like some backup to tell my manager!!!
October 2nd, 2007, 01:23 PM
What card is this terminated on, and is the flow relativley constant?
With a 2-second pulse width the flow would be missed if it exceeded 30 gal/min.
October 2nd, 2007, 01:25 PM
Is this a roto-meter? We used to use those for water, but replaced them with mag tubes because of problems like that.
October 2nd, 2007, 01:33 PM
The flowmeter is terminated into channel 0 (DI) of a MicroLogix 1500 controller. I have two meters hooked up (0, 1) with two more (2, 3) planned.
Yes, the flow is constant and the pulse module's max pulse train is 22GPM (22 pulses per-minute). I'm pretty sure this does not require a high-speed pulse card, besides, my pulse maximum is around 5 per-minute.
I'm worried about the slipping of the pulse count wheel.
I'm wondering if this is mechanics, or because I have 24v applied.
October 2nd, 2007, 01:44 PM
One question I would have:
Does the pulse get shorter in duration as flow rate increases.
If I understand your observation about the skipping or slipping You would need to determine if this could cause the flow meter register to still peg and NOT generate a pulse. That doesn't seem likely but I doubt that the meter could be "perfectly acurate" with this "slippling" or "skipping" going on.
October 2nd, 2007, 01:47 PM
I agree with Stationmaster. I'm not sure how the 24v power would hamper anything. It realy sounds like a meter error.
October 2nd, 2007, 02:14 PM
I have seen A/C voltage damage REED type switches by demagnetizing them over time, I'm hoping you are using 24vdc.
Contact "bounce" can cause problems with accuracy, it is actually quite a common phenomenon. BUT that would cause EXTRA pulses and therefore your PLC would read HIGHER than your meter.
If your meter has an electro-magnetic pick-up, perhaps it is causing magnetic drag on the actuator. Does the meter info include instructions on maximum excitation voltage?
October 2nd, 2007, 05:06 PM
Thanks for all the suggestions.
I am using 24vdc, I'm glad that is acceptable.
What I have found in recent research is that roto flowmeters are less accurate at the lower end of the pulser's flow rate scale. My meter measures 0-22GPM, but I am on the 5GPM scale. Lesson learned!
Indeed, the pulse wheel part of the meter, mechanically, is operating incorrectly.
October 2nd, 2007, 09:55 PM
That rate of pulses shouldn't be an issue with the scan time of a 1500. However, this IS AB we're talking about! It looks like that at your max of 22gpm, that's 22 ppm, or about .33 pps. So, you should get a pulse every 2 and some change seconds at max flow.
I still wouldn't trust it!
Look into setting up your counter routine in a STI section, using an interval of 500 ms or less. Maybe use IIM. You should never miss a pulse.
I try to keep time-specific stuff clocking separately from program scan times; yes, scan is fast, but you will miss things when dealing with frequency type inputs.
ETA - that pulse width of 2 seconds doesn't allow for much trailing edge time before the next pulse at full flow. Is it adjustable?