PDA

View Full Version : Fill by weight


CharlesM
July 9th, 2009, 10:45 AM
We are looking at an application where we will be filling a container by weight. The current system uses a large hopper on top that feeds two vibratory feeders. Then a small hopper with a scale to do the weight. This should be simple to do in a plc however they are currently using a scale system. I am thinking their scale system is running some kind of PID to slow down the feed as it gets near the setpoint. How important is the scale controller? Can I do this in the plc? I have never done this type of application but it looks like I should be able to do it in a plc.

Here is the spec's as I know them.

Part weight 3400 grams
Fill time 30 seconds
+/- 2 grams

panic mode
July 9th, 2009, 11:01 AM
you are on the right track. i've done programming on practically same thing (with 3 or 4 filling heads on same machine and control system). target weight was 7-20kg, +/-30g, fill time was 5-6 sec for each filling head. measurement was done with loadcells, everything was controlled by small PLC. Due short cycle time, and vibrations, challenge was to get accurate weight reading (I wish I had 30 seconds).

mordred
July 9th, 2009, 11:43 AM
I've yet to see a scale indicator that has PID functions possibly its sending an analog signal to a PLC this can be done exclusively in a PLC unless there is an issue with American weights and measures. If so then one route you can do is have a scale indicator that has an optional analog output card 920I from Rice lake has this feature. Or you can send an ascii text on a standard indicator set for constant transmit to send the data to the PLC. perhaps you can post the scale indicator make and model if it has PID I would be interested in knowing what it is

CharlesM
July 9th, 2009, 12:52 PM
Here is the controller I was looking at

http://www.hardyinstruments.com/process_weighing/filler_dispensers+controllers/hi+3010+filler+_+dispenser+controller

The scale part itself is not a problem. I just worry about the fill logic. Can it be turn on output till >= set point or does it need to be an analog output that slows the feeder as it gets close to setpoint.

Controls_Engineer79
July 9th, 2009, 12:57 PM
well you could do it both ways

but with the set points - maybe have it filling at full speed till it reaches 90% then have it slow down at that point till it reaches 98% then slow down even more

have more then one set point will give you better +/- control

Longhorn
July 9th, 2009, 02:53 PM
I am not familiar with your application / equipment but have a good deal of experience with liquid fillers. A common approach is to "fine fill" for the first few percent then "coarse fill" until the container is a few grams (say 200g) from the target weight before switching back to fine fill to achieve the target weight.
Hope this helps!

bernie_carlton
July 9th, 2009, 03:06 PM
Longhorn - just curious, why the initial 'fine fill'?

Banker
July 9th, 2009, 03:55 PM
It can be done in a PLC. But you have be aware of the amount you are filling after you stop. Is it a valve or a belt. If you have a "noisy" weighing signal because of vibration there can be an problem there. To long filter time on the weight signal can cause a signal delay.

Banker

CharlesM
July 9th, 2009, 04:00 PM
Good inforamtion guys. I was waiting to see if anyone said "no don't try this in the plc" or look at one of these controllers.

I think we will be using a vibratory feeder or maybe a conveyor. One thing I have working for me is time. I think I can go through a few steps in the filling process because I have time to do it.

Gummit
July 9th, 2009, 06:07 PM
Good load cell meters have "in-flight" compensation.

In-flight compensation allows the filling system to switch off the filler valve
before the correct weight has been reached. This may be required if some of
the product could be delivered for a time after the filler control valve has
been switched off. The In-flight compensation can be applied to both the
main feed and trickle feed controls outputs.

http://www.datatrackpi.com/panel-meters/tracker-240-load-cell.htm

widelto
July 9th, 2009, 09:25 PM
Longhorn - just curious, why the initial 'fine fill'?
It must be the opposite first coarse fill and fine fill at the end.

widelto
July 9th, 2009, 09:42 PM
Here is the controller I was looking at

http://www.hardyinstruments.com/process_weighing/filler_dispensers+controllers/hi+3010+filler+_+dispenser+controller

The scale part itself is not a problem. I just worry about the fill logic. Can it be turn on output till >= set point or does it need to be an analog output that slows the feeder as it gets close to setpoint.
Charles:
Im in charge of designing a system to fill containers with edible oil ( 14 & 20 liters) maximum time is aprox 20 secs per container, I want to use the same controller HI 3010 or use a compactlogix (1769-WS) with hardy C2 load cell ( no calibration needed ), just waiting for quotations and then compare controllers (3 or 4) Vs. compactlogix solution.

JeffKiper
July 9th, 2009, 10:15 PM
I just climber out of our mix Towers. We have 17 scales from just about every make and model.
CharlesM you did not say what PLC you where using. I have only set up 1 weight system in a PLC (AB SLC 5/05) all the rest I just support. I like using the in chassis card made for scales. We uses Hardy on the one I setup. If you choose to use an analog version scale it to real numbers.

We have a Hardy HI 3010 in 1 of our operations it works nice.

CharlesM
July 10th, 2009, 07:52 AM
I want to use the same controller HI 3010 or use a compactlogix (1769-WS) with hardy C2 load cell ( no calibration needed ), just waiting for quotations and then compare controllers (3 or 4) Vs. compactlogix solution. It would be good to see which one you use.

CharlesM you did not say what PLC you where using.I will be using Siemens.

SMOKE
July 10th, 2009, 08:42 AM
I once ask Pete if Delta made a scale interface. He said that they had made some customs. I would talk to him.I have done lots of filling in the past, using Sartorious, and metler scales, as well as load cells.I think you will find you will have to pulse your feeder when your near your target weight. Update time is issue, and if you use a PID it is hard to compensate for update and material in flight.I typically use a volume gross fill and finish with flat vibe.

Clay B.
July 10th, 2009, 10:27 AM
think I see why you have a ramp function versus a "trickle" function. You are feeding from a vibrating bed. When this is done a ramp down function is prefered because you can better control the feed rate. Vibration tables tend to be non-linear to the speed of the vibration motor. So be ramping the speed down and monitoring the rate of change on the weight you end up with a more accurate final weight.

If you were using a fill valve the most common method is to have a trickle position. This way your flowing at a slow enough speed that you can stop accuratly.

Also Note: If you take the PLC route. Not all PLC's have load cell inputs. You may still need a controller.

Longhorn
July 10th, 2009, 02:35 PM
Longhorn - just curious, why the initial 'fine fill'?

We use plastic bottles and this initial weight stabilises the bottle. Coarse fill would make the bottle move due to the initial high flow!

spaderkung
July 10th, 2009, 02:43 PM
I think the key issue here is to separate the filling into coarse/fine regardless of method to make it fast _and_ accurate.

Filling to 90% of the set-point is probably possible with a fixed filling time. And if it is possible to reach 100% within tolerance after calibration then problem solved with one parameter - fill time.

A "PID" that is fast could also have overshoot. And in this application that is not allowed. If "PID" is used at all then two sets of parameters are necessary - fast/slow (equivalent of coarse/fine using different size of pipes).

An easily adjusted compensation time (see in-flight) will be helpful.

Longhorn
July 10th, 2009, 02:44 PM
It must be the opposite first coarse fill and fine fill at the end.

Fine fill first - to stabilise bottle or container
Coarse fill - to fill bulk volume
Fine fill finish - to trickle increase to target weight.

We use in-flight correction on one of our fillers, the weight of the container at the point of the valve closing is taken and subtracted from the final weight to get in-flight weight.

Next revolution round this is subtracted from the final fine fill weight e.g 200g changes to 195g.

hope this helps!

CharlesM
July 10th, 2009, 02:51 PM
We use in-flight correction on one of our fillers, the weight of the container at the point of the valve closing is taken and subtracted from the final weight to get in-flight weight.

Next revolution round this is subtracted from the final fine fill weight e.g 200g changes to 195g.
Good info Thanks

SMOKE
July 11th, 2009, 10:24 PM
We use in-flight correction on one of our fillers, the weight of the container at the point of the valve closing is taken and subtracted from the final weight to get in-flight weight.

Next revolution round this is subtracted from the final fine fill weight e.g 200g changes to 195g.

That would only work with a single speed fill No PID. You would need a very consistent feed rate to make that work.

Longhorn
July 12th, 2009, 06:57 AM
Smoke - You are absolutely correct, and this is exactly what happens. This is achieved by a seperate PID Loop to control the pressure inside the "bowl" above the fluid. Keep in mind i am talking about a liquid filler here.

LH

widelto
July 12th, 2009, 11:24 AM
Fine fill first - to stabilise bottle or container
Coarse fill - to fill bulk volume
Fine fill finish - to trickle increase to target weight.

We use in-flight correction on one of our fillers, the weight of the container at the point of the valve closing is taken and subtracted from the final weight to get in-flight weight.

Next revolution round this is subtracted from the final fine fill weight e.g 200g changes to 195g.

hope this helps!

I like it, good point, ill try it in my design.

Peter Nachtwey
July 12th, 2009, 01:43 PM
I once ask Pete if Delta made a scale interface. He said that they had made some customs. I would talk to him.
There are others that have off the shelf systems. We are not a good option unless one needs a custom system. However, my business part did design some excellent strain gauge amplifiers in the early 90s.

However, if I had to get the job done quickly I would use a small single axis motion controller with analog inputs, but that is just because they seem to grow around here. Faster scans, faster analog faster digital, easy to use Ethernet interface, better diagnostics. How can a PLC compete with that?
It wouldn't be the first time a motion controller got used for something else.

CharlesM
July 13th, 2009, 02:54 PM
However, if I had to get the job done quickly I would use a small single axis motion controller with analog inputs, but that is just because they seem to grow around here. Faster scans, faster analog faster digital, easy to use Ethernet interface, better diagnostics. How can a PLC compete with that? I might be able to do that. I just happen to have a RMC150 with a scale attached. I need to find something to feed the material then I could test it. Would you use quick moves for the fill?

Peter Nachtwey
July 13th, 2009, 06:09 PM
A quick move would probably work well but I would try using taking advantage of the raw speed and go as fast as I can in open loop. I would compensate for the in-flight material. Eventually you will approach a time where the in flight material plus what you have filled will be very close to the desired fill weight. Think of the in-flight time as a dead time and the in flight weight is proportional to the control output. When getting close to the fill weight I would ramp down quickly to about 10% or so and top it off.

The in-flight material would be proportional to the sum of control output signals for the in-flight time. Also note that if the material drops 1 ft then the in-flight time is 250 milliseconds and if the sample rate is 500 micro seconds then the in-flight buffer would need to be 500 control output readings long.

Don't sum the whole array each scan.
Total:=Total+New-ControlOuputQ[i];
ControlOuputQ[i]:=New;
I:=I+1;
If I>=LengthOfControlOutputQ THEN
I:=0;
END_IF
EstimatedWeight:=MeasuredWeight+K*Total;
Where K is the relationship of the control output history to the in-flight material. You should be able to calculate K from the feeder rate as a function of the control signal and time.

A RMC150 seems like gross overkill and do you have the analog inputs already? At least you can do a proof of concept quickly.

Thim
July 13th, 2009, 06:29 PM
What is "in-flight material"
I checked the net but didn't found anything relevant.

Peter Nachtwey
July 13th, 2009, 07:36 PM
What is "in-flight material"
I checked the net but didn't found anything relevant.
In flight time is similar to dead time. Material that is in the process of dropping on the scale but hasn't hit the scale yet so isn't being measured. If you dropped material from 16ft or 4.9m it would take a second for material to hit the scale after you stopped the feeder or conveyor. If you stopped the feeder only when the fill weight was reached you would find that you have one extra second worth of material on the scale. It takes a surprising long time for things to fall when dealing with computers.

CharlesM
July 13th, 2009, 08:10 PM
My reply was half in jest. I would use anything before suffering with an S7I wanted to say something but I let it pass. I was once asked "Why would you do that in the motion controller when you can do it in the PLC" I get along fine with S7 but I don't get to advanced. Have you looked at the S7-1200? It's their new low end controller but I think it will replace the 300 & 400 in a few years. It is only ladder now but they do have blocks for working with arrays.

A RMC150 seems like gross overkill and do you have the analog inputs already? At least you can do a proof of concept quickly. Yes the 150 is in my demo machine and I have a scale connected and working. If I can find something for a feeder I will give it a test.

Thim
July 14th, 2009, 11:36 AM
Thx Peter,

So it is what we call Afterfall, i wasn't familiar with the term "In-Flight"

Maybe stupid remark, isn't the vibrating feeder a slow stopping something?

Clay B.
July 14th, 2009, 12:26 PM
I wanted to say something but I let it pass. I was once asked "Why would you do that in the motion controller when you can do it in the PLC" I get along fine with S7 but I don't get to advanced. Have you looked at the S7-1200? It's their new low end controller but I think it will replace the 300 & 400 in a few years. It is only ladder now but they do have blocks for working with arrays.

Yes the 150 is in my demo machine and I have a scale connected and working. If I can find something for a feeder I will give it a test.


We have built several weighing systems using the S7-200 PLC with a MS module. It will do everything that is required of this job.

Anyway...using an axis controller seems like overkill to me. You have multiple manufactors that deal strickly with weight control and AB or Siemens or Unitronics for that manner have onboard weight controllers.

When you use an exisiting weight controller, things like Rate of Change, Preact, Auto preact, tare tolerances, zero tolerances..etc.. etc are already built in... so you do not have to create your own.



Maybe stupid remark, isn't the vibrating feeder a slow stopping something?


Short answer: Yes

Vibration feed can be very tricky depending on how the vibration is made. You can get irratic behavior at various speeds. Usually if I have a system that uses vibration feed I will pick a VFD where I have lots of "skip" frequencies or if I am driving the drive with a PLC I will skip those in my program.

Best system when using vibration feed is to have some sort of valve so you can shut feed off before vibration stops and use vibration for rate.

testsubject
July 14th, 2009, 02:47 PM
At the company I used to work for, we did this kind of thing all the time. I used an LC controller and we used 3 setpoints: fast, slow, and final. When the slow target was met, we stopped and checked to see if the final was met in a given timeframe. If not, we "burst" at the slow rate for a small period of time and then check again. The company has been using vibrators for years in just this manner.