Rapid Fill To Weight

Orn Jonsson

Member
Join Date
Apr 2002
Location
Boise Idaho
Posts
125
Hi folks,

I have an application which needs some improvements....

Current app: An Auger type Filler with Servo and PLC controls is used to fill a container to a certain weight by adjusting the RPMs, and total Auger revolutions during a fill cycle.

Problem: The machine is controlling a distance (revs) to obtain a precise weight (indirect process variable). Wear and tear of the augers means a very frequent calibration of the machine.

Proposed solution: Build a new filler that fills to weight by measuring weight.
Problem: Do I need SERVO Control or Could I use a VFD Controlled motor and a PLC (don't mind complex programming), to control the speed during fill and the back off as I get closer to my tail adjusted setpoint?
I can't see using PID for this kind of control!!!!!!!! (unless. I moved the fill weight SP automatically up as I run the fill cycle, sort of a fill weight trajectory)

The material which is being used is very fine powder and I have to fill 400 grams +/- 5 grams in 20 seconds or less.

What is the key to not overshooting your fillweight when doing a very fast fill?

How do you deal with inconsistent tail weight?

Look forward to your replies

Orn Jonsson
 
Without seeing your operation and knowing all the characteristics, I would think that using a signal from the scales you could slow down the auger near the end of the cycle, say 350 grams or so to get the precise measurement you need.
 
Orn Jonsson said:
I can't see using PID for this kind of control!!!!!!!! (unless. I moved the fill weight SP automatically up as I run the fill cycle, sort of a fill weight trajectory)

That is how I would do it. However I would let the motor run open loop until close to the correct weight and then use then ramp the fill rate to 0 at the correct weight. This way you get the most peformance and still get accurate fills. This would be similar to the quick move on our controller.
 
Gennerally + or - 5 grams of 400 grams is a high degree of accuracy. ]

Have you ever seen a progressive reloading machine for loading bullets? The gunpowder is generally measured by volume. There is a disk that rotates under a hopper that provides the correct volume. The disk has a hole in it that is sized for the correct volume for the correct weight of the given type of gunpowder. I gennerally find the accuracy to be within 1/2%.

For a simple solution I would use a hopper/disk dispenser that would drop 5 grams, or less, a a time on your reciver. The disk would rotate a series of holes in the disk from the hopper on the top to a dispenser hole on the bottom. (the disk would be sandwiched between two plates and the top hole would be offset from the bottom hole).

The reciver would be on the load end of a balance scale. On the oposite end of the balance scale a prox switch, (or an electrical contact between the scale and a fixed contact), would detect the proper weight when the scale reaches a fixed height indicating a signal to stop rotating the dispenser disk, and to move the measured load off the reciver. Or you could let the disk rotate and simply close a sliding door on the bottom of the dispenser plate.

Depending on your disk speed, I can see this speeding up your operation considerably, and providing the accuracy you need by weight dispite of wear on the mechanics.

In my opinion this dosnt need to be a complicated PID or servo application. You just need to trash the auger altogeather.

Let us know what you decide on. Good luck with it.
 
Last edited:
Quick Move

There is a bit in the mode field that is set for quick moves. A quick move is similar to a normal move only the acceleration field is changed from inches per second per second to volts per second and the velocity field is changed to volts. The first part of the move is done open loop where the output to the drive is ramped up at the volts per second rate specified by the acceleration field. The output is increased until it reaches the volts specified the speed field. Normally the volts is set to 10 volts for maximum speed.

The trick is to know when to switch to close loop control to make a normal close loop ramp down to the set point. This is determined by the current velocity and specified deceleration rate. At this point the target position is set to the actual position and the target velocity is set to the actual velocity and the integrator term is set to zero so the controller now thinks that it has made a perfect move until this point. This keeps the integrator from causing overshoots due to windup. From this point on the controller uses the target generator so that a normal s-curve ramp can be generated and enters normal closed loop control using the feed forwards and PID.

This works extremely well on hydraulic systems however I don't see why it wouldn't work if you equate weight with position for your application.

I would think that this kind of applicatin has been beat to death and there would be controllers optimized to do this. If not, then I have found a new market.
 
Thinking about material

Part of the problem for high accuracy weighment is the physical characteristics of the material. Peter's solution would work quite well if the material flows smoothly. If there is irregularity in the material flow (clumping, static...) a ramp may not work better than any other method.

What I have used is:
High speed to preact
low speed to flyweight
jog (if necesary) to setpoint
and use an average of the jog cycles (or oversetpoint conditions) to adjust the flyweight setpoint.

It is simple and reliable.
 
Orn,

I don't know your application at all, but I have done a lot of controlled weighments of very fine powdered materials. My personal opinion is that the volumetric disks described by elevmike would not work well for fine powders (they work best for constant shape/packing density course powders/pellets). I would be much more comfortable using an auger based system to dose fine powders. I have used two approaches depending on the application:

One: exactly what tomneth suggested (and similar to what Peter suggested) - run full speed on the auger until a preset weighment is reached somewhere below the target setpoint. At that point, switch to a dribble speed where the auger runs at one tenth (or less) of max speed to reach the final weighment.

Two: Add a mixing/holding hopper between the weighment location and the process. You can then reduce the the batching rate so that you can run your dosing equipment at the rates that will give you the desired accuracy.


If the variation in sloughing (even at low speed) on the last revolution is larger than the +/- 5 grams accuracy that you desire, then you have the wrong equipment - you need a smaller auger. If this is the case, and the smaller auger prevents you from getting the rate, then you could always try two dosers (course/fine) to get what you want.

Another type of doser that I have tested (though never used commercially), that might solve your problem is a vibratory doser (NOT a vibratory tray feeder). Take a look at Peabody SolidsFlow (last I knew they were based in Fort Mill, SC ph: 803-547-7711). I haven't been in that industry for five years now though, and it looks like they may have been bought out by Schenck/AccuRate (http://www.schenckamericas.com/solidsflowfeeders.html ).

Good luck,

Marc
 
marc,

I cant argue with your point on my disk feeder. It is good for some materials but not all for sure.

However I will stick to the balance scale, with contact or phototype setpoints, over the complexeties and expense of an all electronic PID solution. For about $200.00 you can get a 1kg scale wiith a damper to stabelize the scale in high speed operations.

We have a 20 lb balance scale that has a indicator that moves about 12 inches for about 1/2 inch of vertical movement on the load plate. Try to imigane placing three or more sensors that you could adjust along the pointer travel-line that would trigger different phases of your operation (fast / slow/ jog/ stop).
 
Rapid Fill

Thanks a bunch for all of your inputs.
I was a bit facinated by the progressive reloader concept but it seems that it may be complicated to build.
FULL BORE OPEN LOOP.... Thats a new one to me. Is the brake pedal field-bus dependent?
However it seems to be the absolute fastest way. Ramp down to zero with no integral value sounds good. I take it that's my best insurance against overshooting.
Vibratory feeding causes feed material segregation (it's a mix).

Looks like a small VFD controlled horizontal auger sticking through the bottom of a hopper and loading into a load cell mounted preload container with a phneumatic gate on the bottom, might be something that I would try at this time.

Off to look for load cells...

later
 
Vibratory feeding causes feed material segregation (it's a mix).

Orn: That's one of the reasons why I said NOT a vibratory tray. The SolidsFlow feeder is radically different in operation and worth a look.

Marc
 
Orn,

I use this concept all the time for filling, we use vibrators and conveyor belts though. Sometimes a single drive, at others 2 drives side by side (coarse and fine fill)


First target (Fast or Bulk Fill), approx 80% of target with Feed Rate at or near full speed.
Second target (Slow or Dribble Fill), approx 98% of target with Feed rate at 20-50% of full speed.
Third target (Target check), 100% of target, if met stop, if not, turn back on at second target rate for preset time and check again.

This gives me accurate filling with many diverse materials. This should also work with your auger. Just change the rev/second with a VFD controlled by you PLC's 0-10vdc or 4-20ma Analog card.

It's simpler than setting up a PID.

I feel you pain when it comes to a heterogeneous mixes. I see variations all the time from the start of a run to the end of the run due to product settling in the hopper and the lighter particles rising to the top. That's one of the reasons it is tough to do these volumetrically.
 

Similar Topics

Good morning all Someone of you have experience with this toolkit and PackML? Mine is close to 0. Customer ask us for PackML, and I found (or it...
Replies
0
Views
651
Hi everyone, I have a problem with EN2TR during I using the EN2TR web page. I was trying to use EN2TR web via connecting thru E-TAP and go thru...
Replies
1
Views
1,223
Dear All, We are upgrading project where wonderware rapid toolkit module used to develop the canbus communication drivers . Is there any one who...
Replies
1
Views
1,236
i made a modifaction for my project to overcome a problem which is the motor - which controlled by AC drive - rotate automatically when we power...
Replies
6
Views
2,314
Hi all, I have to upgrade a Texas PLC to a Siemens S7 PLC. In Texas PLC I have a 500-5026 module (rapid responde module) with 1 input an 1...
Replies
0
Views
1,784
Back
Top Bottom