AB 1394 Rotary Knife Application

Hi All,

I have a rotary cut-to-length knife application and I am using an AB 1394 drive. Board Speed range 40 to 90fpm. Cut length 8 to 16ft. A sample cut need to be obtained at 4ft for quality control purposes.
Circumference of rotary kinfe is 50.292".

Tha 1394 use PLC5 via remote I/O for batching and sample cut command.

System works good except I have a problem getting an accurate cut on the sample, it cuts 52 to 54".

Any suggestions or does anyone have experience with this using the 1394 GML software?

Any help greatly appreciated!
John

I don't have any rotary knive experience but could your product be slipping?

Also, how is the PLC giving the cut command? Is there enough time and motor torque to bring the blade around in the necessary time. The laws of physics are not to be broken.

I would agree with ndized that there isn't enough torque to make that kind of offset.(assuming the head is what changes speed and not the line )
But is that cut length consistant on the other cuts or does that vary a little also?
What kind of feed backs are involved?
Is there a registration input for the 1394?

More info is needed.

Drewcrew6

Not sure if my last got posted, here goes again!

The boardline is 400 ft long and continuous so there is definately no slippage there, or in the knife gears, they are very tight!

The motor is 15hp and you can turn the knife gears with a small rachet, definately lots of torque.

Programming Brief: I am gearing the knife speed to that of the board when a certain amount of board passes under the knife, (encoder on boardline with rubber wheel that rests on the board). Once the speed of the knife and board are matched, I caculate a position error and then accerlate the knife briefly to compensate for the error, the speeds are then matched (geared). This is all done before the knife makes the cut.

This works fine when the borad to be cut is longer than the circumference of the knife rotation, problem occurs when the board to be cut is shorter than the circumference. circumference = 50.292"
desired board = 48".

I am not using the registration input, not sure what it does, do you know?

Feedbacks are from the motor resolver and the encoder on the boardline.
All equipment that I am using was recommended by AB!

Thanks

When you're cutting sheet lengths shorter than the knife circumference, the knife has to rotate at a higher circumferential speed than the line speed. Could there be something in your logic that prevents that? If possible, run through your calculations manually and check against the drive's intermediate calculations. Maybe there's a sign change somewhere when you cut less than synchronous and the drive doesn't handle it properly.

If all else fails, could you get QC to accept a 5 foot sample?

I don't mean to be sarcastic at all when I say this but, cut-to-length applications are a good example of where not to use AC drives.

Cut-to-length is really a positioning application if you study it carefully. You are literally trying to place the knife in a particular position in the sheet.

My recommendation is to use an AC drive for the main sheet drive and a servo drive for the knife. Use an encoder running on the main sheet to provide the speed/position info for the servo. This always seems to give good results.

More often than not, the knife drum is filled with concrete or is otherwise very heavy. In order to get good cut length accuracy, the knife drive has to be able to make fast position changes at the very last minute as error is detected. Doing that on a very heavy drum will require supprising amounts of short term overload torque. Size the servo drive and motor accordingly.

I don't think an AB 1394 or any other brand of precision drive is good enough without positioning capability.

DickDV said:

I don't think an AB 1394 or any other brand of precision drive is good enough without positioning capability.

Click to expand...

Never used one, but I thought the 1394 WAS a servo drive...

This application should be simple. DickDV is right on except on the concrete part. At least I KNOW that there is NO concrete in a veneer clipper. The blade or cutting device should be as light as possible so the knife can be easily accelerated and decelerate. Actually this is never easy as the accelerations can be extreme. The key here is that the knife must be accelerated to get into position and then slowed down to match the belt or conveyor speed when doing the cut.

There is no way you can compensate for position error by calculating a correction on-the-fly. By the time you make the calculation the positions have changed and the calculations are no longer relevant. The knife should be executing a cam profile that is geared to the main feed system/conveyor. The cam can be scaled to adjust for different intervals of cut or the cam can be trigger so that the cut occurs at the correct place. The cam profile is calculated so that the trajectoy will cut on exactly the correct spot. With proper tuning and design, the servo will follow the cam profile and cut within a few thousandths of an inch.

What am I missing here?

I have never seen a flying knife in person, though I understand the basic concept. I think Steve brought up a VERY good point. John is not able to cut a length shorter than the circumference of the knife wheel. Remember, longer distances between cuts works just fine.

The knife requires 'x' number of degrees of rotation to make the actual cut. Let's say it takes 90° to make the cut. He needs to make 2 cuts, so he needs 180° of 'at speed' rotation (knife in contact with the board). He has 180° remaining between these 2 cuts to speed up and get ahead enough to prepare for the next cut.

What if the cut requires 180° (or more) of 'at speed' rotation? There is simply no room left to speed up.

My point is that there is going to be a minimum board length that can be cut with a 50.292" circumference knife, and this will be based on the number of degrees required to make a cut. That minimum may in fact be 50.292"!... :nodi:

beerchug

-Eric

I am not using the registration input, not sure what it does, do you know?

Click to expand...

We use a rotary knife to cut product but the knife head is relativly close to the desired length (within about 3/4" ) and only about 10" in diameter. The head also has 4 or 5 blades depending on the product.

I have yet to see the actual program in the 1394 servo drive . From what I'm told the registration input veifies the position of the blade at about 50 degrees before the actual cut. So I'm assuming that if in correct position at that point no additional adjustment needed if it is not in position it allows a little time to make the needed adjustment.

From the operation of the machine I would say that that is right. We had a lot of problems that rooted from the registration input.(originally was a 24vdc switch to fiber-optic back to 24 vdc, now is a straght 24 vdc with a resistor on the input to tame any stray noise and/or reflections)

For as many times the I worked on this piece (actually 4 identical) The thought of the knife needing to match conveyor speed at cut time never crossed my mind.

Amazing what you can learn without asking a question!!!!!!!!!!!!

If ignorance is bliss welcome to heaven!!!!!!!!!!



Drewcrew6

The 1394 is a servo drive, so I am doing this,

"My recommendation is to use an AC drive for the main sheet drive and a servo drive for the knife. Use an encoder running on the main sheet to provide the speed/position info for the servo. This always seems to give good results."


I welcome and appreciate all the comments but i think that I very may well be limited to the 50.292".

I may try the cam idea with a scale factor though.

Thanks
John

I didn't sleep much last night after reading all these comments! Here's what I am going to try!

Once the knife is finished cutting a board it will stop its current routine and go into a sample cut subroutine. The sample cut will be based on a cam table not gearing. The encoder count for board length is set to zero when the cut is made and starts counting again to whatever the board length to be cut is, so that will be 48" for the sample cut routine. The knife will be just jogging forward after the cut for 4.02336" at which point the cam table will take over.
The cam table is as follows
For anyone who don't know camming it is position matching!
CAM TABLE
Board Knife
4.02336 4.02336
4.979373913 5.0292
5.935387826 6.03504
6.891401739 7.04088
7.847415652 8.04672
8.803429565 9.05256
9.759443478 10.0584
10.71545739 11.06424
11.6714713 12.07008
12.62748522 13.07592
13.58349913 14.08176
14.53951304 15.0876
15.49552696 16.09344
16.45154087 17.09928
17.40755478 18.10512
18.3635687 19.11096
19.31958261 20.1168
20.27559652 21.12264
21.23161043 22.12848
22.18762435 23.13432
23.14363826 24.14016
24.09965217 25.146
25.05566609 26.15184
26.01168 27.15768
26.96769391 28.16352
27.92370783 29.16936
28.87972174 30.1752
29.83573565 31.18104
30.79174956 32.18688
31.74776348 33.19272
32.70377739 34.19856
33.6597913 35.2044
34.61580522 36.21024
35.57181913 37.21608
36.52783304 38.22192
37.48384696 39.22776
38.43986087 40.2336
39.39587478 41.23944
40.35188869 42.24528
41.30790261 43.25112
42.26391652 44.25696
43.21993043 45.2628
44.17594435 46.26864
45.13195826 47.27448
46.08797217 48.28032
47.04398609 49.28616
48 50.292

Once the cut is made, board length encoder set to zero again and I will park the knife (have to park to wait for a certain amount of board to pass under) I then go back to the main cut routine.

I may add more points to the cam table to get more accuracy if needed.

Anyone see a problem with this?

Thanks
John

On the right path

Never ever reset the encoder? How many times must I say that???
Instead subtract 50.292 inches from the the feed position.

WHY? Think about this. I realize that 50.292 may not be exact and reseting the encoder is good enough. However, if there are marks on the material 50.292 a part and you must cut on the mark you must not reset the encoder.


You shouldn't need more points. The motion controller should be able to fit a smooth polynomial between a few adjact points or do a spline.
It appears there are enough points.

This is the enocder I use for measuring how much board is passing under the knife.

I actually don't reset the encoder, it is 3200ppm so it rolls at 3200 ppm, there is a 12" circumference measuring wheel on the encoder, so it rolls over to 0 at 25600 for an 8 foot board and will roll over at
1200 for the 48" board.

I never ever reset the knife circumference encoder, that is actually a resolver not an encoder. Once it is homed to zero position it stays there!

I assume you are talking about the motor resolver??

Thanks for your help!
John

Sorry, it will roll offer at 12800 for the 48" board!