Need input for servo press feed, I need to become educated

[bigmark1972]

Hi,
I am a machine builder and I am going to be building a press feed for a new job we hope to get. I am very good with PLC's, Robots, Controls in general but I have not had the need to get into servo controls until now.

My application is really pretty simple, with only a few different motion profiles needed. I will also be using a load mounted encoder for absolute part position verification. I'd like it to be a stand alone system, linked to a plc for control. The only thing I need to be able to change via input is the feed distance and perhaps speed.

Although I have a basic knowledge about how various servo systems work, I was wondering if anyone could recommend a good book or two that will get me up to speed.

I like to completely understand something before I jump into it so any input is appreciated.
Thanks

 

[SERVOCAM]

no simple book that I have seen that will get you up to speed. Many mfgs have references on their sites that help. There are several I saw, but mostly out of date or get into design rather then application.

There are many single-axis stand alone drives out there that could do what you want. Well maybe, depends on is your "load mounted encoder" an absolute encoder, or an incremental encoder? And then some single-axis drives won't do dual loop (accept 2nd encoder to close position loop around). Many single-axis drives will have some communications way to a PLC: MODBUS, DeviceNet, Profibus, Serial ASCII. Is it possibly the the press feed does not have that much I/O and you can just use the motion controller/drive to control the whole thing? Then use an OIT - Operator Interface Panel to enter your variables.

 

[blacksea]

If you don't need many IO for control, motion controller with build-in driver maybe suggested - Elmo (

http://www.elmomc.com)

or Galil (

http://www.galilmc.com)

products maybe useful for your application.

Regarding books: "Control system design guide" by George Ellis, "The Electro-Craft Engineering handbook", "Step by step motion Control" by Jacob Tal (see Galil website above)...

 

[bigmark1972]

Thanks for the replies!

I do not need an absolute encoder on the load, an incremental would be fine. I am going to be feeding material in 15-20' lengths and am planning to "register" each new length with a through beam sensor or similar means.

I considered using the drive controller for the entire operation, but it may get a bit complicated for that. The PLC is going to control a very high speed hydraulic press and I will be processing several different parts/processes on each press. I will need to control an automatic bar strip feed, bundle loader, etc, etc.

I am already pretty good with RLL and PLL plus I think it would be easier for me to just call for a motion profile when I need it. I may be wrong (probably am) but aren't most current drive controllers programmed via a PC and in a different way than your average PLC? I know galil and others seem to have their own versions of how to get things done while rll is always rll.

Regarding books: "Control system design guide" by George Ellis, "The Electro-Craft Engineering handbook", "Step by step motion Control" by Jacob Tal (see Galil website above)...

I have the second book and have absorbed much of the information, I will check out the first as well.

You guys are a great asset to us guys out there that are forced into being a jack of all trades master of none.

I have a weird situation where I run a manufacturing company with my brother and father and have to do "all" of the steps in bringing a production line into production all by myself.

Design, cad, machining, wiring, hydraulics, pneumatics, controls, everything down to getting painted is my responsibility. I have self taught myself most of what I know except for robotics which I went to a short tech course for.

I love what I do, I love to learn and people like the ones here that give their hard earned advice and educated opinions are greatly appreciated.
Thanks,

Mark

 

[SERVOCAM]

I would suggest taking a look at Emerson Control Techniques EN Sereis / FM module or Unidrive SP. There are many others I know of that can also do this application, but I think you will be happy with their products.

http://www.emersonct.com

This is a feed-to-length appliation with regristartion sensor and secondary encoder feedback.

Depending on what mfg PLC's you use, possibly they have a motion control module that can do what you want.

>I may be wrong (probably am) but aren't most current drive controllers programmed via a PC and in a different way than your average PLC?

Yes, they are typically programmed via a PC and each manufacture uses their own motion language. Typically a very basic language. A Motion controler is also state logic - reads line by line and executes that line, then moves onto the next. Now some motion controllers have PLC packages for them for I/O ladder logic and other IEC-61131 interfaces. The only real standard but not adopted for general motion control is RS-274 G&M CNC Lanugage. It is genneraly pretty easy to program motion controllers.

 

[blacksea]

Mark,
I don't sure that PLC has "high-speed position capture" feature as modern motion controllers. The feature is exactly intended for precise length messuremrnt.

Srvocam: Do you have more info about refered RS-274 G&M CNC Lanugage?

 

[SERVOCAM]

Right, you have to be sure on the PLC that it is a motion control card, not an index card or step/direction card...that it has the funcions you need. For example, like GE Fanuc's DSM-314 - which is a 4-Axis servo module for the PLC for that would have the functions you need.

Try this:

http://www.manufacturing.net/ctl/in...482158&spacedesc=news&nid=2361&rid=883149483&

 

[SERVOCAM]

Use this for the standard:

http://www.iso.org/iso/en/CatalogueDetailPage.CatalogueDetail?CSNUMBER=13540&scopelist=

 

[bigmark1972]

Thanks for all of your help, it is very much appreciated. I have been considering using emerson they seem to have some nice products and seem quite user friendly. Now I just hope we get the work so I can get started.


Mark

 

[snoogie]

Don't use a photocell for registration. Use a "press" clear signal (prox or limit) that is set at the point where the press is just open enough to get the material through so you get a head start on the feed. Most mechanical presses have CAM switches on them to provide this signal which can be finely adjusted to get the most out of the feed window but being hydraulic it won't have a CAM. You said the press is high speed so I am guessing that you want to start the feed as soon as you can, hence the prox "feed" signal as soon as the press is open enough (do not consider accelertion above one G as all sorts of things start to go wrong if you do!). Be careful of the press cycle times quoted, especially with hydraulics, if you are not using a hydraulic servo valve to control the press stroke you will have real problems controlling the stroke. Overtravel is the main problem. You may size your system based on a cut time of 1 second, but it can end up more like 2 seconds once the time taken up by the overtravel is considered so now to maintain the same production rate you have to double you accel rate! Just some precautions to wach out for.

Snoogie

 

[bigmark1972]

Thanks for the information snoogie. By registration I mean I need to read actual bar position once for every new length and enter that value into the controller. After the loading sequnce feeding will be a simple to length index confirmed by a load mounted incremental encoder.

This is the basic cycle of operation required everytime the machine loads and cuts a new length of material: (2 stage prog die)

1. use the new length to push the remaining part from the last length into an accurate position through the prog die. (this move will need to be slow enough not to let momentum carry part out of position.)
2. retract new piece of material,
3. stroke the press cutting that last part
4. push out the final good part from that last length.
5. push out more still and seperate the scrap piece at the end of the length.
6. re-feed new length into correct stage in the first die station.
7. continue to feed to length and stamp until material passes sensor/switch indicating last possible piece.
8. new length sequence will begin again.

Obviously I will write a subroutine and the system will know if it is feeding the very first length and would skip all of the push-out processes and simply put the first length into the correct starting position and go from there.

I will begin each new index as soon as I have clearance to do so. I am custom building these presses for this job I need approx 1/2" of stroke and will move the press full travel both ways, possibly tweaking the stroke with stop blocks and reading an early travel confirmation for end of stroke anticipation and a faster return stroke.

I plan on keeping moving mass to a minimum to help limit inertia, I estimate I will need approx 20 tons of force but I will nail that down before construction.

Proportional valves have made great leaps in the past few years, and are gaining on servo valves while drasticall less complex and expensive (I am building 7 presses so cost is a factor). I would think a proportional valve would suffice here, if I were to use one it would be for a "soft" start and stop but I am unsure if this is neccasary. I am mainly concerned with fast response and flow capacity.

This is going to be a roller feed type of system (eliminates clamp/unclamp delays) are you saying that tuning is an issue over 1 g? I only need to feed approx 2-1/2 inches so speed while important does not need to be extreme. I do need positional accuracy of +/- .030" I see no problems obtaining that. I will be feeding lengths around 15' and weigh between 7-20 lbs. (some ferrous some non ferrous)
Thanks for your input.

Mark

 

[snoogie]

I see, now what you are doing. This is very different to my first impression, my experience is with feeding flat contiuous strip into a press or cut off. The accel problem over 1 g is more of a mechanical thing than drive tunning, in that the drive/motor will do it but it kind of leaves a trail of nuts and bolts behind it if you know what I mean.

Hope you get the order.
Snoogie

 

[mc5w]

1. You need a 1.5 axis system so that you can have encoders on both the motor and the load. However, an encoder on the motor has sufficient accuracy as long as the pinch rolls do not slip.

2. You need 2 sets of current limits, both 80% and around 160% of motor full load current. If a die has a positive stop into which the material runs and then stops, you need the 80% current limit so that the motor does not overheat. You would also need to program the software to ignore the matter that it cannot reach final position because the positive stop is controlling the situation. Most drives are built so that at standstill the motor can only exert 80% torque because the unwanted current from the carrier is at 60% of full load current.

3. You most definitely want the input modules for the signals from programmable limit switches and so forth to be replaceable. There is a way for the noise suppression to somehow fail and when it does it usually fails for all inputs.

4. You servo computer ( computer numerical control ) needs to use buffer gates in between the 5 volt digital logic an the analog to digital converter. A/D converters use a 12 to 15 volt version of transistor-transistor logic that tends to short the 12 to 15 volt analog power into the digital logic when it gets to be 1 to 10 years old. Gates that have high voltage open collector outputs with a pull up resistor going to the positive side of the analog power should do the trick. If you canot get a CNC that is built this way you might have to build a daughter card that inserts in place of the A/D converter. Also, the lowest grade of A/D converter is sold with the least significant bit not necessarily working. The next grade up only costs about $17 (1998 price) than the least grade. The A/D converter and the buffer gates should be socketed because A/D converters onlynlast about as long as industrial vacuum tubes did.

 

[bigmark1972]

Geat info thank you!

I have no plans to use a dead stop, I do plan on using both the motors encoder and the load mounterd encoder. I am aware that my performance will be limitied by the "traction" I have on the bar for both the drive system and load encoder. Luckily the parts family shares a common profile. I am planning on building custom drive rolls profiled to match the part perfectly out of hardened tool steel. With no carriage to return and no clamp and index to move moving mass is very little. As we say on our softball team I am going to "grip it and rip it" I'm shooting for a cycle time of under 1/2 second.
This job is big, almost 30M annual so these presses need to be lightning fast.