[Brian2007]

I am looking for help with servo control design. Let me try to explain what I am trying to accomplish.

I am wanting to control the rotation of a servo motor in either/both half and full revolution steps. The input will be from encoders on the X and Y axis. Now it gets a little more complicated, when no motion is detected the servo should be at rest and when either or both encoders detect motion the servo should run in relation to both speed and distance. That is, selection needs to be provided as to how many revolutions are required per inch of distance be it in X, Y or a combination such as a circle. It gets better. The speed needs to keep up with the motion as it will vary but again the rotational count can not. The servo size is in the 1/2 to 3/4 hp range and the load is light and smooth.

Can anyone give me some direction on this? I'm stymied.

Any help will be greatly appreciated.
Thanks
Brian

[Al_The_Man]

Reading and re-reading the requirements, I gather you want a servo motor geared off of one or two separate encoders?
If this only involves one servo, but two axis feedback, is the servo expected to move both X & Y?
There is the capability in such systems offered by Galil Motion control of gearing of a servo which can be achieved from off of more than one encoder.
As well as electronic gearing, the electronic Cam feature can also be utilized.
Maybe a bit more detail as to what the mechanism involves would help.
Al.

[Brian2007]

Ok Al your right, I was vague. There needs to be two encoders one for the x axis and another for the y. If you have ever seen a quilting machine you will know where I'm going with this. If not think of a machine that can travel freely both in the x and y axis at the same time, thus getting input from both encoders. This input needs to be converted into output for the motor. If the machine moves 1" in any direction the motor needs to rotate a determined number of times, regardless of speed of travel. If more information is required I will do my best to provide it.

Thanks

Brian

[Al_The_Man]

I am not clear on what is driving and what is being driven?
Are the encoders being driven independant of the motor?
Is it anything like this? http://www.galilmc.com/support/motio...t_stitcher.swf
Al.

[vger]

I think you may need to take both X and Y encoder outputs into a computing circuit to achieve this function. If you have access to the drive pulses (step / direction ?) for the X and Y axis you could use that rather than seperate encoders. I am assuming here that you want equal length "stitches" produced with the servo. If X moves +1" make 10 stitches, if Y moves +1 " make 10 stitches, if both X and Y move +1" make 14.14 stitches. Am I correct in my assumptions ?

[Brian2007]

I agree that the encoder must be input to a microprocessor. Your are partially correct if x is +1 it should make 10 stitches (or what ever it is set to), if y is +1 again you should get 10 stitches, now if both x and y move on a diagonal +1 you still get 10 stitches. Here is a link to one of the best in the business http://www.intellistitch.com/ this unit has loads of features. some of which I would not need, but maybe it will better explain what I would like to accomplish.

Thanks

[vger]

My wife and I were looking at machines like this at the state fair a couple of years ago and we actually considered buying one. Then other things got in the way... grandbaby... cruises... work... etc. I remeber being interested in the method of stitch regulation (being an electronics tech) and realized that the motor must be a DC servomotor under computer control. I looked at the intellistitch site and now understand I think what you are working toward. As described here http://www.nolting.com/article.php?i=6 they have 3 encoders on the machine, one one each axis and a third on the motor. The one on the motor is used to control it's stopping position. Some A/B encoders have an additional output that provides a pulse just once every revolution as an index pulse. That would be good for auto setting the needle up position. Now to the X/Y. If you travel the frame at a 45 degree angle for 1" at a 10 stitch/inch setting you should get 10 stitches. When you make that 45 degree move for 1", each encoder see's only .707 inches movement. I'll have to think on this one for a bit... controlling a machine that is driven in the X/Y axis so that the feed rate (velocity) is constant is one thing... sensing uncontolled movement in the X/Y and deriving velocity is another. Gotta go do some work.

Steve

[gar]

080104-1153 EST USA

D = sq-rt ( X squared + Y squared )

Each time X or Y increments recalculate D. You might apply a scale factor to D. In whatever increments D works in you move when that incremental change occurs.

.

[Brian2007]

Thanks Guys,

As you can tell electronics is not my strong point, I am working on a machine design of my own and have most of the CAD work done, down to doing the tedious part of double checking the dimensions and doing revisions. Hope to get prints to the machine shop later this month. I really do want to try and incorporate a form of stitch regulation.

[Brian2007]

Here is a thought, I realize that it will take a microprocessor to make this thing work. Would it not be possible to input info from the encoders to a low end pc and let it crunch the numbers? Of course there will be the need for software, but won't any processor have to have instructions? Ok I'll get back to the mechanical design end of this thing, I understand that.

Thanks

[neilw20]

Gar has the right idea.
You need some form of numerical processor. You could do analog, but thats last century stuff.

1). Get the X and Y encoder counts/direction/etc into the processor.
2). Calulate H = sqrt (X squared + Y squared)
3). Stitches = H multiplied by a number related to stitch length required.
4). Digital to analog conversion output to control speed of motor proportional to the error from Stiches required minus stiches done.
5). Output to enable motor to be driven.
6). An input pulse for each stitch done.
7). The Cpu enables the motor output until the number of stitches have completed.

Something like that will get it going.
Figure out the easiest way to read the encoders at the maximum speed.
A PC will end up the most expedient, unless you have a microprocessor development system.

A few things to consider.
Is the table is pushed manually, or is it to be driven, like with Mach3
If it is driven, then the maximum speed of movement can be controlled.
If manual, what stops it being fed too fast.

If it was all driven with Mach3 in CNC mode, a whole new strategy is needed.

[Brian2007]

Thanks neilw20,

Ok, the machine is manually powered and yes you can over run the regulation at which time an alarm sounds. An optical encoder on the motor shaft or one of the driven shafts can generate the needed feedback counts for completed stitches, it would keep track of needle position. A servo motor will have to be used that is for sure. When I think about it either the upper or lower machine shaft would be ideal for a position/count encoder as either will turn 1 revolution per stitch and the maximum rpm will not be over 2000. I will admit that I am relying on the genius of you all, I can handle thrust loading, gear box design and the stresses involved and only have an overall view of the electronics involved in this venture.

Thanks Again
Brian

[neilw20]

You can put eddy current brakes on each axis, and let the micro stop people exceeding you stitch rate.

[Brian2007]

neilw20
Dynamic brakes? That's an idea,but I'll be satisfied just to get the basics