CNC Mill Conversion Questions.

[bladehunter]

Yep that's the one Tom, guy to contact is Chai. Linearmotionbearings have been on ebay for a while (could be the longest) almost everyone seems happy with his service.

 

[jumps4]

Thanks Gary,

I am now leaning towards using a closed loop stepper system, with rotary encoders on the steppers. This seems to be the newest approach and offers the best attributes of steppers and servos. The linear encoder approach seems too costly for now, as the only dual-encoder input controller/driver, I could find, is north of $2,000 USD...

rotary encoders on the stepper motors will not compensate for backlash, they will only pick up missed steps from the motor. the backlash in the table movement will not be seen by the encoder because the movement is all after the motors output shaft. even placing them on the end of the ballscrew will not reflect true table movement. I started out chasing the perfect setup and it is really not feasible on these types of machines. after rethinking what I will be making and the fact not everything on a part has to be in perfect specs just certain points. I went with the basic setup to make the part leaving the really precision points undersize and then come back with measurement's to locate and machine those areas. any machine has to be learned by the operator and they all have hi and low points. I was driving myself nuts chasing perfection when It wasn't really required for one off or short run hobby machining.
steve

 

[Eddyde]

rotary encoders on the stepper motors will not compensate for backlash, they will only pick up missed steps from the motor. the backlash in the table movement will not be seen by the encoder because the movement is all after the motors output shaft. even placing them on the end of the ballscrew will not reflect true table movement. I started out chasing the perfect setup and it is really not feasible on these types of machines. after rethinking what I will be making and the fact not everything on a part has to be in perfect specs just certain points. I went with the basic setup to make the part leaving the really precision points undersize and then come back with measurement's to locate and machine those areas. any machine has to be learned by the operator and they all have hi and low points. I was driving myself nuts chasing perfection when It wasn't really required for one off or short run hobby machining.
steve

Thanks Steve,

I am aware that the rotary encoders won't tell the true table position, that's why I was initially thinking of using linear encoders instead. However, it seems in order to do a successful setup using linear encoders, they must be used in conjunction with rotary encoders, as a second encoder input. My first investigation into "dual encoder input controllers" could only find very expensive options "north of $2,000". Fortunately a gracious member wrote me privately and turned me on to some Galil controllers on ebay that have dual encoder capability and can be had for south of $1000, So I think will invest in one of these controllers. For now, I will go with the steppers & rotary encoders then add the linear encoders later on, as I will have the capability.
Let's just say I am an "accuracy freak"
Cheers
Eddy

 

[JimDawson]

I agree with Steve. An encoder on a lead screw is a useless endeavor in either a stepper or servo system, unless you have a high end machine with double ball nuts, and high precision ball screws. It will pick up lost steps, but that's about it. The positioning will be pretty close, but the backlash and lead screw error will not be compensated for. IMHO the measuring feeback device needs to be on the load, in this case the table or the quill. In the hobby shop, an occasional Oh S&*^ moment is probably acceptable, but if not, then a bit more hardware is needed.

 

[JimDawson]

........
However, it seems in order to do a successful setup using linear encoders, they must be used in conjunction with rotary encoders, as a second encoder input. ........
Let's just say I am an "accuracy freak"
Cheers
Eddy

I have to disagree that theory. If you have a linear encoder (any kind of a linear measuring device) attached to the load (table), and that will output a quadrature signal, it can and should be used as the primary encoder. That will achieve the maximum accuracy. The secondary encoder is not needed unless the application absolutely must not fail. Think aerospace and military applications.

My mill holds a 0.0001 repeatable position, not that my parts come out that close, because there are other factors at work also. But actual tolerance is generally in the range of +/- 0.0005 if I don't get too greedy with the cut.

 

[Eddyde]

I have to disagree that theory. If you have a linear encoder (any kind of a linear measuring device) attached to the load (table), and that will output a quadrature signal, it can and should be used as the primary encoder. That will achieve the maximum accuracy. The secondary encoder is not needed unless the application absolutely must not fail. Think aerospace and military applications.

My mill holds a 0.0001 repeatable position, not that my parts come out that close, because there are other factors at work also. But actual tolerance is generally in the range of +/- 0.0005 if I don't get too greedy with the cut.

Okay maybe I have it wrong, I was looking into a "closed loop stepper drive" like this:
http://www.leadshine.com/category.aspx?type=products&category=easy-servo-products
I thought it would be necessary to use a "dual encoder input controller" such as the Galil, to make it work with linear encoders?

 

[JimDawson]

Okay maybe I have it wrong, I was looking into a "closed loop stepper drive" like this:
http://www.leadshine.com/category.aspx?type=products&category=easy-servo-products
I thought it would be necessary to use a "dual encoder input controller" such as the Galil, to make it work with linear encoders?

I read through the product specs, they are closing the loop with the encoder on the motor. This will work fine in most applications, it just does not account for lead screw errors. It will keep the system from loosing pulses, and is probably adequate for the home machine shop.

It looks like they have ''smart'' drives that will insure that the motors reach the commanded position by reading the encoder feeback. Not a bad system. I have never seen these before. I need to look into this further.

I can see where the confusion is. In this case, it would really not be necessary to have both an encoder on the motor and an encoder on the load.

 

[Eddyde]

I read through the product specs, they are closing the loop with the encoder on the motor. This will work fine in most applications, it just does not account for lead screw errors. It will keep the system from loosing pulses, and is probably adequate for the home machine shop.

It looks like they have ''smart'' drives that will insure that the motors reach the commanded position by reading the encoder feeback. Not a bad system. I have never seen these before. I need to look into this further.

I can see where the confusion is. In this case, it would really not be necessary to have both an encoder on the motor and an encoder on the load.

Okay so if I don't need encoders on the motor and on the load then why do I need the Galil controller that accepts dual encoder inputs? Or would that be necessary only if i used servos?

 

[JimDawson]

Okay so if I don't need encoders on the motor and on the load then why do I need the Galil controller that accepts dual encoder inputs? Or would that be necessary only if i used servos?

Well, actually you don't need dual encoder inputs, in most cases. If you were to use a Galil controller to operate stepper motors and were using it in step and direction mode, then you would need the Aux encoder for position feedback if you want to close the loop. This is because Galil uses the main encoder counter to generate the step output.

You can have the encoder feedback generated from an encoder on the leadscrew, motor, or the load.

Dual encoders are not normally needed on servo systems, as long as the system is tight. If the feedback is from the load it's self, then really only one encoder is required.

 

[bladehunter]

Regarding Chai's ballscrews.

Got them yesterday (Tuesday), screws machined to my specs, nut ground as per drawing, ground section finished as per the rest of the nut, looks like an off the shelf item.

Cost was US $120.00 for parts and $49.00 shipping,great value for a hobby machine I reckon. Doubt very much if I could get just one screw for that amount locally. Just prepping the machine for installation, as I figure it'd be a good Father\Son thing.

I'll report back after the weekend regarding operation.