[SINTORMAN]

I am helping install an Acer vertical mill with a millstar add-on that drives table x and y and the quill as the z. Position feedback is linear scales on the x and y with the z using the encoder on the motor (will most likely swap this to a scale as well).

The system is.... old. The cnc system is running windows 95 which while miraculously still seems to be running fine, probably isn't long for this world. We want to bring the system into this century.

Everything works to my knowledge and I think the current route is to keep the glentek dc brushed servos and glentek drives and then utilize linuxcnc to control it (I'm not super worried about the servos themselves as they most likely have very low hours but just the age of the drives does concern me. Someone mentioned making my own with big SSRs. Is that a viable option?)

I have looked into other systems but the lack of position confirmation via the scales seems to be limiting. The masso g3 for example seemed like a sweet solution until realizing that I would need to create my own control loop because the hardware cannot support it (is this a viable option?)

I suppose the ultimate question is... Does anyone have experience with retrofitting this system and/or what would you recommend we do?

This is a hobby grade setup that we want to learn with primarily. Budget is first, ease of use second and installation complexity last (I'm an EE or at least almost lol)

Secondary question is how do most people deal with a powered quill? Assuming it is switched to a linear scale, is there a fast way to switch from powered to manual feed? On this cnc you must unbolt the screw from the quill via the drive nut which isn't fast by any means. Or would switching the motor to the knee be a good option? (I have read the issue with that is if the knee doesn't raise/lower perfectly vertical then you can't really account for that whereas you can adjust the quill via the head assembly.)

[mactec54]

I am helping install an Acer vertical mill with a millstar add-on that drives table x and y and the quill as the z. Position feedback is linear scales on the x and y with the z using the encoder on the motor (will most likely swap this to a scale as well).

The system is.... old. The cnc system is running windows 95 which while miraculously still seems to be running fine, probably isn't long for this world. We want to bring the system into this century.

Everything works to my knowledge and I think the current route is to keep the glentek dc brushed servos and glentek drives and then utilize linuxcnc to control it (I'm not super worried about the servos themselves as they most likely have very low hours but just the age of the drives does concern me. Someone mentioned making my own with big SSRs. Is that a viable option?)

I have looked into other systems but the lack of position confirmation via the scales seems to be limiting. The masso g3 for example seemed like a sweet solution until realizing that I would need to create my own control loop because the hardware cannot support it (is this a viable option?)

I suppose the ultimate question is... Does anyone have experience with retrofitting this system and/or what would you recommend we do?

This is a hobby grade setup that we want to learn with primarily. Budget is first, ease of use second and installation complexity last (I'm an EE or at least almost lol)

Secondary question is how do most people deal with a powered quill? Assuming it is switched to a linear scale, is there a fast way to switch from powered to manual feed? On this cnc you must unbolt the screw from the quill via the drive nut which isn't fast by any means. Or would switching the motor to the knee be a good option? (I have read the issue with that is if the knee doesn't raise/lower perfectly vertical then you can't really account for that whereas you can adjust the quill via the head assembly.)

If this system is working changing the control won't make much difference, you still have the old motors and drives, windows 95 will keep running until someone messes it up.

To modernize this machine, you would be using AC servo with any of the CNC controls that are available Eding CNC / Acorn / Mach4 / UCCNC / Planet CNC and many others to choose from

[SINTORMAN]

If this system is working changing the control won't make much difference, you still have the old motors and drives, windows 95 will keep running until someone messes it up.

To modernize this machine, you would be using AC servo with any of the CNC controls that are available Eding CNC / Acorn / Mach4 / UCCNC / Planet CNC and many others to choose from

... interfacing with the machine will be massively different.... The most up to date communication is serial. Not only that but the interface also sucks just for daily use. Can't seem to find a way to do a relative home/zero and getting to different functions is tedious using the stock UI. Old servos yes but again minimal use. Unless you are saying that they won't be as accurate as current drives. I got the impression that with the right control, they are still great units. Windows 95 sure but I have very little confidence in the control hardware as well for longevity and would rather update now rather than later.

And as long as the DC servos work we will stick with those as that's too much cost to justify upgrading at this point.

We want not only better CAD to CAM implemntation closer to how 3d printers operate nowadays with just sending gcode over the network but also daily use of a great DRO with lots of built in functions like patterns and shapes and modern stuff like touch probe and auto tool height.

[joeavaerage]

Hi,
can you tell us a bit more about what servos/drives you have and how it is the linear scales close the position loop?.

There are of course controllers that do that, and its very much the way things were done years ago, but these days simpler open loop motion controllers are the norm,
the servo drive that closes the position loop using either the rotary encoder of the servo or in some cases a load mounted linear scale.

Depending on how your current system is arranged you may be faced with the choice to replace your existing servos and drives with more modern AC types, therafter all the
motion control solutions that Matec mentioned are feasible.

If you wish to retain the old servos that may require a feedback capable motion controller....and they are rather more expensive and tricky to set up.

Craig

[SINTORMAN]

Hi,
can you tell us a bit more about what servos/drives you have and how it is the linear scales close the position loop?.

There are of course controllers that do that, and its very much the way things were done years ago, but these days simpler open loop motion controllers are the norm,
the servo drive that closes the position loop using either the rotary encoder of the servo or in some cases a load mounted linear scale.

Depending on how your current system is arranged you may be faced with the choice to replace your existing servos and drives with more modern AC types, therafter all the
motion control solutions that Matec mentioned are feasible.

If you wish to retain the old servos that may require a feedback capable motion controller....and they are rather more expensive and tricky to set up.

Craig

So the X and Y utilize the Glentek GM4020-39-02006301 which means 39V/krpm, no brake, 7VDC tach, no encoder, and no resolver.
The Z uses the GM4020-39-00406401 which means 39V/krpm, no brake, no tach, 1000 PPR encoder and no resolver.
I believe all drives are Glentek SMA7115-1

Manual for the motors is found here: https://www.glentek.com/download/gm4...et/?wpdmdl=244
Drive manual: https://www.glentek.com/download/sma...al/?wpdmdl=371

They are mitutoyo linear scales that feed into the millstar control box. I don't know how it's using them but I know that's the only way it's receiving position feedback as the x and y only spit out a tach signal.

I could add encoders but it seems like having actual table position is far better (with proper PID tuning) to eliminate backlash.

Correct me if I am wrong but it would appear that this is quite feasible with linuxcnc utilizing a mesa 7176E as well as the 7185 for capturing the linear scales output (utilizing the stock servos and drives). Then the other option that I was looking at was the Allin1DC controller which is a bit pricey but appears to handle that as well.

Can you ballpark a price for changing all the servos out for new stuff and match the power/torque to drive it as well as stock? They are 1.3 HP, 23 lb-in/2.6 Nm

[joeavaerage]

Hi,
the x and y servos rely on a motion controller to close the position loop, ie the motion controller must be PID capable.

linuxcnc utilizing a mesa 7176E as well as the 7185 for capturing the linear scales output (utilizing the stock servos and drives).

Yes, LinuxCNC can do that and is probably the most cost effective solution.

As you say there are other controllers that can close the loop, probably one of the most capable but pricey is Gallil.

I could add encoders but it seems like having actual table position is far better (with proper PID tuning) to eliminate backlash.

Using linear scales to close the position loop is desirable, although many good machines have just a rotary encoder on the ballscrew or servo without undue loss of accuracy.

I use Delta AC servos, good quality, backup, software at fair prices.

A Delta A2 series servo and drive have both an encoder built into the servo ( 1,280,000 count per rev) but also a secondary encoder channel for closing the position loop with say a linear scale.
Using such a servo only requires an open loop motion controller (easy, cheap) which feeds Step/Dir signals to the A2 servo drive, and the servo/drive does the rest. ie full PID type feedback.

It is increasingly common amongst servo manufacturers to use these dual loop servos.

I use 750W B2 series servos which do not have the secondary encoder channel, and they are $438USD for servo/drive and cables.The 750W A2 series is more again, around $670USD for servo/drive/cables.

Using these more modern servos would allow you to retain the linear scales but use simpler and cheaper open loop software motion controllers like Mach4 and UCCNC.

If you wish to retain your existing servos then you'll need a feedback capable controller, and accept whatever expense or complication that comes with that.

Craig

[SINTORMAN]

Hi,
the x and y servos rely on a motion controller to close the position loop, ie the motion controller must be PID capable.



Yes, LinuxCNC can do that and is probably the most cost effective solution.

As you say there are other controllers that can close the loop, probably one of the most capable but pricey is Gallil.



Using linear scales to close the position loop is desirable, although many good machines have just a rotary encoder on the ballscrew or servo without undue loss of accuracy.

I use Delta AC servos, good quality, backup, software at fair prices.

A Delta A2 series servo and drive have both an encoder built into the servo ( 1,280,000 count per rev) but also a secondary encoder channel for closing the position loop with say a linear scale.
Using such a servo only requires an open loop motion controller (easy, cheap) which feeds Step/Dir signals to the A2 servo drive, and the servo/drive does the rest. ie full PID type feedback.

It is increasingly common amongst servo manufacturers to use these dual loop servos.

I use 750W B2 series servos which do not have the secondary encoder channel, and they are $438USD for servo/drive and cables.The 750W A2 series is more again, around $670USD for servo/drive/cables.

Using these more modern servos would allow you to retain the linear scales but use simpler and cheaper open loop software motion controllers like Mach4 and UCCNC.

If you wish to retain your existing servos then you'll need a feedback capable controller, and accept whatever expense or complication that comes with that.

Craig

That's good information on the delta stuff. Well considering my labor is free for this then the owner will probably opt for the linuxcnc option lol.

A few more questions come to mind.

1. Obviously you sacrifice production speed and rigidity by downsizing your servo power/torque but how much lower can you go on a 3hp vertical mill with hobbyist use in mind? Obviously what you suggested is about half the power but looks to be similar torque and speed but could you theoretically go with a much lower power ac servo/drive combo to save cost? I could maybe talk him into new A2 combos if all in for the servos and drives was a grand or less.

2. Do you know of any dc brushed drives that have linear scale input feedback? If I just wanted to replace those and then go with something like the masso g3 for simplicity.

3. I haven't looked on here yet as I'm not sure where to look but I am an electrical engineering student that loves to create stuff. Has anyone on here created their own closed loop drives utilizing a microcontroller taking in step/dir and then using a PID loop with a linear scale input to control a h bridge to drive the servo? I know PID control on small scale is quite popular but not sure what translates and what doesn't to large scale.

4. Where would you recommend I look to purchase the delta gear? I'm only seeing alibaba-esque retailers.

[joeavaerage]

Hi,

Obviously you sacrifice production speed and rigidity by downsizing your servo power/torque but how much lower can you go on a 3hp vertical mill with hobbyist use in mind?

I am not advocating that you downsize.....it would be a shame to end up with a less than satisfactory result. My new build mill is about 800kg, has three 115kg cast iron axis beds,
32mm C5 ground ballscrews and 750W Delta servos on all three axes. It moves scary fast....still slow by comparison to a production mill certainly but for a hobbyists machine
750W or 400W servos should be plenty. You should do the inertia calculation to arrive at the expected axis acceleration and max velocity for a given level of torque.
Then you can truly ask and answer the question....'is this sufficient for my needs?'

As an example I had thought that 400W per axis would have been enough but when I looked I found that a 750W servo cost $438 and a 400W $390. Why would I try to save $48
when I could have double the power? Of course I got the 750W units, and very glad I did, 400w would have been ok but 750W is better!.

Do you know of any dc brushed drives that have linear scale input feedback? If I just wanted to replace those and then go with something like the masso g3 for simplicity.

No, I don't. That does not mean that they aren't out there, just I don't know about them.

There are any number of posts and threads on this forum where people have tried to get old servos to go by adapting drives from a different manufacturer. At the very least
its a challenge and many I suspect never succeed. My advice would be to buy a servo and drive as a matched pair. There are plenty of new things to learn when getting into CNC
and I'm dubious that 'mixing and matching' servos and drives is a good use of your talent.

I'm not sure where to look but I am an electrical engineering student that loves to create stuff. Has anyone on here created their own closed loop drives utilizing a microcontroller taking in step/dir and then using a PID loop

I too am an Electrical Engineer and like you love designing and making things. I have for instance a 3kW remanufactured servo from the early 90's in perfect order that I wish to use as a spindle motor.
Given that it has a resolver rather than an encoder, what drives are available are really expensive, so I decided to make one. I'm using a Texas InstrumentsF28069M Launchpad
which has a TMS320F28069SM microcontroller that is just the bee's knees for rotating machinery. I've had a lot of fun and learnt a lot in doing it but I'm glad my machne does not depend on me
finishing this project!

I use my mill for work daily, and I just cannot afford to have it down for any length of time, it has become an indispencable par of my work.

By all means adapt or otherwise build a controller if that's your interest, but my recomendation is get new servos and matching drives, one of the many perfectly excellent and cheap
control solutions and get on with it.

[joeavaerage]

Hi,

Obviously you sacrifice production speed and rigidity by downsizing your servo power/torque but how much lower can you go on a 3hp vertical mill with hobbyist use in mind?

I am not advocating that you downsize.....it would be a shame to end up with a less than satisfactory result. My new build mill is about 800kg, has three 115kg cast iron axis beds,
32mm C5 ground ballscrews and 750W Delta servos on all three axes. It moves scary fast....still slow by comparison to a production mill certainly but for a hobbyists machine
750W or 400W servos should be plenty. You should do the inertia calculation to arrive at the expected axis acceleration and max velocity for a given level of torque.
Then you can truly ask and answer the question....'is this sufficient for my needs?'

As an example I had thought that 400W per axis would have been enough but when I looked I found that a 750W servo cost $438 and a 400W $390. Why would I try to save $48
when I could have double the power? Of course I got the 750W units, and very glad I did, 400w would have been ok but 750W is better!.

Do you know of any dc brushed drives that have linear scale input feedback? If I just wanted to replace those and then go with something like the masso g3 for simplicity.

No, I don't. That does not mean that they aren't out there, just I don't know about them.

There are any number of posts and threads on this forum where people have tried to get old servos to go by adapting drives from a different manufacturer. At the very least
its a challenge and many I suspect never succeed. My advice would be to buy a servo and drive as a matched pair. There are plenty of new things to learn when getting into CNC
and I'm dubious that 'mixing and matching' servos and drives is a good use of your talent.

I'm not sure where to look but I am an electrical engineering student that loves to create stuff. Has anyone on here created their own closed loop drives utilizing a microcontroller taking in step/dir and then using a PID loop

I too am an Electrical Engineer and like you love designing and making things. I have for instance a 3kW remanufactured servo from the early 90's in perfect order that I wish to use as a spindle motor.
Given that it has a resolver rather than an encoder, what drives are available are really expensive, so I decided to make one. I'm using a Texas InstrumentsF28069M Launchpad
which has a TMS320F28069SM microcontroller that is just the bee's knees for rotating machinery. I've had a lot of fun and learnt a lot in doing it but I'm glad my machne does not depend on me
finishing this project!

I use my mill for work daily, and I just cannot afford to have it down for any length of time, it has become an indispensable part of my work.

By all means adapt or otherwise build a controller if that's your interest, but my recommendation is get new servos and matching drives, one of the many perfectly excellent and cheap
control solutions and get on with it. There is still huge scope for experimentation and building without having to build your own drives.

4. Where would you recommend I look to purchase the delta gear? I'm only seeing alibaba-esque retailers.

https://www.fasttobuy.com/

Craig

[SINTORMAN]

https://www.fasttobuy.com/

Craig

Yeah that's the first one I found but I wonder if this is worth the gamble lol....

https://m.alibaba.com/product/62574629070/Original-Brand-ASDA-A2-Series-Delta.html

[Jim Dawson]

A Dynomotion Kanalog/Kflop system will work fine with your Glentech drives & motors. And also works with your linear scales if they are quadrature output. Nothing wrong with LinuxCNC either. I'm still using the old DC drives and motors on my machine, they are working just fine.

If you want to buy Delta drives & motors, Automation Direct is a good option. https://www.automationdirect.com/selectors/sureservo2

Automation Direct also has the less expensive LS Electric line of servo motors and drives https://www.automationdirect.com/selectors/ls-servo Not all of the features of the Delta drives, but work just fine.

DMM servo drives & motors are another good option. https://store.dmm-tech.com/

All of the above have good factory support, good documentation, and easy to use setup software. You won't normally get this when buying from non-distributor sellers. I have used all of the above motors and drives in various systems and have no complaints about any of them. I have both Delta and DMM on my lathe, and I just installed a couple of LS Electric servos on another project.

As far as a quick switchover from CNC to manual on your Z axis. It's pretty hard to do with a ball screw, but with some clever engineering it might be possible. I can switch mine over in a few seconds with a lever flip and a mouse click. My machine started out life as a 2 axis CNC/manual and I added the Z axis to the quill, originally had an Anilam M controller. I went a bit different route than most folks and designed a gear system that uses the original pinion shaft. I added an air spring backlash compensator and am able to consistently hold 0.0001'' or better in the Z axis.

[joeavaerage]

Hi,

Automation Direct also has the less expensive LS Electric line of servo motors and drives https://www.automationdirect.com/selectors/ls-servo Not all of the features of the Delta drives, but work just fine.

I don't think, or at least can't find that LS servos have a secondary encoder input like a Delta A2. If you were contemplating using an open loop software/controller solution like Mach/UCCNC then that feature is
REQUIRED if you wish to retain linear scales and therefore LS may not be the right choice.

A Dynomotion Kanalog/Kflop system will work fine with your Glentech drives & motors. And also works with your linear scales if they are quadrature output. Nothing wrong with LinuxCNC either. I'm still using the old DC drives and motors on my machine, they are working just fine.

I had forgotten Dynamotion, but it would indeed be a cost effective solution. Dynamotion has a Mach3 plugin, but not a Mach4 plugin and hence I tend to overlook it. Both Dynamotion and LinuxCNC would allow you to
roll up your sleeves and get stuck in. Programming and tuning PID loops should be right up your alley and within your field of expertise. When I first started contemplating CNC my focus at that time was isolation routing
of PCBs and for that purpose Mach3 was, and still is, highly applicable, and that is the direction I went, progressing to Mach4 a year or two later. Mach4 is still an open loop software/motion control solution and as a consequence
I tend to focus on solutions where the feedback loop is closed by the servo drive.

In the early days of CNC, read DC servos and servo amplifiers (non PID), all CNC controllers were required to close the loop and enact PID control. Ever increasingly the servo drives have become smarter, to whit they now ALL close the
position loop, either with the servo encoder, or as we have talked about, around a secondary encoder channel like a linear encoder. If you have researched any of the bus protocols like Ethercat, Profibus or CANOpen you'll find
that each servo drive enacts a portion of the motion control for 'distributed motion control'. Thus the controller is no longer the motion control, its just a trajectory planner, each servo enacts its portion of the overall motion control
solution. The modern control paradigm is for ever increasing capability of servos and/or control nodes but ever decreasing smartness of the central controller which in effect is becoming a supervisor only.

The question is 'do you wish to go the modern smart servo drive/simple motion control' or the 'old school linear servo amp/PID controller' route. Both work but have different philosophy.

Craig