Need help brainstorming a solution for this mill turn spindle

[mactec54]

Unless you changed the digital gearing on the fly - you don't care about positioning so much when you're using it as a spindle, and you don't care about speed so much when you're doing fiddly precision. How you get Mach3 to spit that out is another question altogether...

He could only do that if he had to separate Servo Drives and motors which is what some do, you can't change electronic gearing on the fly, if he used a motor that is big enough for the job, one setup will work as well too 0- 3000RPM and any increment that he wants to index the spindle there is no complication needed, this is very basic stuff that is done everyday

Look at the machine you are waiting for I don't know what the indexing increments are for that machine but they are doing the same thing without anything that is complicated

[Jim Dawson]

It is easy to achieve the resolution and RPM goals without any gymnastics using off-the-shelf hardware, but not in keeping with the additional goals of cheap and Mach3.

Delta servo drives will accept step & direction inputs up to 4 mHz and have 20 bit encoder resolution. The problem is that to get into the power range required for the spindle the cost is in the $1000+ range.

There are several free or inexpensive alternatives to Mach3.

The better system would be to use analog control of the drive, but this requires more sophistication (and cost) on the control end and of course a drive that will accept an analog input. There is no really good compromise for the overall system.

[QuinnSjoblom]

He could only do that if he had to separate Servo Drives and motors which is what some do, you can't change electronic gearing on the fly, if he used a motor that is big enough for the job, one setup will work as well too 0- 3000RPM and any increment that he wants to index the spindle there is no complication needed, this is very basic stuff that is done everyday

Look at the machine you are waiting for I don't know what the indexing increments are for that machine but they are doing the same thing without anything that is complicated

If you're talking about a different drive with 3 or 4mhz max frequency, then yes, you can step .005 degree and have 3k rpm. We are talking about a 500khz dmm drive and you can not drive it with step/dir to 3k rpm if its set for .005 step (full encoder resolution). Its not up for debate, it's the way it is. If you believe otherwise, fine, let's agree to disagree. I'm done arguing about it. Please allow those of us that understand the frequency limitation to continue discussing it.

[QuinnSjoblom]

It is easy to achieve the resolution and RPM goals without any gymnastics using off-the-shelf hardware, but not in keeping with the additional goals of cheap and Mach3.

Delta servo drives will accept step & direction inputs up to 4 mHz and have 20 bit encoder resolution. The problem is that to get into the power range required for the spindle the cost is in the $1000+ range.

There are several free or inexpensive alternatives to Mach3.

The better system would be to use analog control of the drive, but this requires more sophistication (and cost) on the control end and of course a drive that will accept an analog input. There is no really good compromise for the overall system.

Yep, exactly. Either higher frequency drive, or something other than step/dir. In my case, it will be fine for what I want to do within the limitation of 500khz and step/dir control. With 2-1 mechanical and 8-1 electrical I'll have 1500 rpm at the spindle with a step resolution of .02 degrees and 8000oz/in of torque. This will allow turning and milling in the same program without stopping to change pulleys or settings. If I need higher res for hobbing gears, all I have to do is plug in the servo usb to take out electrical gearing and have .0025 step with max rpm of 187.5. If I have a batch of smaller diameter parts to turn at higher rpm, I can swap pulley for 1-1 and get 3k rpm with step res of .04 degrees.

[QuinnSjoblom]

Finished with the design and components are picked out. Decided the 1kw servo is a bad decision. The whole thing is costing over 2500, so cutting the horsepower and torque in half to save 130 bucks is pretty pointless. Going with the 1.8kw. Pulley system is 8m polychain 36mm wide with carbon gt belt. Pulley on the spindle is a 50 tooth (about 5 inch diameter), pulley on servo is optional 25 or 50t. The 2 pulley options will use 2 separate belts to avoid the use of a tensioner or long mounting adjustment with servo. Pulley system is not cheap. After a lot of reading, the gt2 8m setup with the carbon reinforced belt should be about as good as it gets for rigidity, torque, and backlash. To buy all this stuff new, it would be about 500 bucks for the 3 pulleys and 2 belts with taperlock bushings. Found all of it on ebay, unused old stock for about 200 bucks total.

After thinking about it, I decided this setup will live most of its life at the 1 to 1 mechanical ratio, 8 to 1 electrical, 3000rpm turning. Here's what i realized, 90 percent of the work I do will be turning, then 90 or 45 degree indexing with an irrelevant starting azimuth. As long as my mechanical ratio is a whole number (1-1,1-2,1-3, etc), and my electrical gearing is a binary multiple (1,2,4,8,16, etc), i will get precise indexing of 180deg, 90deg, 45deg, 22.5deg, etc., even though step res is .04 degree. Those indexing points will land exactly on full steps. A couple limitations to this, I can't load a part that already has a flat surface and dial it in to sweep perfectly flat, it would have to be dialed in to nearest .04 degree, or ease up on the drawbar and tap it into flat. This is not something I'll be doing though. All indexing happens at once starting with a round part. Other limitation is hex features. The binary gearing won't put full steps on 60 degree divisions. The workaround is easy, when I need a hex pattern, or dial in an existing flat feature, or anything else that needs higher res, I just drop the electrical gearing out and have .005deg step with 375rpm limit. Plenty of resolution for anything I would need to do including hobbing. If for some reason I want more, throw the 25t on the servo and get .0025deg step. As for holding torque with 1-1 mechanical, it's way more than enough after calculating machining forces at given radius with the work I do. I think the setup should work quite well, and since I work with almost exclusively aluminum, 3000rpm will definitely be an advantage over 1500rpm.

[spumco]

Sounds like you've got a solid plan. Please keep us in the loop as it progresses - I may pirate some of your work if I get around to motorizing my tailstock spindle!

[QuinnSjoblom]

Second guessing the belt style a bit now. After talking with a gates representative for a while, it seems there's a trade off with the carbon gt belts. Apparently the regular gt3 belts have the least amount of backlash, but will stretch a bit more than carbon gt, carbon gt has a bit more backlash than regular gt3, but no stretch. Hard to quantify which is gonna lead to more overall deflection when applying loads in alternating directions. I would say minimal deflection during finish passes is most important so gt3 might be better. Probably just gonna have to do some testing to get actual deflection numbers and mill some test parts to compare surface finish.

Edit: easy solution. Going with original design using 36mm wide pulleys which is direct fit for carbon gt. Gt3 belts use same type of pulley but available widths of belt are different. Goes from 30mm to 50mm wide. Ill order 50mm belt and build a jig with razor to cut it to 36mm width. That way I can test both belts.

[Jim Dawson]

If you don't have to move the spindle during a cut, a disk brake works really well for holding in position.

[QuinnSjoblom]

If you don't have to move the spindle during a cut, a disk brake works really well for holding in position.

Yep, it's definitely possible i could end up adding a brake, but I'm hoping I won't need it as code kind of becomes a mess. I'll need a custom post of I don't want to dig through every gcode file and add enable/disable before and after every A move.
The servo definitely has enough torque to hold position for the type of machining I'll do, so it just comes down to how rigid the belt is. That's why I'm spending so much time making sure i have the best possible belt style for this.

Everything is ordered. Should be starting with the build on a week or so

[spumco]

If you put an idler pulley on the belt OD to take up slack rather than a swining motor mount, you'll wind up with more teeth engaged - especially on the small pulley.

Should cut down backlash and an idler is easy to machine yourself. Just make sure to check the minimum pulley OD for back-bending installations.

EDIT - regarding the brake...

The gent who posted the youtube videos how me makes the Mega/Ultra 4th axis has some interesting tidbits on brakes buried in his videos. He mentions that while servos can hold pretty well, servo dither can cause surface finish problems and recommends a brake. He also mentions adding a brake - at least for some drag - to the tailstock as well (assuming you're using a trunnion table or tail mounted chuck).

[QuinnSjoblom]

If you put an idler pulley on the belt OD to take up slack rather than a swining motor mount, you'll wind up with more teeth engaged - especially on the small pulley.

Should cut down backlash and an idler is easy to machine yourself. Just make sure to check the minimum pulley OD for back-bending installations.

EDIT - regarding the brake...

The gent who posted the youtube videos how me makes the Mega/Ultra 4th axis has some interesting tidbits on brakes buried in his videos. He mentions that while servos can hold pretty well, servo dither can cause surface finish problems and recommends a brake. He also mentions adding a brake - at least for some drag - to the tailstock as well (assuming you're using a trunnion table or tail mounted chuck).

Good point on the idler. Would indeed give more tooth engagement. As for swinging mount, I don't think I would need that even without the idler. My 2 gearing options only differ by 4mm center distance so was just going to have slightly elongated mounting holes, but that would mean through bolts with nuts. Tapping bolts directly into the mount would be easier and could do that if using adjustable idler. Ill consider that for sure.

I've been talking back and forth with dmm and specifically mentioned servo oscillation and it's affect on surface finish with not using a brake. The representative said he's never seen that issue with rotary tables using direct drive with one of their servos, but I guess we'll see. I'm definately keeping the addition of a brake in mind while designing this.

[QuinnSjoblom]

Wanted to give an update on the belt situation in case anyone follows this thread for a build at some point. Originally the plan was to test both polychain carbon gt belts and gt3 belts since the carbon gt has no stretch/more backlash and gt3 has more stretch, less backlash. I thought they both used the same pulley profile but that was incorrect. They have specific pulley types. However, after to talking with gates a bit more, it sounds like the backlash of carbon gt is more significant than the stretch of gt3, especially with a short center distance. Anyway, I returned the polychain pulleys and ordered gt3. Thing about the gt3 pulleys, goes from 30mm wide to 50mm wide, So i went 50mm wide. Pretty substantial pulleys and belts. Almost everything is here now, just waiting on the servo which I ordered a week ago. No shipping notification, been trying to contact cnc4pc without any luck. Hopefully it shipped already

[mactec54]

Wanted to give an update on the belt situation in case anyone follows this thread for a build at some point. Originally the plan was to test both polychain carbon gt belts and gt3 belts since the carbon gt has no stretch/more backlash and gt3 has more stretch, less backlash. I thought they both used the same pulley profile but that was incorrect. They have specific pulley types. However, after to talking with gates a bit more, it sounds like the backlash of carbon gt is more significant than the stretch of gt3, especially with a short center distance. Anyway, I returned the polychain pulleys and ordered gt3. Thing about the gt3 pulleys, goes from 30mm wide to 50mm wide, So i went 50mm wide. Pretty substantial pulleys and belts. Almost everything is here now, just waiting on the servo which I ordered a week ago. No shipping notification, been trying to contact cnc4pc without any luck. Hopefully it shipped already

That is a wide belt when a 20mm wide belt would most likely be all you need, look at a Harley and see they are using 1.500 wide belt for there drive, that wide will be a lot of drag and not very good at speed 50mm wide is a big Hp belt

[QuinnSjoblom]

That is a wide belt when a 20mm wide belt would most likely be all you need, look at a Harley and see they are using 1.500 wide belt for there drive, that wide will be a lot of drag and not very good at speed 50mm wide is a big Hp belt

Yes, but this is a precision rotary axis, not a Harley davidson. a much smaller belt would obviously handle the horsepower. all belts stretch some small amount and teeth deflect some amount. The wider the belt is, the least amount of tooth deflection and stretch there will be. I'm not worried about drag. It's a 1.8kw servo and I doubt I will ever pull more than 1hp turning my aluminum parts. The goal is to not use a brake. If using a brake, i would go with a much smaller belt, but since I'm trying to without, an oversized belt is going to reduce deflection when applying machining loads during index milling

It's hard to know exactly how big of a belt is beneficial, but it definitely can't be figured just on horsepower requirement. I figure as long as I'm staying within the hp limits of the servo with turning horsepower plus drag, a bigger belt can only help with rigidity during indexing. If a 25mm belt allows x amount of angular deflection from belt stretch and tooth deflection, a 50mm belt will cut that in half.

[mactec54]

Yes, but this is a precision rotary axis, not a Harley davidson. a much smaller belt would obviously handle the horsepower. all belts stretch some small amount and teeth deflect some amount. The wider the belt is, the least amount of tooth deflection and stretch there will be. I'm not worried about drag. It's a 1.8kw servo and I doubt I will ever pull more than 1hp turning my aluminum parts. The goal is to not use a brake. If using a brake, i would go with a much smaller belt, but since I'm trying to without, an oversized belt is going to reduce deflection when applying machining loads during index milling

It's hard to know exactly how big of a belt is beneficial, but it definitely can't be figured just on horsepower requirement. I figure as long as I'm staying within the hp limits of the servo with turning horsepower plus drag, a bigger belt can only help with rigidity during indexing. If a 25mm belt allows x amount of angular deflection from belt stretch and tooth deflection, a 50mm belt will cut that in half.

No the bigger wider belt won't help with rigidity, This is how you should be doing it and you have no belt stretch and deflection, to worry about not with GT series belts

This is how Mike Everman does it with his commercial 4th axes drives

[QuinnSjoblom]

No the bigger wider belt won't help with rigidity, This is how you should be doing it and you have no belt stretch and deflection, to worry about not with GT series belts

This is how Mike Everman does it with his commercial 4th axes drives

First of all, that's a 4th axis, not a mill turn spindle. I can tell from the huge gear reduction. Much lower rpm than what I'm looking for. Irrelevant to this "mill turn" project. I can understand why that 4th axis is built that way, to increase tooth engagement and reduce length of unsupported belt to minimize stretch. For a high reduction, it's a great idea, but irrelevant for a 1 to 1 mill turn.

Second of all, do you have any basis for stating "no the bigger wider belt won't help with rigidity"? Can you back that up with any actual fact? Its simple physics. Belts stretch, teeth deflect/compress under load. Doesn't matter what type of belt it is, everything deflects some amount. Increasing the width of the belt spreads the load across more tooth area, which means the deflection/compression of the tooth is reduced. As for belt stretch, I think it's pretty obvious why a wider belt takes more force to stretch a given amount. Just to be clear, when I say "rigidity" I'm talking about angular deflection of the spindle when torque is applied from machining forces

[mactec54]

First of all, that's a 4th axis, not a mill turn spindle. I can tell from the huge gear reduction. Much lower rpm than what I'm looking for. Irrelevant to this "mill turn" project. I can understand why that 4th axis is built that way, to increase tooth engagement and reduce length of unsupported belt to minimize stretch. For a high reduction, it's a great idea, but irrelevant for a 1 to 1 mill turn.

Second of all, do you have any basis for stating "no the bigger wider belt won't help with rigidity"? Can you back that up with any actual fact? Its simple physics. Belts stretch, teeth deflect/compress under load. Doesn't matter what type of belt it is, everything deflects some amount. Increasing the width of the belt spreads the load across more tooth area, which means the deflection/compression of the tooth is reduced. As for belt stretch, I think it's pretty obvious why a wider belt takes more force to stretch a given amount. Just to be clear, when I say "rigidity" I'm talking about angular deflection of the spindle when torque is applied from machining forces

You are not going to have enough Hp or loading to cause any stretch, it's free spinning most of the time most likely you could do what you want with a 16mm wide belt designed like what I posted will handle any machining you are going to do

He makes all sizes and is a proven system, I guess you will find out that what you are doing is not needed, I have built them also with belts up to 75mm wide for big Hp drives, you will find that the 50mm wide won't be easy to work with

For any drive design like this you need the 2 tensioning rollers this helps keep the belt the same with any directional changes

If you want 1:1 then you want to make it the same design as I posted, that is if you want accuracy as well

[QuinnSjoblom]

You are not going to have enough Hp or loading to cause any stretch, it's free spinning most of the time most likely you could do what you want with a 16mm wide belt designed like what I posted will handle any machining you are going to do

He makes all sizes and is a proven system, I guess you will find out that what you are doing is not needed, I have built them also with belts up to 75mm wide for big Hp drives, you will find that the 50mm wide won't be easy to work with

For any drive design like this you need the 2 tensioning rollers this helps keep the belt the same with any directional changes

If you want 1:1 then you want to make it the same design as I posted, that is if you want accuracy as well

How do you figure there is no stretch or tooth deflection on belts? This in incorrect, it's always a factor. Like a said a couple times, it has nothing to do with the horsepower I have. It's for preventing tooth deflection and stretch during machining loads while the servo is holding a static position. A 16mm belt will obviously allow more tooth deflection. These belts are not rigid objects. The teeth deform some amount under load. Yes, the 50mm could be overkill, but the next size down is 30mm and I believe there is an advantage going over that width when it comes to the servo holding a position. The gt3 belts DO stretch, doesn't matter if it's wrapping around an idler or not, that's a smooth surface and still allows belt stretch under load. This stretch as well as tooth deformation is reduced with a wider belt. I agree with you that the wider belt is going to create extra drag. If the drag of 50mm wide belt puts me over my horsepower capability, then yes it will be a problem and I will go down to 30mm wide, otherwise it's only helping to reduce angular deflection during index milling. You can't argue that belts have no stretch or tooth deformation, that's simply wrong and a wider belt reduces this. Once again, let's agree to disagree. I think I'll take the word of the gates engineer I've been talking to over the last week. I would assume he knows a bit more about these belts than either of us.

[mactec54]

How do you figure there is no stretch or tooth deflection on belts? This in incorrect, it's always a factor. Like a said a couple times, it has nothing to do with the horsepower I have. It's for preventing tooth deflection and stretch during machining loads while the servo is holding a static position. A 16mm belt will obviously allow more tooth deflection. These belts are not rigid objects. The teeth deform some amount under load. Yes, the 50mm could be overkill, but the next size down is 30mm and I believe there is an advantage going over that width when it comes to the servo holding a position. The gt3 belts DO stretch, doesn't matter if it's wrapping around an idler or not, that's a smooth surface and still allows belt stretch under load. This stretch as well as tooth deformation is reduced with a wider belt. I agree with you that the wider belt is going to create extra drag. If the drag of 50mm wide belt puts me over my horsepower capability, then yes it will be a problem and I will go down to 30mm wide, otherwise it's only helping to reduce angular deflection during index milling. You can't argue that belts have no stretch or tooth deformation, that's simply wrong and a wider belt reduces this. Once again, let's agree to disagree. I think I'll take the word of the gates engineer I've been talking to over the last week. I would assume he knows a bit more about these belts than either of us.

Your small machining loads won't even be noticed

If you have a problem with tooth deflection then you don't have enough teeth engaged with would be a design problem not a belt problem

Any teeth deformation is taken care of by the number of teeth engaged, timing belts have been used for CNC machines for 40 years this is not something new you are doing, Bridgeport is one that used a 25mm wide belt and never had an accuracy problem

The idlers are there to keep the belt tensioned in both directions

If your belt is correctly tensioned there will not be any stretch while it is operating only over loading the belt beyond it's design is when you will see stretch, it all comes down to how you have calculated your loading for your use of the belt

I know some of the Gates application engineers, and I machine a lot of there Timing Pulleys to there spec's when needed, I have all the GT series Profiles

[QuinnSjoblom]

Your small machining loads won't even be noticed

If you have a problem with tooth deflection then you don't have enough teeth engaged with would be a design problem not a belt problem

Any teeth deformation is taken care of by the number of teeth engaged, timing belts have been used for CNC machines for 40 years this is not something new you are doing, Bridgeport is one that used a 25mm wide belt and never had an accuracy problem

The idlers are there to keep the belt tensioned in both directions

If your belt is correctly tensioned there will not be any stretch while it is operating only over loading the belt beyond it's design is when you will see stretch, it all comes down to how you have calculated your loading for your use of the belt

I know some of the Gates application engineers, and I machine a lot of there Timing Pulleys to there spec's when needed, I have all the GT series Profiles

Saying that belts only stretch unless overloaded is incorrect. Same with tooth deflection. Any load at all will cause tooth deflection and belt stretch, it's just a question of how much. Everything in this universe deflects with any amount of pressure. If you place a grain of rice on the corner of a 12 inch thick chunk of granite, the granite bends. There is no such thing as "no deflection". Tooth deflection and belt stretch is proportional to belt width. The fact that you say things like "there will be no stretch" makes me think you don't understand these facts. There will always be stretch and tooth deflection. The goal is to reduce it as much as possible. There is no point where it's eliminated completely. It's the laws of physics

And once again, the 50mm may very well be overkill and beyond what's going to give significant improvent, but I definitely dont believe there's no reason to go bigger than 16mm. The more deflection there is, the worse chatter will be during machining.