Design Discussion: High Performance CNC Spindle

[macardoso]

Low TIR

I think the golden standard is to have your entire tooling assembly below 0.0003" TIR. Since most tooling will tend to add at least 0.0002" TIR, the spindle must be very accurate to avoid tolerance stacking.

Ultra precision spindle bearings are very expensive so I have long considered hard turning the spindle taper in place on the machine to be a viable option. The bearings would still need to be very high precision, but maybe you could sneak by with something less than ABEC9/ABEC7.

Spindle repair companies will sometimes hard turn a commercial VMC spindle in place so this seems reasonable to me.

Belt vibrations and pulley concentrically will play into the runout at these levels so the transmission systems needs to be well balanced.

 

[RJSakowski]

My machine is directly under the living room and the better half is sensitive to noise from below. However, I find the the noise due to machining far exceeds that of the motor and spindle. Because of this, I find myself running less aggressively than I might otherwise.

 

[macardoso]

My machine is directly under the living room and the better half is sensitive to noise from below. However, I find the the noise due to machining far exceeds that of the motor and spindle. Because of this, I find myself running less aggressively than I might otherwise.

Sounds very familiar Funny how engineering requirements come from all over the place

 

[macardoso]

I'm going to call it a day for putting ideas up here, hopefully give @shooter123456 a chance to share some of his design as well. I'll follow up with some of the other topics perhaps tomorrow.

 

[RJSakowski]

One of the issues with the Tormach design was a increase in spindle length due to frictional heating. One approach has been to use Belleville or wave spring washers to control the bearing preload. Perhaps a better way would be to use opposed angular contact or tapered roller bearings at the bottom of the spindle and a third bearing that has axial play to control runout at the top.

 

[ACHiPo]

Interesting thread!

 

[spumco]

Macardoso,

You know I've been on the same hunt for about 18 months as I plan my upgrade from R8 to BT30...

There is a guy who has replicated the Fadal 2-speed poly-V system. There's a quick Youtube video of it working, and more on his instagram posts. He's using air instead of hydraulics but it looks sweet. That system give you so much more flexibility with motor & spindle selection.

Do want. Not so sure I want to go through the hassle of building one, however.

I started off with a huge "3HP" 1110mm frame BLDC on my mill from the factory. Pathetic, proprietary driver would cut out, no low end torque. That spindle motor & drive are probably what killed Mikini more than anything. Figured I had a motor good for scrap only - who wants to pay $900 for a Tiawan BLDC drive on a suspect motor? Motor is the same thing (more or less) as on the Tormach 770.

So I switched to a 2HP induction motor with encoder feedback to the VFD and motion controller. Works very well, but is simply too big & heavy for my mill at about 80lbs. Can rigid peck-tap despite 1:1.56 ratio because there is an index sensor on the spindle to trigger the Z/spindle synchronization.

I am about to get started on a servo upgrade to cut down on weight and provide C-axis positioning. Originally got a Parker Gemini drive & servo which would get me to about 8kRPM with some belt ratio. I was prepared to live with (some) lost low end torque. Got tired of waiting for Ebay to cough up an A06B-0853 2.2kw 10kRPM from a Chiron for under $500.

As you mentioned most 240-1P drives top out about about 8-10 amps, and there are a few at 10-12. Getting a 5kRPM servo and driving it at 1:2 means your little 1.5HP motor is going to be a dog drilling steel at low speeds.

But... I read an old post on the 'Zone about servos & BLDC's recently, and discovered that some of the old 'analog' servo drives had some serious amps. Jim Dawson is probably responsible in some way for that thread...

AMC, Copley, and others made 'generic' drives which top out at 20A continous - now we're talking 5hp servos! They take 0-10Vdc analog input for velocity control.

However, most servos for sale on Ebay are resolver types, not encoders with halls. And quite a few have proprietary encoder or resolver outputs (Fanuc, Yaskawa, Kollmorgen (some)). No-go on the drive to motor communication - playing mix-n-match with servos & drives is not for the faint of heart.

So...enter CUI Devices (and others). Programmable commutation encoders with auto hall phasing. Find a motor with speed rating and continuous amps at the max of your drive and just replace all the proprietary signal components with a $40 CUI commutation encoder.

Now you can get a cheap old analog drive, and a big 130mm frame 5kRPM servo and have a 10kRPM spindle with positioning capability. And because it's 1:2, you don't need a separate spindle encoder (if using timing belts) - just tell the controller you've got a 2x encoder count with 2 index pulses per rev. And the servo should still have home shop grunt at low RPM.

Or duplicate the Fadal thing like Mr. Genius above and now you've got proper low-end torque and whatever top end you want. I think a 4HP servo running a facemill at a 2:1 reduction would push the limits of a BT30 spindle. Definately need the drive dogs for that one.

OR... and this is what's sitting on my bench at the moment... I'm going to try taking that huge BLDC and fit it with an encoder. Then drive it like a servo with a Copley XTL-230-40 20A drive. The Copley has auto phasing, and auto hall alignment to boot. I hope & suspect that the old 'dog' BLDC will perform rather well and it's got a 6kRPM mechanical rating. We shall see. The Copley fell in to my lap, so to speak.

If not, I also scored another Gemini & a 4kRPM rated speed 3KW servo for a silly price. One of them will go on the mill, and the other will go on a to-be-designed CNC lathe.

Summary - VFD & induction motor (get a sheet metal one, not cast-iron) is inexpensive and easy. Otherwise one can probably piece together a servo system with some patience.

I will report back on the BLDC to servoectomy experiment shortly.

-Ralph

 

[shooter123456]

The spindle I am working on has slightly different design goals, but the majority is similar to what macardoso posted above. The overall goal is to come up with a spindle design that is smoother, quieter, and faster than the stock design, that can be made for reasonable cost using machines available to hobby users. I don't love outsourcing the work to professionals because then I have a nice end product without the pride of making it.

My goals for the spindle are (at the moment):
1. A smooth 10k RPM
2. Rigid tap capable
3. ATC ready
4. 2-3 HP

For the taper, I am going with BT30 for a number of reasons.
1. Tooling is readily available and inexpensive (TTS is restricted to Tormach and a few knockoffs of questionable quality)
2. Tooling is more likely to be sharable with future machines
3. Substantially more variety available for tools
4. Better retention with a pull stud (compared to TTS using R8 collet and 3/4" shank)
5. Much easier to implement an ATC
6. Fits well for the size of the machine

I would like to be able to fit the spindle on the machine with minimal permanent modification to it. The biggest restriction there is going to be the bore that the outside of the housing needs to fit into (2.36" on my machine). That restricts the size of the upper bearing slightly as well as the spindle wall thickness. The alternative is either making a new head for the machine (Either with welded steel or aluminum screwed together), which is a possibility, but one I would prefer to avoid if there isn't a significant gain in performance. However, the way the head separates from the Z slide, a new one would not be out of the question.

Right now I am looking at using a matched pair of 7007 bearings on the bottom and a single 6906 bearing at the top. The reason for this selection is largely size and cost. If I decide to go ahead and make a new head for the machine, I can fit a larger set. The lower bearings are the same size as the current lower bearing (though now there is only 1) and the upper bearing is slightly larger (30x47mm vs 25x47mm). A matched pair of 7007 bearings is relatively inexpensive and they are easy to find. I have found quite a few match pairs for under $150 that will work. Stepping up in size to a 7208 would also be possible as those aren't prohibitively expensive, but that would need a new head to make it worthwhile.

The top bearing would be free to float axially to allow for thermal expansion while the bottom pair will be fixed.

I have a few pictures of my current design compared to the stock spindle. I am still early in the planning and designing process so everything is still open to suggestion and change.

 

[bakrch]

Very exciting thread.

My home shop/garage is pretty far away from the house ... not terribly concerned with noise levels.

I am far from a formally educated engineer, but if I can lend a hand in any way I will do so.

Pending approval from management, I have access to a fully equipped machine shop for personal projects (manual and CNC). This includes ovens for heat treatment that are large enough to fit anything we can dream up. Precision grinding would be my limitation, absolutely none of that equipment in our shop.

I know we want to keep this in the realm of a hobbyists ability to produce, but I do not mind prototyping.

 

[macardoso]

You know I've been on the same hunt for about 18 months as I plan my upgrade from R8 to BT30...

I forgot all about that but yes! Never knew details of what you were looking for.

There is a guy who has replicated the Fadal 2-speed poly-V system. There's a quick Youtube video of it working, and more on his instagram posts. He's using air instead of hydraulics but it looks sweet. That system give you so much more flexibility with motor & spindle selection.

I've seen this one and it is awesome! Bit noisy, but otherwise really clever. Exactly what I'm thinking of.

I am about to get started on a servo upgrade to cut down on weight and provide C-axis positioning. Originally got a Parker Gemini drive & servo which would get me to about 8kRPM with some belt ratio. I was prepared to live with (some) lost low end torque. Got tired of waiting for Ebay to cough up an A06B-0853 2.2kw 10kRPM from a Chiron for under $500.

They're always on my watch list. So expensive though

I started off with a huge "3HP" 1110mm frame BLDC on my mill from the factory. Pathetic, proprietary driver would cut out, no low end torque. That spindle motor & drive are probably what killed Mikini more than anything. Figured I had a motor good for scrap only - who wants to pay $900 for a Tiawan BLDC drive on a suspect motor? Motor is the same thing (more or less) as on the Tormach 770.

I thought all the Tormachs were AC induction motors. Huh. It's amazing how much Tormach did right from the get go - even with all the issues they've had. Nobody else has been able to stick in that market.

So I switched to a 2HP induction motor with encoder feedback to the VFD and motion controller. Works very well, but is simply too big & heavy for my mill at about 80lbs. Can rigid peck-tap despite 1:1.56 ratio because there is an index sensor on the spindle to trigger the Z/spindle synchronization.

That's amazing. What controller are you running?

Getting a 5kRPM servo and driving it at 1:2 means your little 1.5HP motor is going to be a dog drilling steel at low speeds.

My 1:1 11Nm 2.25 HP servo hits 100% on a mildly aggressive 1/2" drill through aluminum. It still has some significant overload reserves though. Steel is a different beast.

So...enter CUI Devices (and others). Programmable commutation encoders with auto hall phasing. Find a motor with speed rating and continuous amps at the max of your drive and just replace all the proprietary signal components with a $40 CUI commutation encoder.

I've seen their stuff and it looks great! Always have gone with US Digital for low cost encoders.

OR... and this is what's sitting on my bench at the moment... I'm going to try taking that huge BLDC and fit it with an encoder. Then drive it like a servo with a Copley XTL-230-40 20A drive. The Copley has auto phasing, and auto hall alignment to boot. I hope & suspect that the old 'dog' BLDC will perform rather well and it's got a 6kRPM mechanical rating. We shall see. The Copley fell in to my lap, so to speak.

Can't wait to see how that works out. Hopefully the BLDC is up to the task here. Gave the Copley a quick look over. Nice drive. It can do BLDC motors?

I will report back on the BLDC to servoectomy experiment shortly.

Can't wait to hear!