Design Discussion: High Performance CNC Spindle

[spumco]

What controller are you running?

Software: UCCNC
Controller: UC400ETH
BOB: PMDX 126 w/107 0-10v spindle board

Gave the Copley a quick look over. Nice drive. It can do BLDC motors?

EDIT - don't go searching Ebay and snapping up all the drives I just discovered!
The XTL (and some others in the Xenus line) can do halls, differential encoder w/ halls, and diff encoder w/o halls.
The XTL-/S model can do analog encoder or analog halls /digital encoder
the XTL-/R model can do resolvers
Control signals: analog 0-10v, pwm, step/dir, external encoder for gearing/camming, CANopen, ASCII

Also has a secondary input for a load encoder (dual-loop)

buy a finished Tormach spindle cartridge rated for 10K.

Or a 12k version. Tormach is chinese after all..
https://www.aliexpress.com/item/323...chweb0_0,searchweb201602_4,searchweb201603_55

get my hands on a RF45 sized mill that I've been stalking.

Um...
https://cleveland.craigslist.org/tls/d/euclid-cnc-vertical-mill/7045473138.html

I'm blanking on @spumco 's machine.

Mikini 1610L. About halfway between a 770 & 1100 Tormach in size. Linear rails & steppers (for now).

Mounting the encoder to the spindle is also tricky

Offset the encoder and drive it with a timing belt right off the spindle main pulley. Drive it at some sane ratio (1:1, 2:1, etc.) and adjust the PPR in the controller. This method has been used on plenty of lathes (Omniturn & other 'factory' retrofits).

Stolen off the 'net:


Determine your desired resolution needed for a spindle/C-axis (can't be that small, right?) and get one with a PPR that doesn't blow the optocouplers away on the BOB or drive.

A 1024 line encoder is 4096cpr in quad. So lets say a 5kRPM motor at 1:2 (for the magical 10k spindle) and a 2:1 reduction to the encoder.

4096cpr / 2 = 2048cpr at 10k spindle rpm.
2048 * 5000 /60 = 170khz. This is do-able most low-end BOBs which seem to all be using the same 200kHz optos.

And 2048cpr gives you 0.176 degrees per encoder count resolution. Not great for a 4th axis, but would be fine for a spindle to position a broach or line up for an ATC.

This also means you don't need to use a timing belt from the spindle motor to the spindle - any poly-V belt slip will be taken care of by the direct connection from spindle to encoder. Quieter belt drive at the expense of higher belt tension and top/motor bearing radial loads. You can get a cheapo USDigital, CUI, or import with whatever output style you want and not spend a fortune on a SICK or Heidenwhatever.

Most hollow-bore encoders I've seen top out at about 32-50mm bore - they aren't really suited to installation over a spindle.

-Ralph

 

[spumco]

Interruption for Public Service Announcement

I (re)discovered the forum thread describing the two-speed pneumatic belt selector construction. It's from a German hobby CNC forum - must use Google Translate. Great thread, but the belt thingie starts about page 20.

https://forum.zerspanungsbude.net/viewtopic.php?f=50&t=26643&start=190

Even some of the little details are interesting - check out the shaft locks he's using to mount the pulleys. I'd not seen those before and they look extremely useful.

Might be helpful - something like this opens up many options for spindle motors. As previously mentioned, one can either spend huge money on a high-speed (10Krpm) motor, or run a 1:2 (or whatever) ratio and live with reduced torque at low RPM. Meaning you'd need to over-size the motor 2x to make up for lost torque.

Sure you can fit a 6"-8" frame motor that takes 30A @320vdc to run, but what amplifier will drive it off 240-1P? The absolute max amplifier I can find is 20A/40A continuous/peak off single-phase - and that's with big heatsinks and fans.

With the above belt selector you can go with a smaller, lighter motor that takes fewer amps and still have plenty of low-end torque. Less money, less weight for the head to move.

EDIT: found his Instagram site:

View this content on Instagram

View this content on Instagram

End PSA

Still waiting for some stupid Wago connectors to test out the Copley drive. Hoping my 'Lego' servo motor/drive plan works out:

1. Any motor with the desired speed & amp rating
2. Ditch the factory feedback for an AMT or US Digital commutation encoder
3. Drive it with a 'generic' amplifier.

Should open up many, many possibilities on the used servo market. Fingers crossed.

 

[macardoso]

EDIT - don't go searching Ebay and snapping up all the drives I just discovered!

I would never I have a sizable supply of AB Ultra 3000's hanging around for my needs. I'm always worried when I tell people about my secret suppliers that they'll go grab it all!

Also has a secondary input for a load encoder (dual-loop)

You should definitely use that.

Or a 12k version. Tormach is chinese after all..

OK, wow. I honestly hadn't thought of that, but this seems to make a lot of sense. a BT30 spindle built from scratch would cost roughly:

• $200 in AC and radial bearings
• $150 in stock
• $100 for pulleys
• $100 for a drawbar and belvilles
• Some amount for a gripper for the pullstud

So maybe $600 and then you need to actually machine it to tolerance. I mean if that's what you're looking for then perfect, but I'm more open to an out of the box option. The 90mm diameter is pretty big and the $1000 takes a big bite out of the $1-2000 budget... but it is a fully complete spindle cartridge.

Do any of you have experience buying from AliExpress or Alibaba? I have always gone for the Chinese sellers on eBay, but I'd love to have another option. Are you protected as a buyer?

Um...

Nice find. I'm still holding out for HGR to dump a mill in my lap real cheap. It's actually quite a bit bigger than an RF45. I keep lowballing them, and I'm pretty sure people are going to snatch up the last couple, but I can't help get my hopes up. When I'm done crying over those when they're all gone, maybe I'll go look at that one.

Mikini 1610L. About halfway between a 770 & 1100 Tormach in size. Linear rails & steppers (for now).

Cool! Have you been happy with it? Guessing you did some serious modifications.

I (re)discovered the forum thread describing the two-speed pneumatic belt selector construction. It's from a German hobby CNC forum - must use Google Translate. Great thread, but the belt thingie starts about page 20.

God that awesome. Will be reading it today.

Sure you can fit a 6"-8" frame motor that takes 30A @320vdc to run, but what amplifier will drive it off 240-1P? The absolute max amplifier I can find is 20A/40A continuous/peak off single-phase - and that's with big heatsinks and fans.

You can oversize most 3 phase drives and derate them 0.578 for single phase operation. Hypothetically you could run a 100HP motor on single phase, but you'd need a 175HP 3 phase drive. It gets expensive doing that, and you have to check with the MFG to see if they officially support it.


A big downside of the franken-motor idea is that you lose special information about the motor that helps the drives run them really well. For example, on my AB drives, it expects you to be able to enter the flux saturation ratio of the windings at various current output levels and you also need to know the thermal resistance and capacitance. The drive uses these values to generate field weakening and correctly model the thermal loading of the motor.

I've done exactly what you're describing, but I've never had a company willing to share that information with me about their motors.

 

[spumco]

Cool! Have you been happy with it? Guessing you did some serious modifications.

Happy? Yes, but it's been more of a learning experiment than a useful part-producing tool. It met all the requirements I had at the time - cheap, single-phase, can still (barely) move it without paying riggers, and no gibs/ways to deal with. No mechanical conversion issues. Came from the factory with ballscrews & HiWin rails.

Modifications? Mechanically, sort-of. Electrically, yes. It currently has the original Nema34 steppers and drives, but everything else electrical has been replaced. I made a kludged-up PDB because the induction motor I have is too big to fit a normal PDB multi-stage cylinder, and I built the 4th axis discussed before. But no serious mechanical mods.

And those will be gone soon, too. As I mentioned I have a BT30 spindle, DMM 750w servos, PDB, and a new motion controller & BOB to throw in. I'm setting it up to eventually get an ATC. Sadly, the head has to come off as my R8 spindle is 80mm and the BT30 is 90mm. Boo.

You should definitely use that.

No point on a spindle motor, right? But for the next build (lathe?) I'll probably do it.

Do any of you have experience buying from AliExpress or Alibaba?

Not directly. I bought it from an importer. Its all about risk-tolerance.

you lose special information about the motor that helps the drives run them really well.

I suspect you're right. However, my current project as proof-of-concept is to take a really, really obscure motor with very little available data and see if the Copley will make it sing.

When I say little data - I know the pole count and the advertised power & continuous amps + DC bus voltage. That's about it. Mikini was painfully secretive about technical specifications: no motor plate, all electronics painted black, part numbers ground off the chips, etc. All I have to go on are some bits of info they let slip years ago in the 'Zone forum. They even had the nerve to call thier BLDC spindle motor a "servo" - I guess technically hall feedback counts, right? /snort/

Anyway, the Copley manuals and videos indicate their drive & software can figure out all the variables needed to make it work. They even have a 'nameplate' tuning calculation mode where you can bypass all the fussy values and just enter rated speed, amps, torque and power and it'll do the magic. Watching them do coarse positioning with halls only, and no-halls commutation (wake-and-wiggle) is pretty cool.

If the drive can figure this particular BLDC motor out, it should have no problem with a brand-name Baldor, Parker, Siemens, ABB, whatever where I have all the motor data needed. If I can get 80-90% of the performance of a matched-pair from an OEM ($$$) from the Franken motor approach, I'll be happy.

Or some smoke will come out.

And I was wrong earlier about the Tormach 770 - it is an AC induction motor, not a BLDC. I was confusing it with some of the other smaller mills - memory failing.

Hope I'm not hijacking the spindle thread too much.

I figured any spindle discussion needs to start (or be closely followed by) with a motor discussion.

Since, as you said earlier, the spindle is the heart of a machine tool - how do you drive the spindle? I'm perfectly happy with induction motors but you can only get so much power on top of a hobby-class/size machine before the motor weight becomes an issue. On a lathe, who cares? The spindle motor just sits there - go get the biggest motor & VFD your power company will let you have.

On a mill, however, a 3HP sheet-metal motor seems about the limit. I ran across what's termed a 'feather picker' motor - it's a Nema 56C framed 3P induction motor that's a bit longer than the typical 2HP version. That coupled with a VFD that does homing (Hitachi) and a spindle encoder could be a decent combination. Problem is that it weighs 60lbs, and a new motor & VFD & good cables will get you close enough to $800-$1000 to start thinking about a PM motor.

That's why I've been going down the AC servo/drive rabbit hole for the past year trying to find something cheap and less weight/size than an induction motor.

Another option is to use a built-in ATC spindle (S30?) for a router. Expensive, but so much easier to install or integrate than a separate spindle/motor/gears or belts. I just wish the folks making router spindles would get the word that a low-speed spindle - say 8/10kRPM - would be a winner. Some of us want to make holes in steel, and I can't interpolate a 2" deep 3/8" hole with an end mill. Not happening, and running that 3/8" drill at 6kRPM is a good way to experiment with friction-stir welding.

-Ralph

 

[macardoso]

Sadly, the head has to come off as my R8 spindle is 80mm and the BT30 is 90mm. Boo.

So you did buy that spindle cartridge? I'm super interested to hear what you think of it!

No point on a spindle motor, right?

No, not really. But it would help a lot on an axis motor with linear scale feedback.

Not directly. I bought it from an importer. Its all about risk-tolerance.

I'd be really interested in hearing more about that process if you don't mind. You can PM me if you don't mind sharing.

I suspect you're right. However, my current project as proof-of-concept is to take a really, really obscure motor with very little available data and see if the Copley will make it sing.

I think it will work just great! Wasn't trying to suggest that your setup won't work, but rather to warn others that trying to mix and match servos can be really difficult unless you know what you're doing.

Hope I'm not hijacking the spindle thread too much.

Not at all. Very interested to hear about all of this. Didn't really have a direction in mind for this thread, more of an open discussion board for all things spindle related.

That's why I've been going down the AC servo/drive rabbit hole for the past year trying to find something cheap and less weight/size than an induction motor.

It is tough. There are only a few options out there that are really good. Still not ideal though. AC servos have a huge benefit for size and weight. My 2.25HP 5000rpm servo is only 85mm square and weighs 10lbs. That size induction motor would weigh 60lbs or more.

Another option is to use a built-in ATC spindle (S30?) for a router. Expensive, but so much easier to install or integrate than a separate spindle/motor/gears or belts. I just wish the folks making router spindles would get the word that a low-speed spindle - say 8/10kRPM - would be a winner. Some of us want to make holes in steel, and I can't interpolate a 2" deep 3/8" hole with an end mill. Not happening, and running that 3/8" drill at 6kRPM is a good way to experiment with friction-stir welding.

God I wish someone would have a metal cutting spindle offering. I always cringe when I see the router guys trying to get through steel at 18krpm on one of those little water cooled spindles.

 

[jbolt]

Great discussion! I went down this road a few years back for my PM932 conversion. In the end I decided it was not worth the blood & treasure for a hobby mill. 6K is fast enough for what I do which is mostly one-off parts. I look forward to seeing what comes of this.

There are metal cutting spindles out there but not something for the hobby market. When we built the CNC router for cutting aluminum plate up to 1/2 thick for the high school robotics team I spoke with manufacturers all over the world to find a spindle that met our requirements and budget. A lot of interesting things out there you won't find on a website.

I always cringe when I see the router guys trying to get through steel at 18krpm on one of those little water cooled spindles.

Hey! It wouldn't fit on the mill!!!

 

[spumco]

So you did buy that spindle cartridge?

Yes, I bought it. It's not very interesting, but here's the deal:

The Mikini is/was the predecessor to the SkyfireCNC VM2 (big one). Same designer. Mikini was imported by some guy in California, and he specified all the electronics (I think). So there is some commonality of mechanical parts (ish) between the Mikini and the Skyfire.

I saw that Skyfire was selling their BT30 spindles, so I got hold of their US importer buy that didn't work out. So I tried their Canadian distributor and he got the spindle and PDB imported for me. I got spanked on duties (twice) but the spindle is nice. I haven't installed it...

Because it turns out Skyfire modified the original head castings in a few important ways. I knew before I bought it that the new spindle was 90mm (vs 80mm), but the head is also now taller by about 3/4" inch and has a few other changes. So fitting this thing is going to be a chore. I have to make a top adapter plate as the new spindle is also taller. The Skyfire head is open on top - mine is not - so I've got to cut the top open and make sure the adapter plate is also a reinforcing/structural item.

Sigh. That's why I'm hunting for the 'perfect' motor. I'm not going to do such massive surgery (including pulling the column to fit a new Z-servo) and not put exactly the spindle motor I want on it. And adding the two-speed belt drive selector discussed earlier would be a royal pain - half the spindle pulley (HTD-5m) is down inside the head.

So I'm on the hunt for a massively oversized servo that I can drive at 1:2 and get 8-10kRPM and still drill or face-mill steel. I've got 20A continuous to play with and that gets me in the 3-4kw neighborhood.

But it all depends on whenther this franken-motor experiment works. If it does, I'll invest in a big servo and have a go.

 

[spumco]

Franken motor experiment successful!

1. Ripped off the old Hall sensor and installed an AMT-312D (differential A/B/Z + Halls) on the motor.
2. Used the CUI/AMT phasing software and it auto-phased the Halls to the windings. Sweet.
3. Soldered up an encoder cable from the AMT to the Copley drive.
4. Plugged in the drive & went through the Copley auto Hall phasing, motor tuning, and so forth.

The old BLDC is now spinning like a grown-up servo. With the OEM Hall-only BLDC drive it would hunt +/- 200RPM without a load - and cut out randomly above 3500RPM. You could stop it with your hand below 1500RPM.

Now I can run it at 50RPM and the leather glove test indicates it has gobs of torque even at standstill.

It can also ramp up from 0-6kRPM in a blink. Need a braking resistor to stop it that fast, but that's down the road.

More importantly, it can now be driven in positioning mode just like a 'real' servo. Watching it whip back and forth was very satisfying.

So... this opens some doors for used motors being retrofitted as spindle drive motors. No more 'treadmill' motors or insanely heavy induction motors.

PM servos all the way now for me - even if they show up with some odd-ball resolver or proprietary encoder signal, the encoderectomy plan will take care of that.

There are plenty of ABB, Kollmorgen, Siemens, Baldor, and other PM servo motors with max speeds well in excess of the DMM Tech 1.8kw servos I've seen some folks using. If you can find a 2 to 4kw servo with a top mechanical speed of 5k-8kRPM you no longer need to gear it to the point where there's no low end torque.

The AMT encoders are programmable, so you can set the resolution to avoid overwhelming your controller max input speed.

I'm going to shoot for about 1:1.6 and a 8k+RPM spindle top end, and trying for about 10Nm of continuous torque.

Good luck, let me know if anyone has questions about this setup. Remember - I'm not an expert. I just got lucky poking away at this thing.

-Ralph

View attachment 20200125_125710 (2).mov
View attachment 20200125_124855.mov

 

[shooter123456]

I have been trying to decide if BT30 would be affordable/practical with the price of tool holders.

Comparing cost, I quickly noticed that there are a million BT30 options out there which makes it a little bit difficult to compare apples to apples. With TTS, which BT30 holder is comparable in performance? It is also tough to compare because there are few details about the TTS holders. What are they made of? They only say alloy steel, which could mean a wide variety of materials. Where are they made? Mine don't say made in the USA so it is probably a safe bet they are made overseas. So a direct comparison is difficult, to say the least.

Some of their prices at the moment are:
Set screw holders - $22
ER20 holders - $45
Boring arbor - $49
Face mill arbor - $80

Their set screw holder prices dropped recently. I recall them being about $35 when I checked last and I am not sure how long they will stay at that price. Might be worth grabbing a few before they go back up.

For BT30 holders, the hardest part is just determining which holder would have roughly equal performance with all other factors not considered (ie BT30 spindle being better than R8)
On Amazon I found some of the following prices:
Tegara BT30 1/8" set screw holder - pack of 2 - balance to 20,000 RPM - $42 each
No name import BT30 ER16 holders - advertised .0001" TIR - pack of 4 - $22 each
ZXHAO ER16 holder - No details - $31 each

With various other options ranging from $20 - $200 each, there are a lot to choose from. A nice side note is that if any of them don't meet the advertised specifications, returning to Amazon is easy.

Tormach has the set screw holders for BT30 around $50 each, which is significantly more expensive than their TTS products. For the ER20 holders, they are slightly more expensive at $55 each.

That also does not take into account the pull studs needed for an ATC. Those add anywhere from $7-$20 to each tool holder.

That also ignores the fact that many people already have TTS holders. Starting from nothing, it may not be quite as bad tooling up for BT30, but having a significant investment in TTS holders already would certainly make it more difficult.

At this point I am still leaning toward BT30 because I think the performance will be better and I like the wide range of tooling available. Being able to share tooling with future machines is also a plus in my mind.

 

[macardoso]

That was my concern with tooling for BT30. It is attractive, but I have a far amount of TTS at this point. Probably $500 into it. Would suck to just walk away from that investment. Measuring tool offsets offline with BT30 is tricky too.