Pancake Motors

[JerryFlyGuy]

So I'm looking at building a 5 axis head for my machine. I've now got the software to make tool paths and so, time to get the mechanics down.

I'd been thinking about various ways to do it and am still investigating. One of the directions I've been looking is at pancake motors.

What are the negatives to using a motor of this type [prefer AC but open to other options]. I'm looking for a low speed high torque solution. Low cogging yet high accuracy.

I'm going to be needing ~ 180rpm [max] and ~100N-m torque at stall [cont] and would like to see ~15arc-sec of accuracy. Is this all realistic? Is there something about these styles of motors that preclude them from these types of positioning applications? 240VAC at reasonable amperages?

Open frame vs Framed? [my preference is actually open frame due to my configuration [Head/Head design]]

Sources of reasonably priced motors?

Anything special about Drives to operate these motors?

Thanks in advance guys/gals, for any direction or help that can be offered on any areas. I'm a newbie when it comes to these things.

JFG

[mike_Kilroy]

check out:

http://kollmorgen.com/website/com/en...ary_motors.php

3 types of house/unhoused models.

yes, your specs fit these type of motors

position accuracy is almost totally a function of the feedback device you pic, not the motor.

If you want ballpark prices on any of these motors for reference, pm me.

[vger]

Jerry,

That's a pretty tight angular spec, 15 arc sec. For an encoder operating in 4X mode that would take an encoder with at least 21,600 PPR.

I'm assuming you are wanting to use pancake motors due to their profile. I've got several small pancake motors that I was thinking of using for my 4th and 5th axis but I'll have to belt reduce to get the needed torque. They do start and stop amazingly fast.

Looking forward to your build.

Steve

[mike_Kilroy]

We use the KBM Kollmorgen (and other Kollmorgen series) with Hiedenhain encoders on rotary tables for coordinate measurement machines; Heid has 90,000 & 180,000 real sine wave lines output that we then multiply by upto 4096 more so we can get .5um resolution at 1.5m diam on a part..... Yes, 15arcsec is tiny but everyday doable. My customer ships approx. 100 of these machines a year. this is 512,000 counts per degree IIRC....

[ihavenofish]

i have been curious about pancake motors too for this application, dont see much info about them outside of wind turbine generators. most heads are worm drive that i can see. the hobby guys are using big belt reductions with many complaining they dont have the holding power.

ive just finished a rotary module design that will work as a 5 axis head. mines very small (80mmx125mmx160mm long), obviously for small machines.

what i did was put on a direct drive 2.5" frameless ac servo that has a mere 1.2NM torque continuous. ive coupled that with a very powerful mecanical disc brake setup that is released when the axis is commanded to move. this was my compromise for allowing heavy fixed axis milling, and light coordinated finishing, and getting as tiny a package as possible. its also how tarus does they big 5 axis mould machines that require heavy roughing ability. you can up the torque by using a longer motor and adding a water jacket, making up to 5nm continuous possible on the same frame diametre to get a bit more coordinated milling ability at the sacrifice of another 2" in module length.

basically, ive given up on the 100nm direct drive all purpose solution that is small enough to fit in a hobby class machine head for now.

[JerryFlyGuy]

Steve the 15 arc-sec comes from some basic numbers. If I have a tool tip radius [from the pivot center of the head] of 12" then 15 arc-sec = 0.0008 of accuracy of the tip [barely under the 1/1000"] If it isn't that do-able then it probably isn't worth doing. As Mike noted there are ways of accomplishing this.

ihavenofish I'm curious to know how your going to actuate the brake between moves? Doesn't a brake take some degree of movement [a tiny amount] to actually 'grab' and stop the motion? What about when doing continuous contouring? 1.2Nm = .88 ft-lbs. That would take some serious reduction to bring it up to a usable torque level wouldn't it? Even my spec'd 100Nm is only 73ft-lbs. For my work it's enough however [wood, foam and Alum] for anything heavier I'd think a person would need 5-700Nm to hold position.

From some of the feedback that I'm getting [cost vs performance] it seem's that maybe the benefit isn't there. The motors for this would cost in excess of ~$10K while a similar performance head using conventional motors would be ~$2.5K w/ a full blown AC servo setup.

Ah well, it was worth a kick at the cat

JFG

[ihavenofish]

ihavenofish I'm curious to know how your going to actuate the brake between moves? Doesn't a brake take some degree of movement [a tiny amount] to actually 'grab' and stop the motion? What about when doing continuous contouring? 1.2Nm = .88 ft-lbs. That would take some serious reduction to bring it up to a usable torque level wouldn't it? Even my spec'd 100Nm is only 73ft-lbs. For my work it's enough however [wood, foam and Alum] for anything heavier I'd think a person would need 5-700Nm to hold position.

my brake is a bit of a secret for now, it makes the whole thing special i guess. but basically, mach has to command it to release, then move the axis, then re-engage, which is something that im trying to make automated so you dont have to put a line in the g code. the brake is "normally" on. as in, with no power, the axis wont move. so it will work well for inexing, which being direct drive will have no backlash and the precision of the encider, which in my case is only 32768 divisions using a small renco encoder, but thats adequate here for most work this little machine will do giving .0015" resolution at the tip of a 4" tool in a tool holder. a higner resolution resolver could easily be added but it drives the price up alot.

as mentioned, my machine is small. the max recomended tool length is going to be about 4", which puts the tool tip about 8" from the pivot centre.

that means you can put about 1.3 pounds of continuous cutting force. a quick check of the trusty calculator shows that is:

6061 aluminium
0.25" ball nose
3 flute
15000rpm
0.125" depth of cut
0.01" stepover
.001 feed per tooth
234ipm
0.015 nm torque at the spindle

and that actually a fair finishing pass in aluminum. in wood, foam, tooling board etc it should provide very useful coordinated cutting.

with the brake locked, your good for anything that wont stall a 2HP spindle.

[Larken]

A closed loop Stepper is the lowest speed highest torque motor for its size. I've got one in the works, and Gecko does too.

Are you planning on direct driving the rotory axis's or using a wormgear or planetary ?

[ihavenofish]

A closed loop Stepper is the lowest speed highest torque motor for its size. I've got one in the works, and Gecko does too.

Are you planning on direct driving the rotory axis's or using a wormgear or planetary ?

steppers have more torque relative to an equal ac servo, but still nowhere near what hes looking for. they also cant provide the resolution by a longshot. even with 1/64 microstepping like a vexta system, thats only 12000 steps per revolution, and the way steppers function, they arent precision index points whether you have an encoder or not.

if you are going with a worm drive, none of this matters, and you just need to deal with backlash in the gears.

theres lots of encoder enabled "step servos" available though, they look interesting, but for price vs performance, ac servos are a better buy i think.

[Larken]

You could built a high polecount hub motor with rare earth magnets. It could either be a 2 phase or 3 phase. But it would be a lot more work than a pancake style gearhead and small motor.

theres lots of encoder enabled "step servos" available though, they look interesting

Are there ? I only found 2.

[ihavenofish]

You could built a high polecount hub motor with rare earth magnets. It could either be a 2 phase or 3 phase. But it would be a lot more work than a pancake style gearhead and small motor.

Are there ? I only found 2.

not looking hard enough

vexta makes them, dynomotion, granite devices has it in their latest drives. there should be a half dozen more out there.

its very common, but like servos, the encoder on the motor tends to be a driving price element.

xzero will have an ac servo kit soon that will be very cheap - cheaper than any of the available step servo setups - but again, still not really ideal for what the OP is looing for, which is a direct drive, high torque, high resolution rotary system.

[mike_Kilroy]

A closed loop Stepper is the lowest speed highest torque motor for its size. I've got one in the works, and Gecko does too.

Would you give an example of this with a link to a part number with specs and dimensions please?

When you say highest torque, do you distinguish between continuous stall torque or peak torque or do you mean both?

thanks.