Servo Direct Drive?

[Contract_Pilot]

Can servo's be Direct drive I read some place they need to be geared! Some say if sized right they can be direct drive.

Looking at Hybrid or Actual Servo I can go with Either I have a 7I77 and a 7I76 mesa board in the bin.

Planning on my 704 retro or a 770 locally that has a smashed electronics cab so was removed and discarded.

Similar Threads:

• HELP!Direct Drive? Stepper/Servo?
• Newbie- CNC Lathe - Servo direct drive or pulley drive?
• Automation Direct 1KW AC Servo Motor with Drive
• Direct drive a servo...is it possible
• Direct Drive Servo or pulleys

[hanermo]

Direct drive works fine, if the servo is big enough.

About 400W, 1.3 Nm continuous, and == 3.9 Nm peak, and up, should do ok on linear guide machines.
Because of the high peak torque, it should/might also be fine on guideways as long as there is no binding.

For comparison, 3.9Nm is what you can twist with your hand, but is fairly hard to do.
Clearly harder than normal machine handles are turned.

I use much bigger servos, and I gear them down 1:2.
I do so to get more resolution and accuracy.

10.000 count servos, 750W, 220V, at 1:2, with HTD8-30 mm wide belts.
I don´t need all the accuracy, nor the top speed, but the extra resolution is desirable to me.

Step size is 0.2 um.
Step size is =/= actual machine resolution or repeatability.
Actual incremental move size is less than one micron with 4 mm screws (32 mm D).
Lathe.

The smaller servos mentioned, 400W, I use on a VMC.
At 1:3, HTD 5-15 mm.
5000 count servos, 60V.
A mill does not need anywhere the resolution that a lathe does.

Industrial machines are mostly direct drive these days ..
but they use 1 kW (older) to 2 kW (newer) servos.
About 10.000 counts (Haas, since ==2012), older Haas was 2000 count.

Industrially they are looking for "speed" in acceleration and top speed .. totally not needed for non-production machines.
And the tables are easily 1000 kg or more in mass.
My VMC table is 200 kg now, and 400 kg+ soon.

For calcs:
3 Nm at 1:3 == approx. 600 kgf push force on the table.

Example for bigger servo, 750W.

The torque is 10 Nm (peak at servo) x 2 = 20 Nm.
Push force is T = Fl / 2 x pi x efficiency
T - torque in Nm, 20 Nm
F = push force in N
l = screw lead in m, 0.004 m
efficiency 96% (ballscrew)
F = 2 x 3.14 x 0.96 / 0.004
= 15072 N = 1500 kgf.

A 32 mm ballscrew is rated at about 1400-1500 kgf push force.
This is std on industrial machines, medium size VMCs, and is what I used for myself.

[SCzEngrgGroup]

In many/most cases, using direct-drive with a servo will mean using a motor that is larger than it needs to be to get the necessary torque. Servos typically have considerable RPM capability (4000 RPM is not unusual for DC servos). With direct drive, this would result in ridiculously high rapid speeds (800 IPM with typical 5-pitch screws) which is completely unusable in a small machine. Using belt reducers (which are typically around 3:1 or 4:1 ratio) allows the use of smaller, cheaper, lower torque motors (because the reducers multiply torque, while reducing speed), and brings the peak speed down to something more reasonable. So, instead of an 800 oz-in direct drive servo capable of more speed than can be used, you can instead use a 200 oz-in servo with 4:1 reducer, and get the same usable performance for a lot less money.

Regards,
Ray L.

[arizonavideo]

Sure you can.



I have been running the Electrocraft for 4 so years direct drive with a Gecko drive and a 5mm pitch double nut ball screw. Its a 90V servo running at 75V.

I can stall it but not with normal cutting. I wanted it to stall before breaking the bit or breaking off the end mounts and that is how it works.

I do say that now I'm leaning toward 900oz steppers for the X and Y just because they work just as well and don't have brushes. I don't get 300 IPM but 125 IPM is just fine for there type of mills.

Don't forget the Clear path servos from Teknic. I grabbed a CMP-SDSK-3421S-RLN to try on the z of the new PM-940M mill. If it is little weak for that it will work great for the X.

[CNCMAN172]

Servos can be used to directly drive a ball screw or some other method of going from rotary to linear motion. The reason you often see gearheads on servos is the fact servos run at higher speeds than stepper motors. Most servo motors can run up to 3000 RPM. At low speeds the servo motor has less torque than a stepper motor, but at higher speeds the servo motor has much more torque where at high speeds stepper motors have much less torque. The idea is to get the servo motor geared to the middle of its speed range so you get the best of both worlds.

Russ

[Al_The_Man]

A little misleading, for a given frame size steppers are usually have more torque, but the torque curve for a servo and stepper of the same torque, the servo is superior, this is due to the relatively flat continuous torque curve that is maximum at zero rpm right up to the max. rated rpm.
Whereas the stepper torque curve drops off rapidly as rpm increase, there are steps to overcome this in the drive, by maintaining the rated current, but still lower top end rpm torque compared to a servo.
As previously stated, the higher max rpm of a servo allows a more economical frame and drive size to be applied when some kind of gearing is used.
Al.

[quickjam]

I have 1kW 2500rpm 2500ppr AC servos (lcmt-10m02-80m04025). I'm trying to decide on the rest. I'm looking at 20mm 5m pitch ball screws. That would be max 492ipm. That would be about 500rpm to run 100ipm. 400rpm to run 80ipm. The machine gantry would weight out about 250lbs with dual y drives supported with sbr25 rails. That should be way enough power to travel. The force will be a spindle motor not yet selected for cutting hardwood/alumnium at times. Process time isn't important to me. My question is can i run direct drive to the ball screw? Or would I be disappointed and wasting my time?
My other option is using 10mm pitch ball screws with a 5:1 gearbox. That would be max 196ipm. @ a more than double the cost.

[CNCMAN172]

quickjam,

Those are very large servos. You can connect them directly to the ball screw with a coupler. The picture below is a 750W Panasonic motor attached to one of my ballscrews. The second picture is a 200W motor connected to a harmonic drive with a 11:1 reduction on one of my machines. What helps determine which way to go is the torque and speed requirements. A stepper motor has more torque at slower speeds, but a servo motor has more torque and more at higher speeds. Most servos range between 2500-3000 rpm. In a direct connection depending on the pitch of the ballscrew you can sometimes limit the speed of the motor to 200-300 RPM. For optimal performance again based on your ballscrew pitch you need need a gearhead like you said. In my case I started direct connected and still have that arrangement on my smaller 24"x24" machine and it works perfect. On my larger 60"x60" machine the ballscrews I found on ebay were designed for rapid movement so to get in the mid-range of the servo I need to add a gear head. If you do with a gear head you need to find one with no backlash. Those are more expensive and that could determine you approach on price alone.

Russ

[SCzEngrgGroup]

I have 1kW 2500rpm 2500ppr AC servos (lcmt-10m02-80m04025). I'm trying to decide on the rest. I'm looking at 20mm 5m pitch ball screws. That would be max 492ipm. That would be about 500rpm to run 100ipm. 400rpm to run 80ipm. The machine gantry would weight out about 250lbs with dual y drives supported with sbr25 rails. That should be way enough power to travel. The force will be a spindle motor not yet selected for cutting hardwood/alumnium at times. Process time isn't important to me. My question is can i run direct drive to the ball screw? Or would I be disappointed and wasting my time?
My other option is using 10mm pitch ball screws with a 5:1 gearbox. That would be max 196ipm. @ a more than double the cost.

At roughly 1.3HP, those motors are MASSIVELY over-sized for ANY benchtop machine - probably by a factor of 2 or more. One problem you're going to have is, if you ever crash it, you WILL do serious damage. Using reducers will do little more than increase the amount of damage.

Building a GOOD CNC machine means actually doing the engineering calculations to DESIGN an appropriate drive system matched to the machine hardware and performance requirements. Simply buying parts based on guesswork, and bolting them together, particularly grossly over-size parts, will not yield a good result...

Regards,
Ray L.

[quickjam]

Oh this will be a rare build for sure. But no table top will hold. 3 tons and counting. I'm sure to find a weakest link somewhere. And that I will learn from. TX

[SCzEngrgGroup]

Oh this will be a rare build for sure. But no table top will hold. 3 tons and counting. I'm sure to find a weakest link somewhere. And that I will learn from. TX

OK, so you're building a 3 TON machine, and asking for design advise on a forum where a G0704 is considered a large machine? What could possibly go wrong....

Regards,
Ray L.

[quickjam]

Tx CNCMAN.
I will try direct with 2505 screws and see how it goes from there.

[CNCMAN172]

quickjam,

Make sure you get quality ballscrews. I have seen people on the zone that got some from some of the Chinese suppliers and with very little pressure the spring seals on the ends released allowing balls to escape. Depending on what you are cutting the actual force on the ballscrew varies, but I think you will be fine. I also recommend you go with a controller that can accommodate error signals from the Servo amplifiers. Why? Well like others pointed out you are using large servos which are powerful, so you want to make sure you have limit switches in place, and have configured your estop circuit to avoid any possible damage to the mechanical structure of your machine. Most servo amps have a signal that comes out that says "Servo Ready" and another one that says "Servo Error", if you take advantage of these you can halt things if something starts to go wrong. I use an ESS/MB2 combo to control one of my machines. The smoothstepper is common on many machines but it is clearly not the only solution. The MB2 is just a breakout board designed for the ESS, which provides for differential inputs, relays, spindle control, and some safety circuits. I just finished a build with this and for the first time took advantage of the safety circuit by feeding the Amplifier Error output into the MB2 safety circuit. The machine will not enable if there is any kind of error. So as soon as the servo amp sees a current overload it shuts down everything instantly. So when I ran my machine accidentally into a fixture on my table the high current on that servo fired and shut everything down. Anyway something to consider.

Russ

[hanermo]

Quickman ... the direct drive is the best option You have.

The good part of direct drive is easy, fast, simple, relatively cheap.
At 2500 ppr, You are already much better than needed, for processing wood/alu on a router.
You have stuff good enough for milling, about 5x more than a router.

The other good part You have, is You already have servos big enough for this (and more than big enough).
Servos 2x-5x-10x too big don´t hurt anything.
You adjust it in the sw parameters to whatever You want.

For You, even 400W servos would work.

Examples:
I am using commercial, new, 750W ac brushless servos on my lathe.
At 10.000 counts/turn.
With 1:2 belt drive.
Yours are more powerful, if less accurate, at 2500 vs 10.000.

Why?
Because I am looking for max resolution, ie the smallest possible increment I can get, down to submicron resolutions.
I may switch to direct drive, and a custom ground screw on the x, for the resolution.

The very strong belt drive is not working well for me, for what I want (special unusual case).
I suspect irregularity in belts and pulleys.
This would not be your case, and would not affect anyone with a router, your case.

A 1.3 kW servo will be about 3.5-6Nm cont, 10-16 Nm peak. +/-
At 1:1 => 10 Nm => 1800 kgf push force.

Data and example.
With 1:2 belt drive I have 20 Nm torque at screw, from 10 Nm peak x 1:2 = 20 Nm.

My saddle assy on the 12x24 lathe is 230 kg mass, max.
4 tool toolchanger (20 kg), 7 tool revolver toolchanger (70 kg), tooling plates, everything.

At 20 Nm, I have about 3600 kgf push force.
This is about 3600 / 200 = 18 G acceleration.
This is about 5-30 times too much.

18 G x 10 (m/sec is 1 G) = 180 m / sec acceleration.
x 1000 mm (in m) / 1000 ms in meter = 180 mms/sec2.
It means it can accelerate to 180 mm / sec in 1 ms.
0.001 sec.
Travel from midpoint is about 120 mm.
So it will hit the end, x axis, in about or less, than 0.002 secs. 2 ms or less end-end.

At 3000 rpm servo / 1:2 drive = 1500 rpm at screw.
I want about 1000 rpm max at screw, and mostly 700 rpm or less.
Faster than 1000 rpm kills screws fast.

I set the acceleration rates low in the sw controller, and the max speed to 1/3 in the controller, and it all works perfectly.
My step size is 0.2 microns, and I am GUARANTEED very low following error due to the massive push force,

For now, actual following error == 200 steps max based on the led readout at the servo controller, with crap, soft low factory basic values, in PID parameter tuning.
I could drop this to 50 at any time.
Or to 20.
Probably to 10 or less.
I could probably get even lower servo error values, with maybe jitter at screw and mechanical system jitter errors.
Belt drives are great for eliminating these, mostly.

Still tuning/getting set up, and I want soft parameters, for now.
200 steps x 0.2 microns = 0.04 mm max error. Ever.

Positioning is always 0-1 step, 0.2 microns theoretical max error, + mechanical error at screw and mounts etc.
Ie total motion control train error.

So my fast moves, max power cuts, have less than 0.04 mm error at tooltip, and final cuts are always way under 1 micron error at servo.

Servos are:
Faster than hobby/diy guys need.
Accelerate faster than we need.
Extremely accurate, and this is sometimes/often needed.

The right choice, if funds allow, is 3-10x more resolution than best-desired accuracy.
Your step size should be 10x - 3x better than your desired resolution.

Direct-drive big servos of very high accuracy are the best possible choice.
Yours are exactly in the sweet spot and will give You fantastic results and huge reliability.

You really really should wire in the error-signal from the servos, for crashes.
Really recommended.
This stops the motion at == 100 counts error == 0.02 secs in a crash.
If You don´t, the damage will be catastrophic everywhere in a crash.

A 1,3 kW servo x 1:1 => 800 kgf x mass impact.
Imagine lowering your car on your x axis components at feed rate.
Twisted columns, ways, bent/damaged gibs, etc.

Broken spindle (bearings), spindle mount, ballscrew axis mount(s) (load), ballscrew (brinelling), and most probably major ancillaries until something stops.
This takes about 0.1 - 1 mm distance after things catch hard, I think, on a router.
The system bends about 1 mm, and then it blows up.

Anecdote/example:
My lathe is perhaps 10x stronger or more rigid vs. a router, and did bend ,aybe about 0.5-1 mm on the total system/QCTP, on overloads, before faulting.
From what I have seen.
I faulted it twice so far, so about 1800 kgf force (more/less within peak force for dynamic=rotating ratings for 32 mm screw).

On work moves, ie not crashes, just too-hard work.
Deep drilling stainless, broaching.

[quickjam]

Hey CNCMAN,
I've finally have gotten to the point you have mentioned.
I already have a ddmmv2.1 board installed. That ESS/MB2 looks nicer but I"ve went with this other.
Your point is clear on the logics with an error signal. I'm planning that now. Its seem to more complicated than I was thinking.
i do have an Estop loop directly to the BOB set on MACH ESTOP input. But i do not have hardwired to cut the line voltage to the drivers/power supplies.

Servo Drive:
SERVO READY(Enable)- Input
SERVO RESET- Input
SERVO ERROR- Output

I have planned to use a pushbutton for each drive RESET(Which now typing "Hey I can just tie them together and use one")
I have six SERVO READY(Enable)drivers and 6 enables in MACH3 so i plan to terminate each to the BOB inputs.
As for my SERVO ERROR output. I only see inputs 1-4 in MACH3. With this output I can change default into active or non active during a fault. I'm going to try to fit a relay in to cut the enable signals.through normally closed contacts and ERROR setting to output when happens to cut the enable signals.

Does that sound about right to you?

[CNCMAN172]

Quick jam
Yeah that will work. I like to monitor the error outputs from the servo drive, because in the past I would have one servo fault and the machine would keep running the gcode and screw up the work item. I would not notice a given axis stopped moving until it was too late. I have three machine and two just use the enable signal and to not cut AC power to the servo drive. Never had an issue using this scheme on home built machines

Russ

[Uguessedit]

There is always that guy who says they’re too big, lol.

[quickjam]

Direct drive worked good. With 2500 ppr I have plenty on resolution. My motors don’t even warm up. With my setup 2500rpm motor they only run about 700rpm at rapid. 200 ipm rapid travel which has been fast enough for my work.