Open vs Closed Loop (steppers vs servos)

[EverydayDiesel]

Hello,

I am curious the advantages of using a (closed loop/servos) over what seems to be the most common (open loop/steppers)

All of the high end machines use closed loop/servo setups and I am curious why most people use steppers on this forum. The cost is not THAT much different (relative to the cost of a high end machine)

So why does everyone use open loop/steppers?

[acourtjester]

I Have not used servos, the difference is that a servo has feedback to be sure it moved. There may be a small amount of wiring and circuitery added with servo setups.

[SegoMan DeSigns]

Clearpath Nema 34 SDSK's here 375 lb gantry with a traverse speed of 1200 IPM I can traverse 15' in 10 seconds..

[adbuch]

PlasmaCAM uses closed loop servos with optically encoded feedback for x and y, stepper motor for Z. Here is some information expanding on this concept.
David

Servo vs. Stepper - Jim Colt.jpg

Why does PlasmaCAM use servo motors instead of stepper motors?

It's true that this machine could be made a lot cheaper if we used stepper motors and drives instead of high-tech servo motor systems. But when it comes to something this critical, quality cannot be compromised. Experienced machine tool users know that servo motors are vastly superior to stepper motors, because newer, high-performance machines (like CNC mills) use servo motors, whereas older, more troublesome machines used stepper motors. Yet the real reason for the performance difference requires some explanation.

PlasmaCAM's servo motors use optically-encoded feedback, so the controlling software always knows the true position of the machine. A simple stepper motor controller must "trust" that the motor has moved exactly as requested each time a step current is made. Without feedback, the controller cannot identify and correct whenever the motor misses a few steps (like during jolts, vibrations, hang-ups, etc). Since a single shape may require millions of steps to trace, errors in position will continue to accumulate unbeknownst to the controller, until the machine is finally re-zeroed against a physical stop. Hence, errors in stepper position are both unpredictable and unreported.

To solve this problem, optical encoders can be coupled to stepper motors so the controller can determine the true position and make corrections when needed. However, the added cost and complexity prevents the system from being a low-budget alternative to servo systems. The more common solution involves overrating the motors (using larger, higher-geared motors that are under utilized) to reduce the probability of slipping. This extra rotating inertia (typically much greater than the mass of the moving parts) brings undesirable side effects: diminished acceleration and speed. For plasma cutting, the machine's ability to cut intricate shapes (which require abrupt changes in direction during high-speed cutting) would be greatly limited.

Powerful, lightweight motors are a must, because their torque is needed for speeding up and slowing down all the moving mass - including the spinning motor armatures. (To picture the importance of this, imagine pushing a 100 lb cart at a walking pace and trying to make a sharp 90 degree turn without overshooting at all.) That's why the Model DHC2™ fully utilizes the 300 oz-in capability of its servo motors, providing over 1.5 G of acceleration. (An object traveling at 100 inches per minute can totally reverse direction in only 0.002 inch at this acceleration.) See the photos of intricate sample parts throughout this web site and watch the demo video to see what this machine can do for you.

The Model DHC2 does provide some cost savings by utilizing a stepper motor for the Z axis only. This motor is slower and weaker than the X/Y axes motors. However, it's not nearly as critical because the motor only has to control the height of the torch. Accuracy is not affected, less speed is required, and the position is automatically reset by the controller between each cut.

[Rodw]

Well somebody has to put an argument forward for steppers. Having spent a few weeks working with an engineer developing a stepper motor selection model, it is possible to achieve similar acceleration results with a carefully designed stepper system. Developing our model was also hampered by mathematical errors in so called authoritative white papers. We concluded its a lot harder to design for steppers than to design a servo based system. And thats why in our view many of the motor packages available on the market are not well designed.

But for a DIY kit, its much easier for the end user to install stepper motors and not have to worry about the tuning that comes with servo motors.

Because there are no cutting forces involved with plasma, a well designed system is highly predictable so if the motors and drivers are matched to the task, its not possible to ever loose steps. Plus good design can utilise the massive low down torque steppers have to provide neck snapping acceleration right where its required and thats low down overcoming inertia to get things moving.

We are not fans of the closed loop stepper systems becasue we know that the only way a stepper will loose steps is if our design parameters are exceeded (by a long shot). And if that happens, the poor old motor has no available torque left to catch up the missing steps.

Instead, a smarter way to deal with this situation is to use a more sophisticated stepper driver that has a boost feature. When boost is enabled, the driver will increase the motor current and in turn the motor responds by providing more torque.

So how might boost work? Well in one of our designs, we have selected a 5 amp stepper motor that our model says only needs 3.1 amps to achieve our required performance. But the manufacturer says this motor can handle 8 amps peak. So lets use a stepper driver that can deliver 8 amps but set it at say 4 or 5 amps. So lets configure the driver to boost to 8 amps if the motor is struggling a bit. By reference to the torque curve for the motor, we know that this boost will increase torque by up to 20% and once the motor gets to its cruising speed/RPM (we picked 400 in/min), the additional amps makes no further difference to performance so we might as well turn the boost off and let the motor run cooler.

So how do we enable boost? Well one way might be to add an encoder and monitor the following error and if it gets too high (before it misses a step), we can feed it that 20% more torque. And the marketing department can talk warm and fuzzy's about closed loop systems. But encoders are expensive so if we can we'll avoid that approach and rely on the fact we are dealing with a predictable system. So we can use other parameters to boost the motor when the motion profile tells us it needs it.

Of course there is no point having a model without physical confirmation and thats why I have a bunch of hardware to retrofit my table as proof of concept.

[robertspark]

not convinced by boost or trying to catch up on following errors / lost steps.... if its lost its lost.... start again.... lower your parameters....

Improve the acceleration generation to be s-curve or a modified trapezoidal profile and you can accelerate faster....

I'll sit here with the beer and bar snacks and see how this thread unfolds...

....An object traveling at 100 inches per minute can totally reverse direction in only 0.002 inch at this acceleration.....

the math is wrong..... (physics!)
See Example number 1 https://www.khanacademy.org/science/phy ... celeration

Velocity Squared (in/sec) / Acceleration (in/s/s) = radius (inch)

0.0048 inch.... (but I'm splitting hairs really as what is 2.8 thou really...)

you would need a little over 3.6g to decelerate from 100ipm in 0.002 inch and reverse direction.

there is always one smart$%$£

[robertspark]

rod,
with regards to those authoritative white papers and knowledge in general

One good book that explains what the author calls DDAA and its endless cycle of learning and development
Discovery, Debate, Acceptance and Arrogance

https://books.google.co.uk/books?id=03_ ... ppy&f=fals, I oue

you begin to realise that what you, I and everyone actually knows is very little in the whole scheme of things.

(I listen to the audio book whilst commuting to work / driving to meetings and inspecting work progress [when we still did those...])

[adbuch]

....An object traveling at 100 inches per minute can totally reverse direction in only 0.002 inch at this acceleration.....

the math is wrong

Take it up with Plasmacam.
David

[robertspark]

....An object traveling at 100 inches per minute can totally reverse direction in only 0.002 inch at this acceleration.....

the math is wrong

Take it up with Plasmacam.
David

They'll be a charge for it.... it'll be part of their upgrade package...

[Rodw]

Robert, for once we agree. 0.0048 inch. I'm actually a bit doubtful that a small servo can accelerate at 1.5 g. I'd love to see a motion profile plot time v's velocity to confirm. I am aware of some machines that can do 0.7 g but they use 3.5 kW servos!
I also know machines with steppers which can do 50 m/min (about 1970 in/min) with 2 nema 23's and quality drives and motors.

[robertspark]

Robert, for once we agree. 0.0048 inch. I'm actually a bit doubtful that a small servo can accelerate at 1.5 g. I'd love to see a motion profile plot time v's velocity to confirm. I am aware of some machines that can do 0.7 g but they use 3.5 kW servos!
I also know machines with steppers which can do 50 m/min (about 1970 in/min) with 2 nema 23's and quality drives and motors.

Not to blow my own trumpet here (but who else will...)

I am running these 4.2A stepper motors on 20mm wide HTD 5mm belts with 20Teeth pulleys and the machine can do over 1g of acceleration at about 350ipm..... but it will sway so much so that when I've got my heavy fat $%%^ on it it will still sway and the pulleys can begin to slip above this so I tone it down to 0.3g at 400ipm and don't put unnecessary stress on components.

The power supply is 50V AC (so equivalent to 70Vdc) a little beyond the stepper voltage rule of thumb....
32 * square root (L) = VMAX
https://www.geckodrive.com/support/step ... asics.html

https://cnc4you.co.uk/resources/Stepper ... 03.1Nm.pdf

So it is very much possible to run nema 23's at 4.2A that don't need to be 3.5kW servos if you build the gantry light....

what I did not realise until 2 weeks ago was that I'd set my stepper drives up as 4.2A as the stepper motors are 4.2A..... but when you look closely at the drive software it says "peak"..... not RMS...... which will explain why the stepper motors are always cool and are running below their current rating so they could be pushed more...... to a point as again I am using HTD belts which are obviously rubber with steel wire braid in them.... and they can be slipped when the acceleration gets too high (above 1g)

I must admit though I've not tried to cut at 1g acceleration just shunted the axis back and forth to a dial indicator so it would be interesting to cut a test piece out to see what it actually looks like with belt stretch / flex from the HTD belt teeth as they compress

[Rodw]

I don't think it will matter much. Whats been interesting with our engineered model is that in most instances the motors never need to be run at their rated amperage anyway. I'm planning on using a 5 amp stepper but it only needs 3.1 amps to do the job and there is barely any difference between peak and RMS. It will only go up 11 degrees C and thats 100% duty cycle.

[SegoMan DeSigns]

1- Plug laptop into Clearpath servos
2- Launch tuning software
3- Hit enter & set back and watch the show
4- If problems exist go to Teknic for an online tuning session

Yup it's that difficult Oh waitress another order of those if you would ...

[robertspark]

1- Plug laptop into Clearpath servos
2- Launch tuning software
3- Hit enter & set back and watch the show
4- If problems exist go to Teknic for an online tuning session

Yup it's that difficult Oh waitress another order of those if you would ...

[robertspark]

.... I'm planning on using a 5 amp stepper ...

Thats a big + heavy stepper, Nema 34? (I use something like that on the mill, but its not moving at 400/600ipm


Sounds like the gantry needs a diet ...


Its a bit like car designing .... if you want a higher top speed..... sometimes its cheaper to make it more aerodynamic that just ploking a 427ci engine in it.
https://www.arielmotor.co.uk/style/atom ... -atom4.jpg
https://www.bac-mono.com/the-new-mono


A bit like this one in my opinion (well built but man does it look like and over engineered gantry for a plasma cutter)

[ScottRa]

Built by a press/brake designer!

[robertspark]

You may be right either that or the torch is heavier than it looks

It looks the business though and I'm sure it moves but the lights probably dim when it moves all axis simultaneously.... Does it have to be that big

[Rodw]

.... I'm planning on using a 5 amp stepper ...

Thats a big + heavy stepper, Nema 34? (I use something like that on the mill, but its not moving at 400/600ipm

Sounds like the gantry needs a diet ...

Sometimes Robert you have no idea. We were actually considering a carbon fibre gantry.

[ScottRa]

Shoot, they could have the whole plasma riding around up there. Save 500$ worth of cable. I wondered why it runs the long axis of the table. Cutting huge pieces by moving them across the table and doing a section at a time? Interesting base construction as well.

[EverydayDiesel]

Other then price, is there any down sides to running a servo setup?

[robertspark]

.... I'm planning on using a 5 amp stepper ...

Thats a big + heavy stepper, Nema 34? (I use something like that on the mill, but its not moving at 400/600ipm

Sounds like the gantry needs a diet ...

Sometimes Robert you have no idea. We were actually considering a carbon fibre gantry.

I'd like to see the pressure vessel that the carbon fibre gantry is going to be cured in. I've don't a bit of work with a few companies who manufacturer carbon fibre.

[robertspark]

Other then price, is there any down sides to running a servo setup?

None, when selected correctly for the application.

[ScottRa]

Not exactly what I expected but a search found this. Aluminum gantry. Not a cheap build!

https://www.cnczone.com/forums/plasma-e ... tware.html

[Rodw]

Thats a big + heavy stepper, Nema 34? (I use something like that on the mill, but its not moving at 400/600ipm

Sounds like the gantry needs a diet ...

Sometimes Robert you have no idea. We were actually considering a carbon fibre gantry.

I'd like to see the pressure vessel that the carbon fibre gantry is going to be cured in. I've don't a bit of work with a few companies who manufacturer carbon fibre.

You still have no idea. No pressure vessel required. If you know what you are doing (I don't) you could do it in your home garage for about $150.

[beefy]

Interesting thread.

First of all I've never used servo motors (yet). If I'm not wrong, one benefit to servo motors is they always have their maximum torque available at all speeds. On the other hand, a stepper motor loses torque as it's rpm increases.

So does that not mean that as a machines rapid speed increases, it acceleration capabilities decrease (when using steppers). Rod, unfortunately most guys don't have access to a stepper engineer to help them out but I have no doubt what you say is true. So one advantage I see to servos is much higher rapid speed capability for a similar sized motor. On a big nested cut, or arty cut, I can see quite a bit of time saved having faster rapids.

One note of experience - the quality of the stepper drive and motor can make a huge difference to speed and acceleration. Learnt that the hard way.

I also agree with Rod about the lost steps issue. If a system is designed correct you'll never lose steps with a stepper motor system. The basic "tuning" method is to keep ramping up speed and acceleration until the steppers slip, then back things off by a decent safety margin. Did mine that way and never had any lost steps.
However, let's say something happened that did cause a tight spot, or whatever, and the stepper did lose position. Most likely the operator would not know about it until it's too late, whereas I believe most servo drives will fault to let you know. At least closed loop steppers should give you that advantage.

These drives that momentarily increase current to a stepper motor sound good but doesn't that mean you need larger/heavier steppers from the start in order to take advantage of this feature.

Rod,
love your idea of a carbon fibre gantry. I've often thought of 4140 thin wall steel tube spaceframe for a gantry design. That gets used in aeroplane and racecar frames, but I think carbon fibre would be the ducks nuts.
Looking forward to seeing your tables on the market, I think they will be quite special.