plasma table retrofit: smart servos or closed loop steppers?

[Josh Leber]

Hey all, I'm fairly new to the forum so please be patient with me but I could use some expeirienced enlightenment.

My company bought an older stepper driven VICON 5'x10' plasma table for retrofit. (yes we can justify the time & the money) The machine has no controlls just the drive motors.

From what I've read, I'm pretty set on not using a straight open loop stepper system for the standard reasons. We've been looking into 3 options:

1. Standard servo system: Servo motor with gear box reducer (10arc/min or better) seperate servo drives. This option I think I know most of the pros & cons except gear box life expetancy? I talked with Walt at dynatorch and he claimed to have a customer with better than 7500 hours of use on his gearboxes with virtually no wear (under 6 arc/min)

2. smart servo motors: example: Dynatorch system. This is very apealing but aparently requires software specifically designed for smart servos which unfourtunatly are not all that common yet. Not very modular. The dynatorch software will work but that dos'nt leave room for a future change. The fact that the servo does not require support drives is very apealing. The motor connects to the controll PC via a serial cable (clean and simple or non-modular and obsucre?)

3. This option I know little about so any input here will be especially valued. Steppers with piggyback encoders and drives capable of making slip/skip corrections based on the encoder output. Closed loop operation right? Sounds like the best of both worlds. The stepper's synchrounouse controll of shaft movment and high tourq at reletivly low speeds coupled with a servo's ability to recalculate movement in real time incase of a slip/skip.

Servos require a big speed reduction for my application wheras steppers do not (2000-3000 rpm ~ vs ~ 100-300 rpm)(I see this as a signifigant atvantage in favor of the steppers)

whatda ya think?

What ever we go with needs to be mainsrteam enough to find upgrade and support info in the future.

One last question: Anyone here know what the hypertherm phenix cnc plasma controll software goes for?

[Al_The_Man]

Originally Posted by Josh Leber 3. This option I know little about so any input here will be especially valued. Steppers with piggyback encoders and drives capable of making slip/skip corrections based on the encoder output. Closed loop operation right? Sounds like the best of both worlds. The stepper's synchrounouse controll of shaft movment and high tourq at reletivly low speeds coupled with a servo's ability to recalculate movement in real time incase of a slip/skip.

In my opinion, if you are going to fit encoders, then you may as well use the much better advantage of servo's.

Originally Posted by Josh Leber Servos require a big speed reduction for my application wheras steppers do not (2000-3000 rpm ~ vs ~ 100-300 rpm)(I see this as a signifigant atvantage in favor of the steppers)

You are going to need the same motor/torque characteristics in order to meet the required inertia motor-to-load matching, that is required by your accel/decel requirements as well as max feed rate.
The use of reduction enables efficient motor sizing/specifications.
Al.

[Torchhead]

There have been a lot of stepper based plasma systems out there for a long time. If designed correctly stepper systems are reliable and don't lose steps. If you have a situation where the table is losing position the servos will try to correct but sometimes to the point of their own (or the part your cutting) self destruction. If you are seeing position lag something is wrong. Either the load is too heavy or the machine is too weak and the servo's can only correct up to a point. The main thing is that steppers will not require reduction whereas servo's most certainly will.

Don't get me wrong. My plasma table has servos on X and Y (actually dual on Y) and a stepper on the Z which is under control of the software (Mach3) and an integrated THC. I also have two other smaller tables that are stepper based one of which I use to route 15 mil runs on PCB's. It never misses a step.

There are cost effective ways to do the speed reduction for servos without using low backlash ($$$$) gearboxes. Toothed belts and pulleys are very effective and you can double step them using a jackshaft arrangement to get higher multiples.

It appears to me that the Dynatorch solution is an all or nothing approach since it takes their software to run their motors and their ATHC is integrated in with their software, etc, etc.

About the encoders on steppers: Since it's difficult to control the exact postion of a stepper, encoder feedback would tend to not be as accurate as on a servo. So what do you tell the electronics to do if the position is off from the number of steps? Add more steps? How far do you have to get behind/ahead before something figures out it's a lost cause? If the stepper can't get there in the first place how will adding more steps in help?

There is a legit use for encoders on a stepper and you can do it with MACH3 and a card from Rogers Machinery. It monitors the real postion via encoders and the commanded position via the screen DRO's and if there is too much variance it halts the machine. It's there to prevent scrap parts. Closed loop control has a lot of challenges. Most of the (affordable) servo systems like the Gecko's close the loop at the drive and the software runs open loop.

Do some asking over on the MACH3 section (there are over 4000 on the support group site). Lots of us have built and retrofitted all kinds of machines.

[Josh Leber]

Torchhead, Thanks for your responce

I had a feeling the whole "propriotary software/ hardware" sounded limited. I am at this point invested in the whole CNC machine thing and as our facility grows its only inevitable that we must know more on the finer workings. If we went with the smart motor thing and wanted to make some changes later to individual components we'd be stuck with a complete retrofit again.

On the note of software, everyone seems to have "special magis plasma software" designed specifically for the plasma and oxyfule cutting industries. They all have a special feature or two like arc monitors to signal a blow out, arc start conformation etc. but they are still controllers which (with the exception of the hypertherm phenix program) dont do file preperation, i.e. kerf compensation, lead in, lead out etc. which you must do beforehand. That being said, you mention mach3 (I'll check it out in a minute). Will it cover all the more important features specific to plasma cutting with the proper configureation?

Also one last question. What was your personal reasoning with going servos on the x and y?

Thanks again, Josh

[Al_The_Man]

Originally Posted by Torchhead . The main thing is that steppers will not require reduction whereas servo's most certainly will. .

I don't really understand that statement, I assume you are comparing motors of equal torque?
Al.

[Torchhead]

You cannot compare "torque" as easily between steppers and servos. If you look at the curves and ratings steppers develop their max torque at stall and the curve starts to drop off with higher RPM. They are relatively low RPM devices and it's common to see them directly connected to leadscrews (where there is an inherent reduction) or with about 3:1 reduction on rack and pinion where reduction is minimal or even non-existant.
The rating on servos is more realistic and done in the center and upper end of their RPM range so reflects torques in the range the motors will be used. Couple that with the added gear reduction for servo's and your force at the load is going to be greater with a smaller motor. It's the trade off of speed VS torque. Also servos don't do extremely low RPM very well (a condition call "cogging") so the goal is to try and operate them at higher motor RPM. Steppers love low RPM. In the end the goal is to have a machine that provides the feedrate and power to move the load.

[Torchhead]

I like my software "smart" and my motors dumb (:-)

On the software: You need to think in terms of the CNC process and the three main pieces. The process to get the design from your head to the table is:
Drawing(CAD)->Toolpath(CAM)->Controller.

The CAD process can be anything from a simple 2D CAD or drawing program (I use CorelDraw since all of our stuff is decorative and normally has a lot of text)

The CAM process takes a CAD file and allows you to define HOW you want it cut on your machine. What kind of tool, kerf width, offsets, lead-ins, pierce height, inital cut height, etc) and then using a "Post" to generate a g-code file that is specific to your controller. For a very cost effective CAM solution for plasma or router see SheetCAM (www.sheetcam.com)

The controller can be hardware or software or a combo. the most popular in the lower cost options is to use a software controller (that is where MACH3 comes in) and a PC as the "engine" driving motor control modules (Gecko) through the parallel ports.

You need to temper my advice with the understanding that I sell products largely for use with MACH3 but I got there by having a commercial plasma cutting operation (for profit) and learning the pieces I needed the hard way. I like the concept of having options in each of the three parts of the process.

I obviously have a financial angle but on our site there is a lot of information and links to resources for plasma.

www.candcnc.com

Commercial Decorative Steel cutting at www.fourhillsdesign.com

[Al_The_Man]

Originally Posted by Torchhead You cannot compare "torque" as easily between steppers and servos. If you look at the curves and ratings steppers develop their max torque at stall and the curve starts to drop off with higher RPM. It's the trade off of speed VS torque. Also servos don't do extremely low RPM very well (a condition call "cogging") so the goal is to try and operate them at higher motor RPM.

Well the servo motors I use, both AC and DC all have a fairly flat torque curve, in fact in both case the torque is maximum at zero speed, with very little tail-off up to max. rpm.
The only cogging at very low rpm I have experienced is with BLDC style.
DC servo's with skewed rotor design and sinusoidal AC both can be used down to zero speed with no cogging effect and max. torque.
Direct drive ball screws have been used for many years with excellent results, as long as the inertia ratio is calculated correctly.
Al.

[Josh Leber]

I joined the mach3 yahoo group last nignt, and thanks for the link. Currently we use AutoCad for our technical drawings but are planning a switch to solid works so we have the drawing part down. I'll check into sheetcam for file prep.

Now as far as auxilliary controlls go:

1. how do you controll the signal to start the arc, then confirm that it has started and operational? Is this just done with some kind of on/off type of inputs and outputs but not considered another axis controll? If so what is the interface device? It looks as Mach3 can be taylored to do all of this, I'm just unclear on the physical part of it.

2. Aside from the cut height being controlled being by a voltage sensing THC, how does the z servo/stepper know to lift betwene cuts?

[Torchhead]

1. The g-code starts the torch using an M03 command. The Function of the CAM (SheetCAM) is to know when and where the torch has to fire. It's slightly more complex than that, since you really need to have a lead-in in the scrap area because piercing creates a "divit" about 2X the width of the kerf. You should do the initial pierce at a height higher than your cut height and we have found that if you let the THC take over too soon after the pierce (which creates a voltage surge) the head will dive, and if the circuit is quick enough, recover and move back up....sort of a head bounce. There is another factor/feature you need to shop for: The unit should provide some way to hold table motion until there is a valid arc. Some plasma units have a machine interface and provide a signal when the arc is actually cutting metal. If you don't have that you have to program in a long enough delay at pierce to handle any delay in the arc starting, usually 2 - 3seconds, and then on 90% of starts it just sits there burning a big ole hole in the metal. Your THC should have a way to hold table motion until it gets a valid "Arc Good" signal. We send a current sense transformer for plasma units that don't have an intergal Arc Good signal.

2. The same way it knows how to start. The CAM program knows when the cut is finished turns off the plasma (you need a slight pause here to let things settle) and lifts the Z. One problem with a 2 axis + stand alone THC is that the THC has to know when to lift the head. If you have a full 3 axis machine the head lift and all other positioning is under control of the motion software.

Another "gotcha" in plasma cutting is that you pierce at a precise height above the material. Too close and molten steel blows back into the tip and causes it to short (and fail, requiring the machine be stopped and the tip replaced). To high and you get a bad pierce and no cut or worse, a long ugly gouge in the top of the metal. Your machine needs a way to "know" where the top of the material is. It's no big deal on perfectly flat material that never buckles and your cutting surface is perfectly level. One solution is to not worry where you are in the Z plane and just "probe" for the surface each and every pierce. That works good on big pieces with not many pierces. It's a real time consumer on small detailed cuts. SheetCAM in tandem with MACH3 and our THC has a feature we call touch-n-go where you can set a value in inches and Z will reference on the next pierce after your set number. Since the Z is being controlled by MACH3 at all times (even during THC UP and Down Moves) it keeps up with where it is so intermediate pierces are done at a known height.


Me thinks Solid Works is a fine package. Many times overkill for shape cutting with plasma. Kinda like driving a Ferrari 6 blocks in traffic to work and home. It is of course your companies money so they can do as they wish.

Plasma cutting is not much more that plotting with a little more fireworks. In fact I use the HPGL (plt) file format to import into SheetCAM because it A IT Always works no matter what version of software you use for export and B It's perfectly suited for shapecutting since it was made to draw shapes on paper (not pictures). Others use DXF and complain when the have to go back and fix stuff or save in another version of DXF. To each his/her own.

[Josh Leber]

Wow. That explains alott. I apreciate the in depth explaination. So I'm pretty set on mach3 & sheet cam to run the thing. I was looking at bob campbell's THC, 3 axis combo break out board and steppers running on Gecko drives. But just looked back at your post and found the link to your site. How do your parts compare to those above? Went back and fourth on servos or not and decided that steppers are inexpencive engough that if we don't like them, it wont be a big deal to upgrade to servos and servo drives. I understand Mach3 will run either.

Hav'nt looked in the plasma cutters schematic for a arc good machine interface point yet. The machine is a Miller spectrum 1250. but you say a CT will work the same.


Something else: I'de like to build a decent computer for this thing. We alread have many of the components here so cost i'snt a big deal. I was thinking something with 2gHz or better processor, 1 gig of ram, 40gig hard drive. Nothing special but not sure on what is optimal for a system like this. Do you think I would need to do more on any of these components, missing something or maybe overkill? This would be a workstation soley for running the machine with some in shop drawing changes (most drawing would hapen in the engineering dept not on the shop floor) and would be on the network. So i'de probably be running Acad, sheetcam & mach3 all on win XP and thats about it.

The upgrade to solid works mainly due to more complex 3d modeling needs for development of new products with components that have a higher $$$ comitment for prototyping. Dont want to say have a $10,000 die made for a rotomolded part and later find out some bracket or component interferes with it. The deal seems to be if we set up one engineer with a SW workstation, we think it would be benificial to have everyone on the same platform for ease of comunication & compatibility. Beyond that, I't has been an item that some of out suppliers (most use SW) would like to see as right now they either have to re-draw or convert our 2d drawings to produce new parts. This has been one more thing to suck up development time. So if you have the Farri for the track, what the hey might as well take it to the store. right? er .... maybe not... but you get the idea.

Hey thanks again for all your input.

[Torchhead]

Josh: I designed the original THC that became the THC300. The MP1000 is a second generation design that is processor based and offers the ability to preset target tip volts for different types of cuts, voltage setting from the front panel or MACH3, preset readout, built in THC delay function to stop head bounce, a full Port of buffered outputs for the step and direction for the motor modules, added isolated inputs for a front panel tethered handwheel (MPG) [Free this month with the MP1000-THC] and front panel mapped function swtiches for MACH3 control or Optionally interface to control our power modules : the ACM-100 and DCM-100.

The MP1000 holds specs and has features that units costing 10X don't have. It's real time integration with the control software (MACH3) lets it be part of the toolpath and offers things like anti-dive on tight corners.

So the MP1000 is a fill two port master breakout (you don't have to buy another breakout to run your motors or auxillary cards to have more relays) with lots of input and output (including 3 optional aux relays up on the table I/O card that mounts on the table where Home, Limit and safety switch inputs belong. It is designed to be more than just a THC or Just a Spindle Speed and is the basis for a control panel for the whole machine.

All that and you get the support and expertise of guys that have lots of designs under their belt and actually use the stuff in a commercial environment. (i.e. we cut metal and wood for profit)

Read the detailed information on the website about the MP1000, the MPG-01 and then download the 46 page manual for the MP1000-THC.

Also join our on-line product support group and lurk and see how we support our products. It's all in the open. The group is listed on the CandCNC website. Your support comes from the guys that designed and use the products on a daily basis,

You don't want the MACH3 computer to do anything but run the machine and talk to the network. Do all of your design, revisions and tracking on other machines. A 1.8G processor will work fine (some have used much slower) and 384M RAM. You need a second parallel port and Windows 2000 or XP. Either works fine. Hook 'r up with a network card and you are good to go. MACH3 is free to try but the THC function is dead in the Demo version. Sheetcam will give you a 30day full feature trial.

I don't know about the Miller 1250. If it does not have a machine interface then no big deal. We can help you get it hooked up.

Recommend you also look at getting the Gecko motor modules since it then becomes a nice package that a lot of the 4000+ users of MACH have and you will get lot's-o-help from us all.