[Dougal]

Hi guys, hopefully some of you can spare some knowledge to help a newbie.

I'm in New Zealand and have bought a mill that while rebranded, having an MT3 spindle and running on 240v is essentially an X2. So to save confusion I'll call it an X2 from now on.

I plan to convert this to CNC to make what the machine shop boys call rats and mice. the pieces that are too small to bother them with. Mountainbike parts, small car parts etc.
I'm also hoping that with a probe I can use it as a small CMM to reverse engineer parts.

I've been trolling sites like this for a few weeks and the options available just seem to grow rather than shrink. So I need some help nailing down the best collection of parts.
I want ballscrews, that's easy as the CNC fusion kit has that covered.
I'm on the fence regarding servos vs steppers. The absolute positioning of the servos will be a major advantage (I freely admit to crashing and stalling machine tools) and I like that I can buy braked servos which will go nowhere when the power is off.

I have absolutely no idea what control system to buy and I need help here. I would like to run the system from a laptop and will probably have to build/buy a 24v supply. But I would like the rest to be as complete for plug and play as possible.
I'm fairly savvy with electronics, but I'd rather pay for a kit to avoid the hassle.

CAM is something I'll address later. I've been playing around with demos of lazycam as well as cambam and g-simple thanks to finding out about them here. I'm a solidworks user, I have used G-code for industrial automation but not milling and have never used a CAM package before. But I know plenty of people who earn their living that way.

Any help appreciated.

[Stepper Monkey]

If you are going as far as ballscrews and servos, why a 24v drive and power supply? That would seriously cripple the thing, and only save you a few bucks at most.

BTW, I have brakes on my steppers as well, to both prevent the Z backdriving and hold positions on power off. They aren't just a servo thing.

If you are looking to cut corners anywhere, a servo system might be the better thing to drop. At this size both systems perform the equally well for designed operation, its only when you crash it does it become an issue.
That said, is the fairly significant chunk of extra money worth the occasional instance of having to recut of one of the little 'rats and mice' you might have potentially been able to save after a crash if you had servos? Remember that even with servos few parts can just be picked back up after a crash and resumed. Unless your cuts are very long and your material very expensive or irreplaceable, its hard to justify the cost solely on that basis.
Servos are definitely nice to have, so you could just really want them for other reasons, but the often used reason of recovery-after-crash is not really of much value at all when it comes to this type and size of machining.

[Dougal]

If you are going as far as ballscrews and servos, why a 24v drive and power supply? That would seriously cripple the thing, and only save you a few bucks at most.

I had stumbled across one controller that said it needed a 24v supply, what do most require?

BTW, I have brakes on my steppers as well, to both prevent the Z backdriving and hold positions on power off. They aren't just a servo thing.

If you are looking to cut corners anywhere, a servo system might be the better thing to drop. At this size both systems perform the equally well for designed operation, its only when you crash it does it become an issue.
That said, is the fairly significant chunk of extra money worth the occasional instance of having to recut of one of the little 'rats and mice' you might have potentially been able to save after a crash if you had servos? Remember that even with servos few parts can just be picked back up after a crash and resumed. Unless your cuts are very long and your material very expensive or irreplaceable, its hard to justify the cost solely on that basis.
Servos are definitely nice to have, so you could just really want them for other reasons, but the often used reason of recovery-after-crash is not really of much value at all when it comes to this type and size of machining.

Thanks for your advice, since you have brakes on your steppers, is it possible to put encoders for feeback on them as well?
Or would this require a controller that isn't commonly available?

Are the servos available in this size fitted with incremental or absolute encoders? Absolute would of course be best (power up the machine and it knows where it is), but are they available and at what cost?

[Stepper Monkey]

I had stumbled across one controller that said it needed a 24v supply, what do most require?

Some stepper systems run at 24v, but it is much, much better if they are capable to run at 36 or 48, or even as much as 60 or 72v. I am not a servo expert by any stretch, but they generally run much higher - the ones I have seen have ran at 72 or 90v.

Thanks for your advice, since you have brakes on your steppers, is it possible to put encoders for feeback on them as well?
Or would this require a controller that isn't commonly available?

The brakes are standard Nema 23 items, and it is common and easily done to place a shaft encoder on the rear shaft of double shaft steppers (its generally why the second shaft is there). Mach3 controller software can accept the standard quadrature from encoders natively. There are also a number of commonly available breakout and interface boards to help for this purpose if needed.

One thought though; any feedback system using only shaft encoders is limited, as it does not give any extra benefit beyond more intelligently handling failure modes as you mentioned earlier. Under normal operating conditions they are redundant.

To actually increase accuracy or other performance attributes during normal operation, it takes MUCH more expensive linear encoders on each axis, as shaft encoders only monitor motor position. It therefore cannot reflect actual table position, as backlash or any other errors induced in the components after the motors themselves cannot be monitored by this type of system. This is as true for servos as steppers.
A shaft encoder only system has uses, not to knock them, just to point out that not all feedback systems are equal!

[sansbury]

In terms of crash recovery, won't an accurate homing procedure do as much to salvage the job as a servo drive? You can return to a known position, fix the cause of the crash, and restart the job from there. OK, maybe a *little* more manual intervention but you'll presumably be in there anyway to fix the code that crashe dyou in the first place. And if the crash has caused the part to shift (or more likely run it over) then you're well and @#$!ed no matter what kind of drive you're using...

[Stepper Monkey]

In terms of crash recovery, won't an accurate homing procedure do as much to salvage the job as a servo drive? You can return to a known position, fix the cause of the crash, and restart the job from there. OK, maybe a *little* more manual intervention but you'll presumably be in there anyway to fix the code that crashe dyou in the first place. And if the crash has caused the part to shift (or more likely run it over) then you're well and @#$!ed no matter what kind of drive you're using...

Exactly this. Well said.

[X2cnc]

I want ballscrews, that's easy as the CNC fusion kit has that covered.

I have the CNCFusion ballscrew kit. The X and Y parts are just fine; I'd buy them again without a second thought. Since I plan on setting up a second X2 if things go well, that's not an idle comment.

Unfortunately, the Z-axis bracketry sucks decomposing roadkill. It's nicely made and bolts right up (requires drilling a couple of extra holes in the column, no big deal). However, it covers the Z gib screws. You can't even remove the bracket without removing the head first, so any adjustment of the gibs requires completely removing the head and CNCFusion bits.

I had read complaints about this while searching for comments on the kit, but I didn't really understand what people were talking about.

I have no clue why CNCFusion did things that way, but they could have pulled the screw forward a bit to clear the gib screws, or put it in the middle like some other kits, or even mounted it on the other side. My simplest solution is to make new brackets similar to the CNCFusion ones, except mounting the ballscrew on the other side.

The KDNTools Z-axis kit mounts in the middle of the headstock. The web site shows an Acme screw, but they supposedly have a ballscrew option. The CNCFusion XY bits look sturdier than the KDNTools stuff, and were cheaper; I'd use those with the KDNTools Z kit.

The Hossmachine design also uses a central leadscrew, bit Dan ran the screw up above the machine instead of down along the column. It took a while to figure out why - the table doesn't get close to the screw on the KDNTools layout - but Dan's layout with the overhead screw keeps it out of the path of coolant and swarf at the expense of making the mill a bit taller, though you would only care if you planned to enclose it.

[Dougal]

Thanks for the advice guys.
X2CNC, excellent information about the Z screw on the CNCfusion kit. I'll most likely still get the full kit and make new mounts later like you have. Getting the machine working is going to be the first and most important step.

If I invest in some linear scales or seperate encoders, are there any advantages to servos over steppers?

[X2cnc]

I haven't made my brackets yet, but I'll post a drawing when I'm done. I think I can flip the CNCFusion bracket around to the other side and get away with one adapter at the top of the column.

I asked about encoders a few weeks ago. There were several replies, all to the effect that they shouldn't be necessary with steppers. The count should always reflect the position as long as you're not losing steps, and if you are losing steps, you have a problem you need to fix.

[Stepper Monkey]

If I invest in some linear scales or seperate encoders, are there any advantages to servos over steppers?

That tends to get you into a quality of gear well beyond anything reasonable for a hobbyist, while as rotary encoders can cost $20-30 each, proper linear encoders easily exceed the cost of the entire machine - for EACH one of them. They normally aren't found except on very serious five- and six-figure equipment.
That being said, there are people here who have quite creatively made some reasonably functional ones by mounting cheap Chinese digital hand calipers to each axis and wired the readouts to a custom board that converts the signals into standard quadrature, but even that adds up to quite a bit of a project and cost. I don't know who has them on here but they may pop up and offer more advice on that particular project.
Linear encoders are neat if you have them or can scrounge them up used and cheap, but really probably not the direction you need to realistically plan to go on a budget hobby mill!

[Dougal]

If I invest in some linear scales or seperate encoders, are there any advantages to servos over steppers?

That tends to get you into a quality of gear well beyond anything reasonable for a hobbyist, while as rotary encoders can cost $20-30 each, proper linear encoders easily exceed the cost of the entire machine - for EACH one of them. They normally aren't found except on very serious five- and six-figure equipment.
That being said, there are people here who have quite creatively made some reasonably functional ones by mounting cheap Chinese digital hand calipers to each axis and wired the readouts to a custom board that converts the signals into standard quadrature, but even that adds up to quite a bit of a project and cost. I don't know who has them on here but they may pop up and offer more advice on that particular project.
Linear encoders are neat if you have them or can scrounge them up used and cheap, but really probably not the direction you need to realistically plan to go on a budget hobby mill!

I'm still throwing around ideas at this stage.
The cheap digital vernier type slides that you're talking about would work great for incremental position, but if I went to seperate encoders I'd be looking for absolute position. I've used linear potentiometers for a few projects before and still have two in the 125mm stroke size. That'd take care of the x axis, I don't know if they're available in the size required for y and z but the cost would probably be a few hundred $US for each axis.
I've spent a lot of time around serious sized cnc mills, even helped repair them but unfortunately never operated one. Some of them were large enough to mill half a car.

So it appears steppers are the most attractive option at this stage, what is the best bet for controllers and interface boards? Keeping in mind that I'd like to drive it all from a laptop via USB even though parallel and serial are available. Our power here is 240v but building/buying a power supply shouldn't be a problem.

[Stepper Monkey]

Keeping in mind that I'd like to drive it all from a laptop via USB even though parallel and serial are available.

hrrrm? Why would you want to do that? USB was not designed for real-time functions, and is not suited to it at all. It is good for massive data transfer to 'smart' devices capable of then parsing things and data handling at their end, but CNC machines aren't smart devices. The parallel was designed specifically for direct CNC machine control, its actually precisely why it exists.
Parallel is the right tool for that job. People think it is old and therefore must be obsolete, but for this purpose there is none better - back when it was invented the old line printers that it was designed for had no buffers or internal control systems themselves, they required full direct real-time hands-on control by the host computer. Quite literally they were CNC devices, direct machine control is actually exactly what parallel was doing all along. We just happen to be the only people still using it as it was intended!

USB and parallel were developed for two different jobs, (massive non-synchronous data transfer versus real time control) and they don't interchange.

There are USB control systems for CNC out there, but they don't act like you may think. Using one, your computer isn't then the controller anymore. At that point it simply uses the USB to pass the gcode to a separate embedded controlling computer module mounted in your drive box that then does the work of sending the signals to the drivers - using, um, parallel. There is no way of getting around it, and so if you have parallel already, adding extra hardware and an extra control computer in between seems redundant.

[cyclestart]

steppers, is it possible to put encoders for feeback on them as well?
Or would this require a controller that isn't commonly available?

http://www.microproto.com/MMDSLS.htm

Not sure if something like this can easily be done strictly on the software side of things.
If the software attempts to make up for a lost step by speeding up the stepper the game is lost.
If the software can adapt the feedrate with a lost step it could work in a sort of kind of way.
If the software would stop all motion when detecting a lost step the work could be salvaged and adjustments made to the program.

Just talking hypothetically. I never gave it much thought until you asked.

Re using a digital caliper or similar for feedback. There may be a speed issue involved. Seem to remember reading something about that.

edit/ if the software could somehow add some needed steps it would be a cool solution.

[The Blight]

There is one guy on here who is working on a motion controller that uses feedback from cheap chinese linear scales, but they only have a 50hz update rate, so that will turn out to be a problem. A better sollution would be to use them as possition checkers and not drive the motors out from the signal you get from them. Just to verify the position.

I have been working on a simple device to check for lost steps and stop the machine if it detects any. Most people say that this is overkill, but I say you can never be to safe.

Stepper Monkey is right about USB for cnc. It is used as a serial data line, not as a means to control the machine. You need some logic computer in between which converts the signal into a parallel signal.

[hoss2006]

SmoothStepper USB Motion Control Interface for Mach 3.
They just got a new shipment of 136 boards in July 14th.
No need for a parallel port at all if you are really wanting to use a laptop.
Very impressive results as you can see in the video,
he tripled his speed vs the LPT port.

This is my first test of the Beta version of Smooth Stepper. Compared the LPT port I have increased rapids from 370 IPM to 1,100 IPM. And increased the rate of acceleration by 5X... These parts are being cut @ 150 IPM.

There are other vids too using mills and steppers, just search youtube.
Something to consider when the time comes, or not.
I might have to get on the list.
Hoss

[ame="http://www.youtube.com/watch?v=uaG3pd8xuWw"]Smooth Stepper USB Interface[/ame]

[Stepper Monkey]

Yup, thats one of them. It has the embedded computer on the board to sit between your laptops USB and the stepper drives, and its job is to convert the signal into a parallel signal for real time control. For only one extra piece of gear, a couple of extra cables, and $155 bucks, you can do exactly what you could have done by simply plugging it into your parallel port in the first place, only with less performance.
If your laptop doesn't have a parallel, a PCMCIA parallel card works better and is much cheaper.

This system may indeed allow faster rapids on routers, though pointless on a mill with less than a foot of axis travel, and at the cost of giving up things like the look ahead buffers and CV, and therefore things like actual cutting time and tool life along with a lot of other nice features that are pretty important. Looks like a solution in search of a problem to me.

I can see the uses of these for a number of reasons in some cases, none that apply to this case though. You are actually giving up a LOT in exchange when you trade both your computers power and the very powerful Mach software for a simple embedded control chip and whatever limited software and memory it has available.
Rapids aren't an issue anyway here, as the mill and the motors are the limit. The LPT and Mach can send signals a LOT faster than the mill can move, so signal transfer rate through the parallel is not the short link in the chain by a long shot. Giving up a lot of useful features to fix a problem that isn't one doesn't help anything.
Get a 10' router with a fast screw and it might well be a different story.

[The Blight]

PCMCIA cards are not that much cheaper. Unless I'm missing something? Last time I checked, a true PCMCIA parallel port cost $125. Do you know of any cheaper cards? Not the ones that are actually a usb port with a built in parallel port emulator. Those are no good when it comes to motion control.

[hoss2006]

Are ya done? 2 edits and counting

Dougal,
Just giving you an option that's available out there,
If you have an open mind you can decide for yourself.
They are a new company but reading their initial pdf,
the spindle control and more inputs/outputs than 2 parallel ports
looks inviting. I've already used all of mine, an MPG hogs a bunch.
I would need to see the final PDF with the wiring diagrams and such first.
Don't just take my word, read up on things yourself
so you can make an informed decision.
There are many more options available to you.
If you want to reach me for anything, you can contact me on my website
and I'll be glad to help.
I have shared lots of tips, tutorials and projects free for the asking.
Cheers, Hoss

[Regnar]

"Keeping in mind that I'd like to drive it all from a laptop via USB even though parallel and serial are available"



You might want to keep your eye out on this thread. http://www.cnczone.com/forums/showthread.php?t=59763

And here is the website. www.cncbrain.com I cannot vouch for them but their response to peoples questions is great. Not only that but for extreme accuracy could be within the grasp of a hobbiest.

[The Blight]

CNCBrain does look quite nice. Will have to look into that tomorrow. I had to spend over $200 to get all the outputs and inputs that I needed for my machine. So if you need lots of inputs/outputs then the USB sollution might be best, but if you are just planning on running the machine without any added functions, a simple parallel port would do just fine. I too am curious as to why you have reached that conclusion.