Best components?

[Outlawtaz]

Noob to CNCZone but not to the CNC world. Let me preface my question with a little about me. I am a licensed aircraft mechanic and also spent a number of years as field tech for a (unnamed) CNC manufacturer. So, I am already familiar to with design and functionality of CNCs but was limited to one manufacturer. I'm looking to broaden my ideas...

I am mostly familiar with NSK and Rexroth, everything else was proprietary to the manufacturer including motors, drives, VFDs and controllers.

Anyway, I am finally in a position to build a machine the way I want. I was wondering what everyone's thoughts were about the best components for a home build; best overall and best bag-for-the-buck. Looking for everything from drives to motors to ballscrews to ways...

TIA - I look forward to everyone's thoughts and experiences

Similar Threads:

• 5 Components
• How would these components be made
• DIY with quality components
• Linear Components and More

[ger21]

For linear rails, any of the major manufacturers are probably comparable. THK, NSK, Rexroth, Star. Most people use Hiwin, because they can be a lot cheaper, and are more readily available.

The same manufacturers also make ballscrews. But most people use cheap chinese screws, entirely because of the cost of high quality new ballscrews.

[wizard]

Before you can get detailed answers, it would help to have details about your expectations and the machines size. These two things size and expectations really drive what you should build including the components selected.

To put it another way there is a huge difference in a machine built to process strutural members for a sale boat as opposed to a machine a luthier may want to make guitar parts. One guy will be making parts that get twisted and bent into place the other might want fits held to 0.001" on stiff parts. So expectations are a big concern.

The next issue in my mind are what resources do you hace access to? Any amount of shop equipment can help. In my mind the best solution at the moment for framing is to go with welded steel out of box sections (square or rectangular tube). This assumes of course that you can get access to a welder and machine shop equipment.

For guys with limited tooling extruded aluminum frames are possible but im not a real fan of T-slots. If you go this route make sure quality extrusions are being used.

In any event it is probably best to nail down a mechanical design before even thinking about your controls. Even with mechanical design you have a big decision to make with respect to moving table vs moving gsntry.

[Outlawtaz]

Ger21 - Thanks for the info. What are your thoughts on electrical components?

Wizard - I have access to anything I want: if I don't already own it, I know someone that does. I was a field tech on machine tools for 6 years. I know somebody and can obtain access to more machine shops than I can count.

As for my expectations... I want to build a machine that will hold at least 0.0025". I don't NEED that tight of a tolerance. I only WANT to do this because I know I will not be happy with anything more than that. 1) I spent too many years either working on aircraft with extreme tolerances or dialing in customers machines to be able to repeat insanely tight tolerances. 2) I have too much OCD and will drive myself crazy if I can't achieve a respectable tolerance and repeatability LOL

I agree about using T-slot extrusion. My current thought was to weld a base out of 1018 (most likely) square tubing. Then, I would take the base and face both sides.

I have debated the moving table vs. gantry thing. I have a bunch of experience working on both so I am well aware of design for both. Personally, I am intrigued by the moving table design. I have the room for a large footprint so that is not a consideration. I just think that having the wide spaced trucks on the bed along with the fixed gantry would make tighter machine.

[ger21]

Ger21 - Thanks for the info. What are your thoughts on electrical components?

What are you planning on building? Steppers or servos?
There are a ton of different options, so knowing what you want to build will narrow it down a bit.

[Outlawtaz]

What are you planning on building? Steppers or servos?
There are a ton of different options, so knowing what you want to build will narrow it down a bit.

Wanting to build a router for use in woods, plastics, MAYBE soft metals. Maximum size will probably be in the 4'x4' range. Would like to build with a VFD so that I have cutter rpm control in the g-code. I will use ballscrews and ways since I don't trust belts (at least not in the size used on these machines) and I know how to set backlash and maintain the systems anyway.

As for servos vs steppers: I don't have a need for fast rapids and most movements will be small anyway (engraving type stuff), so steppers will probably work the best and be cheaper. My main hangup is I don't trust the cheap and poorly constructed stuff, chinese or otherwise. I want quality so that I have less chance of breakdowns and better control

[ger21]

For a long time, Gecko stepper drives were considered the best available.
In the past few years, Leadshines digital drives have become very popular. Some say that their EM series drives are better than Geckos. I'm using Leadshine AM882 drives, which are only available from China through Ebay, and are unsupported. But they work great, and are about 40% cheaper than the EM series or Geckos.

For motors, almost everyone is using Chinese motors, because they are so much cheaper. And steppers are pretty much bulletproof, so any failures are very rare.
If you want to pay more for top of the line motors, look at Oriental Motor (Vexta) or Lin Engineering.


Other electronics depend somewhat on your choice of control software.
I'm using UCCNC software, with their UC300ETH controller, and a UB1 breakout board form CNC Room in Thailand.

Centroids acorn control is only a few months old, but is quickly gaining popularity.

Linux CNC is also quite popular, and is free, if you don't mind running Linux on your control computer.

[Outlawtaz]

Ger21 - That's the kind of info I am looking for. To be honest, I don't really mind chinese components so much as a lot of the things coming out of there are crap. If there are quality components like the Leadshine, I wouldn't hesitate to use it. Everything I had read said to use Geckos but if another manufacturer has better bang for the buck, then you better believe I will use it.

As for software, what's is the fascination with Mach 3/4? Is it that great of a software package or is it just that everyone is using it?

[jgwentworth96]

I have yet purchase anything from them, but automationoverstock.com is very interesting and has good pricing. They purchased the remaining stock of the US warehouse of HIWIN rails and they have some NSK ball-screws as well.

[ger21]

Mach3 was groundbreaking when it was created (first as Mach1, then Mach2). It brought affordable, Windows based CNC control to the masses. It was the dominant control for many years. And also factor in that tens of thousands of chinese machines on Ebay are Mach3 controlled, and you have a LOT of people using Mach3. My current machine that's been running for 10 years is still using Mach3.

But my personal opinion is that Mach3 is no longer a viable option.
Mach3 has a lot of minor bugs that were easy to overlook when there wasn't a good (or affordable) alternative.
And development and support stopped over 5 years ago.
But most importantly, there are now several alternatives that are both better, and in some cases even cheaper than Mach3 is.
There's just no reason to buy a buy a buggy Mach3 today.

I switched to UCCNC for my new machine, as I mentioned. I think there are at least 5 solid alternatives to Mach3 at under $400 mark, and a few more if you go up in price to the $1500-$2000 range.

Imo, Mach4 has been a complete disaster. The developers created a control that users didn't want.
From what I here, it's far superior to Mach3, but it's expensive, extremely complex to setup, and still suffers from Mach3's biggest weakness. Mach4 requires 3rd party hardware motion controllers. So Mach4's available features depend on the hardware developers support of those features. Meaning different hardware results in different available features.
Very few Mach3 customers have upgraded to Mach4. Many of us waited years for Mach4 to arrive, and were very disappointed in what we saw. Most of the biggest Mach3 supporters have moved on to other controls.

[Outlawtaz]

I have yet purchase anything from them, but automationoverstock.com is very interesting and has good pricing. They purchased the remaining stock of the US warehouse of HIWIN rails and they have some NSK ball-screws as well.

I will have to take a look. Thanks for the tip.

Ger21 - Interesting history on Mach. I am trying to relate my knowledge of production machines with Fanuc or other proprietary controls (mostly Haas) into my own creation. I remember when we got a Hurco in the machine shop in high school back around 1990. We thought it was the coolest thing ever since we only got to see CNCs in full production shops and we now had one to play with in class.

What do you think are the better bang for the buck programs?

[wizard]

The issue of control software / hardware apparently is a very personal thing so when you ask questions focused on which is best you will likely get widely varying answers.

For example the absolute cheapest route, for a respectable controller, is likely GRBL running on Arduino. It is very limited due to the micro processor it is built upon. Do realize though that ports to ARM controllers exist. My suspicion is that you wouldn't be happy going this route due to exposure to high end controllers but this is the viable low end.

LinuxCNC might be comparable to what you are familiar with capability wise. I do however believe that they have painted themselves into a corner trying to run realtime off commodity PCs. This mainly due to issues if I/O on modern PCs. In any event well worth a look.

I have to agree with the above comments related to Mach 3/4. The only good thing here is support for external motion controllers.

Centroids DIY solutions, including Acorn, are very interesting if you like the Centroid way of doing things. They offer what i would have to say is the closest thing to a professional solution at the asking price point. Several people have gone this route recently so try looking for those threads. All indications are surprisingly good stability right out of the box.

One thing not often considered here is buying a CNC controller directly from China. The costs are surprisingly low for FANUC emulating or "ripping off" controllers. The biggest issue in my mind is communications which has me reluctant.

This is just a quick overview there are probably a dozen more solutions for CNC on DIY machines. This doesn't even consider all of the reprap controllers out there.

Lastly you could always buy a used Fanuc controller considering that you know your way around them.

[wizard]

Ger21 - Thanks for the info. What are your thoughts on electrical components?

Wizard - I have access to anything I want: if I don't already own it, I know someone that does. I was a field tech on machine tools for 6 years. I know somebody and can obtain access to more machine shops than I can count.

This puts you in an entirely different league than most on this forum.

As for my expectations... I want to build a machine that will hold at least 0.0025". I don't NEED that tight of a tolerance. I only WANT to do this because I know I will not be happy with anything more than that.

Your experience shines through here because you aren't asking for one micron over ten meters accuracy. Given careful machine design, stress relief and a large milling machine you can most certainly hit that level of precision.

The one gotcha is low cost ball screws. This is where a controller that supports screw mapping might be helpful.

1) I spent too many years either working on aircraft with extreme tolerances or dialing in customers machines to be able to repeat insanely tight tolerances. 2) I have too much OCD and will drive myself crazy if I can't achieve a respectable tolerance and repeatability LOL

I agree about using T-slot extrusion. My current thought was to weld a base out of 1018 (most likely) square tubing. Then, I would take the base and face both sides.

This is likely your best choice considering cost and performance. Some guys have done epoxy granite and there are a couple of threads from guys doing carbon fiber builds. Im just not ready for these new materials but i watch such builds closely. Maybe one day i will feel comfortable with the new tech but im just as likely to resort to cast iron if money is no object.

I have debated the moving table vs. gantry thing. I have a bunch of experience working on both so I am well aware of design for both. Personally, I am intrigued by the moving table design. I have the room for a large footprint so that is not a consideration. I just think that having the wide spaced trucks on the bed along with the fixed gantry would make tighter machine.

Moving table machines certainly have advantages with the one huge disadvantage being size. If you mount the trucks, carriages, linear bearings or whatever you call them on the table itself you always end up with unsupported gaps under the cutting head. One idea that I've considered in the recess of my mind is to use three bearing blocks per rail mounted on the frame. One set sitting directly under the spindle

Basically you flip the common arrangement and put the rails on the table and the bearings on the frame. Think about how a knee mills X axis works. This has another positive benefit in that the machines frame can be smaller taking up less space when not operating. Like a Bridgeport you risk banging the table into things in operation but do lower the foot print some.

[Outlawtaz]

Moving table machines certainly have advantages with the one huge disadvantage being size. If you mount the trucks, carriages, linear bearings or whatever you call them on the table itself you always end up with unsupported gaps under the cutting head. One idea that I've considered in the recess of my mind is to use three bearing blocks per rail mounted on the frame. One set sitting directly under the spindle.

Definitely an interesting concept. I might have to mull that one too if you don't mind the re-appropriation of the idea... My initial thought would be to make the table out of ATP-5 aluminum. That way, its already ground to a good spec (0.005" thickness and face variation), its strong as hell and the weight savings would be worth the extra $$$. And you could always skim the table after the machine is assembled if you can't live with that tolerance.

The issue of control software / hardware apparently is a very personal thing so when you ask questions focused on which is best you will likely get widely varying answers.

To answer all your points in that post, I would like to build my own control. I have seen some privately (as opposed to commercial) assembled controls that have a touch screen computer monitor and that intrigues me. I will probably go with something windows based, just for the sheer fact that if I don't like the software, I can always format and start over with another software.

[ger21]

To answer all your points in that post, I would like to build my own control. I have seen some privately (as opposed to commercial) assembled controls that have a touch screen computer monitor and that intrigues me. I will probably go with something windows based, just for the sheer fact that if I don't like the software, I can always format and start over with another software.

Part of what Wizard was alluding to I think is that the hardware is tied to the software.

Your stepper drives, and in some cases, breakout boards are interchangeable, but there's a hardware pulse generator that will only work with specific software.

And control software is very much a personal choice. Some use standard g-code, others use a lot of proprietary codes. Some let you customize to your hearts content, and some don't.

[Outlawtaz]

Part of what Wizard was alluding to I think is that the hardware is tied to the software.

Your stepper drives, and in some cases, breakout boards are interchangeable, but there's a hardware pulse generator that will only work with specific software.

Ahhh, I didn't realize that the pulse generator was tied to the software. Thanks for the info.

This is why I'm asking you wonderful people about this project. I need to learn a lot more about stepper motor controls since all my experience has been with servos...

[wizard]

Ahhh, I didn't realize that the pulse generator was tied to the software. Thanks for the info.

They can be and as pointed out that might be a problem. Surprisingly something low end like GRBL has a large number of GUI solutions for different OS's. This is probably the one big advantage of open systems.

Someplace in the middle are products only "supported" via one interface program on one OS. In any event if your interest is experimentaion choose your I/O hardware carefully. Id think long and hard about Windows in any case.

This is why I'm asking you wonderful people about this project. I need to learn a lot more about stepper motor controls since all my experience has been with servos...

Stepper based systems are actually extremely simple so i have no doubt that you will adapt fast. Once you understand step and direction interfacing to the drives the rest is very similar to older servo based systems. I say older servo systems as a way to separate those systems from servo based drives using step and direction interfacing.

Honestly i dont know when step and direction interfacing servo drives first appeared on the market but from the stand point of the pulse generator such drives look like steppers. The position feed back loop goes back to the drive instead of the pulse generating hardware.

Exit here to say im hoping this doesnt confuse you. The take away here is that most steeper based systems are open loop. I say most because these days there are stepper based systems with encoder feed back.

[Outlawtaz]

They can be and as pointed out that might be a problem. Surprisingly something low end like GRBL has a large number of GUI solutions for different OS's. This is probably the one big advantage of open systems.

Someplace in the middle are products only "supported" via one interface program on one OS. In any event if your interest is experimentaion choose your I/O hardware carefully. Id think long and hard about Windows in any case.

Using windows was just an assumption on my part. I just figured there would be more options for Win than other OS. Even the major manufacturers are using Win based stuff in their control software and have been for about 10 years that I know of. Of course, it all falls back to macros in the basic sense...

I just need to learn it. I have never been an applications guy, I always was a mechanic.

Stepper based systems are actually extremely simple so i have no doubt that you will adapt fast. Once you understand step and direction interfacing to the drives the rest is very similar to older servo based systems. I say older servo systems as a way to separate those systems from servo based drives using step and direction interfacing.

Honestly i dont know when step and direction interfacing servo drives first appeared on the market but from the stand point of the pulse generator such drives look like steppers. The position feed back loop goes back to the drive instead of the pulse generating hardware.

Exit here to say im hoping this doesnt confuse you. The take away here is that most steeper based systems are open loop. I say most because these days there are stepper based systems with encoder feed back.

No confusion. In my career, have seen machine tools go from off the shelf control systems to fully independent motion control, just depends on the manufacturer as to when it happened. Some of the older tool changers used stepper motors so I am little familiar them. The problem is/was they used a home switch for Tool 1 and then either had another switch to know a tool was in position or just used a set number of steps. I just need to wrap my head around using it as an axis instead.

That being said, I know how to set steps per "x" distance. It was how I adjusted axes to match on mills with a Renishaw ballbar tester. Unfortunately, unless I can find someone to borrow a ballbar from, I will have to do it the old fashioned way with a set of calipers.

[Outlawtaz]

Ger21 and Wizard - Let me ask, is there a good place on the web to teach myself the electronics and motion control of steppers? Like I said, I have a great understanding of the actual mechanicals and theories surrounding design.

I just need to learn the CNC interface and motors and possibly update my VFD knowledge.

Thanks again, you guys have been great!

[wizard]

Using windows was just an assumption on my part. I just figured there would be more options for Win than other OS. Even the major manufacturers are using Win based stuff in their control software and have been for about 10 years that I know of. Of course, it all falls back to macros in the basic sense...

Windows is one of those things where is can be great for some and a horror story for others. I've seen automated production lines costing tens of million come to a halt be cause of a Windows update. Also it still appears that trying to communicate to pulse generators over USB is unreliable under Windows, this is why you find many of these pulse generators communicating via Ethernet. Then you have the issue of BIOS or remote management hardware interfering with real time software.

Obviously Windows works with many software solutions, likely helped via buffering in hardware. It is just fair to understand that some solutions operating over USB have had issues.

I just need to learn it. I have never been an applications guy, I always was a mechanic.




No confusion. In my career, have seen machine tools go from off the shelf control systems to fully independent motion control, just depends on the manufacturer as to when it happened. Some of the older tool changers used stepper motors so I am little familiar them. The problem is/was they used a home switch for Tool 1 and then either had another switch to know a tool was in position or just used a set number of steps. I just need to wrap my head around using it as an axis instead.

There are differing approaches but it is very probably that a CNC commercial CNC controller is supporting some sort of PLC functionality or communicating with a generic PLC. That PLC may or may not be controlling the stepper with G-Code like commands. One neat thing about LinuxCNC is that it supports a PLC subsystem.

That being said, I know how to set steps per "x" distance. It was how I adjusted axes to match on mills with a Renishaw ballbar tester. Unfortunately, unless I can find someone to borrow a ballbar from, I will have to do it the old fashioned way with a set of calipers.