KMotionCNC instead of Mach4

[Roger]

Aug 22, 2021 18:01:59 GMT jon38r80 said:

Seems that a large pc case standing beside the pc you are using both facing out of your picture as it were might be a solution. I agree you are squeezing a quart into a pint pot. That's the trouble when you grow a system. At some point you outgrow the original and need to replan it.

I can see that you are running into the usual horrendous cable management problems that you get when you grow a system.

Anther possibility would perhaps be to build the PC and the controllers into a small server frame cabinet but that would be a large amnount of work and expense.,

Your solution makes much more sense as it stands for now.

It's tricky to make it neat, the PC is a ludicrous size really. I imagine I could put together a much smaller unit these days. I think it's on its 4th motherboard, now sporting a Quad Core one with a decent graphics card so I can run the CAD and everything else on the one PC. It's on it's third controller in that cabinet, so that's had its fair share of upheaval. I'm keeping the Mach4 control in case I have an issue with the other one. I'm going to simplify the cables shortly, so I'll show how that ends up later.
In the end, it's probably good enough for now. One day I'll upgrade the PC, and maybe that will end up more compact and with less cables. It doesn't help having a network router in there to add a Wireless Access Point so I can get a decent WiFi signal to the mobile phone.

[Roger]

Now I've finally got back to using this software, I thought I'd share a couple of things with you. I've modified the default screen significantly around the jog buttons. I've also hidden many of the things that I considered to be just clutter.

The screens can be modified using a tool provided. It's a bit clunky, but it does work.

Here are the things I've changed using the screen designer...

1) Removed the Inch and Relative Radio buttons, I'm never going to use them. I can set those things in a program if I ever needed them anyway. I've left the MM and Abs radio buttons so I can see that this is the mode it's in.
2) The Jog buttons have been reassigned so that the Red buttons are fast continuous travel and the Orange ones slower. The arrow step the amount indicated by the Step Size radio button. This is so I can use the mouse on a large target while looking at the job while jogging to an edge.
3) The Single Step button has been made large and moved to the bottom LH corner.
4) The Feedrate and Spindle sliders have been simplified and the Rapid override removed. I don't use rapids, although they are used in some peck cycles.
5) The Home button was added and the supplied program was debugged and the various outputs assigned so it uses the limit switches then backs off to use the marker pulse on the Linear scales.

I've also made some changes to the Application itself, which I suspect isn't often done. However, the source code is supplied as part of the download, so you can do what you like with it. This is what I've done to it...

1) Changed the Metric displat to only have three digits after the decimal point.
2) Implemented a default value in the Set for X and Y to be half the displayed value.
3) Changed the override values for the Feedrate and Spindle speed. You can now override the Feedrate to x5 instead of x2. The sliders aren't that good really, they don't work as well as proper Windows ones. I may improve those when I get time.

Some things jump out as being very different to Mach4, most better, some worse.

1) The path display isn't as good as Mach4, that could use some work. I might have a go at that myself in due course. You also have to run it as a simulation. That's not very convenient, but it does execute quite quickly.
2) I've set it up so you do need to home it when running for the first time. It does remember the last offsets though, so I can find my way back to exactly where I left it. That's a big plus.
3) It's nice and snappy with the jog buttons, just like Mach4 used to be before they messed it up.
4) The way the programs load is nicer. It makes a copy of the file rather than hold it open like Mach4. This means you can recalculate paths using CAM and save them while KMotionCNC is running without it complaining that you can't save the file.
5) You can load up to 7 programs at a time and select which one using the radio buttons on the RH edge. That proves to be really useful when you're doing several operations. You can load them in sequence and then just use them from the control instead of reloading them from the folder.
6) Manual command line is like Mach3 used to be, ie a single line, unlike the painful one in Mach4. This one also remembers all of the outputs in a dropdown list which is handy. I might add another button to go to X0Y0Y0F1000 which is something I do all the time. That could go up near the Home button.
7) The Spindle keeps running when the job halts or pauses, I need to find a way to make that stop then resume, because it's a nuisance to do that manually.
8) You absolutely must not press the Stop button unless it's an Emergency. That's because you will have to restart and rehome the program. To be honest, you could probably remove that altogether, but I suppose it's handy when you're shutting down the program. It's the only way to disable the drives. I don't think it disables the drives if you close KMotionCNC, I'll have to check. That needs to be fixed.


KMotionCNC modified screen by Georgia Montgomery, on Flickr

So overall, it's a pretty clean interface now I've done a bit of work on it. I'm sure it will evolve over time, bit it's certainly usable as it is.

[Roger]

This is before the changes...


20210821_104817 by Georgia Montgomery, on Flickr

... and this is today's effort. The hanging flying leads from the CM106-ESS board at the top right have been replaced with simplified shorter cables to the interface board underneath it. The same connectors are now used for either setup which reduces the spaghetti considerably.

Now all I need to do to change over controllers is to plug the Servo cables into the other one, do the same for the Frequency Converter, and move the two DC Brake wires across. In all it's less than five minutes work.

It's still messy, and I need to put the cables into the trunking, but it's good enough for now. At least I have a plan B if something goes awry.


20210823_183659 by Georgia Montgomery, on Flickr

[Roger]

What you can't see from the pictures of the cabinet, is the way the CM106-ESS module is clipped to the DIN rail. I've found these to be very useful for quite a few projects. The idea is that you can slide a PCB in from the end, and extend them with the location pins. There are obviously end caps too.

20210824_095452 by Georgia Montgomery, on Flickr

The backs can be flat or have a DIN rail clip. They aren't cheap, but it's a neat solution.

To be honest, I'd almost certainly design something that could be 3D printed to any length and configuration these days.

20210824_095531 by Georgia Montgomery, on Flickr

[jon38r80]

one of the things that makes your cabinet look worse than it probably is is all those plugs that have a mess of single cables going to them. It might help to put some spiral wind sleeving on them , self almalgamatingf tape or that weird cloth tape used on car wiring harnesses. its probably too late now to use that braided plastic sleeving that fits a range of bundled singles unless you can depin the plugs or you intend cutting and shutting more of the wiring.
Starting to look a lot less like a cats cradle now though.
Much better.

[Roger]

Aug 25, 2021 10:17:04 GMT jon38r80 said:

one of the things that makes your cabinet look worse than it probably is is all those plugs that have a mess of single cables going to them. It might help to put some spiral wind sleeving on them , self almalgamatingf tape or that weird cloth tape used on car wiring harnesses. its probably too late now to use that braided plastic sleeving that fits a range of bundled singles unless you can depin the plugs or you intend cutting and shutting more of the wiring.
Starting to look a lot less like a cats cradle now though.
Much better.

You're absolutely right. I ought to get some of that spiral binding, and put the wires at the bottom in the trunking. I also ought to put another piece of the trunking between the breakers and the DIN rail, although that part isn't too bad.

I've just bought the smallest size of this type from eBay it only cost £3.09 for 5 metres, so it's really cheap.

[Roger]

Adding the Linear Scales has really highlighted how worn the Y-axis dovetails are. I can see it hunt considerably on the Y-position if I put a clock on the spindle and against the work. It's not just a matter of adjustment, the whole Y-axis yaws when changing direction. This is only noticeable now because it's controlling position to the Linear Scale which is offset to the leadscrew. Clearly the carriage isn't being held parallel.

I've tried to find a precision prism to check this and allow me to scrape it in, but I can't find one. So here's what I'm going to make from a piece of Mild Steel bar. I can normalise it first, and then machine it all over.

The piece is 50mm x 30mm x 400mm but I only need 49mm x 25mm x 400mm, so I can take a decent amount off the top and bottom. I did try to find Cast Iron of a suitable size, but couldn't find any. As long as I'm careful with how I hold this down and grind it, I can't see any issues. I've got over 400mm of travel on the Tool and Cutter Grinder, so I'll do it on there.

I can 3D machine the slope to get it pretty close to start with, then set it up at an angle in the tilting vices. At least, that's Plan A

Dovetail prism by Georgia Montgomery, on Flickr

[Roger]

The first step in making the Dovetail Prism is to normalise the Cold Rolled 30mm x 50mm bar. I really wanted 25mm thick, but they didn't have it. I'll machine it down, and in the process it will help remove any residual stresses.

So here's the temporary furnace made from Thermalite blocks.

20210902_094619 by Georgia Montgomery, on Flickr

I've stood the part off the bottom with a scrap of insulation to get the flames underneath too.



20210902_094655 by Georgia Montgomery, on Flickr

This is going to take some time...

20210902_094949 by Georgia Montgomery, on Flickr

But eventually, after about ten minutes, and heating from both ends, it ended up bright red all the way along.

20210902_100910 by Georgia Montgomery, on Flickr

I then squeezed the sizes of the furnace against the piece, and added some insulation on the ends to keep the heat in. It's not as much insulation as I'd like, but it's probably good enough. It's going to take hours to cool down.

20210902_101307 by Georgia Montgomery, on Flickr



20210902_101315 by Georgia Montgomery, on Flickr

[Roger]

Continuing with the machining of the 60 Degree Scraping Prism. this is the first of the faces being skimmed with two 0.5mm deep cuts to remove the skin.

20210906_100158 by Georgia Montgomery, on Flickr

This is the first time I've used these edge clamps in anger. I've had them for years, but never had the need for them. I could have used some of those other neat edge clamps they were selling at Alexandra Palace for the sides. I've just put stops, pressed as hard as possible by hand while tightening the bolts. Not ideal, but good enough.

I don't want to clamp this down tightly, I want it to be unstresses so it's as flat as possible.

20210906_100755 by Georgia Montgomery, on Flickr

Here's a wobbly video showing the feed and speed I'm using. All very gentle and quiet.

20210906_100811 by Georgia Montgomery, on Flickr

The part was flipped over that the other side given the same treatment...

20210906_114235 by Georgia Montgomery, on Flickr

... followed by the edges. I decided to clamp them to some Vee blocks to make sure they were vertical. One will end up at 60 degrees though, and the other isn't that important. However, it will be easier to set up on the grinder if it's all square and flat.

20210906_141210 by Georgia Montgomery, on Flickr

Here's the second edge getting the same treatment.

20210906_152248 by Georgia Montgomery, on Flickr

Then it was back to the flat sides, and a light 0.3mm cut taken to clean it up and get it as flat as possible.

[Roger]

The prism needs a 30 degree chamfer, and usually that would entail setting up the job at an angle. However, with CNC there's an easier option.

Here, I'm using the same 30mm diameter Face Mill to cut a series of steps to rough out the shape.

20210907_122323 by Georgia Montgomery, on Flickr



20210907_140037 by Georgia Montgomery, on Flickr

It's left a bit of a step at the bottom to clean up, but that's fine as this is only a roughing pass.

20210907_161707 by Georgia Montgomery, on Flickr

The 'finishing' pass is just to get the geometry close enough to be able to grind is without removing a huge amount of material.

I'm using a 12mm Ball Nosed cutter, with the CAM set up to leave 50micron scollups. It's going to take a couple of hours even like this, so there's no point in going over the top.

You'll note that I've had to rearrange the clamps to miss the tool holder.

20210907_163147 by Georgia Montgomery, on Flickr

For those of you who are not familiar with using CNC, the tool setup is actually simple for a Ball Nosed cutter. I've set the CAM preference to use the end of the cutter as the reference. Everything is done by dead reckoning from known faces or edges that have been selected on the CAM model. In this case, the back LH edge was the X/Y reference zero and the top was the Z zero. With my simple machine, that means I can treat a Ball Nosed cutter like all of the others, just touching down on the top of the job and clicking zero on the CNC Z-axis DRO. The complex issues of where the tool has to actually go are worked out by the software, you don't have to think about it.

20210907_164555 by Georgia Montgomery, on Flickr



20210907_181523 by Georgia Montgomery, on Flickr

Holding this for grinding is easy for the square and parallel edges, but the 30 degree angle is another matter. I did consider using my angled vice, maybe even using two of them. However, I've opted to use the crude Adjustable Angle Plate that I made decades ago for my Southbend Lathe when I made my first batch of Quick Change Toolpost Holders.

So here I'm adding two M10 tapped holes that will line up with that.

The tap wrench is an Apprentice filing piece from the first year. The job had to be within 0.1mm in every dimension, even the Vees for holding the separate hardened jaws, which were also hand filed. It's taken a bit of a bashing over the years, but it's a very useful tool.
One poor sod never did manage to finish this piece. In desperation they said he could use the Bridgeport Mill to rough out the last attempt. There was a terrible noise coming from the machine, with plenty of smoke, yet he seemed oblivious to it. Turns out he had it running in reverse. Needless to say, he ended up in the office.

20210908_103732 by Georgia Montgomery, on Flickr

The Vernier Protractor ought to be good enough to set the angle, I don't really have a better way to do it.

20210908_110640 by Georgia Montgomery, on Flickr

And finally, a quick sanity check to see it's the right angle and overall size.

20210908_111125 by Georgia Montgomery, on Flickr


So far, so good.

[Roger]

I haven't done anything more to the Mill recently, so I thought I'd better get on with the Precision Prism that I need to scrape the Dovetail Slides.

Fortunately the Jones & Shipman Tool and Cutter Grinder is incredibly flexible, and it serves as a rudimentary manual Surface Grinder with about 450mm of usable travel. That's more than enough for the 400mm long prism.

The crude splash guard is lashed up from bits and pieces and is held in place by a magnet.

20211001_121453 by Georgia Montgomery, on Flickr

It hasn't taken much to quickly clean this up roughly so it can be turned over and ground on the other side. The intention is to grind both sides twice in an attempt to get it as parallel as possible. That will make clocking up the other sides much easier.

20211001_121504 by Georgia Montgomery, on Flickr

[Roger]

I wan't happy with the parallelism of the surface grinding. It normally isn't that fussy because the parts are short. However, this part is 400mm long, so it needs to be pretty good.

Clocking the surface of the swivel table showed that this is where most of it is coming from. Once it had been lifted off, it was obvious why it wasn't that good. It's rusty and generally covered in debris.

20211006_132105 by Georgia Montgomery, on Flickr


That's a bit better!


20211006_133651 by Georgia Montgomery, on Flickr

It's still about 8 microns out over the distance of the chuck, but it probably wasn't any better when it was new. It's not a surface grinder after all. Now, I could spend the next week scraping it, of just live with what I've got. The latter makes more sense. I'll see how it grinds this time.

20211006_135710 by Georgia Montgomery, on Flickr

[Roger]

I've been chewing over some other ideas for maintaining the accuracy of the Mill in the long term, and one idea that crossed my mind was to see if there was any way to convert the dovetial slides to ball slides. I was imagining this as simply being impractical, but this video shows someone doing just that, albeit on a much smaller machine.
Basically, they're completely replacing the Y-axis carriage with a Steel plate that take the side blocks on the Y-axis and the X-axis. In other words, they're replacing both X and Y slide ways with linear rails.

This is definitely something that I could do. The knee might have to stay as it is though, I'm not sure if it's feasible to modify that. Their machine is more like a Sieg, with the head sliding on the column.

Anyway, I'd be interested to know what others think about this idea. It's obviously a lot of work, but it could transform the accuracy of the machine.

[Neale]

I have a Wabeco F1410 series mill with the linear rail option. It uses 25mm rails, which is one size up from what I would have expected. It uses a sliding head with counter-balance weight - possibly needed because of ballscrews which would let the head descend under its own weight when power is off!

Anyway, I believe the manual is available online, and it has exploded parts diagrams showing construction. Probably not directly relevant but it's sometimes interesting to see how someone has done something similar. I did start musing on doing the same thing when looking at a CNC conversion of my Warco VMC, but then the Wabeco came along at an attractive price.

[Roger]

Oct 10, 2021 7:16:27 GMT Neale said:

I have a Wabeco F1410 series mill with the linear rail option. It uses 25mm rails, which is one size up from what I would have expected. It uses a sliding head with counter-balance weight - possibly needed because of ballscrews which would let the head descend under its own weight when power is off!

Anyway, I believe the manual is available online, and it has exploded parts diagrams showing construction. Probably not directly relevant but it's sometimes interesting to see how someone has done something similar. I did start musing on doing the same thing when looking at a CNC conversion of my Warco VMC, but then the Wabeco came along at an attractive price.

Thanks for that, I'll take a look, there may be some useful ideas there.
I fitted an AC Servo with a DC brake instead of trying to counterbalance it. That works really well, because the servo amplifier has built in controls to handle it whenever it's disabled. Without that it would gracefully descend to the bottom.
I think I'd need at least 30mm rails to be rigid enough, preferably 35mm if they would fit.

I've found the online manual you mention, and it looks like they use one extra long unit on the Y-axis each side, and the X-axis has four. They look like they are mounted in a very similar way to the conversion that I found in the video I linked to.

I'd really like to find a way to add a linear slide to the knee, but that's really going to be a difficult one. Maybe I can mount two of those inboard of the Dovetail slides. I could modify the knee inside and beef up that massively with plates to mount then on.