Learning Curves

Brent Dowell

Administrator
Staff member
Messages
16,540
Location
Reno NV
In my OX build thread, Rob asked me about the learning curves I've been working through. I thought I would put together a separate post on it, as it's kind of interesting topic.

Machine Hardware
The Machine itself is a huge variable. A lot of these machines are unique and there are all sorts of considerations when putting it together correctly. I'm still working through this one. Yes, I've got the parts all put together, but do I have everything adjusted right? My z axis seems to have a little flex. I think it's due to how I have the rollers assembled on that. I think they are too loose, so today I'm going to go out and do just a little dis-assembly and see what I can do to tight things up.

Want to use limit switches? That's a hot topic, some people are able to get them working easily, some aren't. I haven't tried yet, but plan on doing so soon. Quite a bit of research there. The research I did led me to use shielded wiring for all my motor/switch wires. Necessary or overkill? Only time will tell, but much easier to do it now and not have to do it later.

Lots of things to look at like:

  1. Roller adjustments
  2. Table flat and level
  3. squareness of the machine
  4. And a bunch I'm probably missing

Control Hardware
Where to even start with the control hardware, Mach3/4, TinyG, Grbl, and lot of other options I'm not at all familiar with. Each of these has a myriad of cryptic settings that each requires careful research and setting to make sure they are right. Fortunately, there are places you can find on the internet to help determine what they should be. Steps per inch, acceleration, inverse operation, and on and on. I was glad I started with that little laser as it gave me insight how to research what they should be and how to set things up.

Fortunately, if once you get them down, you usually don't have to go back and change them. Bad news, you'll probably forget how/why you set things the way you did.

Want to use a pendant? Lots of options there too, but not all of them will work with all the cards. I've got a little surprise I'm working on with mine I'll probably show soon if I get it to work.

Got a PC? Going to need one. I tried to go the Linux route but ended up giving up and going back to windows. Kept having some issues with the gcode sending programs I used on linux. The ones I'm using with Windows are running fine.


Control Software
Now that you have it hooked up, how are you going to control it? Again, All sorts of possibilities. Some of them might even provide you a configuration management solution for your control hardware. I'm using Chilipeppr right now. I actually like it quite a lot, all except for the fact you have to be on line to use it. But it has been working very well for me. But I've got other backups I can use. I'm sure there are other ones as well. It's open source, so in theory I might be able to download the source and run it on a server on my local network. I might do that, but that could require another bunch of time.

There are ones specific to Grbl as well.


Design Software
Now here's a real pickle. Again, options everywhere. You can design and draw in Inkscape, Sketchup, A wide variety of cad programs. And Vectric, probably the gold standard right now as far as I'm concerned. Oddly enough the learning curve for Vectric was not that bad. Not sure if it's because I've used a lot of other drawing programs in the past, or if it's just exceptionally well thought out, or both. But there are a lot of details in there and I've only begun to scratch the surface. I'm just happy that I've been able to use it well enough to see my electronic ideas manifest them self in the real world.

Project Design
And probably the biggest learning curve of all. What to do with all this technology? Right now I'm consumed with making things for it and for making it work right.
I've got a list of projects I want to do with it.
Guitars
Signs
Tap Handles
enhancements to other machines
Really, you're imagination is the limit.
I'm sure the more I use it, the more uses I will find for it.

Well, I hope that gives an overview of the kinds of learning curves I'm been working my way through. All in all, nothing too hard, but a lot of times your choices at one step in the process can limit your choices later on. Nothing that can't be dealt with, but it all depends on how deep you want to dive into any particular rabbit hole.
 
So here's a specific one I'm going through right now. Seemed odd to me that I had to go through so many prototypes to get the diameter of the circles to work out right to hold the spindle and the dust pipe. So I put together a test pattern to try and verify if the parameters I go online were correct.
Specifically there are parameters that need to be set like 'steps per mm' or 'mm per rev'. Grbl uses steps per mm, and there are a bunch of calculators out there that will generate that value for you. For Tinyg you need to use mm per rev, which should be real easy right? I've got a belt with a 3mm pitch, and a pulley with 20 teeth, so that should be 60mm. Well, not so fast, there are other things at play here like the diametery of the pulley, thickness of the belt. So when I measured 3" run, well, I was about 1-2 32 short over 3"! that's kind of significant.

So, I've been playing around tonight and have settled on a number of
58.63806218mm per rev. I've seen other people with parameters of around 59.5mm instead of 60, but through experimentation, my number has now gotten me to a value that I'm now comfortable with.

Now if I really wanted to get detailed, I could measure any variations on the x versus the y and enter seperate values. I think these values might even vary depending on the tension you have on the belt.

But, I'm not striving for .001 accuracy. It's not as if I'm building parts for nuclear reactors or rockets or anything.




 
Last edited:
I have only my 15 year old ShopBot experience to go on here. I am hoping using Mach 3 and Vectric will be much easier. I am also relying on the huge user base for those programs. Otherwise I would be totally freaked out.
 
... I've got a belt with a 3mm pitch, and a pulley with 20 teeth, so that should be 60mm. Well, not so fast, there are other things at play here like the diametery of the pulley, thickness of the belt. So when I measured 3" run, well, I was about 1-2 32 short over 3"! that's kind of significant.

So, I've been playing around tonight and have settled on a number of
58.63806218mm per rev. I've seen other people with parameters of around 59.5mm instead of 60, but through experimentation, my number has now gotten me to a value that I'm now comfortable with.

Now if I really wanted to get detailed, I could measure any variations on the x versus the y and enter seperate values. I think these values might even vary depending on the tension you have on the belt.

But, I'm not striving for .001 accuracy. It's not as if I'm building parts for nuclear reactors or rockets or anything.

That type of measurement got me going for a while, too. My SO2 uses belts and pulleys with 2mm pitch and the system calculates to 40 steps/mm. The chief engineer at Inventables insisted that was the only number that will work. There was quite a discussion about it on their forum and I kept throwing real world back at them. Finally, the CEO of Inventables made the comment that they would do more investigation in their lab. There has been no more pushing of the 40 number! My system ended up at 40.04 steps/mm on both X and Y. After checking my long (X) axis over a distance of 50" and getting a number, I applied that same setting to Y and got correct movements. When I was confident of the calibration, I did a test cut with a square-circle-triangle pattern I created and it came out dead on.

Don't forget to check the Z axis carefully, too. Obviously, there's not as much distance to work with, but I was noticing a bit of a variation in depth of cut from what I wanted. I installed a NEMA23/ACME lead screw update on my machine, which altered the setting due to a 2/1 pulley ratio. The calculated setting for my Z axis was 94.488. After checking carefully, I had to change that to 94.5 steps/mm to get 100mm of movement with a command of 100mm.
 
I have only my 15 year old ShopBot experience to go on here. I am hoping using Mach 3 and Vectric will be much easier. I am also relying on the huge user base for those programs. Otherwise I would be totally freaked out.

Mach3 actually has some functions for calculating the steps, you give it the distance to move, enter in what it actually moves and it will calculate what steps you should run. I did it over as long of a distance as the axis would do (give or take an inch). For backlash, there is a software compensation where you can enter the backlash for the axis and it will adjust for it while running.
 
Thanks Brent and Bill, this is exactly the kind of thing i wanted to know about these machines. Now i wonder how Leos new machine will be in this regard.

BTW when it comes to limit switches I notice many use things like micro switches. But i think a better solution would be industrial proximity switches. I think the point at which they would trigger would be more repeatable once setup and not dependent on a lever activating a switch. Of course they more expensive to deploy than a micro switch but you get what you pay for. Something like this
 
Thanks Brent and Bill, this is exactly the kind of thing i wanted to know about these machines. Now i wonder how Leos new machine will be in this regard.

BTW when it comes to limit switches I notice many use things like micro switches. But i think a better solution would be industrial proximity switches. I think the point at which they would trigger would be more repeatable once setup and not dependent on a lever activating a switch. Of course they more expensive to deploy than a micro switch but you get what you pay for. Something like this

A lot of cnc folks are using the hall effect sensors with magnets, which is pretty much the same thing, but not as expensive.
 
Don't forget to check the Z axis carefully, too. Obviously, there's not as much distance to work with, but I was noticing a bit of a variation in depth of cut from what I wanted. I installed a NEMA23/ACME lead screw update on my machine, which altered the setting due to a 2/1 pulley ratio. The calculated setting for my Z axis was 94.488. After checking carefully, I had to change that to 94.5 steps/mm to get 100mm of movement with a command of 100mm.

I'm actually pretty confident on the Z axis. It's got an 8mm pitch screw with a direct drive NEMA 23. I will double check that with a digital caliper though.
 
A lot of cnc folks are using the hall effect sensors with magnets, which is pretty much the same thing, but not as expensive.

Yeah that will work as long as magnets don't move and circuit making use of the hall effect sensor deals with hysteresis adequately.

What you have to consider with these things is how much vibration you get from machine movement in operation versus hysteresis sensitivity. But a little test and some fiddling should sort it out. Certainly cheaper than industrial prox switch also depends on what is available for input and power.
If the machine were milling steel though then it would be a no no given magnets would attract fine dwarf.

You going to give it a try on your machine?

Sent from my SGH-I337M using Tapatalk
 
I cannot help a great deal on the learning curve dialog as I went through in in the mid 80's and in a much different environment.

Everything is just soooooooo different today. Even from when I got my machine in 2007 the world is different today.

At work, I hand program G-code. I don't even use CAM at my current job. Some of the machines are conversational. I also do a lot of macro programming using logic. It's just a different world. I have LOTS of Cad and CAM background with programs in the 10k to 75k dollar range. Still different. Vectric cannot do what they can do.

***************

About limit switches - I would recommend it. Honestly, I don't know if my new machine has then. Could be yes, could be no.

My current machine does not.

Mach3 has soft limits which work well - when you use it, but if your machine is really whacked out due to lost steps - the soft limits may not help.

I think even simple micro switch's wired in series that triggers the "E" stop would be fine. That is a really simple setup. It will save the machine from driving the axis into the end of the machine. If my machine is not so wired, I will probably do that. I have had machines in industry wired in that way ind it works fine. It's just like hitting the "E" stop. In fact that is the circuit to use.
 
A lot of cnc folks are using the hall effect sensors with magnets, which is pretty much the same thing, but not as expensive.


I use micro switches myself - honestly? they aren't that off as far as repeatability. They're more repeatable than the backlash in my system -- when i tested using a dial indicator, they repeated to less than a thou -- maybe for a milling machine i could see going more advanced... but i just can't bring myself to go much beyond the little micro switches. I bought a handful of HES transistors and just haven't been motivated to install them at this point since the micro switches work so well.

I can home my machine, set work zero, save it in mach3. Shut the machine down, come back the next day, do another job, then re-home and return to that first saved work zero and resume a cut without any visible deviation. So i wouldn't really get too hung up on having the most precise things if lesser precision will work just fine as well, personally.

On the subject of having or not having - they're not optional as far as i am concerned. A machine without them requires a whole different method of work and you can't ever lose steps or stop the job midway unless you have a very precise means of re-setting zero to the work. I find homing and zeroing at the start of the job a much smoother workflow, but there are different strokes for different folks, I suppose.
 
I guess I was confusing the difference between homing switches and limit switches.

My current machine has homing switches - yes.

When I home out the machine the switches are contacted then the axis moves off by a set value. I think they are just micro switches, and it is a manufactured machine.

My comment about using micro switches for limits was a different topic. I was referring to the opposite end of the axis.

My bad.

Still - limit switches in the way I mentioned it - is a good idea.

Homing is a MUST on these machines. If you DO have a situation where you loose steps, homing is the save that you will need to get back into production.

When you set your work coordinate system is is a fixed position from the home position. If you loose steps while in process, you can home the machine and reset the coordinate systems.
 
I guess I was confusing the difference between homing switches and limit switches.

My current machine has homing switches - yes.



One in the same on my machine ... my limit switches are also my homing switches. I home to one end of each axis using the limit switches on those ends as homing switches. Not much point (in my process) for them to be separate.
 
Top