Author Topic: Automatic tool changer for the lathe  (Read 34840 times)

Offline dave benson

  • CNC Jedi
  • *****
  • Posts: 1345
    • View Profile
Re: Automatic tool changer for the lathe
« Reply #105 on: July 26, 2019, 09:09:56 am »
Thanks Bubba, I think I've finally achieved the goals I stated in the first post (only took two years ::)).
It's had it's up and downs (the project), there were times where I ran up against some problems
which seemed unsolvable at the time which caused me some anxious moments (sometimes for weeks and weeks)
I think that in the end it was a triumph of Dogged Persistence over Engineering Flair.

One more thing I'm  going to get the turrets to do their own touch off procedure  for the tools and parts.
I've also had a look at getting the turrets to estimate the surface finish so perhaps they could tell if
an insert has been chipped and also getting the turrets to use the cameras to get an approximate
estimate of position to mark the start of a probing routine for touching off.

Offline EddyCurrent

  • CNC Jedi
  • *****
  • Posts: 4468
    • View Profile
Re: Automatic tool changer for the lathe
« Reply #106 on: July 26, 2019, 09:34:06 am »

As Bubba said, it is most impressive.
The project covers a lot of disciplines, mechanical, electronic, software, etc. and to my mind there's not going to be vast numbers of people who can cover all that to the level you have.
Made in England

Offline dave benson

  • CNC Jedi
  • *****
  • Posts: 1345
    • View Profile
Re: Automatic tool changer for the lathe
« Reply #107 on: July 26, 2019, 12:06:30 pm »
Thanks Eddy
Coming from you, whose opinion (After seeing your build blog and knowing you from here) I esteem to be
credible knowledgeable and honest  means something.

I think the hardest part was, to be able to make a set of ratchet wheels for example, and have them be
interchangeable with each other, this bought out deficiencies in my machining skills and knowledge and  the mill
itself not to mention a lack of adequate measuring equipment.

I watched some documentary videos about Whitworth,Mulsday,Watt,Vaucanson,Whitney,Remengton,Ford and Oldsmobile.
I didn't know the role Ford had with the Johansson blocks.

I have some better quality measuring equipment now and one nice surface roughness tester and the stand for
sensor. made is Sweden but has not been calibrated for 10 years  I tested it with a good set of feeler gauges and
reads OK. Analogue Dials that read out in real microns not internet one's.

The biggest single improvement came 'Above All Else' from using the high preload 25 mm Z Axis linear rails and affixing a backing
brace 12 mm thick by 130 wide by the length of the whole Z Axis Column to the rear of the Z Axis, the improvement was obvious
as soon as I ran a cut the mill was lot quieter and the finish improved markedly, what this improvement meant was I could run
tools (for roughing) on a particular job from 150-160 mm\m to 220-240 mm\m   the ratchet wheels are 1045 so I cut those out at 200 mm\m
leave the right amount material for a finishing cut  (too little or too much) does effect the final sizing.
I've been almost exclusively using TiAn cutters as they are reasonably cheap now and last quite well.
Obviously I had to add the little compressor that runs off the Z Axis to run at those speeds and this works well down to 6 mm depth
When the slots get down to 8 mm the finish goes off as the chips can't get out, so at this point if I have to go deeper say 12 mm DOC
then (If the sizing is important) I'll will use the cut width property and add a mm or two, this really helps evacuate the chips even down to 16 mm
Some good parallels,1-2-3- blocks the 5 inch Kurt clone vice and the tool setter also helped.


Offline dave benson

  • CNC Jedi
  • *****
  • Posts: 1345
    • View Profile
Re: Automatic tool changer for the lathe
« Reply #108 on: August 15, 2020, 04:21:53 am »
An Update.
Three mouths ago,  I was looking around the internet and found this video, where this clever young man used a computer vision system to detect the center of holes on the mill.
I left a comment saying what an impressive job he had done, related some issues I had using
the same camera he was using.
He left this comment.

Kent VanderVelden
1 week ago
Hi David, I'm sorry to be replying late. Thank you for your comment and especially for linking your labor-of-love project! I can appreciate the troubles you have with lighting. I've built and reviewed vision systems used in research. It's common that the greatest variation in results come from differences between "identical machines" or different positions in the same machine or time of day. In one case, sunlight was reflecting around two corners and into the instrument. These problems then dwarf the signal the researcher is looking for. And camera-auto stuff is super annoying when doing analytical work. When time allows, I would like to introduce more machine vision concepts to people with a tool identification vision system. I hope you'll not think that I'm copying your work. I've added your comment to my notes to be sure to include references to your system. Drop me a line if you make an overview video of your project. Reading your post, it's clear that more people need to learn of your work. Best wishes, Kent

After receiving the comment I went back and had a look at the last video I made and it's six generations of the turret
old, so I'll when time permits I'll post another updated one.

The two things I'll say here are:
The "Prime Directive" of the project was "DO NO HARM" That is to say under no circumstances can
these fully autonomous turrets Start the Spindle.
They will keep the spindle running while checking an error condition ( it will try to fix the problem itself), but if
the error is, for example that the called for tool is not in the Alias list or in the turret then it will stop the spindle
and wait and it's always up to the human to start the spindle.

The second thing is, if at all possible do not rely on one sensor input, you should validate or corroborate your data
with either multiple sensors or use logic as these turrets do, for example if these turrets receive a bad tool id from
the vision system then the first thing they check is, "have I checked this tool before while the lathe has been
running this Gcode and the Gcode is still running, then the human could not have changed the tool as they would have
to stop the lathe to do this. So the turret will ignore the faulty video ID and continue on.
In this circumstance the tool may have been covered with swarf.

Anyway I wish Kent well in his endeavors and will post some updates as time permits It'll take a month
or two as I'm doing the multi Axis grinder project now.