Actually Building the Machine!

[Nena]

The first plates are almost done! I hope to continue my build next week but still need to turn the big cylinder. You cant see the details, this is due the light. Ill make new photo's when i receive it!

Christian

 

[vonG]

wow Christian, thats going to be big. Whats the weight of the head?

I see now what you meant with the encoders. Pictures are always better and you know your stuff in Autodesk too I see. Is there anything that you dont know?
Which plate is the one that you have doubt about?

 

[Nena]

Hi there!

From shaft till end of motor +/- 270mm I think its going to weight about 30kg. Hehehe already changed the plate that i had doubt about!

Is there anything that you dont know?

LoL Uhm... hehehehe!

Greets,

Christian

 

[vonG]

haha, humble I see.

Great that you fixed the plate, think you will be better off with this new setup with less axes. Going to be a big machine anyway.

I wish I had your programming skills. By the way, how long have you worked with programming and cnc?

 

[Nena]

Im a programmer for 20 years now i think 8 years electronics 4 years chemestry and 2 years robotics I like robotics more then the rest! I now can put all skills in one box and that was something i wanted to do!

I overlooked the price of the encoders... they will cost around 450 euro's each but ill order them! And i still need a linear encoder but that one is also expensive.

But mayby i order also the linear encoder it makes my programming so mutch easyer! First i need to draw my Y axis ball screw and i have seen some nice aluminium profiles on ebay (90x90mm and 90x60mm) I hope i can get some 3d files from them.

 

[vonG]

Then I understand why this comes so easy for you. I have no doubt about you making this work.
Ive always been fascinated about automation, but lack the complete skills for it.

Is there any meaning in having a linear encoder on the y-axis except for the programming? You are going to have a vertical lap anyway. The x-axis or the z-axis on the gantry is going to be used for the depth of the cut on the lap, or am I wrong?

 

[Nena]

I use the X axis for the depth, the Y axis is the linear guide to the head. The Y axis is only used to move the stone over the lap plate and has no special function its only used to prevent grooves on the lap. The X is importand becourse i can move the head to the plate and that need to be very precies. Im not ready yet with my drawing. I still need to draw the X and the rest. Thats why i need some 3d drawings of the extrusion aluminium. The linear encoder is placed on the X axis, that axis is almost 1 meter long. When it is correct i can let the motors turn to the position and correct the steps to the given position on the linear/rotary encoder. So i let the motor turn till the exact position has been made. Thats why i also want to use Absolute Encoders, they have 1.200.000 positions in 1 rotation. Lets calculate

Motor has 200 steps per revelution
Harmonic Drive is a reducer 50 on 1

So 200 * 50 = 10000 positions over 360 degrees = 0,036 degree per position

But i could use micro stepping so 0.036 / 8 = 0,0045 and 0.036 / 16 = 0,00225 etc

If i say 45 degrees the motor must turn to the +/- 45 degrees and stop, correct and hold.

 

[vonG]

Ok, thought you were going to have the encoder on the y-axis, which gives no meaning.

I saw some calculation about the index axis, needs 616 steps per degree to cover all the angles that the regular 32, 64, 72, 77, 80, 96, and 120 index gear on a faceting machine gives. That is going to be slow.

How fast is your setup going to move?

 

[Nena]

Uhm lets say if i turn the motors with 3000rpm (witch i wont do) will be reduced by 50 is 60rpm for x and y rotation and x linear. y linear is 3000rpm over 5mm per revolution. So that one can go fast. X linear is 25mm per revolution over 60rpm, its length is 900mm 900/25 = 36 revolutions so almost 30 seconds to move from one side to the other ist slow. Lets say im going to treavel a bit slower it would take a minute to move the complete machine. The machine dont have to be fast in the travel. To move the cutting head from 0 to 359 degrees is in 1 second. So if that is doing a 10 second over it with corrections i would be glad

 

[vonG]

Yes speed is of no use, precision is. And precision you got.
Youre not going to use the complete x-axis anyway.

Cant wait to see the machine faceting a stone from rough to finished gem.

 

[Kingsolomon]

I think you will be dissapointed once you finnish it . Not saying it in a bad way . I just think you like the challenge of building something like this . And when it's done , it facets for you ! So apart from the joy of knowing the machine you created can do that , I think you will move on to another machine that may do it better , of course one that you will be building . So I think this will go on forever . No disrespect , just saying , at the end of the day , you just want to make machines . And that's not a bad thing . You seem to be doing well . I'm looking forward to " model 2 " lol . You know you are going to do it .

 

[Nena]

I hope not I want to facet not build anymore, but i need to and i wont spend a another 10k for building a new one lol. I will upgrade the machine, thats a must but i wont build a new one, its too expensive! I want to cut stones professionally! And get my daughter to a university! Building a second wont happen unless.... hehehehe i still have 4kg of rubys to cut lol

 

[Nena]

I saw some calculation about the index axis, needs 616 steps per degree to cover all the angles that the regular 32, 64, 72, 77, 80, 96, and 120 index gear on a faceting machine gives. That is going to be slow.

I think your wrong with that calculation. Indexes are build upon the 360 degrees. Even if you take the 120 the calculation is then 360/120 = 3 degrees per position So in theory your limited in actions. To make a high res sphere wont be able with an index of 96, if you look at a microscope you should be able to see all cutted facets. If tested it with my software if i should hang a index number on my machine then i have a index of 160.000. So i do cover all possible indexes. If i need to use a index of 96 positions it should be 160.000 / 96 = 1667 positions are oke to cut with for each facet. However the best position would be devided by two position number 833.

I think you ment 616 steps per revolution to cover up all the indexes.

 

[vonG]

Im not the one that did the calculation, but it says that 221,760 is the least common multiple of the index values (32, 64, 72, 77, 80, 96, and 120).
That gives 616 steps per degree (221,760 / 360), it will be able to hit every possible angle with this and a lot of points in between.

Think you have to make the calculation with every gear and not only the 96. The positions must match from gear to gear.

Again, Im not the one that did the calculation to begin with.

 

[Nena]

Even when it needs to coverup 616 steps per degree, non of the indexes will get that position.

32, 64, 72, 77, 80, 96, 120

360/32 = 11,25 OK
360/64 = 5,625 OK
360/72 = 5 OK
360/77 = 4,6753246753246753246753246753247 -> Useless! Not a usefull index! Incorrect! Invalid! (Graphical Imposible)
360/80 = 4,5 OK
360/96 = 3,75 OK
360/120 = 3 OK

The 64th has the smallest rounding. 625 thus you need steps from 0,005 to coverup all positions from all indexes so 360 / 0.005 = An index of 72.000 wil make all possebilities. Thats the smallest! But hey! 0.025 works too! 360/0.025 = 14400!!! So an index of 14.400 will cover up every index on a lapedary!

It does not matter anyway, the overal index that is used the most are the 64 and the 96. And dont forget the 5% error rate from a stepper motor and the metal that is used, the temperature, the... the... lol

 

[vonG]

Christian, you are probably right. I havent done the math behind it and the calculations you have given seems to be right.
As long as you can alter the index by microns when polishing the gem, if the facet would needs that to be completely polished, then I guess index 14400 is what works.

The error rate that you speak of and some other things is the reason why most automated faceting machines dont work. Or at least only give commercial cut gems.
But I think you will figure out a way to battle those errors.

 

[Nena]

Already done the most using Harmonic Drives and using absolute encoders. The error rate is now 0.05 degrees in total. Thats far less than 0.1% error. But other problems will come foreward and that is using mechanical parts. I'm happy if im going to cut a stone with an error rate under de 0.01mm. I think that is my goal! Every thing under that will be a problem even the machine from Germany has that problem!

 

[vonG]

Yes, everything can be done to a certain degree. Lower than 0.01mm can hardly be seen trough 10x anyway.
As I said before, I think you will or have figured out ways to make this work.

 

[Nena]

Hello Thomas,

How is it going with your machine?! Do you have some idears to build one?!

Greets,

Christian

 

[vonG]

Have it all made up on paper, but decided to not go along with it. I will never be able to program all the features in it, and it cost to much to have the program made.

Maybe I will build it in the future if I get the knowledge to program it as I which though.

On the other hand its almost as fun to see your machine come to life instead.