Rotary table - automation

[TorontoBuilder]

nice skills to have.

now i'm more keen to see what you do

 

[rabler]

nice skills to have.

now i'm more keen to see what you do

Thanks! Be patient, I have 3-4 projects in progress. Hand scraping in a straight edge for use in rebuilding a couple of lathes. Picking up a VMC tomorrow that will get a new linuxcnc controller. A few others queued up that haven't gotten started.

 

[markba633csi]

My coding skills weren't (and still aren't) good enough to do it from scratch so I tried the Chuck Fellows (RIP) version. Unfortunately I had trouble getting consistent results with it- skipping steps and other erratic behavior. Obviously not OK when you are cutting gears. So I dumped it and went to the "dummies" version which worked flawlessly. Better interface too. Couple small code errors (keypad connector pinout) but nothing major.
I suspect there was something flaky about Chuck's pushbutton servicing routine but I never could pin it down. Poor guy passed away before I could contact him...
-M

 

[TorontoBuilder]

My coding skills weren't (and still aren't) good enough to do it from scratch so I tried the Chuck Fellows (RIP) version. Unfortunately I had trouble getting consistent results with it- skipping steps and other erratic behavior. Obviously not OK when you are cutting gears. So I dumped it and went to the "dummies" version which worked flawlessly. Better interface too. Couple small code errors (keypad connector pinout) but nothing major.
I suspect there was something flaky about Chuck's pushbutton servicing routine but I never could pin it down. Poor guy passed away before I could contact him...
-M

ah I did not know chuck passed away. He helped me more than a few times

I made this so many years ago I can't recall the fix, but I had to make a minor code change to get mine working properly.

 

[markba633csi]

Interesting- I always wondered why Chuck's code seemed to work for other folks but not me

 

[rabler]

I'm just glad these newer microcontrollers are programmable in C (or Micropython, or ...). I spent a lot of time pushing assembly for older 8bit ones with limited fairly limited depth stack. A lot more painful than today's.

I got the Covid bivalent shot yesterday, plus a flu shot. Two in one day, and today I'm feeling it. So I decided to play with some CAD work on this. Here's an initial draft done in FreeCad. Key feature is no parts above the table top, or below the table bottom. This servo runs to about 400 rpm at peak torque and about 2100 rpm at continuous torque. A single belt reduction would be ideal but 10:1 seems to be about right for speed, this would give 9:1 as shown. I want to be able to have it turn while milling, this is a RT not an indexer. That gives about 1 RPM max for peak torque, about 6 RPM at max continous torque, and around 7 RPM hypothetical "no load". One "step" translates to .00125 degrees in this config, or 4.5 seconds of a degree. These servos have a torque % output as a PWM signal, so I'm thinking of monitoring the torque in the microcontroller and derating the feed (RPM) based on the load.

Note: The worm is 40:1
This is based on 2mm pitch GT2 pulleys/belts. I may have to go to GT3, the 20 tooth pulley is not much bigger than the 3/8" shaft on the clearpath.

 

[rabler]

My coding skills weren't (and still aren't) good enough to do it from scratch so I tried the Chuck Fellows (RIP) version. Unfortunately I had trouble getting consistent results with it- skipping steps and other erratic behavior. Obviously not OK when you are cutting gears. So I dumped it and went to the "dummies" version which worked flawlessly. Better interface too. Couple small code errors (keypad connector pinout) but nothing major.
I suspect there was something flaky about Chuck's pushbutton servicing routine but I never could pin it down. Poor guy passed away before I could contact him...
-M

Software debounce on the keypad? With the combination of using 4x4 scanning plus software debounce that can get pretty messy.

 

[markba633csi]

That's what I'm looking for a 40:1 table, then I can use the same Arduino file and just change the dipswitch on the 8825 drive chip for different microsteps between the indexer and the table. Unfortunately most tables are either 72 or 90 to 1 I think

The Chuck Fellows version didn't use a keypad, just 4 buttons with a resistor divider; it sent voltages representing the buttons to an analog Atmel input. THAT was messy- he used up all the I/Os for the parallel LCD display.
The "dummies" version uses a regular 4x4 keypad and a serial LCD interface- mucho cleaner and better

 

[rabler]

Unfortunately most tables are either 72 or 90 to 1

I better double check mine tomorrow. I’m thinking the bottom of the table looks like more than 40 teeth. Unless it is a multi-start leadscrew/worm, that would mean it isn’t 40:1. For the servo rpm, with 90 to 1 I’d ditch the intermediate pulleys and live with 3:1 there.

For my 10ee DC controller I used a rotary encoder rather than a keypad. Tempting to couple that with a simple touchscreen.

 

[markba633csi]

That would be cool. Yeah I think 40:1 tables are not common (I'm not sure) so I'm curious if yours is/isnt
Although 12" would be too big for me- I want maybe 10" max
You would think the tables that can use divider plates they would be geared like dividing heads: 40 to 1