Hello Guest it is April 27, 2024, 03:10:35 AM

Author Topic: Plywood Router w/ closed loop servo control  (Read 122 times)

0 Members and 1 Guest are viewing this topic.

Offline koep

*
  •  23 23
    • View Profile
Plywood Router w/ closed loop servo control
« on: March 11, 2024, 01:44:43 PM »
Hi Everyone!


This is a story of my recent router build and how it almost turned to personal battle.

A while back friend of mine was retiring and figured he wanted a small cnc machine as hobby.

He sent me a pic of a 3018 router with price tag of app.$300 and
another, only slightly strudier asking for around $1300.

Having built 5 machines that are run daily on business production I was quick to get on
my high horse and told him that those are ridiculous prices and that he'd get better and cheaper by building
it himself. I promised to help

I had already done a working closed loop speed control for brushed dc motor based on stm blue pill board and a carton
of reliance-electric e240 motors with heds encoders I saw as error purchase for my future uses so I figured
this should be a breeze.

I figured that the dynomotion kflop controls that I have used on previous builds are overkill for this 12"*16" router
with plywood frame so I went looking for something way less expensive and ofcourse I came across the bitsensor red board.

First build went ok only took me 8 weeks instead the planned 6 evenings.


First test cuts were ok so my friend picked the router up and I told myself welldone and started looking for other projects.

Boy was I wrong!

After a month my friend told me that the machine is getting increasing amount movement errors that at the end made the router
unusable.

I got the router back and dove in to the rabbit hole :(

After 18 months of pulling my hair, 1 usb isolator, 2 different controllers, 2 toasted heds encoders and 3 full wiring jobs
trying to find cause for what I suspected to be emi noise/ground loops
I finally found that my controller PID loop was able to read the position wrong due to how the over/underflow was handled.

After fixing that it now seems solid, runs 6 hours without loosing position, following error on diam 8 mm circle stays within
+-20 counts @ 1000mm/min.

It was much longer and much more educational project that I expected  :)

Some pictures below

Rgds,

Tapio