rebuilt 2.5 - stepper to servo conversion

[multiplex]

for the past few months i've been agonizing over how to improve the speed a reliability of my Y axis.

Theres the back ground.

Swapped all screws to 1/2-10 5 start to help improve the speed, but was not happy with the lost resolution

Typically what i find while running production jobs is the origin slowly starts to drift - after about 20 -30 parts its off .020".

Beyond that, my rapid on the Y axis was only about 100ipm because the motors would sometimes stall.

The motors / drives when i first built my machine were poorly selected - live and learn.

Keling 425oz Nema 23's. Require 84v but i'm feeding them with 38v
Keling Stepper drivers which are limited to 40v max

I was up against replacing four drives, and getting a larger power supply.

Even though that would have been cheaper, I decided to try my hand with servos

Here is the machine



Here are the two Y axis motors



After much debate between the Gecko and Viper servo drives, I decided to go with the Viper 95 drive



Servo Motor I selected

Keling KL23-130-60 (NEMA23)

Skewed Rotor Design, 60V/20A
Constant Torque: 50 oz-in
Peak Torque: 350 oz-in
No load Speed: 4700 R.P.M
Terminal Voltage : 60 VDC

I'm still going to be driving these motors with my 38v power supply to help limit the RPM.

My goal is 4:1 reduction, and a max Rapid of 300 IPM. Based on the screw pitch, that will require 2,400 RPM

I searched high and low for an afforadable gear reduction system, but finally settled on this belt system i found on ebay

http://cgi.ebay.com/ws/eBayISAPI.dll...m=250563161378



I'm hoping all these parts show up soon, so i can get things going.

I typically plan my upgrades around new production jobs and I have one coming up in a week or so that I need to be ready for.

First plan is to switch the Y axis, and then continue to convert the X and Z as money allows

[asuratman]

Why not using bigger motor such as KL34-180-90. If you use belt for driving 2 motor in x axis may have problem with synchronizing speed?

[multiplex]

i'm pretty confident the 425oz motor is enough if i was able to power it properly.

That would mean 84v power supply - roughly 150 bucks

4 gecko drives, 150 each

So thats 750 bucks to stay with stepper technology

I decided i'd rather move to servos on this build, and save the motors and drives for a less taxing plasma table i'm working on at the same time

[ger21]

You probably could have gotten better performance by going with smaller steppers. The 425oz motors are notorious for offering poor performance with low voltage, inexpensive drives. They are a very poor choice for small routers.

And lowering the acceleration would probably get rid of you're .020 error. You must have been losing steps somewhere.

I get 170ipm on my Y axis, with 1/2-8 2 start acme, a Xylotex, and 250oz steppers at 24V. And I've run it for 12 hours with no loss of of position.

[multiplex]

I totally agree Gerry. In fact before i went down this road, i bought some smaller motors which better matched the voltage I had available

I found that I could drive them much faster, but they still lacked the torque to drive the gantry as fast as I wanted.

The error really showed up when i switched from 2 to 5 start screws. That reduced the number of steps required and really made the resolution course. I did then bump up the micro-stepping, but thats not always the best solution. I think i'm currently using 1/16 stepping.

So yeah - i definitely could have stuck with steppers, matched the drives and power supply and kept going.

just thought servo's might be fun

I just finished a nice job here at work using Allen bradley drives and motors, and a nice THK ball screw assembly. That thing seriously hauled ass. so i'm a little bit addicted. next i'll be swapping out my acme screws for ballscrews

[multiplex]

got a few items in today, and installed

first half of the belt drive



all bolted up with the Keling Servo motor



I went with 500 count US Digital encoders, here's a close up of the wheel



so with two revolutions per inch, and 4:1 reduction - i get 4000 pulses per inch. which equals .00025 per step

i got anxious and powered up the one motor directly to see how quick the table moved. of course i disconnected the slaved screw first. It traveled 48" in about 10 seconds, which gives me just under 300 ipm. I'll be happy if I hit that number once all is done

these motors are rated at 60v, and 4700 max rpm. my power supply is 38v and i'm estimating around 2400 rpm at this voltage.

[ger21]

500 count encoders need 2000 steps per rev. At 2400 rpm, that's 80,000 steps per second. Do you have a PC that can do that?

[multiplex]

Yes and no.

My pc is good up to about 75k, but I wouldn't want to push it that hard

Viper servo drive has a built in step multiplier, which defualts at four. From there you could go higher, but would start to loose resolution. At four no resolution is lost, an the drive has some overhead to position itself

It was one of the main reasons I went with viper over gecko. Beyond that, it has a few other variables that should make the setup really nice

[multiplex]

well i waited what seemed like forever for my viper 95 servo drives to show up. They finally made it here friday, and a few hours later i was making things move!

please excuse the mess in the photos, the next step will be to redo all of the wiring on the machine which has slowly gotten out of control over the years

I experimented with two drive ratios - 4:1, and 5:1. Ultimately 5:1 won out, due to the slightly higher torque it produced. With 4:1 i was still able to stop the table by pushing on it. 5:1, I was not able to stall

This limited my current IPM to 220, which is plenty fast. To go faster, I need a higher voltage power supply. I currently have 38, and would possibly step up to 58 - however I wouldn't be able to use all of that voltage - the machine would be way too fast

Along with the motor conversion, i also removed, cleaned and dry teflon sprayed my screws. I also figured out why my X and Y were slightly out of square and was able to fix that easily.

good day in the shop!





"http://www.youtube.com/watch?v=BuqncJi2eEQ"]YouTube- cnc router - y axis servo motor conversion

[multiplex]

ok so like every project, this keeps going further and further.

I've decided to move up to a larger power supply (56v), and switch all axis to servo. At the same time, i'm redoing my control box to clean things up a bit

cable carrier for the x axis



one end of the electical cabinet, in progress



I picked up this sweet 8 port usb to serial box on ebay for 40 bucks. Plan is to wire up all 4 Viper servo drives, plus my VFD.



power supply, spindle drive and servo drives



finished end panel, with servo drives and limit switch connectors



powder coated, and then laser engraved



here is the mill i do most of my work on. Milltronics Partner 5. works great!, although it did need some repair to get it going



Hoping to get the panel done this weekend, and some more wiring squared away. maybe by next week it will be up and running

[multiplex]

i guess this thread isn't very exciting

finally finished up all the wiring on the machine this weekend. took way longer than i had expected! but everything fired up without a hitch

best part is the 8-1 usb serial box. i can have a hyperterminal window open with each servo driver connected. I'm hoping that larken comes out with some tuning software for the servo drives, or perhaps i'll write a simple polling routine to grab some data and graph it















only wish now is that the screws were more accurate. maping across the x axis, the beginning of the screw gains about .015" over 10 inches but then settles in, so it makes calibrating the screw hard

[stevespo]

Hey, I think it's pretty exciting. Keep the photos and updates coming. The enclosure and engraved panels look awesome. Glad to hear the performance has worked out well for you. I'm curious if you needed to install any type of filter when the VFD went in the cabinet?

My own stepper to servo conversion has been slow in happening. I finally have the last few enclosure components I need to get my wiring job finished. Everything has been tested on the bench and it looks great. I'm using 300W BLDC servos and the Granite VSD-E drives. I'm going direct drive and looking forward to the performance/precision boost.

How are you mapping the screw accuracy? Do you have a glass scale, or is there a way to do it with the optical encoders and some type of dial indicator? One inch at a time?

Steve