Servos jumpy n missing steps....

[bigal]

Hi,
I am currently rebuilding a cnc milling machine which was built using 3x gecko 320's and servo drives. I have rebuilt the control box and generally rewired the whole machine as it was cobbaled together a bit before, however i am encountering a few problems.
I am fairly new to servos, my last CNC router used steppers which i never had any problems with at all, however these servos seem to have alot of issues going at any slow speed.

I started using TurboCNC as i have always used it, but moved onto Mach 3 as i was running out of ideas and found that the drives were occasionally triggering the EStop line (interfearance ?) and also the drives occasionally jump, stutter and refuse to behave in general. I have adjusted the gain and damnping till the cows come home and am frankly out of ideas!

Is there anything you guys can reccomend?

i am wondering weither having long flying loom (2M) from the machines junction box to the control box and routing of the wires may be interfearing? also the drives fault out alot, if they really start to misbehave...

[planeflier]

I am also making the change over to servo’s and experiencing some of the same problems that you are describing. I have gone through and soldered all my connections at the terminals. I thought that maybe I was creating noise or interference at these locations. I can get the motors to spin smoothly one direction, but when I try to reverse the directions the motors spit and sputter. I talked to Gecko and they recommended that I run the motors in both directions too help seat the brushes to the armature. This has helped but not cured the problem.

[bigal]

well this mill was origionally owned by kong by the looks of it, and i am sure, reasing his posts, that it worked fine back then. I have two spare geckos and servos that i may try, but i rather think the problem is either interfearance (there is alot of wires tie wrapped from the machine to the control box and they are all bundled together, encoder and 50 odd volts for the motors etc.. weird one....)

[planeflier]

Nobody else is having problems with tuning servos and Gecko drives?

[Torchhead]

It's not advisable to run the encoder feedback on single-ended encoders alongside any power wires (especially those carrying the PWM frequency from the drives). Shielded wires for the encoders and in a separtate bundle. Ground the shieldeds at one end only.

Make sure any table connected switches (Homes, limits etc) are opto isolated with a separate "floating" power supply (split grounds). Noise at the frequencies involved conducts poorly in the air but quite well through common grounds.

The fact you are getting random e-stops indicates noise. You can set the input debounce in MACH3 to ignore noise but it's just masking the problem.

The cure for encoder noise is to use a differential driver pair. It has no common ground and any noise is "common Mode" and rejected. You can get differential cable sets (they have active devices on each end) to use with your single ended encoders. Check the US Digital site. If the encoders are the root of the problem then the special cables will cure it.

[bigal]

excellent, have separated some wires out and it is alot better, also goign to sort out the shielding.

Just listed my two spare geckos and some other bits on ebay if anyone interested (shameless plug approaching!)

http://cgi.ebay.co.uk/ws/eBayISAPI.d...MESE%3AIT&rd=1

[amorris]

I have recently had the same problem. I went from 43 volts to 76 volts on my power supply and had all kinds of trouble trying to tune the motors. I now have x and y tuned, but z now acts really strange. It works for a while, then goes berserk. I found last night, that i had to untie the shield wire for the z axis encoder from ground and it worked fine. It appears I was getting noise into the encoder signals going to the Gecko. I am using US Digital single ended encoders, which are not nearly as noise resistant as differential encoders.
Before untieing the encoder shield, I tried all kinds of things like replacing the encoder read head, power cable to motor, re-routed cables, tightened connections and nothing worked until I disconnected the z encoder shield from ground. This tells me I have a ground loop, but at least it's working better now.
Never underestimate the problems random noise can cause. You can have a machine that rund fine for days, weeks or months, and all of a sudden you can start getting noise.

Just thought I'd share this info.
Allen

[planeflier]

are you using the cables that came from US digital? I purchased their 6' cables with connector on one end. I was wondering if I should try disconnecting the ground also.

[amorris]

I am using the US Digital cables. It won't hurt to disconnect the shield and give it a try. Ultimately, you do want them hooked to ground but you will have to find your ground loop problem if indeed you have one.
I have seen this on several systems I have built. Sometimes it works and sometimes it doesn't. Noise is a pain to find.
I have also tied the shield to the ground of the power supply to the encoders before. That is not the preferred way, but whatever works.