Printable View
I will be using G320's with these Keling Technology KL34-180-90 servo motors:
http://www.kelinginc.net/KL34-180-90.pdf
Can anyine tell me how to set the limit, damping and gain pots on the G320's? I do not have a scope for tuning. I was wondering what other user's of this motor had their pots' settings dialed in at.
Tuning will vary based on the table mechanics and motors. One person's settings my cause your table or motors to break into violent oscillation. :eek: When you get everything hooked up and mounted turn the current limit full sup and set the gain and damping at at 11:00 o'clock position.
Turn the gain up until it gets unstable. Turn it back down and advanced the dampening a few degrees and turn the gain back up again to instability; keep walking it up. After you get it to a point that has good "stiffness". Run the axis through several rapid moves with direction change. That will tell you if you have the gain too high. Too much dampening can cause problems so it's a dance with both controls and they interact.
Unless you know what you are doing with a scope and don't mind mounting remounting the drives (to get the covers off and on), then tuning by hand is the best way.
It's not so critical that it could cause any damage and you may find the default 11 o'clock starting place is fine. You have to have them on you machine to tune them correctly. You can do it sitting on a bench but it's not much better than just running them at default.
For those size motors I turn the current limit to about 2/3 fully clockwise.
Tom Caudle
www.CandCNC.com
Thanks for the feedback Torchhead. I've read about "tuning by hand", but wasn't sure as to how to go about doing it. Thanks again.
Using G320's with Mach3
Can anyone please tell me if the steps per (motor tuning, default 2000) effects the oscillation and creeping that i just cannot seem to get rid of?
I have 250 line encoders and from my reading this get multiplied by 4 times so I have set my steps per to 1000? Is my logic correct or do the reduction gears im using to attach my servos to my lead screws also effect this setting?
Because if I run the axis calibration tool (settings tab) I notice if accepted it changes the steps per setting also, explanation please?
I originally assumed that the software had nothing to do with it as when i power the unit on without Mach3 running i still get the oscillation and creeping across all three axis.
I have tried and retried the above method with no joy at all, i did have X and Y tuned but in an attempt to tune Z all three axis now have the jitters.
Please help.
For anyone interested in helping out on this one paste mms://media1.skynetnz.com/mill into your address bar to see it in action.
cheers
Servo motors will "dither" between two line counts on the encoder which presents it's self as a hum and if you try to move the motor shaft it opposes the move qand you will hear the 'hum' change in frequency. Even poorly tuned drives should not creap or move. The two things that would cause the movement are the two things that cause the motor to move under control: Step pulses or Encoder changes. Noise in either place can be seen as the drive as commands to move. Try putting a .01 cap across the step and common pin at the drive and see if the problem goes away. Make sure you have a filter cap close to the drive across Mtr+ and Mtr -. You did not mention if you have a buffered breakout card on the PC. Trying to run straight off the port will give poor results especially in the 300 series Gecko's
The final approach is to deal with the encoders. Depending on the distance between the encoder and the drive you may be picking up the motor PWM signals or other outside noise. Even shielded cables may not fix it in all circumstances.
On all of our EZPlug Gecko cards made for 320/340's we put differential receivers and provide a differential driver 'pigtail' for the motor. It raises the noise immunity of the encoder circuits by a factor of at least 10. We use unshielded twisted pairs (standard network cables) as the interconnect.
You can't tune out noise with the Gain and Dampening. You can only make the drive less responsive and that will effect operation.
Your calculation for steps is 250 * 4 * gear/belt reduction ratio. If you use rack and pinion you have to calculate in that ratio as well. It's ususally a distance INCREASE which will cancel out some or all of the belt reduction.
In the end you want to know how far the load travels in one rev of the motor. Divide that into one inch and multiply by 1000 and you have your answer.
TOM CAUDLE
www.CandCNC.com
Hi, thanks so much for your reply, I’m using a Sound logic breakout board like this one
http://www.campbelldesigns.com/files...de-rev-1-8.pdf
Now that i know where to look i have a couple of different ideas:
Im using 6 core alarm wire for my encoders, im thinking my cables are to long (about 1.5 meters, approx 2 yards) will making them sorter help?
I will pick up some .01 caps today, what size caps should I use to go across the M+ and M- terminals on the gecko?
Also I have worked out that with the reduction gears and the lead screw one revolution of the motor give me 1mm of movement on the table (we work in metric in this part of the world) so my assumption is 1/1x1000=1000? Sound right?
thanks so much
Run the encoder wires away from the Motor cables. Use shielded wire and ground the shield at one end ONLY. 1.5 meters should be close enough that you probably don't need differential signaling.
Use 450 MFD 100VDC caps or larger on the mtr+ and Mtr - at the Gecko. The .01 are for testing only they may cause problems at higher pulse rates but they will pinpoint the area you need to concentrate on. If you find that it's noise in the step & Dir signals from the Campbell card then be mindful of the way the wires are strung between the Gecko inputs and the BOB. They too need to not run close to the motor outputs. shorter is better.
There has been much debate about if it's better to ground the motor supply (NEG) to the controller box chassis. We come down on the side that motor noise can get into the chassis and into the circuit ground unless the chassis and Gecko drives are well grounded (not just the AC safety ground). We build all of our power controllers with the GECKOS and heatsink isolated from the chassis and the power supply ground not tied to anything but the Gecko Mtr - .
The numbers still work the same way. Just put every thing in MM and seconds. If 1 rev of the motor = 1mm of movement then you have 1000 pulses per mm
. Depending on your target speed you need to determine how many pulses per second you need. Lets say you want to move at 2500mm/min or 41.666 mm/sec. You need to supply 41,666 pulses per second to move at that speed. MACH will do that (and more) but you need a faster computer than the 1G min processor.
Shorting cables will help but more important is how they are routed. Need to keep them several inches away from any motor power wires.
tomCAUDLE
www.CandCNC.com
BOB's
Digital THCs
DXFTool Software
Complete Electronic Packages
Stepper & Servo Systems
Hand Controllers
Have replaced the encoder cabling with shielded microphone style cable, the creeping is alot slower and less jerky and now runs in the opposite direction, no idea what is causing this. Moving the power wires around inside the box does not seem to effect it in any way. I will post a picture of the control unit incase something is obvious.
Is there a chance that the Campbell board is faulty, it is the only thing that was not new and I do seem to remember that when assembled on the bench top running the motors on a 24v battery bank there was still creeping?
thanks alot mate you've been a big help.
http://www.skynetnz.com/temp/controller.jpg
Ok here is the picture of the controller, you can see the space for the other two geckos, the power circuit is in parallel hence the two caps and bridge rectifiers, 1000va total.
Another symptom is that when the x and y are both connected if I move the y to the right the x will move in the positive direction by itself also.
Back to the steps per question, I set it to 1000 and set the motor up, then when I measured 100mm on mach3 the table had only moved 64mm so I used the axis calibration button under settings which set the steps per field to 1500, after tweaking acceleration and velocity I have the y axis moving perfectly (less the creeping) and 100mm of travel on the table reads 100mm of travel on the mach3 DRO have I set this up correctly? I thought it was supposed to be 1000 steps per is all?
thanks very much
if the image dont come up http://www.skynetnz.com/temp/controller.jpg
Before i forget ive added a couple of ferrite core clamps to both the motor and controller ends of the encoder cables, cant hurt?
Give me a call at 1-714-771-1662 during Pacific Coast biz hours.
Mariss
For anyone following this thread it was discovered that the breakout board was faulty. Thanks to Mariss, another happy customer.