Hello Guest it is August 08, 2020, 03:09:58 AM

Author Topic: DC Servo run away  (Read 4804 times)

0 Members and 1 Guest are viewing this topic.

Offline les

*
  •  5 5
    • View Profile
DC Servo run away
« on: April 29, 2006, 05:10:24 PM »
While routing on wood ,every now and then the x or y axis runs away over the router table. I have Win XP & Mach2 installed on a 1.3ghz Athlone,I did setup XP according to XP optimization txt file. I have tried with another PC 3Ghz and 500MHz Intel and I have the same problem. I have taken a storage scope to monitor the step pulse and found that there is a inrush of pulses during the moment the x or y axis runs away.These pulses have about 3x the frequency of the normal step pulse.Does anyone know why this could be happeningĀ  .
Thanks
Les

Offline fdos

*
  •  326 326
    • View Profile
Re: DC Servo run away
« Reply #1 on: April 29, 2006, 06:03:48 PM »
Hmm What drives are you using for these servo's?

Sounds more like encoder loss runaway, or if it is on the step input side then you are getting a lot of noise on the step lines.

Wayne...

Offline les

*
  •  5 5
    • View Profile
Re: DC Servo run away
« Reply #2 on: April 29, 2006, 06:49:43 PM »
Hi Wayne

Thanks for your quick reply....!

The drives a homemade . I had pcbs made , I am using a Pic18f252 with encoder circuitry , the final output drives a h-bridge mosfet irfz48 x 4 .The 12vdc 10amp motors are connected via shielded cable aswel as limit switches and encoder using separate shield cable and a star point grounding to earth.
The code I wrote using ccs c . I also wrote a vb program that I use to setup the pid and all relavant settings for each drive via RS485 addressable.
I have run tests using the vb program for hours with no errors at all (basically jogging forward 100mm then back repeatedly).Then whether or not with the isolated RS232/RS485 cable plugged in to the PC and drive, I get this run away problem .This happens in random and sometimes not at all.

The step pulses and inrush pulses are clean clean. I seems to happen only when I mill wood in a long straight line of +-300mm.

I am open for any suggestion ....Thanks.

Offline Graham Waterworth

*
  • *
  •  2,113 2,113
  • Yorkshire Dales, England
    • View Profile
Re: DC Servo run away
« Reply #3 on: April 30, 2006, 05:19:53 AM »
Hi,

This sounds a bit like running out of numbers e.g. 65536+1 =0

Graham.
Without engineers the world stops

Offline les

*
  •  5 5
    • View Profile
Re: DC Servo run away
« Reply #4 on: April 30, 2006, 09:05:02 AM »
Hi Graham & Wayne

I did test the number problem thats fine..!  thanks Graham.
I looked a the step pulses and found the printer port sends out 3.3v for a hi level ,this is in an unstable area because my PIC runs from 5v. I will put a buffer in series to get the step and direction pulse up to 5v.

The noise level is 100mv , this could give a false trigger if the step pulses peaks a 3.3v .....

I will let you know if my idea works....
Thanks.

Offline fdos

*
  •  326 326
    • View Profile
Re: DC Servo run away
« Reply #5 on: April 30, 2006, 09:18:01 AM »
Well thats would have some affect.   There seems to be a lot of these 3.3v ports around now, even some on desktop PC's.

Even better would be to drive your step/dir with differential signals.   Just requires a couple of chips and makes the signals so much more noise immune.  If your drive has opto's on the inputs these may already be able to take dual ended input, as long the the opto's led cathodes are not tied together.

Wayne...

Offline les

*
  •  5 5
    • View Profile
Re: DC Servo run away
« Reply #6 on: April 30, 2006, 12:50:04 PM »
Hi Wayne

I have used a CD40106 hex inverter chip for x,y and z step/dir into the PIC . Now I get 5v signals that look promising.I will test this tomorrow.
You don't perhaps have a relavant circuit for interfacing the step/dir pulses via optos.

Les

Offline les

*
  •  5 5
    • View Profile
Re: DC Servo run away
« Reply #7 on: May 03, 2006, 03:13:15 PM »
Hi

It seems like my problem is fixed .....!

Thanks for the support

Les