Re: [CAD_CAM_EDM_DRO] Insulating D-shell from chassis?

Hi Randy;

To bad I'm living in NY, this sounds like a problem that being there
would simplfy, you do however provide some important information.

First I will restate what I said before, it is not good practice at all
to isolate a metal connector from the casing it is mounted in. You do
not want to do this on either housing. Not only do you have a safety
issue, you present to your system with other potential noise issues.

The first thing I'd do is to make sure that the two housings are at the
same ground potential and that proper internal gournding has been
implemented. That is the use of star arrainged ground runs within
each panel. This should help with some issues. If the problem
doesn't go away they you will need to check the cable running between
the two devices.

If there is an electrical connection between both shells of the cable
then this has to be broken. This is not easy on a molded cable.
Another gotcha, that may be a problem if you don't know the origin of
the cable, is the potential that one of the pin conductors is also
attached to the shells. These connections may be intentional or a
cable fault. This of course has to be dealt with.

At times when running into ground loop problems I've taken the approach
of buying a molded cable and cutting off one end and reterminating.
This is often the quickest and safest way to get a ground loop free
cable. If you're looking for a high quality cable connection consider
getting DB connectors that use the MILSPEC circular crimp inserts. The
crimper is a little expensive but in the end seems to result in the most
reliable connections and keeps the individual wires well isolated.

It should be noted that you are not the first to have an issue with a
spindle drive causing servo issues. Years ago I had an issue on an
old bandit controller that would cause the axis to loose about 2.5um
everytime a custom spindle controller was shut off. Off course this
did not happen on every machine. In the end it was determined that a
relay mounted on the spindle drive PC board was causing the problem and
there was some distance between the spindle drive and the old bandit CNC
controller. In the end putting a snubbing diode across the relay
solved the problem. This was after, of course, an expensive effort to
rearrainge grounds, reroute cables, add isolation transofromers and all
sorts of other mods. In the end a simple and VERY CHEAP diode solved
the problem completely.

This points out several important things. One; don't rely on your
independant vendors to sell you equipement suitable for automation.
Two; every effrot should be made to apply good electrical noise
reduction practices at the source. Three; what appears to be complete
electrical isolation on the schematic does not mean that in reality.
Four; sometimes you can appear to correct a problem, through different
stradeges and yet over time be proved wrong again and again. Five;
the final resolution to a problem is usually too cheap to tell your boss
about!

Thanks
Dave



zephyrus@... wrote:

>At 11:14 PM 11/16/03 -0500, Dave wrote:
>
>
>
>>The last thing that you want to do is to isolate a chassis mounted
>>component from ground. My question is: do you know for sure that you
>>have a ground loop?
>>
>>
>
>Oh yes, indeed. This is from the "random motor steps" discussion from a
>month or two back. After properly star grounding the driver box, using
>shielded motor cables (terminated only at the box end), making sure I was
>using a proper parallel cable with all 25 leads and good shielding, I was
>still getting occasional random steps on all 3 axes, especially with the
>spindle motor running (oh, and I retrofitted the Sherline spindle control
>box with a shielded power cord).
>
>Isolating the DB25 connector from the driver box chassis did the trick. It
>is rock solid now.
>
>Best regards,
>
>Randy
>
>Randy Gordon-Gilmore ,----.___________ ______________ _________________
>ProtoTrains // = = === == || == == == = || == == == = == =|
>Rio Vista, CA, USA /-O==O------------o==o------------o==o-----------o==o-'
>zephyrus@... http://www.prototrains.com
>
>
>
>
>
>
>
>