Specific EMI/RFI interference problem

[glentner]

I have searched the forums and found close to, but not the exact problem I am having:
I can run a program without the spindle running and everything is fine. When I turn the spindle on and run the program I'll get either a limit switch or E-stop trigger after running for a little bit. So, that has to be an EMI/RFI interference problem from the spindle motor, right? I already have a EMI/RFI filter hooked up to my spindle motor which fixed a previous problem of the stepper motors "jumping" steps while running with the motor on. I also have a choke (ferrite, etc.) around the power cord right before it enters the speed control for the spindle motor. I have most recently rewired my conduits to separate all DC wires from AC wires but I still see the problem.

My next thought was to swap out the e-stop and limit switch cables with shielded ones. I also have another EMI filter that I thought about putting in before the DC power supply???

Has anybody else had this problem and if so, what fixed it??????

My setup is a home built CNC mill with a xylotex controller, steppers, a sherline spindle (motor and speed control) and PC running mach 3.

Any help/input is greatly appreciated. Thanks.

[Al_The_Man]

If it is a Universal Spindle motor, they tend to be very noisy, electrically, but the main cause I suspect of many of these types of problems is insufficient grounding or poor grounding practice.
If you search these forums it has been covered many times before, I always set up a common ground plate and ground ALL power suppy commons, including the PC PS common.
A ground wire should also be taken from the motor case to controller and then to the ground plate.
Al.

[glentner]

Yeah, I've seen many threads suggesting proper grounding to fix EMI problems.
I'm not saying that I'm skeptical of that fixing my problem......... but I guess I am. I've checked all my grounds with a multimeter and they are all within .6 ohms to my main ground.
I guess I have a hard time convincing myself that that should be the next thing I try. But that's why I asked the question on the forums right?

So if I did that next what are we talking about?
I have a 220V (single phase) line coming in a main disconnect box. I split the 220 into two 110v lines running to two separate electrical boxes. One box is for the spindle and has the EMI filter and some electronics for a tach and future use. The other box has the DC power supply and the stepper drive.
Would I run a grounding wire out of the main disconnect box to a grounding strip attached to the mill frame, then run wires out of both other boxes to that main strip, along with a wire out of the PC? From reading other threads on grounding it sounds like you should also run a wire from your shielded cables to that same grounding strip. In my setup, the cables are attached to the electrical box, so is that good enough?????

I was new to EMI interference a little bit ago, now I'm new to good grounding practices. If good grounding fixes my problems, I will be completely happy and astonished at the same time (but isn't that what keeps us going?).

Thanks for the input.

[pminmo]

Are your limit/estop switches normall open? If so change the logic to normally closed.

[gar]

071013-1132 EST USA

glentner:

You may have ground problems, but since your current problem seems to be related to limit switches, then we need to know what your limit switch circuits look like.

If you have a mechanical contact limit switch, then ideally this should be normally closed prior to being made. This means you have a low impedance, near zero, at the switch when the switch is closed and you are moving and running the spindle. When the switch opens it is a very high impedance. Once you make the limit switch there is probably less concern about noise.

Neither terminal of your limit switch should be grounded at the limit switch. You should use a twisted pair cable, probably shielded, from the limit switch to the control input. Further at the control input should be a shunt capacitor to ground with a low resistance pull-up resistor, and a schmidt trigger input gate.

A resistance of 390 ohms from +5 V and a 0.1 mfd ceramic capacitor will give you a time constant of 390 * 0.1 * 10 exp -6, or about 39 microseconds.

If you insert a 1000 ohm resistor from the non-ground lead from the limit switch to the capacitor, and move the pullup to that node, then time constant for incoming noise is 100 microseconds.

The maximum magnitude for the 1000 ohm resistor will depend upon the input current to the gate. For more filtering (longer time constant) increase the capacitance.

.

[glentner]

My limits and e-stop switches are all normally closed. The cables currently aren't shielded and I don't have any capacitors on them.

[Al_The_Man]

What you need to achieve is what Siemens call uni-potential bonding that is trying to achieve the same potential at all ground points. This is usually done by a common ground plate, and all grounds and power supply commons and shields are taken to this point, it is almost impossible to prevent parallel paths to ground, but if at least there is one point that is common to all systems, it helps.
I do not use the through the parallel ports systems, I do however use PC based sytems and I always ground all the supplies, including the PC supply, the servo power DC common is also taken to the same ground, this plate also terminates the service ground.
Keep in mind the systems I install are normally in a very noisy industrial environment and so far this method has worked for me.
Al.

[gar]

071014-0722 EST USA

glentner:

Can you do this?

Effectively disable your limit switches. Disconnect all the wires to the limit switches at your computer. Short the inputs at the computer to simulate your normally closed limit switches. Now run a small back and forth or closed path program with the spindle on. Do you still get the interference problem?

If so, then disconnect all wiring to the spindle motor from the computer and derive AC power to the spindle motor from an extension cord from somewhere else. The AC line filter common should go to the ground wire in the extension cord. Also this filter should be as close to the motor controller as possible. The motor control ground for this experiment should not go to the motor but only to the extension cord. The frame of the motor and any shield around the wire to the motor should go to the common ground of the machine. The wires to the motor from the motor controller should be twisted inside a shielded cable. The intent of this experiment is to try to isolate the path of the noise getting into your computer.

If this does not stop the problem, then remove the spindle motor from the machine and any wires between the motor and the machine (should only mean the said ground wire to the motor frame), and determine how far away you have to move the motor and its circuitry to eliminate the problem.

If you get all of the spindle motor stuff 50 ft away from the computer and the problem still exists, then put a low pass RFI filter on the AC line to the computer.

Basically try to find out what isolation is needed to eliminate the problem. When one can find how the noise is entering the system, then you can work toward a solution.

Basically there are four ways to get the noise into your computer --- conducted, magnetic, capacitive, and radiated.

.

[glentner]

Gar, before I read your posts I did in fact try one of your suggestions. I disconnected the e-stop cable and wired it closed at my electrical box. I ran a program with the spindle on and it still triggered the e-stop. Also, to make things worse, I noticed that one of my stepper motors was "jittering" when the spindle motor was on. I thought that I had that problem already fixed by putting in my filter, but I guess not.

So being that the e-stop triggers without a cable attached to it, what is actually getting the interference????? Is the EMI/RFI getting into my electrical box and into the circuitry there? Or possible into my parallel port cable that runs from the box into my computer?? Would a simple choke on the parallel cable fix all my problems if that is the case?

I'm still not discounting the proper grounding theory...

Thanks for both of your inputs. I'm a mechanical engineer, not electrical so this is all beyond me.

[gar]

071014-1347 EST USA

glentner:

The signal line or lines that need to be shorted should be at the computer, not X ft away. If all your signals go in thru the parallel port, then all these should be shorted at the computer parallel port connector for this experiment. If there are other signals that have to change state into this port, then we need more information. The noise can get into your computer a number of ways.

.

[pminmo]

Physical layout of wiring is very important. Do you have a picture of you contol wiring?

[glentner]

Well, I think I figured out the problem... I un-plugged my wireless network adapter and ran my test program with the spindle running, and no false triggers!
I am using a D-link USB 802.11G wireless adapter that sits right in between my PC and my electrical control box.
It just came to me one minute that 802.11 uses RF to transmit and receive data, and one type of interference that I've heard of is RFI, so maybe it's related?
I still should rewire my system to provide proper grounding and swap out all my cables for shielded ones, but as of my little test last night, I might be good to go. Hopefully I won't be back on this thread in a week saying the problem came back.
I'm not in the forums as much as all of you guys who replied to me, so keep this in mind if someone else has a similar problem to mine.

Thanks for all the help.