How to cancel electrical noise interfearing with switches?

[Witsenburg]

Hi,

I have an MDF wood router up and running, but I cannot get the limit switches to work. It all functions perfect with the drivers and motors disconnected, but under power the computer gets a tripping signal.
I have 6 NC switches in series. A 10K resistor from 5V to the port, the switches from port to 0V. All signal wires are shielded, shielding earthed.
I work with Mach3 software, if I set the debouncing rate really high, everything works but you can shortly trip a switch without the computer noticing. Not good...

Some ideas: should I make 0V and earth (from the power net) the same?
Should I put in some capacitors somewhere to 'clean' the signal? If so, where, howmuch and why?
Any other ideas how to solve this? I'm getting a bit desperate, it's the only thing keeping me from staring some serious work with the machine!

Thanks,

Hugo

[pointcloud]

Few questions...

what type system? not software (gecos galil)

What makes you think it is electrical noise?

[Witsenburg]

I think it's electrical noise, produced by the drivers, since the problems disapear when I disconnect the drivers. And when increasing the debounce rate, the problem starts later. I've meanwhile put a capacitor across the resistor, which again postpones but not eliminates the problem.

What do you mean by 'system'? I have a pentium3 with windows XP home running mach3. I have 3 MSD542 drivers, two PS407 power sources in parrallel, all from Motion Control Products in the UK. The machine is home built out of MDF, see here for the thread. And here for a youtube movie.
The limit switches are simple NC/NO microswitches. The NC is used for limiting, the NO will be used for home-ing the axis. I run two big HY 200 3438 motors for the X and Y axis, and one small 57BYGH728 for the Z. The driver for the small motor is of course tuned back. It's all 4 wire motors, bipolar stuff. The breakout board is the C10 by CNC4PC. It gets it's power from the PS407s, with a 10V zener and a 7805 to give it a stable 5V.

Well, that's all the things I can think of, now.
All input is welcome!

Cheers,

Hugo

[kreutz]

Look at this thread http://www.cnczone.com/forums/showthread.php?t=34974

[Cruiser]

That sounds like "Ground Loop" to me ! Better look for one of those obvious connections you know had to be connected that shouldn't have been connected !

[HuFlungDung]

FWIW, simple micro switches may bounce enough from vibration to actually open. I recall having a problem like this on a knee mill retrofit, particularly the Z axis limits on the quill were prone to nuisance tripping. There is a lot of vibration from a milling cutter!

The solution for my case was a better quality, heavy duty microswitch.

[Witsenburg]

Well, I solved the problem! Maybe not very conventional, but it works, so I'm happy.
I started with unshielded wires, parrallel to the motor wires, running from 5V to port. No pull-up resistor, no shielding, no noise suppression capacitors, nothing. I then had the problem, and started with the pull-up. Then I added capacitors. Later I rewired, using a different route and shielded wire. But leaving the resistor and capacitors in place.
Today, I went back to the original connections, without resistors or capacitors, but using the new route en shielded wire. Problem solved! Debounce is now set at 500.

Cheers,

Hugo

[pminmo]

You originally did a big no-no, parallel control signals with motor wiring. Motor wiring (assume a chopped driver) is a nasty emi and rfi generator. Parallel wires are great antenna couplers. Physical distance is an important consideration when locating signal wires in relation to power wires, the further from noise the better. Shielding helps reduce that problem assuming the shield has a proper path to shunt the induced noise.