On 11/27/2014 08:32 AM, Karlsson & Wang wrote: > A small value capacitor decrease the problem so I guess the > commutation spikes are the problem but driving a capacitive load is > not good. > > Nicklas Karlsson
(My general view on noise) One should not expect to simply run a wire from a signal source to an input in a machine environment and expect it to just work. Machines tend to be electrically and magnetically noisy, and signal inputs tend to have very high impedance. It's like the cable is a bottle and wind is a noise source. Wind going over the end of a bottle will create a sound which can drown out any sound you need to hear. To fix this one can reduce the noise and/or increase the sound level of the signal. I found that AC line filters on VFDs or any switching power supply is pretty much required. I had spindle encoder noise on the far side of my lathe which was cleared up using AC mains filters on the VFD power inputs and ferrite beads on the motor leads, thus reducing the noise source. Sometimes an AM radio or oscilloscope probe can be used to scan for noise sources, but I usually just add the filters anyway. After reducing noise sources, the susceptibility for the bottle or rather cable to be affected by noise can be addressed. Shielding is the most obvious approach, but more often, line conditioning is the real issue. A bottle will make sound due to a pressure wave, starting from the top of the bottle and travels down to the bottom and bounces back. A wire will do the same thing, and the source of the reflected wave could be the signal itself as well as noise. One way to stop the pressure wave from reflecting is to put a hole in the bottom, changing the bottle's bottom from high impedance to low. There are many ways to configure a wire to handle a signal, generally through termination and filtering. Much of this has been worked out already, so the methods for handling the type of cable being used should be studied and put into practice. Another way to deal with noise is to boost the signal above the noise level. That is why RS-232, RS-422/485 run with a much higher voltage than USB, SPI or I2C. RS-422/485 use two wires, one with a positive signal, the other with a negative version of the signal. Noise affects both wires the same way, but the difference stays the same, so the noise is ignored. So for each signal: What is the nature of the signal? (driver and input) What is the nature of the wire? (impedance, configuration, termination, filtering) What is the nature of the noise environment? (suppression, shielding) What method best addresses the above? (which industrial standard) -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/ ------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users