John, I'd still take a close look at your existing system to see if it is robust.
Perhaps an example from a different project my give you an idea. Bear with me. The story is kind of long. I designed an LED lighting system that ran of a 48V bank of batteries charged by generator. Charging could raise the voltage as high as 56 volts. The voltage regulators on each LED lamp was good to 70V so I had plenty of leeway, or so I thought. The power distribution system was in the centre of a long barge. The genset and batteries at one end. The first time we powered up the system and flashed the lights on and off a few times something odd happened. The lights went a funny colour and then many of them went off. Since I'd already built one system that ran on 24V and had been operational for 8 months we were perplexed. True, this one had different boards with 48V instead of 24V. On the way back out we were discussing what could have damaged so many lights. The Chief Engineer asked if the coil of wire would have any impact. I asked what coil. Well it turns out the wire running the DC from the battery pack and the control panel was very expensive. So to avoid cutting out a section that would end up useless they coiled it under the cabinet. A change from 70A to about 10A was much like an ignition coil when the points open. The inductive kick was probably well over 100V especially since the generator was running at the time. We cut the cable to the exact length needed. It still ran in a metal trough under steel plates on the barge deck but looking at the power with a scope showed the problem had gone away. One wouldn't think a coil of about 5 turns would make a difference but it did. The second issue that was prevalent on both systems was that each string of lamps was up to 35m long with a restriction of 8A current on the power conductors. The lamps were all in the last 10m so there was 25m worth of feed cable. That means at full lights on the first lamp after the 20m had the highest voltage and the last lamp 10m further down had the least voltage due to the multiple voltage drops as each lamp used power. I'll introduce another term here. It's called common mode voltage. The power through the 24V or 48V connection to the lamps has a voltage drop based on current consumption. There's an identical drop across the ground power wire. The one on the 24V line isn't that important as long as it can still run the load. But the voltage drop across the ground lead means the local lamps see a different potential between ground and the communications bus than at the other end of the 35m cable. Most communications chips have a specification called maximum common mode voltage. That essentially means you can't have more of a difference between the grounds when the signal is also part of the connection. So for example if the common mode is a 6V drop across the ground conductor then the voltage at one end between tx and gnd might be 2.5V. At the lamp at the other end the signal is also 2.5V since the lamps sees the local ground. But that 6V shows up as soon as there's enough current on the ground to cause the 6V drop. Now one end sees 8.5V relative to ground. This is static and measurable with a simple meter. OK back to CNC. This is the biggest reason you run 4 servo or stepper motor drives with the power and as a single point on the power supply rather than ground to ground to ground to ground. If each Servo draws 8A and the last one sees the ground shift up because of voltage drops along the ground bus and the signals to the motors (or encoder feedback) may be compromised. Now let's look at the AC side of things. You run all your power to a single point doing it right. Each wire to each driver has inductance. And there's capacitance between the wire and the cabinet. At the right frequencies the noise ground return path is now through the capacitance to the cabinet rather than along the wire. Since each driver and each wire is further away from the power supply we're back to the daisy chained power distribution system with more common mode voltage at the far end higher than at the end close to the power. And again that can affect the encoder or control signals. But now it's sporadic. Hard to track. Stepper winding current collapsing creating noise spikes at the same instant as a message to a VFD. So shielding power cabling to your devices with the shield terminated at the distribution end and open at the other can suppress those kinds of problems. Now the high frequency signals and noise are brought back to where you want. Does that all make sense? John Dammeyer > -----Original Message----- > From: John Thornton [mailto:j...@gnipsel.com] > Sent: December-20-15 3:06 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Grounding Issues > > > Well it works fine till I plug in the VFD so I guess I'll have to rewire > the machine and my shop again :( > > JT > > On 12/20/2015 2:25 PM, John Dammeyer wrote: > > Hi John, > > > > It sounds as if you have two AC lines coming into the machine. One for the > > controller and one for the VFD. The comments about a single star ground > > point are standard industry practice. One breaker at the panel which > > protects the wire to the machine. Another breaker inside the machine > that > > protects the equipment. > > > > > > > ---------------------------------------------------------------------------- -- > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
