On Tue, May 8, 2018, at 7:20 AM, andy pugh wrote: > I have concluded that I have a problem with noise from my (shared by 4 > drives) servo power supply. > The STMBL drive controlling the 4th axis suffers from CRC errors with > the smart-serial control signals and shuts down. > > By the simple expedient of running it from a separate PSU this problem > seems to be eliminated. > (It ran fine all night yesterday). > > I don't see this as a good long-term solution, so I want to get rid of > the noise in the DC supply. This seems like a job for a choke, but the > cables have a braided shield.
Which cables have a braided shield? The ones taking DC power from the power supply to the drives? There's nothing magic about those cables - replace them with two suitably sized single-core wires. I would give each drive its own pair of wires. Twist each pair. They should meet only at the power supply terminals, so current for one drive doesn't flow in the power leads of another drive. Running each pair thru one (or more) ferrite cores might help. If you are using two single-core wires, then there are no bulky connectors to deal with and you can use regular ferrite cores instead of the expensive split ones. Is your drive DC supply floating (transformer) or rectified mains? You need to provide a path for the common mode currents that _will_ flow thru the motor-winding-to-frame capacitances. Each time a drive output terminal switches, charge Q = CV will flow from the drive DC bus, thru the stray capacitance, into the motor frame, and then back to the DC bus of the drive. You want to provide a path for the current that doesn't flow through any sensitive ground. I thought there was a PE terminal on the STMBL board, and a pair of small caps connecting it to positive and negative DC bus. But I don't have my boards handy and I can't find the artwork on github. (I can find the kicad files, but not a PDF or other readable snapshot of any particular board version - pet peeve of mine.) If such a PE terminal exists, I would run the a PE lead from each motor frame tightly bundled or twisted with the motor power leads and back to the drive PE. (If the drive doesn't have a PE, I would make one using a pair of 0.1uF or so film caps with suitable voltage and safety ratings.) That provides a low inductance path for common mode current to return from motor frame to drive DC bus. The PE terminals of each drive could be connected to a common control chassis near the drives. The ferrite beads around the DC bus connections to the common power supply would encourage common mode currents to stay in each drive/motor loop instead of wandering off towards the supply and the other drives. > > Should the shield pass through the choke, or should I strip the area > where the choke is installed? If only one end of the shield is grounded, then it can pass thru the choke. However, the shield is at best marginally useful on DC bus wiring, and if grounded at the wrong place, the shield can actually be counterproductive. Are your motor leads shielded? That shield can DEFINITELY be counterproductive, since it acts as additional "motor-lead-to-ground" capacitance. Common mode current WILL flow thru that capacitance every time the drive output switches, and it WILL find its way back to the DC bus. You need to make sure the path back doesn't take it anywhere it shouldn't go. Motor cable shield should be grounded only at the drive end, and directly to the drive PE terminal (capacitor coupled to the DC bus). The motor cable shield can be connected to the motor frame at the motor end (serves as the motor PE conductor). In the ideal case, the motor frame is not connected to the machine frame, so that winding-to-frame capacitive currents are forced to flow back through the motor PE wire or motor cable shield and can't flow thru the machine frame. In many cases, floating the motor from the machine frame is difficult, hence the need to bundle PE tightly with motor leads. Tight bundling reduces the area of the motor-lead-to-PE-lead loop, making that the lowest inductance path and encouraging _most_ of the common mode current to flow in the PE rather than a longer, higher inductance path through the machine frame. Since Q = CV, many of these issues can be glossed over on the typical 48V hobby servo. But as the DC bus voltage goes up, the issues become more severe. At rectified 120V line (170V DC) it matters. At rectified 230V (330V DC) it matters more. And at rectified 480V or 690V (my day job), it becomes huge. -- John Kasunich jmkasun...@fastmail.fm ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
