Andy, I think you need to add a sealed in relay on your input power contactor circuit. You start the machine by pushing a power on button which has a seal in contact to hold the contactor in place after the button is released. When the power drops out the contactor drops out and will not pull back in when the power recovers unless you push the "power on" button. To make it even safer, you can add an auxiliary contact off the crowbar relay such that if the crowbar relay is stuck on, the start circuit will not function ( a NC axillary contact off the crowbar contactor should work).
I'd stick all of this into a metal box just in case something catches on fire.. the box should contain the flames. Contactors are typically much more robust than relays. Regarding DC vs AC contactors, I would use a regular AC contactor for the crowbar contactor. DC contactors are specialized in that they can break DC current which is difficult to do. You don't need to break DC current, you need to conduct DC current.. a different problem that should cause little to no arcing. Just interlock the crowbar contactor with the infeed contactor so that if the crowbar contactor fails stuck on, you don't drive power supply current into the resistor. Over here (in the USA) we have cheap Fuji contactors available from Automation Direct. I'd find out who sells them in the UK. A 25 amp IEC contactor with a 24 vdc coil is about $40.00 US. http://www.automationdirect.com/adc/Shopping/Catalog/Motor_Controls/Fuji_Contactors_-z-_Overloads/9_to_25_Amp I have used hundreds of them and they seem to be very durable. It looks like http://www.lamonde.com/ sells Automation Direct equipment in the UK but I don't see Fuji Contactors in the list of items sold. Dave On 10/2/2013 5:54 AM, andy pugh wrote: > The PSU on my milling machine blew up again last night. This is the > second time it has happened. I haven't pulled the box out of the > machine yet, but I expect to see this again: > https://picasaweb.google.com/lh/photo/xxfP_tT7Ae0op6GxUhSDvtMTjNZETYmyPJy0liipFm0?feat=directlink > Which is a power resistor that has got so hot that it has exploded. > (quite an odd failure mode for a resistor). > > The PSU is home-made and consists mainly of a bunch of capacitors and > rectifier. > > Also in the box are two relays, two power resistors and a timer > circuit. One resistor is a soft-start on the AC input side. When power > is first applied the power passes through one of the resistors to > limit the surge current, and then the timer closes a relay that > short-out that resistor. I am fairly happy that that part of the > circuit is reliable. > > A second relay is used in change-over mode. This has a 240V coil > connected to the incoming power. When the PSU is powered up the relay > closes, and supplies power to everything else inside, when the power > goes off the relay opens and the NC contacts connect a crowbar > resistor across the capacitors to discharge them. > > I think that the failure mode is that, in the case of a power glitch, > the crowbar relay switches and is discharging 300V DC at a fair > current, then the power returns and the NC contact tries to break 300V > 10A and welds the NC contacts closed. Then the NO contacts close and > we now have the crowbar resistor connected directly across the > rectifier output. The crowbar resistor is not sized for continuous > operation. > > I think that there are two problems here. If the crowbar contacts on > the relay are welded shut then the power contacts should not be able > to close. I think I need a more explicitly interlocked relay. > Currently I am using: JG64U here: > http://www.maplin.co.uk/round-base-10a-relays-2567 > > I think that the term I need to search for is "force guided" but would > like some confirmation that that means what I think it means. Are > "contactors" more dependable in this sense? > > The real problem is that the relay is not capable of breaking 300VDC. > However, DC rated contactors are pretty expensive, and seem to jump > for 220 to 44 and then 690V. I don't think I have room in the box for > the big ones. I wonder if > http://uk.rs-online.com/web/p/contactors/4111687/ would be OK? It > really should never try to break with the capacitors charged > (interlocked in HAL, which can see the DC bus voltage, and with the > crowbar in circuit the voltage will drop pretty quickly anyway) > > My inclination is to look for a solid-state solution, but solid-state > isn't very good at doing anything at all on the basis of the power > going off.... > > ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users