On Wednesday 03 February 2016 04:26:07 Erik Christiansen wrote: > On 03.02.16 03:11, Gene Heskett wrote: > > So the only way to do this relatively non-timing critical psu > > switching is to use the carge pump, and a charge pump detector to do > > that control, it can absorb the noises without triggering the > > outside world for 1/4 second while its getting its shit in one sock. > > > > A crapload of such problems could be avoided if the 5i25 failed to a > > logic zero, same as powered off, when its disabled. > > Now I'm not grokking the problem, Gene. The first component in a > simple charge pump is a capacitor, so the DC state of an inactive > input is quite irrelevant. Unless the input is hopping up and down, > there's no charge pumping possible - that's what it's designed to > detect, after all. > > Erik Huh? Have I not explained in enough detail?
I have a charge pump, running at servo-thread, so its square wave is nominally 500 Hz for a 1 kilohertz servo-thread. From my .hal: (word wrap off) loadrt timedelay names=delay-soft,pwr-off-delay loadrt charge_pump addf charge-pump servo-thread addf and.vac servo-thread addf and.spndl-pwr servo-thread addf and.spndl-soft servo-thread addf delay-soft servo-thread addf pwr-off-delay servo-thread setp charge-pump.enable true # just leave it running. setp pwr-off-delay.on-delay 1 # seconds, quick on setp pwr-off-delay.off-delay 12 # seconds, slow off so resistor is in circuit by then setp delay-soft.on-delay 10 # seconds delay till the 51 ohm is SSR shorted setp delay-soft.off-delay 2.0 # seconds in case I have it bass ackwards setp hm2_5i25.0.gpio.011.is_output true net spndl-power motion.motion-enabled pwr-off-delay.in delay-soft.in and.spndl-pwr.in1 setp hm2_5i25.0.gpio.011.is_output true # pin J3-8 net start-spdnl-pwr and.spndl-pwr.out hm2_5i25.0.gpio.011.out #pin p3-8 net spndl-power-on <= delay-soft.out => and.spndl-soft.in1 # crowbar resistor # set up this output on p3-09 for soft start bypassing setp hm2_5i25.0.gpio.012.is_output true #is p3, bob pin 9 # that sets gpio.012 to output, and this crowbars the soft start resistor net spndl-power-soft and.spndl-soft.out hm2_5i25.0.gpio.012.out #P3 pin 9 So when enabled by the time-delays, both pin 8 and pin 9 on the db25, p3 on the 5i25, are being pumped by the charge pump module. In an outboard box, bonded to the main box, is a teeny power transformer feeding a small bridge and filter capacitor to make about 34 volts out of a 12-0-12 winding rated at 200 mils. This feeds a buck switcher to knock it down to 12 volts, regulated. I could adjust it as high as 30 volts, but 12 does the job. This +12 volts is tied to the SSR's + control terminals. Cobbled up on a radio shack project board in that box are 2 charge pump detectors. When they have seen the pump signal long enough to charge up the output capacitor, cF, the gate of a mosfet is driven high enough to turn the mosfet on. The schematic for the detectors is attached, drawn in geda, not eagle. Those 2 pumping signals are sent to the 2 pump detectors, with the drain of the mosfet output transistor of each connected to the SSR's - control terminals. The net result is that the first hexfet to turn on, enables a 40 amp SSR to connect the hot leg of the input power to the primary's of all 4 tordoid transformers THRU a 51 ohm 200 watt resistor to limit the inrush surge that was tripping a 20 amp breaker in the service box. The delay times are such that nominally 10 seconds later, after the filter capacitors have charged well, the second signal starts pumping too, and that pump detector then turns on the second SSR, putting a crowbar across that 51 ohm resistor. In the .hal setup shown above it should also turn off by first turning off the second SSR so that the resistor is restored to be in series with the transformer primaries, and about 3 seconds later, the first SSR is also turned off, disconnecting power from the toroids completely. But, if my clicking on either of the 2 upper left axis buttons to disable the machine, also kills the 5i25's ability to output those pump signals, which I now suspect to be the case simply because nothing else makes sense, then the SSR's will be turned off quickly with zero regard for the off timings set in the .hal pastes above. If the main one goes down first, with the crowbar still enabled, I can see some odd goings on. So when I have injected some starter fluid, aka coffee, and go to the garage to check, the first thing I'll do is see WHEN the pumps stop. And I'm betting they stop, going high & staying there, the instant I click on either button. Now, does this explain it? Something is killing a 20 amp breaker when I click on the axis disable buttons. NOT after the timeouts set in the .hal. A 30 amp breaker survives the surge with all this stuff bypassed. But a 30 amp, feeding a 4 plex, even if 10/3wg is used to feed the 4-plex, is still sick bird according to the NEC. And my fire insurance claim will be null and void. If the resistor crowbarring SSR goes off first, problem solved. So changing out those cobbled up chargepump detectors for the new ones with schotkey diodes, improving the hexfet drive and doubling the cF in the one driving the first SSR will make sure that even if its only for 2 or 4 cycles of the 60 cycle power, the desired off sequence will be maintained such that the resistor is restored to the circuit before the main one goes off at the next zero crossing after its turned off. If that sequence can be established, the turnoff will have the resistor already in circuit to control the surge. Am I making sense, or blabbering into a bowl of morning oatmeal? Now, to go check this theory. I'll post later when its been confirmed or disproved by my ds201. Its there, plugged in to keep it charged. Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>
charge-pump-bucket.sch
Description: Binary data
------------------------------------------------------------------------------ Site24x7 APM Insight: Get Deep Visibility into Application Performance APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month Monitor end-to-end web transactions and take corrective actions now Troubleshoot faster and improve end-user experience. Signup Now! http://pubads.g.doubleclick.net/gampad/clk?id=267308311&iu=/4140
_______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
