On Saturday 10 October 2015 05:58:16 andy pugh wrote: > On 10 October 2015 at 06:42, Gene Heskett <[email protected]> wrote: > > In setting up to use a charge pump, needed to insulate an output > > from the fact that a 5i25, when lcnc isn't running, sets all outputs > > to inputs and pulls them high. > > I am still not convinced that a charge-pump is the only solution to > this. I did some experiments, changing the 5i25's gpio pin around, and the only way to achieve the desired result of starting the vacuum cleaner only when the flood coolant checkbox is checked was to use the charge-pump.
The other, more direct methods all failed and turned on the vacuum when lcnc was stopped, or was all powered down. > > More important, the loadrt line cannot contain either a count= or a > > names= number-of-the-beast control mechanism, its a show stopper > > syntax error. > > Why do you need more than one charge-pump signal? Surely the single > pulsing io line can fan-out to all the hardware it controls either in > HAL or in copper? In hal, using a bank of mux2's or and2's it can be done. In that event, my question degenerates into which takes fewer cpu cycles, an and2, or a mux2? > (Not that I see a good reason for charge_pump to be a singleton > component, but it is explicitly created as one in the .comp file) That was what's puzzling. The charge-pump is a useful module, and making it multi-instance would simplify the logic trail thru hal, making it more readable to the next set of eyeballs that might have to try & figure out just what the heck I was trying to do when the real problem is the failed electrolytic caps in the detector network. I do not have a handy radio shack anymore, closed up a few months ago, so I was forced to use what I had on hand, which was a bag of 47ufx35v caps, a bag of 1n914 diodes, and a small resistor to dole out the charge to the base of a TIP31 transistor, which in turn ran the coil of a 24 volt DPDT icecube relay. The vac draws 6.5 amps running, so both sets of that relays contacts are paralleled. I took a plastic duplex box and broke the link in the duplex so that one of the duplex's sockets was hot full time, and one was switched by the relay. A short run, the plug end of that 14 gauge is plugged into one socket of a 4plex I put on the wall roughly behind the mill. More of that same piece of what we'd call SJ cabling runs from the duplex I'll screw to the wall beside that 4plex, over to the top of the "furniture" I built for the computer, and the huge box with all the mills drivers and psu's in it, to the socket the icecube is plugged into currently mounted to the BoB boxes lid. The NEC illegality is that the white wire in this SJ cable is hot. Its not supposed to be white because that is supposed to denote powerline neutral on this side of the pond. The green wire is the static ground, and I will, after checking the box this BoB etc is in, might supply a static ground to the box. The box itself is made from a couple feet of double sided PCB material as I didn't want to cut up what I have left of the sides of a wrecked ups van. However, had I known that I was going to need room for the relays and 3 of those switching buck regulators I bought a 5 pack of on fleabay for less than $40 USD, and that amazon was going to renege on selling me a small 24 volt switcher, I would have made the box bigger so I could have hidden all this away from wandering fingers. That is not an unfixable problem as I have about 40 feet of that material 6" wide, so I can make a bigger box and transfer stuff. The control transformer that makes the nominally 36 volts to feed all those sub regulators is the sole source of obvious heat, that things core runs at 135F at zero load! Very inefficient. I knew that, which is why I tried to buy a smallish switcher from amazon. NEC legal on this side of the pond? Hell no, thats why it can be unplugged. Even that 4plex is sick bird, its fed with around 6 feet of 10/3 right from the service, which mandates a max breaker rating of 20 amps, and sick bird to feed a regular duplex socket with, so you have to drill out the plugin entrance holes in the back of the duplex, or half strip the terminal screws which will not back out far enough to accept more than a 14 gauge wire. But the breaker feeding it had to be jacked up to a 30 amp, else it tripped on turning on the mills motor supplies with one switch. Actual draw according to an amprobe is about 2.75 amps, but the capacitor charging inrush of a kilowatt+ for spindle motor use tripped a 20 amp breaker 100% of the time, and no breakers are locally available that would absorb that, they are all magneticly tripped, cheap, and as instant as they can be. That really should be a 15 amp breaker, with a couple cycles of thermal time delay to trip. They can be ordered special, cost around $75, whereas the 30 amp magnetic is less than a tenner, you can buy them in carton lots of 10 for an even bigger discount. Obviously everyone knows what will be used in actual practice. What Works(TM). Which is a huge sigh because the insurance people will use any excuse they can find in the ashes to deny a claim... Gotta go make some morning plasma to get both eyes open simultainiously. I' a quart low. Thanks Andy. 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> ------------------------------------------------------------------------------ _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
