On Saturday, May 05, 2012 12:15:29 PM Dave did opine: > On 5/5/2012 8:46 AM, gene heskett wrote: > > On Saturday, May 05, 2012 07:32:05 AM Andy Pugh did opine: > >> On 5 May 2012, at 03:41, gene heskett<ghesk...@wdtv.com> wrote: > >>> arg[3], arg[4] arg[5] etc of a "net" commend can be repeated to add > >>> sending something from arg[2] to more than one load. But I can't > >>> name a previously used output and send it to the 2nd place it needs > >>> to go. Its s show stopper error. > >> > >> Argument order is not important except that the "signal" name is the > >> one directly after "net". Any HAL "signal" (the arbitrary name) can > >> only be netted to one output "driving" pin. > >> > >> So > >> net signal1 out1 in1 in2 in3 > > > > Which seems to say that the out1 name string in the first format, > > becomes the signal1 name string in the second format? > > > >> net signal1 in4 in5 in6 > > > > Humm, in this latter case then we have a bunch of inputs only, so I'd > > have to assume that "signal1" would then be named as the out1 in the > > first format on a different line, effectively becoming a "driver > > pin"? > > > >> Is probably the format you are looking for. There is no need for the > >> driver pin to be in the first net statement, incidentally. > > > > Where the "driver pin" name is an arbitrary substitution of an "out1" > > named in the first format above? > > > > In my present .hal, I have all the loadrt's at the top, and all the > > addf's next. Followed by the long list of setp's, then the net's are > > generally at the bottom of the list. > > > > It occurs to me that the addf's using the servo-thread could > > potentially result in additional servo-thread sized delays if they > > are in the wrong order, and they are processed in the order given in > > the hal file. > > > > Is that true? In the html docs, under halshow at > > <http://www.linuxcnc.org/docs/devel/html/hal/halshow.html> > > the text quite a ways down the page uses "linksp" which I think has > > now been replaced with "net" for several years, but has the actual > > function description been changed such that they are not directly > > interchangeable? > > > > But the net statements would seem to be more sensibly grouped under a > > function label, in the order of signal flow, in order to make the > > signal flow more obvious to us humans. > > > > Am I on the right track? > > > > Does linuxcnc have a utility that can scan a .hal file and draw a flow > > chart? HalShow would appear to be similar, but demands a fully legal > > hal file so that linuxcnc can actually load up and run, so would seem > > to be of no use for troubleshooting a broken .hal. And that is my > > instant problem. > > > > I got nfs setup again on that machine last night late, so I can pull > > that hal file in and print it, killing trees is my most effective > > debugging tool it seems. One could say I am using paper as a sub for > > my ever poorer short term memory. :( > > > > Thanks Andy. > > > > Cheers, Gene > > Gene, > > Are you trying to select one of two or more sources? If so use a mux > component. > Actually, not mux, but scale and sum2, scale to set the expected value, and sum2 to add or subtract the diff as required in order to correct the spindle speed under load. But to do that, I need to take the float output from spindle speed requested, and add or subtract the scaled output of the encoder.0.velocity to determine the error, which is them fed to sum2 along with the requested value, and the sum2 output then passed to pwmgen.0.in. Or at least that is how my understanding of it goes. Where I fell over was in trying to send encoder.0.velocity to all the places it needs to get sent with the 'net' command. Do everything but disconnect pwmgen.0.value from its current signal, the output of net spindle-abs-speed <= abs.0.out => pwmgen.0.value
to something resembling net spindle-feedback sum2.0.out => pwmgen.0.value Up to that point, no feedback exists, but the scaling can be adjusted such that the error signal from the rest of the chain when the spindle is running is relatively small in terms of the output of abs.0.out. This is from the section in the docs showing a hal snippet to put the spindle speed under active feedback control. It sags about 50% at the relatively light cuts and slow rpms of doing the threading. Then I have to grasp the nuances of G76, which seems to work in relative radii from where you start it for nearly everything BUT the radii of the threads min diameter, where the rest of the interpreter seems to work with diameters. Greatly Confusin. And its long list of options is something that could be readily calculated from the thread pitch, clipping off the top and bottom 10% of to fit the usual USS and USF thread profile, so ideally G76 would only need to know the Z stop, which end to ramp out on and the OD of the finished thread. Not really helped by x needing to be set with a touch off after measuring the diameter of a fresh cut, but that is a separate setup procedure problem. Humm, maybe that is the first one of my many problems. Question: I measure the diameter of a fresh cut, BUT I'll bet I should be touching off to the radii, or half the diameter, and I think I've been entering the measured diameter in the touchoff, which then shows as the radii. Duh! Dumb even. Something sure as hell isn't working right. And that sort of an error would cause all sorts of oddities. > If you don't like how the hal works in certain ways.. you should try > "Comp" to make your own component. > Once you learn how Comp works, it seems very simple. > > Dave > I was wondering about that. I've read that piece 2 or 3 times thinking about that possibility. I find that not even a big stick can make either of my printers list that file in landscape mode. I've waste about 50 sheets of paper trying, even rebooted this machine. I even tried loading it into librewriter, which has a font selector when no file is yet loaded, but open the file, and the font selector disappears, and it loads in about a 5 point font. Very frustrating. So I've posted a small fuss on the pclos software forum. This all worked well when I installed it 2 years ago, and nothing in the print category has been updated since last Sept. It has worked well many times since then, but started failing maybe 6 weeks back, and I've now verified that it effects both of my printers. I believe the last time landscape worked was when I was printing that encoder board pcb from eagle. And just 2 weeks ago I did a screen snapshot, loaded it up into gimp and that worked, so whatever has changed, its quite recent and effects both printers. Makes me want to shoot the neighbors cat or something... Thanks Dave. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) My web page: <http://coyoteden.dyndns-free.com:85/gene> This fortune was brought to you by the people at Hewlett-Packard. ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
