Note: This discussion about long lead screws started before the scheduled meeting time, so only part of it was logged. Right. I've heard some screws that had problems this way. I never go past about .75 critical speed on any leadscrew How do you find critical speed? math or tables most manufacturers have nice graphs for critical speeds and Euler buckling loads I see the 50 ratio on the 8' by 50 mm screw. Okay. I'll look from now on. How do you feel about spinning the nut rather than the screw? yes...my big machine is about that too...hence the need for dual angular bearings on both ends spinning the nut has much lower inertia so it's a good thing How much stretch do you figure you put on the long screw? But...it's big $$$$$$$$ I have the screws on mine preloaded about 400 lbs the stretch is almost nothing though one end has to be free to move a bit axially while still being constrained in bending ballscrews get warm and grow quite a bit I understand that. So most of the value of the dual bearings on both ends is to constrain the screw along it's axis? Probably not saying what I mean. yes...you get dramatically higher critical speeds with dual bearings on both ends Are these the angular contact pairs or are they spaced a bit. Spaced How far apart? not much...one bore diameter Greetings... your 400-lb preload, this places the screw in tension? yes thalx Okay. Do you worry about bore allignment end to end? Bore alignment is very critical yes How'd you do it on that machine? a grouted ground pad...and bearing housings with precision ground surfaces that are good datums All this becomes much less critical if you just don't let the ballnut get close to the end bearing I can see that. but I do run it quite close so I had to be careful I've done a bit of thinking about a quilting machine. Long screws on x there. In the small low cost machine I keep thinking about manufacturing I would jut keep the nut ten or so leadscrew diameters away from the end bearing it would save a lot in cost also fixed-fixed is very expensive to do How about a rotating nut. I suppose that we'd want to keep it away from the ends as well. Its good....but the nut has to be balanced (most aren't) and designed such that the grease is not slung out When you look into rotating nuts it turns out to be very expensive Hadn't thought about balance. Good point. but great performance I was thinking of using kerk teflon screws for the sewing machine. Beyond eight feet or so rack and pinion starts to look better how fast do you want on the quilter? 99 percent of the time it's sewing but rapids still need to be there.\ I could imagine it needing pretty fast speeds With about 10' of travel. *** Signoff: davidp ("Leaving") yeah for that I would be doing r&p... leadscrew would work but it would be very high lead and THICK And costly! unless you use our telescoping leadscrew support thing actually that might work real well for you get the idea? For rack and pinion how would you gear the motor down. I can see sliding supports along the screw. A pretty expensive transmission i'm afraid right sliding supports it works morning guys hi john Hi morning.... Gee, since John's arrival seems to have quenched the long leadscrew discussion, should we move on to how to complete the board of directors election? haha yup That would be closer to emc topic. didn't mean to interrupt anything We can talk about big sewing machines later. I think Les and Ray were all talked out anyway.... I am...I'll just sit back for a while I captured a snapshot of the mailing lists back when we first discussed the election (about 3-4 weeks ago) but there have been a few adds and drops since then. Should I grab a fresh one? Are we worried about who votes? More worried about who tabulates. All our worker bees are nominated, and should not be counting the votes. we were planning on having "someone" send out ballots, and tabulate the returns that someone would need a list (unless we simply want to post the ballots to the list) I don't like that idea tho, cause it's virtually guaranteed that some folks will hit reply and send their votes to the list instead of to the tabulator person And the electorate is all on the emc list? I'd prefer "someone" sending and receiving and tabulating. That way reply works. right - hence the need for a list of names I'd think we'd freeze the list when we close nominations. which is today Is that 8 pm your time? something like that I have not heard from Dan midnight GMT, right?.... that's 7pm here I don't think I've ever heard from Dan... is he still active with EMC? Mauch? Falk He's been busy moving a big machine business from shasta ca to portland or. oh ok he's been nominated and seconded, but hasn't indicated whether he would serve Talked with him a month or so ago, before nominations, and he was about to set his machines up again. I just sold him a microscope...he is around but pretty busy can you contact him and ask him if he's interested in serving on the board? (either one of you) yeah...let me check my contacts here Ideas for process.... sending out ballots, not sensitive, one of the list admins can do it. Tabulator.... Jon Elson's wife, thalx, other possibilities...?? the ballots should be mailed from the email addy of the tabulator Easier to reply. Easy to "fake" reply address. easy for you maybe wonder if tbl could/would help That is a good idea. Maybe he could even automate it with a script. Probably would, we could roust him later today. Since Jon is a candidate, having his wife count might be mis-understood there's gonna have to be some "honor system" about voting... I know for a fact that some folks are subscribed at two different addresses the email should remind folks who get two ballots to only return one of them What's the size of the list? Can reduction to "#people" be done in advance? about 300 on users, 120 on dev I will merge the two lists and remove duplicates at that stage (folks that are on both lists with the same addy) but that won't catch people with two addys on one list some folks want to be able to post from both home and work, for example, so they subscribe with both addys, then turn off delivery from one of them I am certainly willing to help out. And I'm still unemployed, so I have no excuse not to! ;-) thalx - Can you easily set up a new e-mail address just for the election? Sure. "emc-elections (at) aldor (dot) com" - I can do it right now. Takes care of that! Great, I would hate to get someone's relatively spam free e-mail address publicized too much. found my old snapshot. It's from 6/20, there were 316 on users, 125 on dev, and when merged and duplicates removed, the total was 362 I can repeat that process after 7pm and forward the new list to thalx Takes care of that! Standard marking of ballot? hanging chads? The ballot should list all the names Return just the ones you vote for? It is kind of a hanging chad issue. right the ballot itself can include a paragraph of instructions Now... does Ray once again put words into Matt mouth to announce the ballot proceedure. SteveStallings: Yep -- seems like I'm doin that again. If folks put a check on the line in front of the name, thalx could use something like diff to count. like..... Please vote for four board members by deleting those on the list below whom you DO NOT wish to vote for. If you leave more than four nominees in the list, your vote will not be counted. If you vote for fewer than four nominees, all your votes will be counted. one thing you can count on - returned ballots will be messed up by mail clients - diff is unlikely to work Steve's paragraph sounds good to me. at a minimum, '>' will be added in front of the lines where is the emc subscriber list....I have lost dan's email yeah, deleting the unwanted names could work it's secure, only admins can read it 'Spose some sort of string comparison would work. I'll look for his addy thanks he is at chris king (but real names aren't attached, so if it's something other than danf or dfalk, etc, I won't find it ...I discoverd I didn't have contacts backed up and then had a hd fail...duh no joy I ran my 6/20 list thru grep looking for dan, chris, king, fal the only dan is xxxxxxxxxx Nope. hmmm...cad_cam subscriber list...i'll try that I last had dfalck (at) sbcglobal (dot) net but that may be old. we could try it.... if he's not subscribed to the lists, he knows nothing about the elections, etc DFALCK at sbcglobal (dot) net is the last I've got. as far as counting votes... I could write an awk prog that could go thru a message, do regexp matches for each candidates name, and report which ones were found (including dealing with cases like more than four names) no joy on the cad_cam list so far if all the replies were stuffed in a directory, then for $MSG in *, do awk prog, blah, blah might work I don't see Paul here today. He nominated and may know. For that matter, is Paul logging so he gets to read today's traffic? I dunno.. he hasn't been logged in for a few days I'm logging, and it will wind up posted at linuxcnc.org jmkasunich: If the awk program listed those who voted but voted for less than four, it might allow thalx to catch mispelled or truncated names. yeah... also list those who voted for more than four, so he can check to see if they didn't read the instructions and put X's in front of their choices instead of deleting the unwanted ones, etc Sounds good. The email account is all set up. thalx: Thanks in advance. yes, thanks So we can do a little advance testing, if you think that would be helpful! (hint hint) Jon's wife would probably have a much harder time with the scripts yeah - testing is good can you send mail from a script? Sure. I did a party invite like that once. cause it might be a good idea to send the messages one at a time, rather than using a big CC or BCC list jmkasunich: I'm with you on the wife thing. less likely to trip the spam filters My twisted brain had thought of generating pub/priv keypairs to keep ballotting in order. But for 400 entries that would require some horsepower. I guess I'm twisted too... I wouldn't go that far, but some kind of protection against accidental (or intentional) duplication of votes, etc some kind of unique hash in each ballot, the hash is detected by the awk prog, and if the same one appears in two msgs, it flags it We could do a list of hashes of each to: address. But then if it doesn't match "from", it doesn't work. hash the to, and embedd it in the message - then no matter where the reply comes from, the hash authenticates it actually doesn't need to be a hash of the to, just any unique identifier numbers from 1-400 would work as well Ok. and if we get dups of the hash...discard all? Accept only the first? look at them and apply the rule of least astonishment most likely cause of dups is accidentally hitting the send key more than once so count them once Heh - I love that phrase! I don't expect any hanky-panky, just trying to avoid accidental errors Yeah, and in-between mailers can certainly send out dups, if intermediate connections were flapping. (There's a T1 here to the house that occasionally goes down during storms...) now if you get two with the same hash but different votes, things get interesting that might be worth an email back to the voter asking for clarification but that's pretty unlikely Who wants to be on the sample testing list? I will Sure, me too... You bet. Whoever sends email to the voting list by some arbitrary time today... you'll get added. actually, one nice thing about using a hash... you can send me (or yourself) 10 ballots with 10 different hashes, as if you were 10 different people... then reply 10 times, intentionally making various mistakes, quoting differently each time, etc... good for testing Ok, that implies it's not tied to To: address. Which is fine. thalx: don't most mail progs store all messages in one big file? My system does. I can't speak for anything modern. for the real election, the hash might as well be a hash of the to addy for testing the hash could be anything yup. how hard is it to extract individual messages from that Not too difficult. Depends more on the content. if each message can be automatically extracted to a file that helps I only have two messages thus far in the inbox.. But yes, that can be done. then a smallish awk prog can read one msg, and output a line that looks like " 0 0 1 0 1 1 0", where hash is the hash from the msg, and the 0's and 1's are the votes for each candidate (first candidate in first column, etc) then you wind up with 400 lines like that - sort the file to find duplcated hashes, and summ each column for vote totals Ok, I've got Ray, Steve, and John now. So... When should we shoot to have this all ready to send out ballots? Any time after the close of nominations. Should not be a big rush. Next weekend would be a first target. (I'm a captive at home tomorrow watching a youngster, so have all day to test it..) Matts message said the deadline for close of voting is Aug 1 Right, but we should have at least a week of votable time- maybe more since it's prime vacation time. right - sooner is better for sending out the ballots send practice ballots out later today, hopefully everybody responds quickly, test counting tomorrow, send out real ballots by mid-week? *possibly* - I have maybe another 45 minutes here today before I split for the afternoon. ok, practice ballots can go out tomorrow Just looking to define the boundaries - ballots out by close of business Wednesday, or should it be sooner? I'll mess around with some awk today wed is fine, IMHO if it's later, it's later... if needed, we can change the aug 1 date - the real deadline will be included in the ballot Good job. Sounds like we have the most of the election procedure worked out. I have my Dans seriously confused....I could try calling chris king tommorow or is that too late Firm deadlines are better, at least for me. as long as we know before the ballots need to go out I've got Dan F's phone number around someplace. I'll call before deadline and try to get him to commit on list. ok he doesn't have to commit on list - a msg to steve or thalx is OK Steve has the list of candidates on linuxcnc, thalx is responsible for ballot Even if he doesn't commit, people can still vote for him if he's seconded... ;-) Ah, I see that he has been! thalx: Good thinking. true, but if he's elected and can't serve, then what - another election? Presidential appointment? ;-) um...just how many positions and nominees do we have at present? 4 positions, 7 nominees ok need to bookmark that page les: it's accessible from linuxcnc.org homepage... go to history,news,events, then first link on that page It is also linked at top of main page. I propose that when the results are in we simply add another column to that table with the vote counts got it Sounds like we've exhausted that topic. Got another? dead silence..... ;-) Y'all come on out to Iron Fever in 3 weeks or so! In wonderful York, PA. I've got a quick one. The current cvs will not compile under bdi-2xx. guess it's time for me to go out and attack that 1966 ford bronco sport in the barn emc1 I presume? Right. Problems with a couple of board drivers. Vigilant and vti stuff. I think paul was working on those files Probably. But in the mean time those of us that want to compile have to drag them out of the makes in both drivers and emcio. I guess I'd rather see them pulled from the makefiles in the sf and put back when the maker of the board certifies that they work. if you want, comment them out and commit That would do the job. Is that a good pattern for future similar issues? Should we give some sort of notice? first choice for broken code - talk to the guy who's working on it a good clear comment in the commit message serves as notice. I'm a believer in allowing broken code to continue to exist in the sf but in the case of a driver for some external device that is proprietary and commercial, we should expect that "business" to handle keeping it up. is it broken because they changed the hardware, or because Paul is in the middle of something? I don't know. I believe that these did compile some time back. Perhaps fest work broke them. I don't think it was fest... for sure Fred and I didn't touch them I was thinking more like work in emcmot that made them incompatable. fred and I worked on homing and backlash only, I strongly doubt that we did anything to break them Okay. Can I help at all with HAL development? I'm having mental blocks about getting the rest of emc2 working extension/revision of HAL itself is on hold until emc2 works (we already have enough HAL stuff to do a stepper based machine) once emc2 works, then I'll move back to hal, probably start with a STG driver Where are we with emc2? free mode works (jogs, homing)... next step is to reconnect the coordinated mode traj planner that I disconnected earlier. Some of that code is very confusing, and I've been trying to simplify it, but I probably should just say "screw it" and just do something We'd have to get traj going in order to run a stepper mill with it now? part of the problem is that the interfaces to the user space code are not well documented, so I don't know what has to be duplicated exactly and what can be changed or simplified if you want to run g-code, yes I know that confusion. there are dozens of loose ends to be tied up, and I've been having trouble deciding where to start... burnout maybe... I've started writing some documentation as a way to get back into it Okay. You suggested just hooking the convoluted stuff back in. Would that get us a way to build a Sherline ready version? yeah I'd very much like to work with that and test it here. theres about 240 lines of core code in control.c that need to be hooked back in, not just drop in though, cause the surrounding world has changed that _should_ get us a working coord mode TP though after that, we need to deal with some config parameters that have been added but don't have corresponding NML messages and I/O hasn't even been started (not that sherline uses much IO) We need to connect the IO stuff to HAL? hello there, anyone here? hi rayh: not really... for testing so far, I'm using a modified version of the IO module that ignores estop and does nothing at all with any other IO quick question although it's about emc not emc2 is it possible to use estop with minimill? alex_joni: external estop button? jmkasunich: nice work with HAL, I've read HAL_Introduction.pdf, and found a few problems, should I mail you the comments? or pull an external estop relay with the estop from emc? yes rayh: external estop circuitry. rayh: actually both Okay. IMHO no. i've seen the way it's done in brigeportaux.cc alex_joni: I'm planning some revisions to HAL in a few months... your input would be most welcome But using bridgeporttask and bridgeportio will work if you overlap the parport addresses. i don't think it would be that hard to recode minimillaux.cc that was one of my questions, both use the same parameter in the INI I overlapped them for Smithy. Turn on spindle and coolant and watch estop. how about the other signals that overlap the step/dir signals? We didn't use axes 4 or 5 so that address is not used from motion. The input address is able to be read anyway. i see... i'll give that a try another question... i've manged finally to get my setup working (sbc, 2 geckos, 2 DC motors, 2 - 2500 encoders) it runs now with freqmod, i can do some G01 You just need to set the locations of the output and input signals where they will be for the mixed setup. but if I enter a G00, i always get an axis following error what speed is used for G00? sounds like rapid speed is too fast Probably asking for to many pulses for the speed of the processor. rapid speed, is that max speed on the axis? Yes. hmmm... How fast is the processor on the sbc? because I went in G01 with higher speed 300 MHz What is the value of INPUT_SCALE? it's an NS Geode GX1 i used 1000.0 that gives me 10 units/turn i will adjust this eventually on the mechanics, right now I'm testing just motors/encoders how hard is the type of steel used in hearboxes before you harden it.. gearboxes.. alex_joni: you'll never get real axis following error with the open loop stepper systems. The only way you get it is if the pulse rate generator can't keep up with commanded feedrate. oh, i see well, i guess i will have to use the multiplier in the gecko You could try a shorter PERIOD value. good point, about the PERIOD Set that down until the display begins to lag noticably. does anyone have a clue? i noticed the machine does work even with PERIOD commented out from the ini I don't know where the hold and setup variables need to be for Gecko drives. Matt did some work on this. somthing that can handle 1000nm after hardening jmkasunich: short question about HAL, how EMC/oriented is it? could you use it for other apps? I tried to make it pretty generic i think HAL could be a very good tool in writing apps especially Hardware-oriented ones with an HALGui, something like LabView I don't know how to write GUIs very well well... I'm pretty new to linux coding, mostly microcontroller programming so far, and some ANSI C how far do you see EMC2? alex_joni: I don't understand the question well... how far before production use oh, you mean when will it be done? yeap a few months i see possibly ready to be experimented with in a few weeks is there a TO_DO list? but completely production ready probably several months maybe I could help with some coding no TO-DO list yet i see, well... I'll send you an email with the comments to HAL_Intro first step is just to start understanding how EMC works... this is not easy ;-( well... I've been reading through the sources of EMC for some weeks now :) what parts do you understand? task? interpreter? motion controller? GUI? IO? I only understand part of the motion controller well... so far I was mostly interested in motion (so far to write a driver for a board I made, for encoder feedback), and now I went roughly through IO, I wanna use estop with minimill but... the code is huge ;) yeah :-( what would be of use? where should I begin? hard to say right now - I'm not sure where to work myself right now what kind of a gui will there be? for emc2? yes same as emc1 - we aren't redoing the GUIs (at least not anytime soon) aha, what gets redone? the GUI sends commands to "task" which in turn sends commands to the motion controller, which in turn generates position commands and gives them to HAL for motor control most of the EMC2 changes are in the motion controller i see Paul Corner has also done lots of work on libnml, which is the messaging library that passes commands between the parts it is written in C++ and I don't really understand it at all so no more rcslib libnml is a reduced version of rcslib cool. is NML usefull for anything besides message passing, shared memory, etc.? IPC I don't understand enough to answer that question ;-)