#linuxcnc | Logs for 2013-04-29

[00:02:57] <pcw_home> Yes field power --> limitswitch --> 7I77 input
[00:04:16] <pcw_home> This does mean you need to use the inverted input pins in HAL for limits
[00:25:55] <ssi> yeah that's fine
[00:26:03] <ssi> I was more concerned about just how much current those pins will sink
[00:26:10] <ssi> but I guess they must be limited internally :)
[00:27:37] <ssi> by the way, I discovered with the help of a kind gentleman in here earlier that I can remove a 0R SMD resistor in my drive and make my /INHIBIT lines into /ENABLE
[00:27:43] <ssi> so I have much saner enable lines now :)
[01:32:03] <ssi> I'm short a drive, but I think the wiring is complete
[01:32:25] <ssi> https://sphotos-b.xx.fbcdn.net/hphotos-prn1/941097_10100131793347582_2132713241_n.jpg
[01:32:43] <ssi> https://sphotos-b.xx.fbcdn.net/hphotos-ash3/944683_10100131793367542_829866394_n.jpg
[01:34:28] <mrsun> neat =)
[01:39:05] <ssi> and now I'm going to bed
[01:39:12] <ssi> it's been entirely too long a weekend
[01:39:19] <mrsun> hope i get my toothed belts today so ic an start making parts for the holding etc of them =)
[02:07:19] <DJ9DJ> moin
[07:22:13] <r00t4rd3d> "Vermont Telephone Company's gigabit internet service is live, half the price of Google Fiber"
[07:46:48] <mrsun> http://sphotos-f.ak.fbcdn.net/hphotos-ak-ash3/946773_10151561646648648_1895717031_n.jpg http://sphotos-h.ak.fbcdn.net/hphotos-ak-prn1/940902_10151561646633648_1764937000_n.jpg Z axis taking shape =)
[08:02:21] <mazafaka> hello, can anyone advice a seller of long drills, about 150 x diameter?
[08:02:41] <mazafaka> Or are there any electrodes for EDM?
[08:03:44] <jthornton> like gun drills?
[08:05:36] <mazafaka> maybe, 14 mm 2500 mm long
[08:05:55] <mrsun> ough
[08:05:58] <mrsun> thats a long drill :P
[08:06:05] <mazafaka> yeah
[08:06:30] <mrsun> what are you doing? hollowing out people? :P
[08:11:00] <mazafaka> mrsun: this is for new corporation, called Skynet. It's some computer or so who has evaluated the task and given out a list of necessary things to be in the stock 'for things after the dawn of people'. No one really understands what it means...
[08:11:59] <mrsun> mazafaka, ahh cool =)
[08:14:52] <mazafaka> yeah, I even think this can bring something new to our world! ^)
[08:14:59] <mazafaka> *:)
[08:16:47] <mrsun> mmhmm =)
[08:17:38] <mrsun> hmm question is how to mount the screw for the Z axis :(
[08:18:02] <mrsun> a rotating nut assembly would be neat but i guess it requires some more work =)
[08:19:59] <mrsun> and i dont have to buy end blocks etc if i use rotating nut :P
[08:20:10] <mrsun> but i guess it requires axial berings ...
[08:20:17] <mrsun> or whatever they are called, thrust bearings
[08:25:46] <syyl_ws> mazafaka, 150 times dia is not uncommon
[08:25:52] <syyl_ws> but in 14mm?
[08:25:53] <syyl_ws> phu :D
[08:27:53] <syyl_ws> for edm thats also a pretty deep hole :D
[08:29:34] <syyl_ws> sounds like something to outsource ;)
[08:29:40] <syyl_ws> "someone else can do that."
[08:30:41] <archivist> gun drilling is a special operation, the machine runs with very high pressure cooling to clear the chips
[08:33:16] <syyl_ws> and the drill gets a support
[08:33:25] <syyl_ws> like a moving steady rest on a lathe
[08:35:01] <jthornton> one more post and Andy hits the 4k mark
[09:02:22] <mazafaka> syyl_ws: http://www.unisig.com/gundrilling-education/what-is-gundrilling.php says about 200:1 and 400:1
[09:09:46] <syyl_ws> cool
[09:09:53] <syyl_ws> that outdates my knowledge :D
[09:10:07] <syyl_ws> 400:1 is crazy
[09:13:45] <Jymmm> cRaZY I tell ya... ¡¡¡ʎzɐɹɔ
[09:14:29] <PetefromTn> Those gun drills are sweet... I have also used the gurhring drills and they are amazing if you have TSC. My VMC unfortunatley does not but it was an option LOL...
[09:18:15] <mazafaka> Jymmm: just about 0.6''x100''
[09:18:43] <Jymmm> mazafaka: =)
[09:19:16] <mazafaka> Jymmm: =)
[09:30:31] <PetefromTn> Boy do I ever love tig welding up tiny rust pinholes in my coolant troughs....LOL
[09:30:46] <jdh> duct tape is quicker
[09:32:02] <PetefromTn> honestly man if I thought it would work or last I would have already tried it !!
[09:33:44] <PetefromTn> This is about as much fun as smashing your thumb with a hammer...
[09:35:46] <jdh> having smashed my thumb a time or two, I think you are wrong.
[09:35:46] <syyl_ws> :D
[09:37:16] <syyl_ws> i always try to miss my thumbs..
[09:39:27] <mazafaka> http://www.youtube.com/watch?v=3lwLYHrqPYw
[09:39:28] <Tecan> (3lwLYHrqPYw) "Over Powered Powerwheels Snow Jam Barbie Jeep Corvette F150 Mustang Drift Widescreen" by "MechanicalPlanet" is "Autos" - Length: 0:02:41
[09:39:37] <L84Supper> PetefromTn: did you decide on spot welding and then coating with polyester resin?
[09:41:39] <mrsun> ffs stupid radiuses and stuff on everything :P
[09:42:02] <mrsun> i guess i can make the upper radius just filing it off as it shouldnt be as important as the lower .. hmm
[09:43:02] <PetefromTn> L84Supper: Actually I am trying to hit all the bad spots and get it sealed up via tig welding and then I bought some west system epoxy that I am gonna mix up and coat the entire bottom of the coolant trough with it to seal it up permanently. Then I am gonna paint the inside with some industrial waterproof epoxy paint rated for immersion...
[09:44:16] <jdh> Imron(tm) paint
[09:44:44] <L84Supper> imron is a polyester IIRC
[09:45:05] <L84Supper> was the rage back in the 80's
[09:45:30] <jdh> still is on boats, and AwlGrip
[09:45:35] <L84Supper> polyurethane enamel
[09:46:39] <PetefromTn> This epoxy paint is pretty expensive but it is apparently waterproof, coolantproof, resistant to several acids, and most chemicals.... Sounds like just what I need. It is also made to go over bare metal...
[09:47:09] <PetefromTn> mazafaka: What is it that they are feeding those power wheels? Sound mean...
[09:47:24] <skunkworks> PetefromTn, how is the conversion coming?
[09:48:48] <PetefromTn> skunkworks: honestly it is going okay. I still need a braking resistor for the Z axis motor driver since the head is not counterweighted and I have to build an encoder mount and some kinda circuit to allow my encoder to only show one index pulse per spindle revolution but right now the coolant trays are kicking my ass...
[09:49:08] <skunkworks> yeck
[09:50:56] <PetefromTn> yeah they are not as bad as they sounds I am just a whiner.... they are mostly full of surface rust and I am needing to wire wheel all of that off then fix the tiny pinholes in the bottom of the trays and then I can epoxy the crap out of them. Should be able to get them sealed and back inside the machine this week hopefully.
[09:51:48] <skunkworks> can't you pour the epoxy in the coolant then roll the machine down the street... ;)
[09:55:31] <jdh> just use the coolant pump to spray the west epoxy, it will coat all the needed areas.
[09:55:47] <PetefromTn> I have thought about rolling the machine down the street LOL honestly like I said it is going quite well considering what I have invested in it so far.
[09:55:59] <PetefromTn> jdh: LOL yeah man I will get right on it...
[09:56:43] <mazafaka> PetefromTn: music is http://www.youtube.com/watch?v=YC-BUYDW-VA
[09:56:44] <Tecan> (YC-BUYDW-VA) "Reverend Horton Heat - Psychobilly Freakout" by "Steven Edmundson" is "Music" - Length: 0:03:37
[09:57:35] <jdh> I spent the weekend in N. Florida. I think that qualifies for PsychoHillBilly
[10:02:02] <PetefromTn> mazafaka: Music...http://www.youtube.com/watch?v=5vUDmFjWgVo
[10:02:03] <Tecan> (5vUDmFjWgVo) "Dire Straits - Brothers in Arms Mandela Live 1988" by "Gabriele Pelosi" is "Music" - Length: 0:08:32
[10:11:42] * r00t4rd3d takes voice from alex_joni
[10:44:33] <mazafaka> PetefromTn: I first time listened to guitars of Led Zeppelin.
[10:44:52] <jdh> in 1970?
[10:45:15] <jdh> wonder if iron man 3 woud be good in 3d imax
[10:46:37] <mazafaka> jdh: in 1970 I myself hardly was even in plans, besides parents did not know each other
[10:49:22] <mazafaka> And now look whom they have grown up... Oh... A perfect mutha... matha... mazafaka :)
[11:11:18] <ssi> weeee drawing wirebook diagrams
[11:35:07] <L84Supper> http://phys.org/news/2013-04-adapteva-parallel-boards-summer.html
[11:35:15] <L84Supper> http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone/posts
[11:36:00] <L84Supper> just needs some buffers/level translators and some analog to run servos or steppers
[11:39:44] <pcw_home> The Parallela is very neat
[11:41:17] <pcw_home> Also has GigE (and 64 GFLOP parallel processor)
[11:42:30] <L84Supper> Linuxcnc + image processing on the same board
[11:42:55] <L84Supper> but at $99 vs an x86 PC, how does it compare?
[11:45:01] <pcw_home> Power wise and card-card bandwidth wise its a big win
[11:45:10] <L84Supper> I'd like to see a universal open control platform for Linuxcnc, Comedi etc
[11:46:20] <pcw_home> its really optimized for desktop supercomputer applications
[11:46:39] <L84Supper> control machines of all sorts, ARM PLC for low cost
[11:47:09] <pcw_home> weak on I/O, strong on card-card comms
[11:47:45] <L84Supper> i have look at the cache in the Epiphany, to see what grained apps it might be useful for
[11:48:25] <L84Supper> we went though this in the 00's with clusters
[11:49:06] <L84Supper> if you needed really fast IO between nodes how much processing was actually going on?
[11:50:09] <pcw_home> Almost all parallel processors need neighbor information (and fast)
[11:50:35] <L84Supper> you might be surprised
[11:51:07] <WalterN> L84Supper: http://www.sintecoptronics.com/scanlens.asp
[11:51:29] <L84Supper> some of the largest clusters at LANL just used 10Mb Ethernet
[11:51:39] <pcw_home> perfectly partitionable tasks that match the processing element capabilities a rare
[11:51:53] <WalterN> L84Supper: thats the only place I could find that had 808nm f-theta scanning lens
[11:51:54] <L84Supper> 10/100 just because it came with the motherboard
[11:52:05] <pcw_home> so its possible but wasteful unless you have good interconnects
[11:52:11] <L84Supper> WalterN: there were some on ebay
[11:52:26] <WalterN> rlly?
[11:52:28] <WalterN> hmm
[11:52:33] <pcw_home> (I had a Intel cube at one time, ex UCB)
[11:52:51] <WalterN> I guess there is
[11:52:59] <WalterN> no wait
[11:53:16] <pcw_home> coax 10 mbit Ethernet Hypercube
[11:53:24] <L84Supper> the cluster requirements were mostly lots of cores, lots of RAM, highest clocks, no HD
[11:54:06] <L84Supper> the GPU clusters were pretty much considered failures since nobody wanted to spend the time reprogramming
[11:54:32] <WalterN> I cant find 808nm f-theta lens on ebay
[11:55:39] <WalterN> why is 1064nm f-theta lenses so popular (it seems)?
[11:55:47] <L84Supper> 606nm is a oddball so you have to check the specs on the lenses, most of the ~1um lenses should work
[11:56:31] <WalterN> well
[11:56:45] <WalterN> the only place that listed how well it works was thorlabs
[11:57:01] <WalterN> and it only has 88% light transmission at 808nm
[11:57:43] <WalterN> at 1064nm it has something like 98%
[11:57:55] <L84Supper> yeah, so you'll lose 12% in the lense and 2-3% at the mirrors
[11:58:18] <L84Supper> so ~15% loss
[11:58:20] <WalterN> 2-3 watts in one lens
[11:58:34] <WalterN> plus the fiber optic cable
[11:58:40] <WalterN> meh
[11:59:49] <WalterN> but... why 1064nm?
[12:00:16] <pcw_home> YAG
[12:01:15] <L84Supper> http://en.wikipedia.org/wiki/Nd:YAG_laser
[12:06:12] <WalterN> and they are more common than 808nm?
[12:07:46] <L84Supper> I'd still like to see an ARM tablet that may be used as a PLC to control machines and systems
[12:08:18] <jdh> why a tablet?
[12:08:45] <L84Supper> low cost built in display
[12:08:51] <pcw_home> Laser diodes are often used to pump YAG lasers
[12:09:03] <jdh> display dies, then it's worthless
[12:09:17] <Tom_itx> tablets don't have very good interface to the outside do they?
[12:09:17] <jdh> and awkward for IO
[12:09:26] <Tom_itx> ^^
[12:09:30] <WalterN> pcw_home: isnt there losses in such a methodology?
[12:09:30] <L84Supper> jdh: yeah, the you just replace the whole thing for $100
[12:09:32] <jdh> could do modbus/tcp or ethernet/ip
[12:09:52] <jdh> but, I'll stick with traditional PLC's and HMIs as needed
[12:09:53] <pcw_home> Yes sure but gains as well
[12:10:53] <WalterN> does more power come out than what goes in?
[12:11:03] <pcw_home> The problem with that is that the whole tablet ecosystem will change in 3 years and you have an unrepairable machine
[12:11:46] <L84Supper> linux arm will be around
[12:12:02] <pcw_home> I'd stick with Ethernet interface to a MachineKit HAL runner
[12:12:17] <pcw_home> Linux ARM will be , the hardware will have moved on
[12:12:28] <L84Supper> non-ARM?
[12:12:50] <pcw_home> I would separate the HMI from the RT
[12:12:58] <jdh> I did an alignment control system for a 2.5kw YAG. Bad things happen to the fiber if the beam points directly at the target surface.
[12:13:13] <L84Supper> I'm thinking of mostly non real time applications
[12:13:34] <L84Supper> process machines
[12:13:52] <WalterN> jdh: woah... heh
[12:13:55] <pcw_home> With YAG you gain the ability to have very high power short pulses (better for marking ablating etc)
[12:15:57] <IchGuckLive> hi all B)
[12:41:10] <L84Supper> http://racine.craigslist.org/tls/3770766855.html mill with open-air belt drive
[12:41:17] <IchGuckLive> today a nice day in germany
[12:42:47] <IchGuckLive> L84Supper: you loos money if you paint this its historic
[12:43:34] <IchGuckLive> someone may shoot you as you do cnc on it for killing historical equipment O.O
[12:49:51] <L84Supper> heh, I'd shoot myself first :)
[12:50:24] <fragalot> L84Supper: prove it
[12:50:28] <archivist> L84Supper, what it would look like if grown up http://www.collection.archivist.info/archive/DJCPD/PD/2006/2006_04_09_skeleton_base/P4072255.JPG
[12:50:34] <Tom_itx> i had an old buffalo drill press that looked similar to that
[12:51:09] <archivist> other pics it is in http://www.collection.archivist.info/searchv13.php?searchstr=skeleton
[12:51:25] <L84Supper> love it, just like China
[12:51:48] <archivist> that was the basement where I used to work
[12:52:10] <L84Supper> that's the one thing that really stands out, lack of machine guards
[12:52:26] <archivist> guards get in the way
[12:52:40] <Tom_itx> was loss of limb a higher rate then than now?
[12:52:43] <Tom_itx> i doubt it
[12:53:25] <L84Supper> http://chicago.craigslist.org/nch/tls/3765503151.html still there, I half tempted to offer $1500
[12:54:03] <Tom_itx> probably cost that much to move it
[12:55:11] <L84Supper> did they give away Shopsmiths in the 60's in boxes of cereal or something?
[12:55:37] <Tom_itx> cracker jacks prize
[12:56:15] <L84Supper> http://chicago.craigslist.org/chc/tls/3761822816.html supposedly works
[12:58:58] <L84Supper> people must be clearing out their old machines around here, there's never this much available at the same time
[12:59:21] <jdh> perhpas they are ramping up for inreased demand
[12:59:40] <IchGuckLive> pattern a sub > http://pastebin.com/uCRsXYTC
[13:02:20] <IchGuckLive> L84Supper: as the outsell of the new mashines also starts
[13:11:52] <L84Supper> looks more like spring cleaning
[13:13:18] <andypugh> I am sure that presentation can halep a lot with an eBay advert: http://www.ebay.co.uk/itm/Metal-working-spin-lathe-/251267994488?pt=UK_Home_Garden_PowerTools_SM&hash=item3a80bd5378
[13:17:01] <L84Supper> http://chicago.craigslist.org/wcl/tls/3769052303.html "Needs work" an understatement
[13:17:25] <frallzor> haha =P
[13:24:57] <asdfasd> andypugh: they mean scrap but mistakenly called it lathe...
[13:25:52] <archivist> I can see plenty of usable castings and free line shafting
[13:26:12] <andypugh> It's possibly rather a useful metal spinning lathe.
[13:26:56] <andypugh> A picture of the bed would have helped...
[13:26:56] <archivist> I agree, I can see the important bits
[13:27:25] <archivist> the bed is there in front left
[13:27:35] <andypugh> It's worth £50 for the Record vice hiding underneath :-)
[13:28:00] <asdfasd> I see a lots of bits to but all of them look like collected from scrap or museum :)
[13:28:02] <archivist> legs for the bed seem to be extra!
[13:28:15] <frallzor> http://www.youtube.com/watch?v=vvBzxnOdn90&feature=youtu.be When bored, make a crappy video =P
[13:28:16] <Tecan> (vvBzxnOdn90) "RenderCNC_routing_720P" by "ODDNAME" is "People" - Length: 0:01:27
[13:28:19] <andypugh> archivist: Yes, I know you can see some of the bed, but why not frame the photo to show all of it?
[13:28:42] <archivist> useless photographer :)
[13:29:26] <archivist> near you, go fetch
[13:30:23] <andypugh> frallzor: Am I right in thinking that you are in Scandiwgia?
[13:30:49] <frallzor> indeed
[13:32:26] <andypugh> Did you see my message on the mailing list about a guy with some ABB robots? He doesn't seem to want to make them work, so was going to break for parts. As he seemed to be thinking 6 x €100 for the harmonic drives + motors I get the impression that that is the sort of all-in price he is thinking.
[13:32:35] <L84Supper> http://www.ebay.com/itm/1993-VMC-1000-cnc-milling-machine-mill-tool-changer-/200906276833?pt=LH_DefaultDomain_0&hash=item2ec6f27fe1
[13:32:49] <L84Supper> isn't this low for a price? $5k
[13:34:46] <WalterN> not enough lasers
[13:35:44] <Tecan> http://i.imgur.com/jSji0in.jpg
[13:38:00] <IchGuckLive> tjb1: ?
[13:38:07] <tjb1> IchGuckLive: ?
[13:38:33] <IchGuckLive> today full load on china plasma 8hr non tiome off
[13:38:43] <tjb1> nice
[13:38:53] <IchGuckLive> noozle change every 2hr at 180m of cut
[13:39:01] <IchGuckLive> worked
[13:40:43] <IchGuckLive> so no hypertherm 45 at 2400 euros its a stamos 85 at 600 euros
[13:41:58] <tjb1> you need to find better supplier
[13:42:15] <tjb1> I can get one for 1230 euros
[13:42:30] <tjb1> I have to go to graduation rehearsal, cya later
[13:42:37] <IchGuckLive> By
[14:06:53] <frallzor> Ahh I want my ordered wood! =(
[14:07:11] <frallzor> that sounded better in my head
[14:10:14] <DaViruz> frallzor: that's because your head is made of wood.
[14:11:21] * frallzor loves Birch
[14:14:00] <archivist> frallzor, bite on an apple https://plus.google.com/photos/108901587400063562323/albums/5872301069131746849?authkey=COvHqZiHz_e_Jg#photos/108901587400063562323/albums/5872301069131746849?authkey=COvHqZiHz_e_Jg
[14:14:46] <archivist> not mine, seller wants 50$ on the old tools mailing list
[14:15:13] <frallzor> I cant bite since I cant see that link =P
[14:15:31] <L84Supper> the Chinese viewpoint on plasma cutters and welders is: it needs to be carried on a scooter, it's going to be replaced in no more than 2 years, has to be inexpensive
[14:16:38] <WalterN> I thought that was their view on everything
[14:17:36] <frallzor> I thought it was "cheap and crappier than the thing we copied"
[14:17:38] <L84Supper> they buy the same machine tools that we do
[14:17:53] <frallzor> minimum of atleast 1 major flaw on the copies
[14:18:12] <L84Supper> nah, they usually copy the flaws as well
[14:18:30] <frallzor> and add some of their own, you know, artistic freedom
[14:18:36] <L84Supper> the changes come later when they shop for parts
[14:18:44] <frallzor> well practiced in the far east
[14:19:27] <archivist> shop for parts! they reduce the parts count by leaving out seals etc
[14:19:45] <frallzor> nah they remove useless parts
[14:19:49] <L84Supper> if it didn't leak and you didn't notice
[14:20:39] <L84Supper> yeah, that's pretty much it, like they guy that used to make TV set back in the 50's, he would snip out parts until it stopped working
[14:22:10] <WalterN> lol
[15:16:38] <generic_nick|2> wowsers were my gibbs loose on the mill. no wonder why i was having chatter issues.
[15:17:09] <generic_nick|2> good thing every part i make uses the entire travel so it isnt worn unevenly.
[15:20:12] <mrsun> http://sphotos-b.ak.fbcdn.net/hphotos-ak-ash3/575358_10151562256848648_904501495_n.jpg http://sphotos-d.ak.fbcdn.net/hphotos-ak-frc1/482583_10151562256818648_2074860697_n.jpg closer and closer i get =) now most grunt work is done! =) time to realy look at getting it moving =)
[15:25:46] <frallzor> boooh
[15:27:49] <mrsun> haha :P
[15:28:03] <mrsun> love you to frallzor <3
[15:29:08] <frallzor> =P
[15:40:34] <generic_nick|2> mrsun: what is it?
[15:40:44] <mrsun> router
[15:40:49] <mrsun> Z axis of it
[15:43:52] <frallzor> that´s not a knife, THIS is a knife!
[15:44:01] <frallzor> I mean router http://img16.imageshack.us/img16/7077/img4898ql.jpg =P
[15:56:15] <DJ9DJ> gn8
[17:20:44] <andypugh> Bah! I have half a dozen ER20 nuts heading my way from Hong Kong, and half a dozen ER25 collet holders waiting for them.
[17:41:49] <Tom_itx> awesome
[17:42:23] <Tom_itx> you should buy a machine to fit them
[17:42:46] <Tom_itx> can never have too many toys
[17:46:30] <Tecan> http://www.youtube.com/watch?v=jac80JB04NQ << Metal Gods - Jusas Preist
[17:48:31] <ssi> ok so
[17:48:34] <ssi> pid tuning
[17:48:37] <ssi> never easy :/
[17:51:58] <ssi> my current-mode drives seem to want fairly huge values in P
[17:52:01] <PCW> Velocity mode drives are normally pretty straightforward to tune (since you are really only tuning FF1 and P)
[17:52:16] <ssi> yeah, I wish these were velocity mode!
[17:52:27] <ssi> got any hot tips for current mode tuning?
[17:52:36] <ssi> I have P at 3000 right now, and the axis is stiff
[17:52:43] <ssi> but it will oscillate a little bit
[17:52:50] <ssi> can stop it by hand
[17:53:26] <PCW> For torque mode drives I set the D term as large as its stable and then P as large as it stable
[17:53:58] <ssi> what should I have P at while I'm bringing D up?
[17:54:00] <ssi> 0? 1?
[17:54:08] <ssi> or leave it at 3000 like it is nom
[17:54:15] <PCW> 1 is fine
[17:54:18] <ssi> ok
[17:54:39] <PCW> so you know where the oscillation /noise is coming from
[17:54:42] <ssi> there's NO stiffness in the drive with P = 1
[17:55:01] <PCW> also dont bother tuning bare motors
[17:55:51] <PCW> tuning will be totally different with on inertial load
[17:55:59] <ssi> it's in the machine
[17:56:01] <PCW> s/on/an/
[17:56:03] <ssi> I'm moving the table
[17:56:09] <ssi> X axis on a big 10x50 mill
[17:56:26] <ssi> so with P = 1, I'm not really sure what constitutes "stable" D
[17:56:38] <ssi> if I set D to 100, it doesn't do anything until I move the axis a bit
[17:56:46] <ssi> then it growls and starts to move one direction or another
[17:56:57] <ssi> if I set it to D = 50, it's still growly when I move it, but it will stay still if I don't touch it
[17:57:01] <ssi> is the latter "stable"?
[17:57:18] <PCW> it should act like damping (resisting the motion)
[17:57:58] <PCW> also make sure that you are using the feedback-deriv input on the PID component
[17:58:14] <ssi> ooh I'm probably not
[17:58:36] <ssi> is that the only hal change?
[17:58:52] <PCW> that will make for A crunchy D term
[17:59:27] <PCW> yes you wire encoder velocity to the pid comps feedback deriv pin
[18:00:32] <ssi> nothing wired to pid.x.feedback anymore?
[18:00:45] <PCW> otherwise the PID comp uses DP/DT (encoder counts per sample) and you get very coarse velocity feedback)
[18:01:08] <PCW> you dont disconnect anything, just add a connection
[18:01:20] <ssi> oh wait
[18:01:22] <ssi> ok sorry
[18:01:41] <ssi> I had encoder velocity (x-vel-fb) wired to pid.command-deriv
[18:01:54] <PCW> OK thats good
[18:02:18] <ssi> sorry I accidentally changed x-pos-fb => pid.x.feedback to .feedback-deriv
[18:02:20] <ssi> but that's wrong
[18:02:48] <PCW> Yep, dont do that !
[18:04:31] <ssi> ok with D up pretty high, it's definitely resisting my motion,
[18:04:35] <skunkworks> interesting timing... http://www.cnczone.com/forums/product_announcements_manufacturer_news/126952-mycnc_-_computer-numeric-control_control_software-3.html#post1263454
[18:04:46] <ssi> but then with a bit of push, it'll reverse and go off on its own in the other direction
[18:05:28] <skunkworks> makes me think even more that he used linuxcnc in his motion control product...
[18:08:00] <PCW> The small P term should slowly pull it close to 0 error
[18:08:33] <PCW> if you set P to 0 is should be pure viscous damping
[18:08:59] <ssi> what signals would be useful to watch on scope for this?
[18:09:05] <ssi> it's definitely not going to 0 error
[18:09:08] <ssi> it's running away from it very slowly
[18:09:25] <PCW> with P of 1?
[18:09:29] <ssi> yes
[18:09:33] <ssi> maybe that's too much D
[18:10:08] <PCW> No D should only add damping (resist motion)
[18:10:24] <ssi> ok I made D much much smaller
[18:10:28] <ssi> and it's more like that
[18:10:32] <ssi> although it's not very much damping force
[18:10:47] <ssi> and it seems like more damping in one direction than the other
[18:11:05] <ssi> no, that's just because I'm on one side of the setpoint I think
[18:11:19] <ssi> if I run it way past the set point to the other side, it starts getting more damping force
[18:11:34] <ssi> but I mean, I can wind the axis an inch by hand before it starts to develop any significant resistance
[18:11:35] <r00t4rd3d> do I need to tap the end of extrusion to get it to take a 5/16 bolt?
[18:11:42] <ssi> then when I let go, it winds itself bak toward setpoint
[18:12:18] <PCW> Yes you have a weak restoring force proportional to position error
[18:12:23] <ssi> ok
[18:12:45] <PCW> If D is high it will slowly creep back
[18:13:17] <PCW> so set D as high as you can without excessive bussing
[18:13:56] <ssi> seems to be around 40
[18:14:36] <ssi> now raise P until it oscillates?
[18:16:07] <ssi> P=3000, D=40 is getting there
[18:16:31] <PCW> Yes or you can take a step and set P so you have just a little overshoot
[18:16:43] <ssi> how would I effect a step?
[18:16:59] <ssi> with a set in hal configuration dialog?
[18:17:18] <PCW> set too high accel and jog is one way
[18:17:50] <PCW> or use siggen component
[18:18:46] <PCW> in any case next steps will need HALScope
[18:19:31] <ssi> got halscope up
[18:19:48] <ssi> watching x-pos-fb, x-pos-cmd, and pid.x.error
[18:22:18] <PCW> Once you have about the maximum gain in P and D that are stable and not too buzzy you can tune FF2 (acceleration feed forward)
[18:24:00] <ssi> ok I have a not-quite-step response
[18:24:05] <ssi> but it's a big accel
[18:24:11] <ssi> I get three rings
[18:24:27] <ssi> and often I'll get oscillation after it stops, that I can damp with my hand
[18:24:27] <PCW> FF2 compensates for inertia by applying torque proportional to acceleration
[18:24:47] <ssi> https://pbs.twimg.com/media/BJDgpeJCcAALmp3.jpg:large
[18:25:31] <ssi> that move is violent too... it shakes the whole machine
[18:26:22] <ssi> doh
[18:26:28] <ssi> I lost my D term when I restarted
[18:26:44] <PCW> Oops
[18:27:09] <PCW> the calibrate routine is somewhat funny
[18:27:38] <ssi> so with the D in there, rather than an oscillation after one of these moves, I just get a little buzzing sometimes
[18:27:44] <ssi> but I still get the three rings like in that picture
[18:27:51] <ssi> should I turn down the P to get less ringing?
[18:28:53] <PCW> can you show the PID output? (it may be saturated)
[18:30:27] <ssi> it's not saturating
[18:30:41] <ssi> I'm plotting it now, and also watching the saturated pin on the component
[18:31:13] <PCW> you could try more D
[18:31:49] <PCW> though it may get too buzzy
[18:32:35] <ssi> how's this look:
[18:32:36] <ssi> https://pbs.twimg.com/media/BJDij9BCEAAvUxg.jpg:large
[18:33:33] <PCW> Its a start, try with sane acceleration
[18:34:20] <ssi> it looks pretty good with low accel
[18:34:34] <ssi> but it does buzz a little bit unless I touch the handwheel gently
[18:35:57] <ssi> also I'm concerned because I can move the axis enough with the handwheel with light pressure that would fault the axis if I had sane ferrors
[18:37:07] <PCW> what is your encoder resolution (counst/inch)
[18:37:16] <ssi> 40,000
[18:38:01] <PCW> so you may be able to have high D and P gains and get rid of some buzzing with a small deadzone
[18:38:31] <PCW> s/high/higher/
[18:39:18] <PCW> what happens if you double D and P (in that order so it does not go unstable))
[18:40:19] <ssi> I was at D 40, P 1000
[18:40:26] <ssi> doubling D makes it oscillate about 0.010"
[18:40:40] <ssi> k doubled both
[18:40:46] <ssi> it's oscillating slightly
[18:41:19] <PCW> Buzzing a few encoder counts or oscillating?
[18:41:29] <ssi> oscillating
[18:41:58] <PCW> what processor are you running this on?
[18:42:30] <ssi> pc? it's a dell 755 core2
[18:42:46] <ssi> https://pbs.twimg.com/media/BJDkwyyCEAAYfSB.jpg:large
[18:42:49] <ssi> here's it oscillating
[18:43:40] <PCW> You might try setting the servo thread up to 4KHz or so if you have the HP
[18:43:50] <ssi> it's already at 4khz actually
[18:44:10] <ssi> I dropped it from 1000000 to 250000
[18:45:22] <andypugh> PID output looks wierdly correlated to error
[18:46:11] <ssi> andypugh: it's a current mode drive, if that explains the weirdness
[18:46:35] <andypugh> No, pid error and pid output should have a simple mathematical relationship.
[18:46:45] <PCW> Can you watch the PID output when you push the handwheel to get an idea of how much torque you have?
[18:46:56] <ssi> sure
[18:47:02] <ssi> I started with the current limit on the drive relatively low
[18:47:08] <ssi> but I've bumped it up quite a bit
[18:48:05] <ssi> not really sure what I'm looking at, but I can saturate the pid by hand
[18:48:20] <PCW> You will have bizarre behavior if you ever get the drive into current limit
[18:49:08] <ssi> bizarre how?
[18:49:41] <PCW> If you can saturate it by hand either your current limit is too low, you dont have enough gearing, or a big enough motor/drive
[18:50:10] <ssi> I would believe insufficient current limit above anything else
[18:50:20] <ssi> because the motors and gearing were all in place before I started
[18:50:46] <PCW> Yes it will then have non-linear behaviour
[18:51:27] <PCW> Are these the original motor drives (and do you have the motor specs?)
[18:51:27] <ssi> ok drive current limit is now at max
[18:51:50] <PCW> be careful...
[18:52:31] <ssi> I can still overcome it by hand, but it takes most of my strength
[18:52:40] <ssi> I'm still running with D and P doubled, which is oscillating
[18:53:12] <ssi> these are 15A drives, fwiw, and my servos are 3.4A continuous stall, 18A max
[18:53:51] <PCW> OK so pretty close
[18:54:56] <PCW> OK so lower P and D so it stops oscillationg
[18:56:09] <ssi> have it at 2000, 50 right now
[18:56:12] <ssi> and it's ok, but it's buzzy
[18:57:02] <PCW> a little buzzy might be cured with small deadzoe
[18:57:21] <PCW> (assuming its encoder dithering)
[18:57:43] <ssi> I dunno
[18:57:50] <ssi> the DRO is jumping close to 0.001"
[18:58:06] <ssi> the tenths digit jumps a lot, the hundredths digit flips occasionally
[18:58:17] <ssi> also if i set a ferror of 0.005", it faults all the time
[18:58:27] <PCW> so its still oscillating
[18:58:47] <PCW> lower both gains till it stops
[18:59:51] <andypugh> Here's a strange thing. I just checked my PID terms on my mill.
[19:00:46] <andypugh> On the Y axis the D term is negative. On the Z axis it is positive. Assuming that the Y had to be wrong, I made it positive, and the machine went into oscillation.
[19:01:34] <PCW> That sounds bizarre
[19:01:52] <andypugh> Yes.
[19:02:06] <andypugh> It's very small, too. -0.08 for a P of 50.
[19:02:42] <PCW> I would look at the position and velocity from the resolver and see if they make sense
[19:03:07] <andypugh> Maybe it is damping response to step-inputs rather than damping the motion?
[19:04:12] <PCW> it should just cause a PID output that opposes motion, nothing more
[19:05:26] <PCW> And it should be possible to trace how its gotten backwards
[19:06:26] <ssi> I think I'm pretty close to something workable
[19:06:35] <ssi> I am at P=2000, D=30
[19:13:25] <PCW> Next step is FF2 (you adjust this to minimize error during acceleration)
[19:17:41] <ssi> aha
[19:19:26] <ssi> what order of magnitude should FF2 be?
[19:20:13] <PCW> well starting with F= MA...
[19:20:29] <PCW> probably small to start
[19:21:09] <ssi> I don't have a huge amount of accel error
[19:21:19] <ssi> but setting FF2=1 actually causes an error spike at the start of a move
[19:21:34] <generic_nick|2> ssi: do you have gain and feedback controls on the servo drives?
[19:22:14] <generic_nick|2> i could never use ff2, it always made it super rough.
[19:22:36] <ssi> generic_nick|2: yeah but they're in current mode, and the datasheet has suggested positions for them
[19:22:39] <ssi> one's max, the other zero
[19:22:54] <PCW> FF2 is needed for torque mode drives (just as FF1 is needed for velocity mode)
[19:23:07] <generic_nick|2> huh i dont use ff1 either lol
[19:23:52] <PCW> If you have velocity mode drives you will get poor performance without FF1
[19:24:21] <ssi> PCW: 0.1 FF2 doesn't seem to hurt anything
[19:24:26] <ssi> but it doesn't really help either
[19:24:38] <generic_nick|2> i know on any drive i have used if the gain on the amp is set too low it needed a ton of p and never had good holding torque. i have always turned the gain up a little
[19:24:46] <ssi> I see more of a problem with steady-state error during a move than I do on acceleration
[19:24:46] <JT-Shop> generic_nick|2: you might want to look at my velocity tuning tutorial
[19:24:54] <ssi> adding as much as FF2=1 *creates* an accel error problem
[19:25:22] <PCW> Yes with all FF terms its a balance
[19:26:04] <PCW> You may also have accel set so low that the error is too small to see easily
[19:26:28] <generic_nick|2> i have no idea if mine are velocity or torque mode. however they do seem to work fine without any ff values
[19:26:40] <generic_nick|2> i have copley 215 drives
[19:26:49] <ssi> yeah lemme turn the accel back up
[19:26:55] <ssi> generic_nick|2: do you have tachometers?
[19:27:05] <generic_nick|2> yes
[19:27:06] <andypugh> I read somewhere that 90% of PID controlers are nowhere near being tuned right, but still work fine.
[19:27:15] <PCW> and the velocity needs to be high enough that the acceleration phase lasts a while on the screen
[19:27:22] <andypugh> I know none of mine are :-)
[19:27:37] <generic_nick|2> andypugh: that 90% includes mine lol
[19:27:59] <cradek> I read somewhere that 90% of statistics are just made up
[19:28:23] <PCW> not so high that you are more than say 1/2 full torque during accel
[19:28:33] <generic_nick|2> http://www.copleycontrols.com/motion/pdf/200_ScanedManual.PDF
[19:28:33] <PCW> (check PID output for this)
[19:28:48] <andypugh> One of mine has an I-term that is the product of three 8x8 maps, so there are 192 numbers to put in. I can't even explore all the degrees of freedom.
[19:29:51] <generic_nick|2> are drives with tach feedback generally velocity mode?
[19:30:00] <generic_nick|2> or does that not matter much?
[19:30:05] <andypugh> One big issue: This is an engine oil-pressure controller. I can't tune it for correct operation at less than the minimum safe pressure, because I am not allowed there.
[19:30:50] <PCW> Yes If you have tachs you have a velocity mode drive (or drive that can run in velocity mode)
[19:31:25] <andypugh> Anyway, time to call it a day. Night all.
[19:31:39] <PCW> 'Night Andy
[19:32:14] <generic_nick|2> ok there is no mention of velocity or torque mode in the drive manual from what i can tell so i guess they're velocity mode.
[19:32:21] <PCW> So for velocity mode drives you adjust FF1 for minimum error during slew
[19:32:38] <PCW> (see JTs tutorial)
[19:33:17] <generic_nick|2> cool ill play with it some.
[19:33:27] <generic_nick|2> and check out the manual. thanks
[19:33:45] <PCW> without FF1 the drive will always be behind the commanded position when in motion
[19:33:49] <generic_nick|2> i just got my drives and motors swapped out for these ones.
[19:33:58] <generic_nick|2> interesting
[19:34:23] <generic_nick|2> i get really good acceleration but i guess i am a little low on velocity
[19:34:55] <PCW> too much and you will lead position when in motion, and too little you will lag position
[19:35:19] <generic_nick|2> good to know.
[19:35:37] <generic_nick|2> if i can get more velocity out of these motors i will be pretty happy
[19:35:53] <ssi> PCW: somewhere around 0.1 or 0.12 FF2 is pretty good
[19:36:00] <ssi> PCW: any more than that and I get a bang on start and stop
[19:36:16] <generic_nick|2> even happier if i could find the prox sensors i had kicking around
[19:36:17] <ssi> as it sits right now, there's a little burble in the error at teh end of the move
[19:36:18] <ssi> but nothing major
[19:37:03] <ssi> well, maybe more major than it should be
[19:37:08] <ssi> on the order of +/-0.0015"
[19:38:00] <ssi> it looks like it rings on decel
[19:38:05] <L84Supper> http://tinyurl.com/cz9wdrw anyone ever use the Aerotech linear motors?
[19:38:09] <PCW> what acceleration
[19:38:27] <ssi> 20
[19:38:35] <ssi> I guess that's in/sec/sec?
[19:39:24] <ssi> https://pbs.twimg.com/media/BJDx3ZHCQAAGnNj.jpg:large
[19:39:33] <ssi> that's the ring at teh end of the move
[19:40:17] <PCW> may need to adjust D/P ratio
[19:40:20] <ssi> if I turn up FF2 a little bit, I get a much bigger peak, but fewer rings
[19:42:04] <PCW> if you add D does the ringing get better or worse?
[19:43:52] <ssi> does oscillating count as worse? :D
[19:44:01] <ssi> I'm monkeynig with it all right now, getting closer
[19:46:19] <PCW> L84Supper: does that motor even have the linearbearings?
[19:49:36] <L84Supper> PCW: yeah, air bearing on the forcer
[19:50:25] <L84Supper> oh no, thats for cooling
[19:51:12] <L84Supper> BYOAB bring yer own air bearing
[19:51:22] <ssi> PCW: my accel error looks pretty good; is there any way for me to dial out the static following error during moves?
[19:51:55] <PCW> you can try some FF1 first
[19:52:11] <ssi> I think this machine might have rough spots on its ways
[19:52:25] <ssi> during a fast move, it shudders and the error gets bigger
[19:52:27] <ssi> in certain spots
[19:52:44] <ssi> the owners told me they had someone polish the gibs because they were rusty from water based coolant
[19:53:18] <Valen> could be screws too
[19:53:39] <ssi> I've got it jogging 120ipm pretty nicely tho
[19:53:44] <PCW> how does it feel by hand
[19:54:29] <ssi> fine
[19:54:39] <ssi> Valen: I don't see the screw whipping at speed
[19:54:43] <ssi> it's a pretty big screw too
[19:54:44] <ssi> 30mm maybe
[19:55:09] <PCW> You can plot the rough spots with the PID output over a full range motion at constant speed (required torque to move)
[19:55:23] <Valen> it could be sloppy i mean
[19:57:34] <ssi> well I was trying to take a video
[19:57:39] <ssi> but my phone died :(
[19:57:52] <ssi> so I think it's tuned enough for this evening anyway
[19:58:01] <ssi> next thing I need to worry about is homing
[19:58:09] <PCW> If you are fairly happy with the tuning you can add some I term (beware, too much I term will cause wild oscillations so have a reasonable ferror)
[19:58:13] <ssi> I have limits working, but there are no home switches, so I'll need to use limits as home
[19:59:10] <PCW> It only had limits?
[20:00:46] <PCW> anyway bbl
[20:02:08] <ssi> yeah only limits
[20:02:34] <ssi> which sucks cause I'll need to home at one end of the table
[20:02:36] <ssi> I wonder how they used to home it
[20:04:17] <PCW> What machine is this?
[20:04:32] <ssi> it's an Atrump knee mill that was centroid-converted in the early 90s
[20:06:09] <PCW> OK so thats why it has the handles
[20:06:16] <ssi> yep
[20:06:56] <PCW> really bbl
[20:07:03] <ssi> :)
[20:09:04] <generic_nick|2> ssi a simple reed switch works
[20:09:30] <generic_nick|2> limit switch works too.
[20:09:52] <generic_nick|2> i have mine so it homes on the limit switch. just have it index latch in the opposite direction
[20:09:55] <Valen> you can home to limit
[20:10:11] <Valen> to be honest we have run our mill for ages without home or limits ;->
[20:10:24] <Valen> we still don't have limits but I have homing now at least lol
[20:11:31] <generic_nick|2> same here valen
[20:11:54] <generic_nick|2> my mill is too big and the servos too weak to do any real damage lol
[20:12:22] <ssi> my limits on my hnc are awesome
[20:12:26] <ssi> it holds amazing tolerances
[20:13:03] <generic_nick|2> i dont know what limits have to do with tolerances lol
[20:13:53] <generic_nick|2> the hnc just has microswitches for limits i believe
[20:14:15] <ssi> on a lathe it has everything to do with tolerances
[20:14:30] <generic_nick|2> and a prox switch
[20:14:33] <ssi> it's all prox switches
[20:14:35] <ssi> but there's three per axis
[20:14:53] <generic_nick|2> that has little to do with tolerances
[20:15:04] <ssi> ok fine you win
[20:16:27] <generic_nick|2> it uses the resolver for homing accuracy if thats what you mean
[20:17:01] <generic_nick|2> the prox switches help by telling it when it's time to latch, but thats about it
[20:17:22] <ssi> there's a prox in the gear as well
[20:17:28] <ssi> because there's multiple resolver indices per rev
[20:17:51] <generic_nick|2> right, that's so it knows when to use the resolver
[20:18:36] <ssi> without accurate homing, it's practically impossible to use a lathe effectively
[20:18:39] <ssi> is what I meant
[20:18:44] <ssi> mills don't really care as much unless you fixture
[20:20:09] <generic_nick|2> i fixture on the lathe too
[20:20:59] <ssi> everything on the lathe is "fixtured"
[20:21:01] <ssi> at least in X
[20:22:00] <generic_nick|2> i was just a bit confused what you meant, after you home the machine should be using soft limits so i always considered them just as failsafes.
[20:22:53] <generic_nick|2> was wondering why you relied on limit switches for accuracy. didnt know you meant for homing
[20:22:59] <ssi> what I meant was homing repeatability == diameter tolerance repeatability
[20:23:06] <ssi> and my hnc homes very accurately
[20:23:31] <generic_nick|2> yep, they are good for that.
[20:23:50] <generic_nick|2> the gear reduction for the resolver helps
[20:24:01] <generic_nick|2> tons of "counts" per rev
[20:24:39] <generic_nick|2> assuming you have the stock setup i guess
[20:26:28] <generic_nick|2> great little machine, it's the perfect size for a garage shop
[20:28:16] <generic_nick|2> hmmm i wonder if this pulley i welded is steel or iron. looks completely machined, no cast surfaces anywhere.
[20:28:30] <generic_nick|2> guess we'll find out once it cools.
[20:28:57] <generic_nick|2> cracks will be a dead givaway :)
[21:43:29] <ssi> ooo
[21:43:32] <ssi> my beagleboard black came in
[22:05:20] <r00t4rd3d> hah, they made my cnc thread at cnczone a stickey :)
[22:05:43] <r00t4rd3d> http://www.cnczone.com/forums/diy_cnc_router_table_machines/
[22:06:22] <ssi> lol nice
[22:06:38] <ssi> my mill conversion is pretty much done, hooray