#linuxcnc | Logs for 2013-12-20

Back
[00:14:03] <Loetmichel> re @ home
[00:16:45] <Loetmichel> *brrrr* 22hrs @work. (subtract 2 times 1h break) ... 20h work ant i am fully exhausted. i thing i am getting old and weak :-(
[02:12:47] <Deejay> moin
[03:32:26] <someone972> So from what I gather, the motion controller is a separate process from the task and io controllers correct?
[03:38:46] <archivist> all connected together with hal
[03:40:06] <archivist> what are you trying to do
[03:40:52] <someone972> I'm trying to understand how all these pieces work together at this point.
[03:41:09] <someone972> Eventually I'd like to offload the motion controller to a chip over usb
[03:41:49] <archivist> usb is generally silly due to latency issues
[03:43:01] <someone972> It's just for a cheap diy-machine, so I'm not too concerned about latency
[03:43:32] <someone972> It's probably going to run pretty slowly as well
[03:44:23] <archivist> steppers require better latency, things need to happen at the right time
[03:46:38] <someone972> Wouldn't sending the commands from the task process to an implementation of the motion controller on the control chip give plenty good latency for the steppers?
[03:50:32] <archivist> we get one or two people a month thinking they can just shove part in an external controller not realising the amount of work nor the power needed in that controller
[03:51:03] <archivist> just use a beaglebone black as your controller, it fits on there
[03:52:28] <someone972> That's an idea
[03:53:34] <someone972> In the meantime, it would still be useful to know how linuxcnc works
[03:54:33] <someone972> Am I correct in thinking that genhexkins is used for kinematics on a simple 3-axis machine?
[03:54:51] <archivist> the only person I know who has usb in the connection is https://github.com/araisrobo/linuxcnc
[03:57:11] <Mr_Mayhem> I think a 3 or 4 axis machine uses trivkins. however its spelled.
[04:01:49] <Mr_Mayhem> I am no expert, but I imagine the motion planner takes in a unit of gcode, and maps all axis movements in time against a stable time base using the accel and decel curves as set earlier. This description is way simplified, but about what I understand so far.
[04:02:24] <archivist> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?TrajectoryControl
[04:02:38] <Mr_Mayhem> Thanks.
[04:02:44] <archivist> and follow the link at the page bottom for more detail
[04:03:14] <someone972> Ah, never even found that page!
[04:03:31] <archivist> the look ahead will likely change real soon now
[04:05:44] <Mr_Mayhem> Wow, cool way to test the boundary conditions of a zillion combinations.
[04:06:40] <Mr_Mayhem> Just write scitzo-move script and compare actial with intended moves, ha that's clever.
[04:14:35] <Mr_Mayhem> I wonder if anyone has rolled their own linear motor coils, etc. I suppose a cnc-based linear motor winding machine would be a neat but quite complex diy project. One could wind one section at a time, and bolt them together. Just a brainstorm for coil rails, magnets-on-shuttle kind of linear motor.
[04:16:44] <Mr_Mayhem> You'd need a pile of mosfets I suppose, 2 to 4 for each coil? hmm, $$
[04:19:11] <archivist> look at the devel mailing list for the recent tests and upgrading of the planner
[04:20:15] <Mr_Mayhem> Ok, that should be interesting to see what they are planning, forgive the pun.
[04:21:35] <Mr_Mayhem> Seriously, I love reading stuff over my head to get exposed to the ideas being discussed, like how they are going to make it more capable or accurate.
[04:21:49] <Mr_Mayhem> And where the challenges lie.
[04:48:43] <Mr_Mayhem> When I write code, my bugz have bugs.
[04:49:47] <someone972> So when the motion subsystem is referenced, does it mean the trajectory planner, kinmatics, and per-joint controllers?
[04:54:00] <Mr_Mayhem> I don't know yet. Here is the top down view page I know about:
[04:54:01] <Mr_Mayhem> http://www.linuxcnc.org/docs/devel/html/config/emc2hal.html
[04:56:19] <Mr_Mayhem> Oh, and here:
[04:56:20] <Mr_Mayhem> http://linuxcnc.org/docs/html/code/Code_Notes.html#_architecture_overview
[04:58:39] <Mr_Mayhem> So looks like Trajectory and Kinematics are contained within the motion subsystem, according to the chart.
[05:00:53] <someone972> Ok, good to know
[05:02:26] <Mr_Mayhem> Heh, first time I saw it myself; trying to peer inside also to see what's going on in there.
[05:02:54] <Mr_Mayhem> Crash course in control systems, hehe.
[05:24:09] <Mr_Mayhem> More of a computer science question: Does a realtime system have to use shared memory and mutexes to be realtime in essence? Does that by defination rule out functional languages like erlang or others? Or are there other ways to go about it which can get by without shared memory or mutexes? Seems like it worked well for telephone switches running thousands of concurrent calls. But maybe not
[05:24:09] <Mr_Mayhem> good enough for reliable timing on motion control?
[05:25:25] <archivist> timing matters, some languages seem poorly designed for hardware control imo
[05:25:33] <Mr_Mayhem> I mean Erlang ran big switches but maybe is not good for motion?
[05:26:21] <Mr_Mayhem> Yeah, I bet most ideas flop when hard pressed to have tight fast timing constraints.
[05:26:39] <archivist> a switch is less realtime, the data travelling through it is realtime
[05:27:32] <Mr_Mayhem> Yeah, the control signals are not as high frequency as the pcm signals or mux, right.
[05:29:44] <Mr_Mayhem> I wonder what figure of merit is used to compare such systems beyond simple models, like more of a topographic model of sorts with multiple variables or phase spaces.
[05:30:44] <Mr_Mayhem> I guess you have the odds of something being late as the most important thing, like a bit error rate almost.
[05:31:57] <skunkworks> http://youtu.be/oUajH5BCOUQ
[05:32:09] <Mr_Mayhem> Reminds me of a geiger counter far from the source with the occasional click.
[05:32:47] <Mr_Mayhem> Except that click is a messed up part or drill bit :-)
[05:35:50] <Mr_Mayhem> Watched the vid. Is the first one multiple line segments, hard to tell. And that would mean the later run is more of a curve fitting thing
[05:35:51] <Mr_Mayhem> ?
[05:37:12] <skunkworks> the current planner has to be able to stop by the end of the next line segment. (the spiral is made up of small line segments) so that limits top speed. The second one has deeper look ahead.
[05:37:46] <archivist> the latency test measures the timing of the basic realtime loops
[05:38:04] <skunkworks> it pretty much gets up to the centripetal acceleration limit.
[05:38:11] <Mr_Mayhem> Very nice. So it can literally see or detect it all and do it as one motion curve?
[05:38:34] <Mr_Mayhem> Or sweep, as it were?
[05:39:07] <skunkworks> well - the way I understand it is he is using circular blends and can keep increasing the velocity with each line segment/blend.
[05:39:32] <skunkworks> it is actually still 'touching' each line segment.
[05:40:56] <Mr_Mayhem> Oh, so it is incremental, but because he is blending as he goes, he can build on what he just did and accelerate towards the machines capability.
[05:42:05] <Mr_Mayhem> Maybe that was a poor description, but it seems pretty clever. He hits all the points, but dosen't slow down.
[05:42:24] <skunkworks> yes - that is how I understand it.
[05:42:44] <skunkworks> bbl
[05:42:55] <archivist> look at the "Initial tests of circular arc blending" threads on the dev mailing list
[05:43:07] <Mr_Mayhem> Makes one wonder how many other jewels are hiding in the curve math, heh.
[05:43:20] <Mr_Mayhem> ok. will do.
[05:49:30] <Mr_Mayhem> On the real-time nature of telephone digital signals, it was neat to watch a multiplexer break sync lock when slowly turning down the rf power and watching the signal blend into the noise floor on the spec analyzer. It happens all of a sudden, rather than in direct proportion to the signal to noise ratio on the satellite. I.e. it follows the bit error rate curve. I guess that is roughly
[05:49:30] <Mr_Mayhem> equivalent to delayed instructions or steps in a cnc, it starts to shudder or worse.
[05:51:06] <Mr_Mayhem> Muxes I played with had relays so they chattered when re-syncing to the incoming data stream.
[05:59:54] <Mr_Mayhem> The curve fitting seems like the old problem of squaring the circle, except the other way around, it is trying to circle the square, or polygon in most cases.
[06:00:56] <Mr_Mayhem> Cusa would be proud.
[06:15:52] <someone972> Well I've had enough poking around for today, time for some sleep. My thanks to everyone who answered my questions and provided insights.
[06:16:02] <someone972> This board is always quite helpful
[06:35:45] <Tom_itx> http://www.electronicproducts.com/Digital_ICs/Standard_and_Programmable_Logic/LOGi_family_of_FPGA_boards_made_to_interface_with_Raspberry_Pi_Beaglebone_Black_and_Arduino_Due.aspx
[06:36:27] <Mr_Mayhem> saw that. Blends in nicely with what I had been thinking.
[06:36:48] <Mr_Mayhem> It was on Slashdot adn kickstarter, yah.
[06:37:31] <Mr_Mayhem> Oh, what fun that will be for motion control enthusiasts.
[06:37:53] <dgarr> Tom_itx: new for pin hal_manualtoolchange.change_button added to master
[06:38:09] <Tom_itx> oh thanks
[06:38:14] <dgarr> your logs are easier to read, thanks
[06:38:53] <Tom_itx> same bot as the other 2
[06:38:59] <Tom_itx> they could make the edits
[06:39:19] <Tom_itx> early start here.. later
[06:59:32] <eric_unterhausen> is there some correspondence between the cute names that debian uses for distributions and the cute names ubuntu uses?
[07:14:21] <jthornton> LOL
[07:36:03] <eric_unterhausen> there is a crew of jerks on the ubuntu forums that like to assert that random (often barely working) changes that canonical makes are good things, because change is good
[07:38:56] <archivist> I hate change
[07:52:48] <eric_unterhausen> I hate change for no reason, I especially hate change if it's not implemented properly
[07:54:39] <eric_unterhausen> I've been trying to figure out if there was a really, really good reason for the grub developers to do what they have done
[07:54:54] <eric_unterhausen> you used to be able to go in, change one file, done
[07:55:21] <eric_unterhausen> the reason why people went to grub over lilo was that it was much simpler, I guess these guys weren't around for that
[07:55:55] <eric_unterhausen> now you have to change 3 files to do anything, one of them through update-grub
[07:56:21] <eric_unterhausen> maybe they should rename update-grub 'lilo'
[10:34:16] <IchGuckLive> hi all B)
[10:44:34] <IchGuckLive> Question can a pyvcp checkbutton be set as defold
[10:44:54] <IchGuckLive> so i uncheck it
[10:45:28] <IchGuckLive> or do i need to invert the pin in hal
[11:07:09] <IchGuckLive> ok i did a not invert
[11:07:27] <IchGuckLive> so name the button off
[11:07:44] <skunkworks> that is one way of doing it :)
[11:08:09] <skunkworks> I think you can do some of this type of stuff with gladevcp
[11:08:19] <skunkworks> (default buttions)
[11:08:30] <IchGuckLive> i stay with pyvcp as long as i can
[11:09:08] <IchGuckLive> i did get the widget of the turn russion cnc-ckub on the run and understanding the things
[11:09:19] <IchGuckLive> so you can go for a D810 remode
[11:10:58] <IchGuckLive> ;-) talking on lathe Newbie is on emediat
[11:38:35] <IchGuckLive> hi flughafen how is it going
[11:40:28] <moist> is this the pncconf channel?
[11:41:05] <archivist> some use pncconf (not me)
[11:41:38] <flughafen> hey IchGuckLive
[11:41:42] <archivist> but in irc ask the real question
[11:41:43] <flughafen> It's good, yourself?
[11:46:07] <IchGuckLive> yes the party is on here for xmas already
[12:12:55] <flughafen> are you having a party IchGuckLive ?
[12:13:46] <IchGuckLive> naibour
[12:17:51] <flughafen> IchGuckLive: do you have a haus?
[12:18:01] <IchGuckLive> yes
[12:18:03] <Deejay> our haus is in the middle of our street
[12:18:05] <IchGuckLive> and a shop
[12:18:11] <Deejay> (house!)
[13:54:53] <Tom_itx> PCW, i had some really odd behavior with that custom bit file i made
[13:55:14] <Tom_itx> leaving out the B & I channels of one encoder and adding 2 sserial
[13:55:44] <Tom_itx> the encoder count worked ok for a while then it went nuts
[13:56:12] <Tom_itx> then the pendant jog started jogging on multiple axis at the same time
[13:56:23] <Tom_itx> not a desired effect
[13:57:21] <Tom_itx> i'm gonna go back to a more 'sane' one and try it again. by sane, i mean not leaving out any pins from the funcitons like A B & I on the encoders
[14:01:55] <Jymmm> I need $12 more for free shipping from amazon
[14:09:11] <cradek> "free"
[14:09:40] <Jymmm> Heh, yeah that
[14:10:32] <Jymmm> cradek: Free, as in UNLIMITED voice/text/data on your cellphone.
[14:20:27] <ReadError> Jymmm amazon prime!
[14:20:29] <ReadError> its so worth it
[14:20:47] <Jymmm> ReadError: Nah, I dont order THAT much
[14:20:55] <ReadError> Jymmm well you can start
[14:20:59] <ReadError> its super handy
[14:21:08] <ReadError> best investment ive made, ever.
[14:21:22] <ReadError> (well dogecoin is first, amazon prime, 2nd)
[14:21:29] <Jymmm> ReadError: Great, Insert *YOUR* valid credit card here --> [ ]
[14:21:41] <ReadError> do they accept dogecoin?
[14:21:48] <ReadError> you know you get a free month, right? :)
[14:21:54] <ReadError> and you can literally keep doing that
[14:22:01] <ReadError> guy has done that for years
[14:22:23] <Jymmm> multiple accounts?
[14:22:39] <ReadError> i would have to ask
[14:22:55] <ReadError> but i think you cancel and just start another trial on the same account
[14:24:14] <Jymmm> Ah, heh
[14:25:02] <Jymmm> ReadError: It's easier to just add $35 order, plus keeps the spending in check.
[14:25:22] <ReadError> i would say the opposite
[14:25:32] <ReadError> i needed 2 flat cat6 cables
[14:25:48] <ReadError> $5+$5 -> done
[14:25:48] <Jymmm> That's an oxymoron
[14:25:54] <ReadError> didnt pay a dime for shipping
[14:25:59] <ReadError> got them the next day
[14:26:57] <Jymmm> Part of the CAT6 specs is they way they are constructed to prevent crosstalk, so saying a "flat ethernet cable" is CAT6 doesnt make sense.
[14:28:26] <ReadError> well yea, but i needed sheilded
[14:28:32] <ReadError> to run analog video
[14:28:38] <ReadError> so crosstalk isnt a hug deal
[14:28:48] <ReadError> they move and flex better
[14:34:40] <Loetmichel> Jymmm: anyways one can desing a flat cable that meets or exceeds Cat6 specs
[14:35:08] <Jymmm> Loetmichel: Not for $6 =)
[14:35:41] <Loetmichel> i had some "belden multimedia cable" that was formed like a banana-> more or less flat (althoug only cat5E) and good for Gbit ethernet and about 0,35 eur per meter
[14:37:03] <Jymmm> Loetmichel: I was commenting on the CAT6 specifically.
[14:37:06] <Loetmichel> inside were the 4 foil shielded pairs beseide each other. it fits perfectly to an RJ45 plug
[14:37:29] <Loetmichel> i wouldn be surprised if belden has the cable upgraded to cat6
[14:37:56] <Jymmm> Loetmichel: But CAT6 isn't just the cable, it's the connectors too (as per spec).
[14:37:59] <Loetmichel> the time i bougt that 100ft drum there was no cat6 spec ;-)
[14:38:16] <Loetmichel> if you say so.
[14:39:11] <Loetmichel> i dont have enough spre money to buy these specs. in germany anything DIN/IEE related has to be bought. no "open source" there. and no small prices either. :(
[15:13:51] <PCW> Tom_itx: leaving out I/O pins can not possibly cause a problem like you see
[15:18:52] <PCW> if your MPGs are on the 7I84 make sure you have the MPG
[15:18:54] <PCW> ground pin connected to the field power ground _AT_ the 7I84
[15:25:26] <JT-Shop> Tom_itx, I just drilled and tapped some holes with the mill G code generator :)
[15:54:40] <asah> So... playing with the glass scales off my maho.
[15:55:10] <asah> I only have three wires coming out of the connector, two of them give me pulses, the other does not.
[15:56:02] <cradek> if there's an index equivalent it might be very short, or rare
[15:56:15] <cradek> recommend setting your scope to trigger on it
[15:56:23] <asah> I am going into a 7i52 which is expecting differential pairs for the A and B channels. what should I feed QA0/ if I am going to feed QA0 one of the signals
[15:56:53] <cradek> does it have jumpers for disabling differential?
[15:56:55] <asah> the "silent" wire is marked as "marker" on the schematics
[15:57:05] <cradek> yeah that's surely index
[15:57:07] <asah> which could be a home pulse.
[15:57:10] <asah> right.
[15:57:25] <asah> ok. haven't moved it all the way to home yet to check it.
[15:57:26] <cradek> how are you monitoring that?
[15:57:46] <asah> but on the subject of differential signaling, what is going to be the cleanest signal?
[15:57:56] <asah> (I am not heavy electronics guy)
[15:58:02] <asah> have an oscilloscope
[15:58:08] <cradek> differential is very much more noiseproof
[15:58:12] <asah> right.
[15:58:19] <cradek> that's the point of it - common-mode noise gets ignored
[15:58:23] <asah> I see much noise on the lines.
[15:58:35] <cradek> trigger on the index line and I bet you'll see it
[15:58:38] <asah> but I don't have differential signals coming out of the encoder
[15:59:17] <asah> these are phillips glass scales going into an exe pulse shaper board.
[15:59:19] <cradek> you have to either generate them or use input hardware that doesn't do differential. some of the mesa hardware is jumper-selectable.
[15:59:36] <asah> I am getting signal out of the exe board.
[16:00:29] <asah> so I would want TTL basically.
[16:00:46] <cradek> yes sounds like
[16:00:51] <asah> ah... I should rtfm.
[16:00:54] <asah> its in there.
[16:00:57] <cradek> or better, add the line drivers
[16:01:08] <cradek> haha
[16:01:50] <asah> thanks!
[16:02:04] <cradek> welcome
[16:16:10] <asah> ok, have it hooked up and am jogging.
[16:16:17] <asah> into the 7i52.
[16:16:39] <asah> the encoder count increments... YAY! but only in the forward direction.
[16:17:00] <cradek> BOO!
[16:17:09] <asah> when I go backwards I am not getting any incrementation
[16:17:14] <cradek> that's really weird
[16:17:15] <asah> or decrementation.
[16:17:21] <asah> just stays.
[16:17:43] <asah> yes.
[16:17:50] <cradek> I struggle to see how that could be a wiring problem. does the quadrature look reasonable on your real scope?
[16:17:58] <cradek> oh I've gotta run, hope you figure it out
[16:18:39] <asah> it is not really clean quadrature.
[16:18:44] <asah> thanks for the help. best!
[16:27:35] <Deejay> gn8
[17:12:05] <Tom_itx> pcw_home, the MPG is still on the 7i47
[17:12:45] <Tom_itx> i'm not sure what was causing the erratic behavior
[17:13:22] <Tom_itx> JT-Shop, i tried the face generator but i forgot my 1" face mill won't reach the table in the current configuration
[17:16:30] <Tom_itx> happy with the new cutters so far though: http://tom-itx.dyndns.org:81/~webpage/cnc/pendant/engrave1.jpg
[17:21:30] <Loetmichel> you could have put a scale in the pic
[17:21:34] <Loetmichel> bow big is that?
[17:21:54] <Loetmichel> how
[17:21:56] <Tom_itx> the plate is 3 x 6"
[17:22:11] <Tom_itx> approx
[17:22:20] <Loetmichel> nice
[17:22:45] <Loetmichel> i made something in lexan a while ago, turned out nice also
[17:22:50] <Tom_itx> i wanted to try the engraving bits so i did that ahead of the milling
[17:23:23] <Loetmichel> http://www.youtube.com/watch?v=Nx3sT3kQV-8
[17:23:47] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14247&g2_imageViewsIndex=1
[17:23:58] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14250&g2_imageViewsIndex=1
[17:24:40] <Loetmichel> that are M3 screws on the sides
[17:25:33] <Tom_itx> what did you use for the lettering?
[17:25:37] <Tom_itx> to generate it?
[17:27:59] <Loetmichel> normal fonts in corelDraw and then radius correction until it was just a double line instead of a rectangle.
[17:28:24] <Loetmichel> fun fact: comic sans works best for this ;-)
[17:29:13] <Loetmichel> i think i used OCR-A for the company lettering
[17:30:15] <Tom_itx> my cad cam has limited fonts but scale pretty easy
[17:38:44] <nlancaster> howdy all!
[19:23:43] <Mr_Mayhem> Loetmichel nice work. What are the electronics for?
[19:25:02] <Mr_Mayhem> Looks like you did halftone art earlier, looking at the circles on the wasteboard. Maybe just a perf pattern.
[19:28:04] <Mr_Mayhem> Oh I see a network router of some kind.
[19:30:18] <Mr_Mayhem> Connects ISDN T1 to Ethernet, looking at the connections. ISDN T1 is pretty stable for videoconferencing. More so than say shared ethernet.
[19:30:46] <Mr_Mayhem> But it can be a pain to configure. Need to plan ahead.
[19:31:06] <Jymmm> Mr_Mayhem: Um, ISDN is just 2 T1 channels
[19:31:53] <Mr_Mayhem> Yeah, PRI, right? One going and one coming.
[19:32:08] <Mr_Mayhem> 23 channels + signalling.
[19:32:38] <Mr_Mayhem> and then the cheap BRI which is like 2 channels and 1 signalling channel if I remember right.
[19:32:57] <Jymmm> 8Kb control
[19:33:00] <Jymmm> iirc
[19:33:44] <Mr_Mayhem> Right, the last 8k channel carries the control signals.
[19:34:18] <Mr_Mayhem> Lots of flavors of control signals, for voice to data to both.
[19:35:52] <Mr_Mayhem> We often had to work with the switch tech to match settings. Often we ordered it, and it wasn't set right, so we had to go around and around to make both ends the same settings.
[19:36:30] <Mr_Mayhem> But once it was up it usually was quite solid and reliable.
[19:39:57] <Mr_Mayhem> You can get very nice compressed video across a ISDN PRI, the codec math has advanced nicely over the years. I saw $100k satellite boxes the size of college dorm refrigerators that were used back in the day to perform the same compression, and they weren't as good picture quality. And one was needed at each end.
[19:44:45] <Mr_Mayhem> Companies used them anyway because it would cut the satellite bill in half, which is very expensive per hour.
[19:54:27] <Loetmichel> Mr_Mayhem: its a small 2*1W audio amplifier sitting in the frame of a military 24" monitor
[19:54:40] <Loetmichel> oh, sorry, wrong electronicx
[19:54:53] <Loetmichel> thiat was a military version of a ISDN router
[19:56:17] <Loetmichel> Mr_Mayhem: and i dod no halftone art
[19:56:28] <Loetmichel> i did a case for a NAS
[19:56:41] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14184
[19:56:50] <Loetmichel> ... needs some air in it ;-)
[20:02:20] <Mr_Mayhem> Hehe, you can do cool patterns.
[20:03:12] <Mr_Mayhem> Man, some nice gear there.
[20:04:26] <Mr_Mayhem> Looks like it's serious chassis design. Love that stuff.
[20:06:13] <Mr_Mayhem> Nothing like a beefy mil-spec box to put hobby goodies in. But usually too expensive.
[20:06:34] <Mr_Mayhem> Find them surplus is the way to go usually.
[20:07:13] <Mr_Mayhem> In your case you make them for the real deal, nice work.'
[20:08:37] <Mr_Mayhem> Do you fold the sheet metal yourself, or customise them?
[20:15:05] <Mr_Mayhem> I like your cnc dust shoe for the vacuum cleaner. It is similiar to a design I am working on.
[20:15:42] <Loetmichel> i fold myself
[20:15:55] <Loetmichel> its easy if you know how ;-)
[20:16:24] <Mr_Mayhem> Ah. Very nice.
[20:16:46] <Mr_Mayhem> Someday I will get a press.
[20:16:51] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14151
[20:17:01] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14154
[20:17:10] <Loetmichel> no need for a press. all done by hand
[20:17:28] <Mr_Mayhem> Ooooo.
[20:17:43] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=14160&g2_imageViewsIndex=1
[20:17:46] <Loetmichel> you see?
[20:17:49] <Mr_Mayhem> Very clever. Do you use any leveling sofware for z compensation.
[20:18:08] <Loetmichel> no
[20:18:30] <Mr_Mayhem> Just a very flat bed, correctly leveled to the cutter?
[20:18:32] <Loetmichel> the wooden "sacrificial plate" will be faced down from time to time to be lefel enough
[20:18:43] <Loetmichel> level
[20:18:49] <Mr_Mayhem> I see. You do the pattern to level the wood.
[20:18:54] <Mr_Mayhem> First.
[20:19:37] <Loetmichel> pattern?
[20:19:55] <Mr_Mayhem> Then your metal folds up easy into a box. I never saw that stunt before! Very clever, no break needed.
[20:19:59] <Loetmichel> no, just a big wood cutting tool and go for it
[20:20:37] <Loetmichel> http://www.youtube.com/watch?v=dEQZ_TY5q8o
[20:20:48] <Mr_Mayhem> Right, I mean usually the bed level routine goes round and round as it works. I understand i think.
[20:21:02] <Loetmichel> the pattern came from this: http://www.youtube.com/watch?v=NxxhvO6wP08
[20:21:26] <Mr_Mayhem> Hey, that's my machine! hehe
[20:22:19] <Loetmichel> its a cheap chinese 6040 CNC
[20:22:40] <Mr_Mayhem> I have the same one.
[20:22:41] <Loetmichel> boss didnt allowd more money
[20:24:18] <Loetmichel> before that machine i did all the prototyping on the machine i build at home: http://www.youtube.com/watch?v=BXNhsid6PD0
[20:24:45] <Mr_Mayhem> I am retrofitting it to use Mesa 5i25 parallel port + Meas 7i76 for breakout. http://www.mesanet.com/ Plus swap out drives for leadshine or equal type.
[20:25:14] <Mr_Mayhem> Oh, so you have your baby at home too. Nice.
[20:25:33] <Loetmichel> http://www.youtube.com/watch?v=nDFLvk_0UJ0 <- which isnt really made for that kind of work
[20:25:53] <Mr_Mayhem> I do tons of perf drilling in loudspeaker experiments.
[20:26:06] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=12533
[20:26:53] <Loetmichel> i do most of it because a hss drill will be dull after a few hundred holes... one of the NAS doors has 920 holes... -> not a good idea
[20:27:07] <Mr_Mayhem> I wrote a script in sketchup to draw the holes in the 60 degree pattern, but it still needs some work.
[20:27:13] <Loetmichel> the TC milling bits will hold about 5 complete casings
[20:27:31] <Loetmichel> i draw nearly all of my stuff in CorelDraw 8
[20:27:55] <Loetmichel> i sould get some time to learn to use my recently purchased Turbocad, though
[20:28:06] <Mr_Mayhem> I am funny, I use sketchup and a trial copy of cambam. I like it enough to buy the full app.
[20:28:15] <Mr_Mayhem> xmas present.
[20:28:47] <Loetmichel> recently: in january iirc
[20:29:01] <Loetmichel> :-)
[20:30:06] <Mr_Mayhem> hey is that acrylic thing a dust shoe? http://www.cyrom.org/palbum/main.php?g2_itemId=4287
[20:30:24] <Mr_Mayhem> Looks like one.
[20:30:45] <Mr_Mayhem> like for the vacuum cleaner.
[20:32:42] <Mr_Mayhem> So far, swapping my chinese yoocnc 6040 electronics for the Mesa pair has been straightfoward. Even got speed control.
[20:33:04] <Mr_Mayhem> for the spindle.
[20:36:14] <Loetmichel> it is
[20:36:41] <Mr_Mayhem> I still need to buy replacement driver boards, that is where my machine actually failed. They use cheap toshiba chips and are made to look like leadshine boards, but are not.
[20:37:22] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=4403&g2_imageViewsIndex=1 <- was for a mill i made for my ex-boss ;-)
[20:37:40] <Mr_Mayhem> Well, I sure could use a dust shoe design, I am still trying to design one. Multi-layer arylic like yours, but I am wandering around.
[20:37:45] <Loetmichel> the cheap toshiba chips are ok
[20:38:13] <Loetmichel> you just have to know that they dont like sorts OR open circuits
[20:38:27] <Loetmichel> disconnectig the motor when running is a sure death sentece
[20:38:58] <Mr_Mayhem> I had no problems, but I think what happened is, my fan was not connected properly so they fried. But only in a short time, like maybe 30 minutes of idle.
[20:39:08] <Loetmichel> (or invertign the power lines... had that lapse once... ) -> http://www.cyrom.org/palbum/main.php?g2_itemId=12554
[20:39:54] <Mr_Mayhem> Doh@
[20:40:25] <Mr_Mayhem> snap, crackle and pop came around to visit, I see.
[20:40:43] <Loetmichel> there is a bigger dust shoe i used on the ex-bosses mill: http://www.youtube.com/watch?v=-iYhICZEsCA
[20:42:02] <Mr_Mayhem> I need one for the 80mm spindle. I like the magnet design of Kentcnc.com but it is too big for the yoocnc 6040. Bigger than some of the parts I cut, hehe.
[20:42:51] <Mr_Mayhem> I asked if he would design a slimmed-down version, but he is not too inclined to do that.
[20:43:34] <Mr_Mayhem> Cool time-lapse video.
[20:43:50] <Loetmichel> it was the last workday for the model company
[20:44:16] <Loetmichel> the title says: "handover to the now company owner and his worker"
[20:44:53] <Loetmichel> and he and his worker managed to destroy the router in less than half a year
[20:45:00] <Mr_Mayhem> Ouch.
[20:45:17] <Mr_Mayhem> What did they make? I am still playing the video.
[20:45:32] <Loetmichel> one of the HPSTEP motors failed. they decided to run it anyway, relying on the coupling timing belts on the back...
[20:46:14] <Mr_Mayhem> Oh, it used two steppers, coupled together.
[20:46:16] <Loetmichel> a few days later these broke, not up to the job of providing torque all the time, being there just for synchonizing the two y steppers...
[20:46:21] <Mr_Mayhem> And one failed.
[20:46:39] <Loetmichel> right
[20:46:56] <Mr_Mayhem> What did this model company make, exactly?
[20:47:37] <Mr_Mayhem> like rc models?
[20:48:12] <Loetmichel> rc modelplanes and parts
[20:48:36] <Mr_Mayhem> I see.
[20:49:08] <Loetmichel> and lighting for planes
[20:49:24] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=4651
[20:49:28] <Mr_Mayhem> Oh, ok. Hey How deep as a percentage do you cut the sheet metal to enable folding it?
[20:49:39] <Mr_Mayhem> Like half?
[20:50:05] <Mr_Mayhem> Or more like a third?
[20:50:57] <Loetmichel> http://www.cyrom.org/hajo/l39demo.avi
[20:51:10] <Loetmichel> half of the material thinckness
[20:51:23] <Mr_Mayhem> ok.
[20:51:28] <Mr_Mayhem> Good to know.
[20:51:29] <Loetmichel> and 2mm wide for a 1,5mm aluminium sheet
[20:51:41] <Mr_Mayhem> ok.
[20:52:02] <Loetmichel> will give you a bend where the corners of the slut sit exactly on edge after the bend
[20:53:07] <Loetmichel> ups
[20:53:13] <Loetmichel> i meant : slot
[20:53:14] <Loetmichel> ;-)
[20:53:26] <Mr_Mayhem> Oh, I see correct width of the cut, so cuts sit on edge.
[20:53:31] <Mr_Mayhem> got it.
[20:53:43] <Mr_Mayhem> be right back in a few minutes.
[20:54:33] <Loetmichel> ... and remember to add about 1,5 to 2 mm to the part to allow for bending
[20:54:56] <Loetmichel> have macde that error myself many times. so the drills in hood and base didnt fit
[20:56:40] <Tom_itx> http://tom-itx.dyndns.org:81/~webpage/cnc/pendant/profile.jpg
[20:56:43] <Tom_itx> progress
[20:57:27] <Loetmichel> nice
[21:05:32] <jdh> Mr_Mayhem: you have a 3060?
[21:05:48] <jdh> er.. 3040?
[21:12:28] <asah> howdy.
[21:13:08] <asah> Anyone know where a default ini and hal file setup for velocity mode PID are sitting?
[21:13:41] <asah> I got my scales up, now tuning the x axis PID and the one that came out of pcconf does not have any of the PID machinery hooked up.
[21:38:55] <Jymmm> If you shopped at a Target Stores retail location between November 27th and December 15th your debit/credit card data may have been compromised.
[21:47:48] <skunkworks> linuxcnc.org
[21:47:52] <skunkworks> heh
[22:16:58] <Mr_Mayhem> Lionmichel I'm back. Thanks for the sheet metal tip, that method of scoring the metal to enable bending it into chassis boxes will be very useful to me. Neat model airplane actuators too, did he make the actual actuators, or the kit to drive them?
[22:18:12] <Mr_Mayhem> I appreciate that, yeah, correct alignment. Oh, I see the rc plane stuff is lights.
[22:19:16] <Mr_Mayhem> So 2mm wide for a 1,5mm aluminium sheet, and and remember to add about 1,5 to 2 mm to the part to allow for bending, got it.
[22:24:48] <Mr_Mayhem> By the way, I know sketchup has a neat folder/unfolder pattern maker for flat materials. I have not played with it yet, but looks straightfoward to use.
[22:25:14] <jdh> do you have a 3040/3020?
[22:25:45] <Mr_Mayhem> I have a YooCNC 6040 with the newer black control box/
[22:26:20] <Mr_Mayhem> http://www.carving-cnc.com/cnc6040-series/cnc-6040z-s80-new-router-engraver-drilling-and-milling-machine.html
[22:28:23] <Mr_Mayhem> It was fine, but I fried 2 of the drives by running it without the cooling fan, so I upgrade the whole thing but keep the vfd. I am using Mesa 5i25 and 7i76 Plug n go kit to bring in control signals and breakout, and shopping for proper leadshine or equivalent for new drives.
[22:28:23] <jdh> probably different drivers from the 3020 & 3040?
[22:28:59] <jdh> but, no reason to think they are any better.
[22:29:07] <Mr_Mayhem> Yeah, but the chips are still Toshiba. Don't know the number offhand.
[22:29:32] <Mr_Mayhem> They are not as lame as the a axis driver card in my box.
[22:29:46] <jdh> is it also dead? :)
[22:29:55] <Mr_Mayhem> Which also has that rectangular Toshiba chip, but old one for sure.
[22:30:04] <Mr_Mayhem> No, that one works./
[22:31:20] <jdh> ever seen one for sale without motors/electronics?
[22:31:26] <Mr_Mayhem> I was adding limit switches, which required removing the little breakout board from the chassis to solder the wires into the holes. And I didn't put the fan connector back on correctly, so the drivers got too hot I suspect/
[22:32:05] <Mr_Mayhem> They sell the chassis only, no control box, spindle. Comes with steppers/
[22:32:22] <Mr_Mayhem> http://www.carving-cnc.com/accessories/6040mainbody.html
[22:32:54] <jdh> how is accuracy?
[22:33:08] <jdh> assuming decent drivers, etc.
[22:33:15] <Mr_Mayhem> The trick here is, they don't sell the spindle or vfd on the web site, so they make you get the whole thing that way. Maybe if you ask them, you could, but I doubt it.
[22:33:51] <jdh> spindle is ok, I just didn't want normal cheap drivers
[22:34:49] <jdh> I was looking for a 3020 or 3040 though.
[22:34:59] <Mr_Mayhem> But you could get them elsewhere, like http://www.automationtechnologiesinc.com/products-page/cnc-spindle
[22:35:39] <Mr_Mayhem> Oh. Well, I don't see that one around in chassis only form. Hmmm.
[22:36:11] <jdh> how much space do you need for the 6040?
[22:36:46] <jdh> 3ftx2ft looks like.
[22:37:05] <Mr_Mayhem> hmmm
[22:37:07] <Mr_Mayhem> http://www.automationtechnologiesinc.com/products-page/cnc-router/kl4530-desktop-cnc-router-with-3-motors
[22:37:33] <Mr_Mayhem> This company sells solid stuff, from what I have read. No cheapie crap.
[22:37:55] <Mr_Mayhem> relative to ebay.
[22:38:59] <Mr_Mayhem> the 6040 is 580mm x 380mm
[22:39:27] <Mr_Mayhem> travel distance of the axis, that is. z is 60mm, barely.
[22:41:18] <Mr_Mayhem> I love mine. Great machine, but crap electronics. The mini vfd is good, has modbus, analog spindle speed control, spindle-at-speed signal.
[22:42:04] <jdh> do you do PC boards with it?
[22:42:21] <Mr_Mayhem> Yes
[22:44:14] <Mr_Mayhem> I use autoleveler software app to compensate for z alignment issues. It probes and then mills, using the height data gleaned during probing to compensate for z variations. Works nice.
[22:44:57] <Mr_Mayhem> http://www.autoleveller.co.uk/
[22:46:57] <Mr_Mayhem> So first mill the mdf waste board flat, then mount the pcb. generate g code from gerber files. Autoleveler as last step, then mill.
[22:48:05] <Mr_Mayhem> I mill out a pocket to hold the board, like half the board thickness, then use washers and bolts into countersunk threaded inserts in the mdf to hold down the pcb.
[22:48:52] <Mr_Mayhem> I grab it just on the outer 1/8" or so, so as not to intrude into the design or the probing pattern.
[22:49:03] <Mr_Mayhem> using the washers.
[22:49:14] <jdh> I was looking at a 3020 in hopes the Z would be better than my current device.
[22:49:40] <Mr_Mayhem> I don't know the z on a 3020.
[22:49:49] <Mr_Mayhem> Probably 60mm
[22:51:56] <Mr_Mayhem> http://www.kelinginc.net/CNCmachines.html has some. I know other buy from them.
[22:52:55] <jdh> same as the above
[22:53:11] <Mr_Mayhem> A lot of people buy stepper drivers from them. I think maybe same people as automation technology inc which I linked earlier for the spindles.
[22:53:21] <Mr_Mayhem> oh, ok.
[22:53:27] <Mr_Mayhem> Dual websites.
[22:55:33] <Mr_Mayhem> Well the KL4530 has 137.16mm z clearance. Much more than my 6040.
[22:55:46] <Mr_Mayhem> that's 5.4 inches.
[22:58:01] <Mr_Mayhem> I guess all the real metal workers are rolling their eyes at these, but hey, gotta start somewhere. :-)
[22:58:12] <jdh> yep
[23:06:54] <jdh> and I was thinking ebay pricing.. http://www.ebay.com/itm/130999706774
[23:10:54] <Mr_Mayhem> Yeah, that one is cute size and cheap, but the watercooled spindle is the way to go if you can upgrade. The higher rpm is useful in pcb work. Don't know what that one does. Is it brushed spindle motor or brushless? Brushes fail too soon.
[23:12:48] <Mr_Mayhem> A lot quieter on the water cooled spindle too.
[23:13:05] <Mr_Mayhem> The servos are almost louder.
[23:13:16] <jdh> only 8k
[23:14:18] <Mr_Mayhem> Mine 1500 watt, 24000 rpm max. I run it at like 20k to not max it out all the time.
[23:15:04] <jdh> that was the stock spindle?
[23:16:59] <Mr_Mayhem> yes
[23:17:34] <Mr_Mayhem> It comes with a real mini-vfd to power it.
[23:18:22] <Mr_Mayhem> Built into the control box, but removable if you wanted to upgrade electronics.
[23:18:47] <jdh> $405 shipping for the 6040
[23:18:51] <Mr_Mayhem> I am retrofitting mine, and squeezing everything into the original box.
[23:18:55] <jdh> puts it over $2k
[23:19:10] <Mr_Mayhem> Yeah, mine was like 1950 or so from YooCNC
[23:19:26] <Mr_Mayhem> carving-cnc.com
[23:20:07] <Mr_Mayhem> But look, you can go beyond the pcb and make small chassis by scoring the sheet metal, and plastics, etc.
[23:20:35] <Mr_Mayhem> I stretched for the bigger one to not be limited to pcbs.
[23:20:52] <jdh> yeah. Maybe I will just work on my current one.
[23:21:27] <Mr_Mayhem> I found out you really need a few hundred dollars of tools and accessories as well to really use it properly. Bits, touch probe, etc.
[23:22:00] <Mr_Mayhem> Vacuum cleaner, dust-cyclone, dust shoe, cad software, the list goes on and on.
[23:23:18] <Mr_Mayhem> I have like $3000 into mine so far, and I'm still working up to get new driver boards.
[23:23:51] <Mr_Mayhem> Including all the above, of course. Not bad compared to the big boys, but serious for a hobbyest.
[23:24:32] <Mr_Mayhem> It's funny how just one of their tools is as much as my whole setup, maybe more.
[23:25:32] <Mr_Mayhem> Linuxcnc has been excellent to me. Saved me a chunk of change and works nice. Can't wait for 2.6.
[23:31:40] <Mr_Mayhem> If I were you, I would make and sell what it can cut now, then bootstrap up to a bigger better machine. Easier said than done, but doable if you are creative. Can you do pcbs now? That is a good angle I assume, if you have some electronics knowlege to start with.
[23:43:17] <Mr_Mayhem> I figured out kicad and had my first boards cut in a few days. My schematic was very simple, but I had to learn the application and the different steps, of which there are many. Let's see, schematic, schematic components to part modules list, netlist, parts layout on the board, routing the copper pattern, export to gerber, Coppercam to produce machine g-code, gcode file to autoleveler to do
[23:43:18] <Mr_Mayhem> z probing for height variation compensation, gcode with autoleveler probe routine to linuxcnc, and finally the pcb is milled, extra copper hatched away, and finally drilled. Don't forget you need also to make a mdf jig to hold the pcb down. HooRah!
[23:46:06] <Mr_Mayhem> When I was done my head was spinning, but made my very own pcbs. So cool.
[23:47:43] <Tom_itx> i just use eagle and ship the gerbers off to a chinaman
[23:48:15] <Mr_Mayhem> hehe
[23:51:53] <Mr_Mayhem> It gets more routine with practice. It was more of a self-challenge to see if I could actually make a decent board than anything else. Nice to be able to prototype something in house too.
[23:53:02] <Mr_Mayhem> No crazy multi-layering skill here yet. I just maybe could do a board flip over manuver, that's about it. Haven't tried 2 sided yet.