#robotics Logs

Jul 28 2018

05:21 AM rue_bed: mrdata, aha
05:24 AM rue_bed: click
05:24 AM rue_bed: the ring osc, using inverters runs slow, like 800Khz, but then you put a 2.2M in on one stage, and it jumps to like 21Mhz
05:24 AM rue_bed: the key is that if the cmos fets hit saturation, they take lots of time to come out, thereby really slowing down their responce
05:24 AM rue_bed: so, it suggests that doing current limiting on vdd and vss, might make a huge difference
11:07 AM rue_: so, I need to write a cam program that generates drill operations for points in a dxf
11:07 AM rue_: as not even the cam software I have can *just* drill
11:11 AM veverak: lolo
11:11 AM veverak: :)
11:11 AM rue_: then I need to ... oh my, this is quite a list
11:12 AM rue_: ok, so 1 step pulse -> 10 microstep driver -> 200ppr motor -> 1mm pitch leadscrew
11:12 AM rue_: whats the distance per step pulse...
11:13 AM rue_: 1/2000 mm?
11:13 AM rue_: k, and I am doing 4 steps per pendant tick
11:14 AM rue_: so, my pendant, on fast, is doing 250-500 steps/mm on fast
11:15 AM rue_: (different software versions)
11:17 AM rue_: oh its up to 16 steps/tick
11:18 AM rue_: no, no 8
11:18 AM rue_: ok I got this
11:19 AM rue_: ok, so, if I'm going thru the controller, I want each pendant step to be about 0.004mm
11:21 AM rue_: ok, parade, where is my camera
11:21 AM rue_: .. pants are a good idea too...
11:38 AM weyland|yutani: is that a cnc rue?
02:19 PM rue_: yes
02:20 PM rue_: right now, the pendant directly operates the steppers, I want to have an option to do it thru the control software
02:21 PM rue_: its a bit more trickey casue the buffering messes things up
02:21 PM rue_: so has to be done carefully
02:30 PM rue_shop3: too bad I'm not good at carefull
02:31 PM rue_shop3: http://www.micro-machine-shop.com/taig_mill_pre_cnc.jpg
02:31 PM rue_shop3: what I see in that image, is a machine thats never been used
02:33 PM rue_shop3: Tom_L, ok, I lost the notes
02:33 PM rue_shop3: how do I set the local origin to the current position?
02:33 PM rue_shop3: workspace origin? see I cant even remember the terms
02:34 PM rue_shop3: zhanx, ?
02:56 PM Tom_L: that's a big ass spindle motor on that thing
02:57 PM Tom_L: http://linuxcnc.org/docs/2.7/html/gcode/g-code.html#gcode:g10-l20
02:58 PM Tom_L: G10 L20 P xx x y z
02:59 PM Tom_L: or G10 L2
03:03 PM rue_shop3: no, it dosn't ahve support
03:03 PM rue_shop3: I have to do it manually
03:03 PM rue_shop3: oh
03:03 PM rue_shop3: the reset
03:03 PM rue_shop3: right
03:05 PM rue_shop3: hmm looks like G10 isn't supported
03:07 PM rue_shop3: nope, no G10 support
03:20 PM rue_shop3: wow, I been over this so much I alraedy started the code
03:25 PM rue_shop3: hmm interesting code challanges
03:25 PM rue_shop3: I need to define a machine
03:25 PM rue_shop3: and I suppose I'm just pushing commands into the abyss, and letting it work out where they go
03:25 PM rue_shop3: or thats part of the machine...
03:26 PM Tom_L: there was one it supported, which was that?
03:26 PM Tom_L: iirc those are the ones you asked for
03:43 PM rue_shop3: I made notes on a bunch of stuff, they probably still exist, somewhere
03:45 PM veverak: lol
04:35 PM veverak: well
04:35 PM veverak: https://imgur.com/a/86paOmv
04:35 PM veverak: I like this error handling more than exceptions
04:36 PM rue_shop3: where did the coxa go?
04:36 PM rue_shop3: dont you only have 4 legs?
04:37 PM veverak: yup
04:37 PM veverak: leg/rl/coxa is not in the urdf::Model
04:37 PM veverak: I changed the naming of the URDF parts
04:38 PM veverak: leg/rl/coxa is 'link', not a joint, code tries to search it as joint
04:38 PM veverak: leg/rl/alfa should be there (as it is the joint with which leg starts)
04:38 PM veverak: I made the change in URDF, but did not modified code
04:38 PM veverak: but it is good material to make the error handling ;)
04:41 PM veverak: I am pretty happy with Either< LeftType, RightType > datatype to work this out
04:43 PM veverak: (it is basically Either{ union{ LeftType left; RightType right; } ... };
04:43 PM veverak: it took while to get everything right, because union... but it is awesome :)
04:44 PM rue_shop3: I almost have a program to generate a drill cycle on the cnc
04:45 PM rue_shop3: now I just have to write the bit that makes it do something
04:45 PM rue_shop3: I need a tea!
04:46 PM veverak: I feel like making it spill out line number
04:46 PM rue_shop3: Tom_L, another idea for that motor I got, I could make one !@#$@#$ of a power dolley
04:49 PM rue_shop3: it occurs to me that their putting 40A ea on the spade connectors to the driver
04:57 PM rue_shop3: ( GO NOW )
04:57 PM rue_shop3: G00 F85.000000 X10.000000 Y10.000000 Z10.000000
04:57 PM rue_shop3: G00 F85.000000 X10.000000 Y10.000000 Z1.000000
04:57 PM rue_shop3: G00 F5.000000 X10.000000 Y10.000000 Z-7.000000
04:57 PM rue_shop3: G4 P0.250000
04:57 PM rue_shop3: G00 F85.000000 X10.000000 Y10.000000 Z10.000000
04:57 PM rue_shop3: but it was only supposed to go 3mm at a time....
05:03 PM rue_shop3: Tom_L, G00 F2 ... should do THAT position at ... oooh click
05:20 PM rue_shop3: !? am I ok? I couldn't remember the fn scanf
05:24 PM veverak: :D
05:25 PM rue_shop3: I hate scanf
05:53 PM Tom_L: rue_shop3, G00 doesn't require a Feedrate, it's rapid
05:54 PM Tom_L: you want G1
05:58 PM Tom_L: dwell on it a bit :)
05:58 PM Tom_L: .5 is reasonable
05:59 PM Tom_L: enough time for all the flutes to pass over the bottom once or twice
05:59 PM Tom_L: so at 100 rpm it will be more and at 25,000 it will be less
05:59 PM Tom_L: at 25,000 you might start a fire :D
06:00 PM Tom_L: it's a tricky answer
06:24 PM rue_shop3: yea
06:25 PM rue_shop3: well, I could have mentioned it was like a 1/16" bit at like 3k
06:25 PM rue_shop3: anyhow
06:29 PM rue_shop3: the drill cycle generator works, I didn't use the dxf front end, cause this hole list is just a csv, but the code is there now
06:30 PM rue_shop3: I still want to do foil etching tho
06:33 PM rue_shop3: haha, I already came up with an amp for the shop
06:33 PM rue_shop3: tho, this new setup is better
07:57 PM zhanx_: rue_shop3, what was the question? I was working on integrating my garden stuff
07:58 PM rue_shop3: cnc stuff, np
07:58 PM rue_shop3: worked out I think
07:58 PM rue_shop3: I made this encoder wheel, but blew the math
07:58 PM rue_shop3: I offset it in degrees, in the middle on a rads based calc
07:58 PM rue_shop3: :(
07:58 PM rue_shop3: x = r1*cos(o1+a*(3.1415926535/180));
07:58 PM rue_shop3: should have been
07:59 PM rue_shop3: x = r1*cos((o1+a)*(3.1415926535/180));
07:59 PM rue_shop3: if I want to use this disc, it'll take lots more math
08:00 PM rue_shop3: explains why the codes all seem to be out
08:02 PM zhanx_: k
08:02 PM zhanx_: i am working on the new task scheduler for the garden
08:04 PM zhanx_: but the wife green lighted me automating the garden to about 90% today, which is good
08:05 PM rue_shop3: well, I suppose the disc itself isn't the important bit, its the rest of the mechanics around it, if I need to I can make a new disc
08:05 PM rue_shop3: all the software will do that quite happily
08:05 PM rue_shop3: and hey, I have the start of my cam software
08:06 PM rue_shop3: it takes input and generates holes!
08:06 PM rue_shop3: I can 3d print an outter case
08:06 PM rue_shop3: hmmm
08:17 PM rue_shop3: Jak_o_Shadows, did you use a generator script for making the synchropullies?
08:17 PM Jak_o_Shadows: I use autodesk inventor - it has a gear tool
08:17 PM rue_shop3: hmm
08:18 PM rue_shop3: I wonder if there is a openscad thing on thingiverse
08:18 PM Jak_o_Shadows: So they're proper involute gears - until the slicer gets a hand on em
08:18 PM rue_shop3: ah
08:18 PM mrdata-: radians are more natural than degrees
08:18 PM Jak_o_Shadows: I think i changed the pressure angle a bit as well
08:18 PM rue_shop3: yea, my tracks are out by 90-(3.14/2) radians
08:20 PM rue_shop3: its too bad, its a nice encoder wheel
08:20 PM rue_shop3: 3 tracks and a clock, the 3 bits yield 24 absolute positions ]:)
08:21 PM rue_shop3: the clock just ditches any alignment issues
08:23 PM rue_shop3: I bought the wood gears generator
08:24 PM rue_shop3: I could play with that too
08:24 PM rue_shop3: it occurs to me, that I need a 1:1 with the caster wheel I'm tracking
08:24 PM rue_shop3: for the absolute positions to mean anything
08:25 PM Tom_L: ok, turned down a tool holder for clearance against a wall. now maybe i can set the tools and start cutting chips i want to cut
08:25 PM rue_shop3: hehe
08:26 PM rue_shop3: that or sharpen up the edges of the tool holder more?
08:27 PM mrdata-: fractions of a circle or of a sphere are also pretty natural
08:28 PM Jak_o_Shadows: Radians fit our maths better, because of e
08:29 PM rue_shop3: arg, I'm all drowsy today and I'm being useless
08:31 PM Tom_L: that will save one whole fixture position now
08:34 PM * Tom_L goes to edit the code to accommodate the new tool clearance
08:39 PM rue_shop4: oooh I cant update the browser, cause nobody supports 32 bit OSs anymore
08:39 PM Jak_o_Shadows: Yup.
08:39 PM Jak_o_Shadows: My server computer doesn't een have 32 bit anything
08:39 PM Jak_o_Shadows: So you can't run any 32bit windows stuff in wine
08:39 PM rue_shop4: and everyone is moving to 256bit-only keys, so my browser cant connect to anything
08:39 PM rue_shop4: its funny, lynx can access more pages than chromium or iceweasel
08:40 PM rue_shop4: if I go to 64 bit, will any of the avr tools work/ be available?
08:40 PM Jak_o_Shadows: Bugger if I know
08:40 PM rue_shop4: can 64 bit linux run 32 bit programs?
08:40 PM rue_shop4: ugh
08:40 PM Jak_o_Shadows: generally, yes, but you neeed all the 32 bit libraries
08:41 PM Jak_o_Shadows: which I suspect debian/ubuntu still do? CentOS doesn't
08:41 PM rue_shop4: does this mean I need to try to recompile every program I'v ever written?
08:41 PM rue_shop4: arg
08:41 PM Jak_o_Shadows: nah, it should probably work, if you have the 32 bit packages installed
08:42 PM Jak_o_Shadows: ask in ##your-linux-distro?
08:42 PM rue_shop3: its debian, nobody knows about it anymore
08:44 PM rue_shop3: were on the fast-track to slakware, and having to hunt down specific versions of every library you need to compile an app, with the only way to know which library you need next being to wait a half hour for the compile to fail again and bark out the message that tells you the missing filenames so you can try to figure out what they are from
08:45 PM rue_shop3: and after 3 days you find out that they used bills library, and his site is down and there are no other copies on the internet
08:47 PM rue_shop3: I doubt I'm gonna be able to compile nedit for 64bit env.
08:51 PM Jak_o_Shadows: I looked at getting wine setup, so I could run a specific server
08:51 PM Jak_o_Shadows: gave up entirely
08:56 PM rue_shop3: I never imagined that all the programs would just turn to dust after they had reached maturity
08:57 PM rue_shop3: I thought they would split into different versions that had different options
08:57 PM Tom_itx: mmm runtime 4.25 hrs
08:57 PM zhanx_: ugh
08:57 PM zhanx_: my code is to bloated to fit
08:58 PM zhanx_: feature creep got me
08:58 PM zhanx_: break time from that, time to look at the sonoff stuf
09:12 PM Jak_o_Shadows: ok, let's see if that usb serial adaptor is going to play nice with me today
09:44 PM Tom_L: chips r flyin now!
11:07 PM * rue_ tries to keep the chips out of his dip
11:10 PM Tom_L: about halfway thru the front side
11:10 PM Tom_L: back side is down to thickness
11:11 PM Tom_L: program is running better so far
11:11 PM rue_: so, landmark day, the CAM software I been on about for so long did some basic stuff today
11:11 PM Tom_L: spindle is a dull red glow instead of bright
11:12 PM Tom_L: nice
11:12 PM rue_: ... red glow?
11:12 PM rue_: isn't that usually caused by running it the wrong way around?
11:12 PM Tom_L: it's actually staying fairly cool
11:13 PM Tom_L: dumped another bucket of ice in the cooling tower
11:14 PM Tom_L: one more layer on the front now
11:14 PM rue_: http://ruemohr.org/~ircjunk/projects/cnc2/disc2.nc
11:14 PM Tom_L: then starting the center profile
11:14 PM rue_: there is output!
11:15 PM Tom_L: can't view that out here
11:17 PM rue_: it could be briefer, but its a great start
11:18 PM Tom_L: come get this pile of chips to melt