#robotics Logs
Jan 13 2018
#robotics Calendar
12:09 AM rue_: if the system were commercial, I bet the only difference would be that the limit switches would have slots of the plate their mounted to for some kinda size tweek
12:12 AM Tom_L: yeah
12:12 AM Tom_L: usually the switch is solid and the trigger is adjustable
12:15 AM Tom_L: rue_, what you gonna do with that fft code you've been feverishly working on?
12:16 AM rue_: I dont know, its one hell of a ride
12:18 AM Tom_L: all these silly details on this thing sure take alot of time
12:18 AM rue_: yea, finishing is always the langth of the rest of the project put togethor
12:19 AM Tom_L: and i can't just leave the mess and come back to it
12:20 AM rue_: its looking good
12:21 AM Tom_L: when i croak be sure to come and get it, nobody here will know what to do with it
12:23 AM rue_: I can see people standing around it trying to work out if its a modified commercial machine or scratch built
12:24 AM Tom_L: i got a book ~1.5" thick on the control
12:24 AM Tom_L: some current, some past iterations of wiring
12:25 AM rue_: I'm gradually ending up with soemthing like that
12:25 AM Tom_L: i can't toss any of it because i can't always remember which one is current
12:26 AM Tom_L: i did get the pendant fixed though
12:26 AM Tom_L: one of the FPGA pins quit so i moved that signal to another pin
12:27 AM rue_: odd
12:27 AM rue_: what do you think happened to it?
12:27 AM rue_: static?
12:27 AM rue_: raw inputs from teathered devices would usually have .01uf caps to ground on every line
12:30 AM Tom_L: i have no idea
12:31 AM Tom_L: several are used in a similar way
12:31 AM Tom_L: i could get an isolation board for it but i haven't done that yet
12:31 AM Tom_L: i hate doing that because it reverses all the logic
12:32 AM Tom_L: switching back and forth is a pita
12:32 AM rue_: 4N137
12:32 AM rue_: iirc
12:32 AM rue_: not cheap little buggers tho
12:32 AM Tom_L: what are they?
12:33 AM rue_: logic output optos
12:36 AM Tom_L: i could get a board for as much as the parts would cost probably
12:36 AM rue_: not inverted operation iirc
12:36 AM Tom_L: i've got a SSerial card i could use
12:36 AM Tom_L: he gave that to me
12:36 AM Tom_L: http://store.mesanet.com/index.php?route=product/product&path=69_63&product_id=95
12:37 AM Tom_L: i've got a spare FPGA card too i think
12:37 AM Tom_L: 72 io
12:39 AM Tom_L: http://store.mesanet.com/index.php?route=product/product&path=69_70&product_id=123
12:39 AM Tom_L: he gave me one of those
12:39 AM Tom_L: i could use it for the limits and pendant
12:39 AM Tom_L: i had it wired that way once
12:39 AM Tom_L: i really don't have room in the control box for another card though
12:40 AM Tom_L: those are kinda neat io expander cards though
12:54 AM rue_: ah, it a bit breakdown without bitwise operators
02:17 AM rue_: for( m = nn; j ^= m, !(j & m); m >>= 1);
02:17 AM rue_: how to abuse C
02:17 AM rue_: that is a reverse binary incrementer
02:17 AM rue_: ;)
02:29 AM zhanx: rue_ is up
02:52 AM rue_: it gets worse
02:52 AM rue_: for( m = nn; !(j & m) && m; m >>= 1, j ^= m ) ; // reverse binary increment j
02:52 AM rue_: the compiler is shaking and knocking...
02:53 AM * rue_ adds another strap across it to the floor
02:56 AM rue_: http://paste.debian.net/1004944/
02:56 AM rue_: see the difference?
02:59 AM zhanx: that computer is gonna melt
02:59 AM zhanx: in other new rue_ finds a new prime number
03:01 AM rue_: infinity-171, go figure.
03:01 AM zhanx: yep pm'd you rue_
03:33 AM rue_: // fix the order into a state machine by following the sequence thru the indexes
03:33 AM rue_: p = 0;
03:33 AM rue_: for(i = 0; i < (BUFFER_SIZE-1); i++) {
03:33 AM rue_: machine[p] = order[i+1]; // its almost 3am, dont ask me how this works cause right now
03:33 AM rue_: p = order[i+1]; // barely I even know. And I wrote it. ( I M Rue Mohr )
03:33 AM rue_: }
03:33 AM rue_: machine[p] = 0; // thankfully, the last element is predicatable
03:33 AM rue_: heh
03:59 AM rue_: ah, on the complex plant,t he vector length is the amplitude and the angle is the phase
04:27 AM rue_: oh my
04:28 AM rue_: so, a fft of a given table size does a fixed, limited number of cos/sin lookups
04:28 AM rue_: you can make a table of those values
04:28 AM rue_: meaning, no trig calcs at all
04:55 AM rue_: a 32 value fft uses 30 trig values
05:07 AM rue_: ah, but generally, the number of angles used it equiv to the data block size
05:37 AM rue_: oh its better than that
05:44 AM anonnumberanon: blah, blah, blah...
05:44 AM rue_: there are only 16 unique values
05:44 AM anonnumberanon: are you all nighting again or decided to wake up early?
05:44 AM rue_: hah, what do you think
05:44 AM anonnumberanon: former
05:44 AM rue_: hahaha
05:45 AM anonnumberanon: i've been waking up at 5 everyday
05:45 AM anonnumberanon: amazingly I'm finally getting things done
05:46 AM rue_: ahaha and... because its sine, there are only 8 unique values
05:46 AM rue_: so, to process a 32 sample fft, you only need 8 sin values
05:46 AM rue_: hah!
05:47 AM anonnumberanon: the weather will go really cold again soon
05:48 AM anonnumberanon: it's aggravating me
06:35 AM Boobie: Which device like AIDO is the most non-scammy and non-nonexistent?
06:35 AM rue_: is it an analog/digital io baord?
06:36 AM rue_: timeout!
10:05 AM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Mill_Assembly1.jpg
10:05 AM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Mill_Assembly2.jpg
10:05 AM Tom_L: making a little progress
10:36 AM slidercrank: a DIY robotic vacuum cleaner https://cs7.pikabu.ru/post_img/2018/01/13/8/1515850585168090949.gif
10:37 AM polprog: hot glue
10:37 AM * polprog pukes
10:54 AM anonnumberanon: Heh
11:29 AM z64556 is now known as z64555
11:44 AM rue_bed: 4 AM YEAAAAAAAAAAA!!!!!!
11:44 AM * rue_bed collapses back to sleep
11:45 AM Tom_L: been workin all morning
11:46 AM rue_bed: the fft thing, the optimizations.... oh god...
11:46 AM rue_bed: so... so foldable....
11:46 AM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Mill_Assembly1.jpg
11:46 AM Tom_L: see
11:46 AM rue_bed: wow, nice
11:47 AM Tom_L: i like the longer overhang on Z
11:47 AM Tom_L: i can reach the edge of the table now
11:47 AM * rue_bed gives tom some oilcloth to start folding into accordians for the slider
11:47 AM Tom_L: excessive travel on Y but i don't mind that
11:47 AM Tom_L: just allows for overengineered large parts
11:48 AM rue_bed: wow, I had some strange dreams...
11:48 AM Tom_L: it's no wonder
11:49 AM Tom_L: with all that convoluted code you've been working on
11:49 AM rue_bed: it folds, and just when you think the folding is done, its symettric again
11:50 AM Tom_L: i can grab that Z overhang and shake it and the only thing that moves is the saw horses it's sitting on
11:50 AM rue_bed: I still think the tube spacers for the z need endcaps
11:50 AM Tom_L: i may
11:50 AM Tom_L: i'm avoiding warpage at all cost
11:50 AM rue_bed: ah right
11:51 AM Tom_L: i learned about that rewelding the standoff
11:51 AM Tom_L: notice the bottom part is welded all around
11:51 AM Tom_L: i had to hammer it flat again
11:51 AM Tom_L: so i didn't do that on the other half
11:51 AM rue_bed: well, put some fake spacers in for the screw penetration (was maybe) and fill them with .. polycrete?
11:51 AM Tom_L: it's all good now
11:51 AM Tom_L: just a little body shop repair
11:51 AM rue_bed: wax maybe
11:52 AM Tom_L: i'm pissed about the drag chain
11:52 AM rue_bed: oh its hard to proof text with double vision
11:52 AM Tom_L: i should have gotten smaller
11:52 AM rue_bed: lam cable
11:52 AM rue_bed: for the x
11:52 AM rue_bed: between the plates
11:52 AM Tom_L: 18mm tall and i could have used 12 which would have bent around just fine
11:53 AM rue_bed: fill the upright with an antiresonant too
11:54 AM Tom_L: i'll try it first
11:54 AM rue_bed: it'll make it heavy as hell tho
11:54 AM rue_bed: take out the bell effect
11:54 AM Tom_L: with the L welded on it may not ring as bad
11:54 AM Tom_L: otherwise it would
11:54 AM * rue_bed does a spectrum analysis on toms cnc
11:55 AM rue_bed: _____|______|____i______
02:49 PM veverak: deshi
02:49 PM veverak: !seen deshipu
02:49 PM veverak: interesting
02:49 PM veverak: https://www.youtube.com/watch?v=JtMiofUCQhA cute <3
07:01 PM rue_: anonnumberanon, 3 instructions?
07:01 PM rue_: ______ NOP; _________
07:05 PM Tom_shop: broke a 4-40 tap off so i took an old one that was already broke and ground it to finish the job. tapped 4 holes in steel with it
07:05 PM rue_: yup
07:05 PM * rue_ dosn't have many taps with the origional point left on them
07:06 PM Tom_shop: it tapped better than the new one
07:08 PM rue_: :)
07:08 PM rue_: wait,t hats not the second one to break on ya?
07:14 PM Tom_shop: the other one was a 5M
07:15 PM rue_: yea, my experience is that paying more dosn't change the rate they break, it just changes the replacement cost.
07:56 PM SolG is now known as Solgriffin
08:11 PM Tom_shop: yay! all the limit switches are installed
08:11 PM rue_: :) silly things
08:12 PM Tom_shop: that was one of the more time consuming tasks
08:12 PM * rue_ hands tom 7200' of white, unlabeled, military wire.
08:12 PM rue_: :D
08:13 PM Tom_shop: dude
08:13 PM Tom_shop: i've got both white and orange!
08:13 PM * rue_ hands tom a wire label machine which makes labels that stay on for 2 days, then fall off
08:13 PM rue_: :)
08:13 PM Tom_shop: pics on the way. cause i gotta go wash my hands
08:13 PM rue_: most of sat is visitors
08:13 PM rue_: for me
08:14 PM Tom_shop: oh and i moved the shelf over my steel workbench
08:14 PM rue_: I'm making progress on the program that makes my fft
08:14 PM rue_: vert space eh?
08:17 PM Tom_L: spot for the mill
08:21 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Z_Axis_Limit1.jpg
08:21 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Z_Axis_Limit2.jpg
08:21 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Z_Axis_Limit3.jpg
08:21 PM Tom_L: i had to weld those mounts in place
08:22 PM Tom_L: and didn't need to add a trip for this one
08:22 PM Tom_L: now to add some hard limits to it
08:27 PM rue_: its still noninterfearence
08:27 PM rue_: tho, that position might make an accordian fun
08:28 PM rue_: kinda a tight space anyhow
08:28 PM Tom_L: yeah
08:28 PM Tom_L: we'll see how it goes
08:29 PM Tom_L: i think i'll drill a hole at the bottom and run the wires thru the 4" tube out the back
08:29 PM Tom_L: clear thru
08:31 PM Tom_L: i just gotta find the wire
08:31 PM rue_: use a 1/4" plastic pipe?
08:31 PM rue_: ooooh when spring comes!
08:32 PM rue_: huh, this is doing really well as integer math
08:33 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/psu/wire.jpg
08:33 PM Tom_L: if you see that lemme know where it is
08:33 PM Tom_L: it's good stuff
08:33 PM rue_: it looks like heating mat cable
08:33 PM Tom_L: aircraft
08:33 PM rue_: k
08:33 PM Tom_L: uber fine wire and braid
08:34 PM rue_: do they have a plane cemetery there?
08:34 PM Tom_L: they used to but i don't know where they moved it to
08:34 PM Tom_L: they probably have more than one
08:34 PM Tom_L: test planes
08:36 PM Tom_L: i may go clean up the mess and take the night off
08:36 PM Tom_L: been at it since 7am
08:39 PM Tom_L: ok found the wire, just hope there's enough
08:39 PM Tom_L: however the white is 3 cond so i could use that if i had to
09:02 PM rue_: huh, strictly 8 bit isn't plausable, so it seems
09:02 PM Tom_L: still workin on the code?
09:03 PM rue_: the calcs need signed numbers, and that leaves -127 +127 and thats not enough headroom by the time you have multiplied a few numbers togethor
09:03 PM rue_: yep
09:04 PM Tom_L: ok so if my scale is 40000 with a 10 microstep driver and a 20tpi screw, what will it be with a 5mm pitch screw?
09:04 PM Tom_L: same driver
09:05 PM Tom_L: just a different screw
09:06 PM Tom_L: 5.08 turns per inch i think
09:06 PM rue_: 5mm
09:06 PM rue_: 200 spr
09:07 PM rue_: *10 is 2000 spr
09:07 PM rue_: 5mm/2000
09:07 PM rue_: is mm/step
09:07 PM Tom_L: .19685 inch per turn
09:07 PM rue_: why would you bring inches back into it?
09:08 PM rue_: .0025mm/microstep
09:08 PM Tom_L: because i'm setting the config file up in inches
09:08 PM rue_: zippo:/files/programming/c/fftsolver$ units .0025mm thou
09:08 PM rue_: * 0.098425197
09:08 PM Tom_L: so i need to know how many steps per inch
09:08 PM Tom_L: right now it's 40000
09:08 PM rue_: zippo:/files/programming/c/fftsolver$ units .0025mm 1/inch
09:08 PM rue_: reciprocal conversion
09:08 PM rue_: * 10160
09:10 PM Tom_L: and how do you arrive at that?
09:10 PM rue_: 5mm pitch
09:10 PM rue_: 200 steps/rev on motor
09:10 PM Tom_L: yeah
09:10 PM rue_: 10 microstep driver
09:11 PM rue_: 200*10 steps/rev
09:11 PM rue_: 200*10 steps/ 5mm
09:11 PM rue_: microstep = 5mm/2000
09:11 PM rue_: = .0025mm
09:11 PM rue_: then run thru units converter for reciprical conversion to steps per inch
09:11 PM rue_: :)
09:11 PM Tom_L: so 10160 should be accurate?
09:12 PM rue_: completely
09:12 PM rue_: think on it approximation
09:12 PM Tom_L: then on the Z axis i need to multiply that by 3
09:12 PM rue_: units 1inch/5mm
09:12 PM rue_: Definition: 5.08
09:12 PM Tom_L: since it's a 3:1 reduction
09:12 PM rue_: there are about 5 turns per inch
09:13 PM rue_: 5*2000 = 10000 so there are about 10000 steps per inch
09:13 PM Tom_L: so i'll lose a little accuracy
09:13 PM rue_: specifically 10160
09:13 PM rue_: hu?
09:13 PM Tom_L: going from 20 tpi to about 5
09:13 PM Tom_L: 4x less accurate
09:13 PM Tom_L: in steps
09:13 PM rue_: yea, you gain LOTS of speed
09:14 PM rue_: your stil have ~10steps/thou
09:14 PM Tom_L: so 1 tenth accuracy?
09:14 PM rue_: units .0025mm thou
09:14 PM rue_: <rue_> * 0.098425197
09:14 PM Tom_L: those steppers hould drive it ok i think
09:15 PM rue_: the 7 might be rounded
09:15 PM rue_: not sure
09:15 PM Tom_L: 570 in/oz
09:17 PM rue_: omg it worked
09:19 PM Tom_L: i got some smaller drag chain
09:20 PM rue_: in 8 bits the gain is off, but... wow
09:20 PM Tom_L: so what are you gonna use it for?
09:20 PM rue_: its got a range of 64 to work with
09:20 PM rue_: I'm gonna put it on an avr
09:23 PM Tom_L: is that correct to multiply the 10160 by 3 for the Z axis?
09:23 PM rue_: 3:1 yea
09:24 PM Tom_L: should get a belt tuesday
09:24 PM Tom_L: also got a 2:1 set coming to try
09:25 PM Tom_L: and if the other 2 direct drive aren't strong enough, i'll add a reduction to them somehow
09:25 PM rue_: how heavy does z seem?
09:26 PM Tom_L: ahh, i was gonna weigh the pieces today and forgot
09:26 PM rue_: I'm totally amused I shoehorned this down to 8 bits
09:26 PM Tom_L: i bet it's 25lbs or so if i were guessing
09:26 PM Tom_L: maybe more
09:26 PM rue_: 20lbs was about my guess
09:26 PM Tom_L: by the time you add the rails and motor
09:27 PM Tom_L: it could be a little more
09:27 PM rue_: well there is a spindle missing that will atleast double it
09:28 PM Tom_L: i can move 2 axis assembled but i have doubts i can move the whole thing as one assembly
09:28 PM rue_: I think the lower bearing of the spindle on my new machine is getting loose
09:28 PM Tom_L: that's on Z
09:28 PM Tom_L: get a bearing rated for a higher rpm
09:28 PM rue_: I was milling down a new peice of wood today, noticed that a) z is pitched a bit along the Y, and that its chattering in wood
09:28 PM Tom_L: or deeper grooves
09:28 PM rue_: iirc I got some highspeed roller bearings since
09:29 PM Tom_L: i bet it's no worse than my sherline axis are
09:29 PM rue_: ok, in theroy I can put this on an avr
09:29 PM Tom_L: snapped a nice carbide cutter off because of it
09:29 PM rue_: this is halarious
09:30 PM rue_: I need to set up some io code
09:30 PM rue_: hah
09:30 PM rue_: this'll be a laugh
10:44 PM anonnumberanon: Rue what instructions?
10:45 PM anonnumberanon: Do you have memory corruption
10:59 PM rue_shop3: no you just forgot about the thinking challange I gave you
11:11 PM rue_shop3: ok, hmm lets see here
11:41 PM rue_shop3: one day I need to implement a pneumatic bearing
11:42 PM rue_shop3: ok, the avr is running at 16Mhz
11:43 PM rue_shop3: I want to trigger the adc to sample at 7112Hz
11:45 PM rue_shop3: so I need to count off 2249.7 cycles
11:45 PM rue_shop3: at 2250 cycles...
11:46 PM rue_shop3: I get 7111.11Hz
11:46 PM rue_shop3: close enough
11:46 PM rue_shop3: so, I can set a 16bit timer to divide by zero, and rollover at 2250