#robotics Logs

Aug 26 2018

02:43 AM weyland|yutani: https://twitter.com/hardmaru/status/1032762806796312576
02:47 AM rue_shop4: theBear,
02:47 AM rue_shop4: you awake
02:47 AM rue_shop4: !?
02:47 AM rue_shop4: BeArRRRRR
02:47 AM rue_shop4: theBear, cmon
02:47 AM rue_shop4: cmon cmon
02:47 AM rue_shop4: arg
03:21 AM rue_shop4: ahahaha
03:21 AM rue_shop4: ahhahaha
03:21 AM rue_shop4: I spent the whole day making a volume adjustment system thats small so it'll fit in the size of box I want
03:21 AM rue_shop4: but...
03:21 AM rue_shop4: I'm putting a 9V battery in there thats going to bump the min. size of the project anyhow
07:55 AM babaum: Hello, I am implementing a CLIK scheme with redundancy (trying to get far from an obstacle). I calculated the cost and I project it into the null space using (I-Ja+Ja)dq0, where dq0 is the gradient of w (the distance from the object) multiplied by a gain. If I raise too much that gain I get that the end-effector at the end is not in the desired position. Can someone explain me why?
08:06 AM rue_bed: hmm
08:07 AM rue_bed: no, its 6am and I'v not slept yet
08:10 AM Jak_o_Shadows: CLIK?
08:10 AM Jak_o_Shadows: How far off is it?
08:10 AM Jak_o_Shadows: Are you hitting numerical error?
08:10 AM Jak_o_Shadows: How valid is your gradient calculation?
08:12 AM babaum: rue_bed, lol
08:12 AM Jak_o_Shadows: Also, (I-Ja+Ja) ?
08:13 AM babaum: Jak_o_Shadows, I made some approximations because the matlab symengine took of a giant expression, so I removed irrelevant terms (things like e-17 where other terms were 0.5)
08:14 AM babaum: CLIK is Closed Loop Inverse Kinematics
08:15 AM Jak_o_Shadows: So basically the problem is newton optimisation?
08:15 AM Jak_o_Shadows: Or newton's method*
08:15 AM Jak_o_Shadows: You have a gradient, and you are taking a step to it
08:16 AM Jak_o_Shadows: Yeah, that's not known to be stable for giant steps
08:16 AM Jak_o_Shadows: You might be better off taking lots of small steps
08:16 AM Jak_o_Shadows: and if you don't want to do that to your real system, do a bunch of real steps, assuming they work - then give your real system the bigger steps
08:17 AM babaum: Yes, I didn't know it. But that's the method we studied in our course. Thanks for the suggestion :)
08:18 AM babaum: So the effects on the end effector are due to the approximations?^
08:21 AM babaum: If I understand well the redundancy exploitment, the term (I-(Ja+)*Ja) should project q0 in the null space, is it right?
11:57 AM rue_bed: odd, I was up till 6am, why, now, at 10am, am I awake
11:57 AM veverak: well
11:57 AM veverak: that is work to be done and your brain knwos it?
11:57 AM veverak: :)
12:57 PM _unreal: he's prob out cold
03:04 PM _unreal is now known as _unreal_
06:01 PM rue_mohr: _unreal_, so, you need a lot of votlage for most motors
06:02 PM rue_mohr: I'm looking to go to 24V for the 6V stepper to get proper speed out of it
06:04 PM Tom_L: there's a formula for that somewhere
06:04 PM Tom_L: i bet optimum is well above 24v
06:04 PM rue_mohr: I = V*dt/L or soemthing like that
06:05 PM Tom_L: where do they get the voltage ratings for steppers anyway? if we all use choppers with much higher voltages
06:05 PM rue_mohr: I think its the dc brake voltage
06:06 PM rue_mohr: V = icoil*rcoil
06:06 PM Tom_L: my new ones stay cold while idle but get warmer when running which was backward from the old ones
06:06 PM rue_mohr: does the controller do an idle shutdown?
06:06 PM Tom_L: yes
06:07 PM Tom_L: and i'm running them well below their rated current
06:07 PM rue_mohr: the printer ones do that, after a second or so the driver drops to a holding current
06:07 PM rue_mohr: the marlin firmware also turns off the drivers after like 120sec of inactivity
06:08 PM Tom_L: is marlin still the one to use?
06:36 PM rue_shop4: Tom_L, did you preload those ball screws?
06:37 PM rue_shop4: marlin seems to be the end of the line for the cnc stuff
06:37 PM rue_shop4: its a heap, but everyone keeps tweeking it
06:37 PM rue_shop4: sorry, I'm thinking grbl
06:37 PM rue_shop4: marlin is a rewrite of grbl
06:38 PM rue_shop4: Tom_L, I got a ballscrew today, and there IS axial give in it
06:52 PM orlock: does it resolve?
06:53 PM orlock: doh
06:54 PM rue_shop4: there is no adustment, I'm 80% sure its not just tilt
06:55 PM Tom_L: the ballscrews have a nut on the end i tightened but i didn't measure any backlash
06:55 PM Tom_L: if the parts weren't measuring round and in tolerance i would have
06:56 PM Tom_L: where did you find a ballscrew?
06:56 PM rue_shop4: china $24
06:56 PM Tom_L: how long?
06:57 PM Tom_L: and what diameter
06:57 PM Tom_L: the nut took out any end play the bearings may have had
06:57 PM Tom_L: or shaft to bearing i should say
06:58 PM rue_shop4: 11.5mm dia 40cm long, 4mm pitch
06:58 PM Tom_L: hmm
06:59 PM Tom_L: mine are 5mm pitch 16mm diameter iirc
06:59 PM Tom_L: did you get end bearings and carriers with it?
06:59 PM rue_shop4: the ones on the sherline are like 6mm ballscrews, I cant even find any that small from china
07:00 PM rue_shop4: just ballscrew and nut
07:00 PM Tom_L: oh
07:00 PM Tom_L: is it for the sherline or your new mill?
07:00 PM rue_shop4: when you go under 12mm from china, they go to ACME
07:00 PM Tom_L: sherline are acme
07:00 PM rue_shop4: this sherline was upgraded to 1mm pitch ballscrews
07:00 PM Tom_L: oh
07:01 PM rue_shop4: I'm collecting for the new mill
07:01 PM Tom_L: are they wore out?
07:01 PM Tom_L: they must have damn small balls in them
07:01 PM rue_shop4: no, their fine, I'm surprised their ok cause their so small
07:01 PM rue_shop4: yea
07:01 PM Tom_L: don't take the nut off the shaft
07:01 PM Tom_L: ever
07:02 PM rue_shop4: dunno if I'm sold on ballscrews now,
07:02 PM Tom_L: i definitely am
07:02 PM rue_shop4: yea I was doing it back to make sure there were really balls in there, there are!
07:02 PM Tom_L: mine show no signs of play
07:02 PM rue_shop4: hmm
07:02 PM rue_shop4: I'll consider acme again, but I need to avoid the friction
07:02 PM Tom_L: compared to the stock sherline, there really is no comparison
07:03 PM rue_shop4: its 90% of the req'd torque
07:03 PM Tom_L: i wouldn't but that's your call
07:03 PM rue_shop4: this thing is larger than I need
07:03 PM rue_shop4: well, its got no preload nut, so... hmm
07:03 PM Tom_L: long?
07:03 PM rue_shop4: 40cm
07:03 PM Tom_L: you mean it's too long?
07:03 PM rue_shop4: no, dia is more than I really need
07:04 PM Tom_L: is the end threaded?
07:04 PM rue_shop4: I wouldn't mind 8mm
07:04 PM Tom_L: 16mm is nice
07:04 PM rue_shop4: its thread end-end
07:04 PM Tom_L: no, i mean threaded for a nut
07:04 PM rue_shop4: so, I have to get creative to drive/carry it
07:04 PM rue_shop4: which I dont mind
07:04 PM rue_shop4: I'm more worried about it having play
07:05 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Bearing_Bracket/Bearing_bracket1.jpg
07:06 PM Tom_L: i did those last night and they fit snug
07:06 PM Tom_L: actually i made a mistake in the cad and made them the same size as the hole
07:06 PM Tom_L: i should have gone .001" or so under
07:06 PM Tom_L: i had to sand them a little
07:07 PM Tom_L: you will get more play with acme i'm sure unless the screw is just pure crap
07:08 PM rue_shop4: I can put a preload on the acme, the problem is the friction
07:10 PM Tom_L: i know
07:10 PM Tom_L: i had issues with Z on my sherline
07:11 PM Tom_L: backlash will cause cutters to break too
07:14 PM _unreal_: just got back..... at least my daughter got to see some long lost friends since the B day party thing didnt work out
07:16 PM orlock: i drilled a hole in my filament splicer for a thermocouple
07:16 PM orlock: thin long bit broke
07:16 PM orlock: drilling into an angle is hard
07:16 PM rue_shop4: punch
07:17 PM rue_shop4: start square
07:17 PM rue_shop4: use divvet to contunue at angle
07:17 PM orlock: yeah - i should have
07:18 PM orlock: its only copper so it wasnt too bad, i actually got about 90% of the way through before it snapped :-\
07:18 PM orlock: i ended up using shorter, maybe 1.5mm bit , had t odrill in from each side
07:18 PM orlock: eyeballed it
07:19 PM orlock: and the holes actually met up
07:19 PM orlock: about as could as i could have expected for eyeballing it
07:21 PM orlock: so i gpt some filament to properly "fuse"
07:21 PM orlock: and by that i mean
07:21 PM orlock: "made a molten mess"
07:21 PM orlock: biut i recovered a 1cm long bit of multi-coloured round plastic
07:31 PM _unreal_: shoot me! now my mother called 8:30pm and its a school day tomorrow
07:31 PM _unreal_: still on the phone been 45min
07:40 PM rue_shop4: ok, I put the solar panel rails on the roof!
07:41 PM rue_shop4: _unreal_, confrence phone hookup up to the whole house, do normal stuff and not miss anything
08:27 PM rue_shop4: damn the current controlled stepper driver isnt' wroking out, the current control works, but there are harmonics and things screwing it all up
08:27 PM rue_shop4: 800W power supply and a 2 40W lightbulbs seems would outperform it
08:27 PM rue_shop4: damn
08:28 PM rue_shop4: burned out my good DIY power supply somehow too
08:28 PM rue_shop4: it keeps blowing output transistors, nomatter how much I upgrade them
08:29 PM rue_shop4: _unreal_, when you read thru the datasheet for the A4988 you see there is a LOT of junk regarding modes of current slope control
08:29 PM rue_shop4: I cant pack it all into a circuit
08:30 PM rue_shop4: I think, if you want to get the performance, you either need a 12"x12" pcb of parts for each motor driver or scrap the motor and get a 4 wire with a commercial controller
08:30 PM rue_shop4: :((
08:31 PM _unreal_: I do have a bunch of sla7024/26 driver chips
08:32 PM rue_shop4: yea, give me a call when you have the circuits made up as per the application note in the datasheet
08:32 PM rue_shop4: I'v actually not done a speed test on the SLA7044 chips I set up
08:32 PM _unreal_: I came across this the other day https://www.semicon.sanken-ele.co.jp/sk_content/sla7026m_ds_en.pdf
08:33 PM _unreal_: never seen that data sheet before. answered a few questions thats always kept me from building them
08:33 PM _unreal_: I dont have any 44's
08:33 PM rue_shop4: I wonder if its the microstepping that helps get the good speeds
08:33 PM _unreal_: ramping and microstepping if anything
08:34 PM _unreal_: ARG.... I want my parts to be here
08:37 PM rue_shop3: and I want that motor driver to have worked out nice
09:56 PM rue_shop4: I need something not frustrating to do
09:56 PM zhanx: rue_shop4, i blew a resistor also
09:56 PM zhanx: the mega is done
10:00 PM rue_shop4: it frustrates me most because its stacking things against "just go buy a printer" when this is all crap that shouldn't happen
10:00 PM zhanx: yep
10:01 PM zhanx: this was my fault on the burning out of the board, the pin failing was one thing
10:01 PM zhanx: the stepper being bad was another
10:01 PM zhanx: the list goes on
10:01 PM rue_shop4: did the direction pin fail?
10:01 PM zhanx: it did
10:01 PM rue_shop4: solder joint or what???
10:02 PM zhanx: never got that far, you want this board?
10:02 PM rue_shop4: no I want your printer to be working so you can play with it
10:03 PM zhanx: i cant fix this board and i can send it to you
10:03 PM zhanx: i will have a new one here wednesday
10:03 PM Tom_L: are those still the old 2560 mega boards?
10:04 PM zhanx: yep
10:04 PM Tom_L: huh, i got one stashed away here somewhere
10:04 PM zhanx: https://www.amazon.com/UMProb-ATmega2560-16AU-Development-arduino-Starter/dp/B0721SZW2G/
10:04 PM rue_shop4: Tom_L, their arduino baords, and the driver board plugs right into them
10:04 PM Tom_L: probably still has marlin loaded on it
10:04 PM Tom_L: rue_shop4, yeah i knew that
10:05 PM Tom_L: just been so long i figured maybe they were using something else now
10:05 PM zhanx: that will be here in the morning so i can work on wiring and stuff till then
10:06 PM zhanx: well wednesday morning
10:06 PM Tom_L: i'd get a spare for the cost of the shipping
10:06 PM Tom_L: you got prime though
10:07 PM zhanx: yep
10:07 PM zhanx: we order alot living remote as we do
10:08 PM zhanx: if they sold them at the nearest town gas would still cost me 1o bucks
10:09 PM Tom_L: i dunno what rev mine is but it looks like that
10:10 PM zhanx: r3 has been out a long time
10:11 PM Tom_L: you like NY?
10:15 PM zhanx: its ok, i live in the Adirondacks so its better
10:55 PM ace4016: is it as grey as Syracuse?
10:58 PM zhanx: nope
11:12 PM ace4016: that's a plus