#robotics Logs

Sep 07 2014

01:33 Hyratel nevermind the device, LOOK AT THE PICTURES HERE http://www.ebay.com/itm/141304382160
03:53 rue_shop3 yea, its a port 80 breakout
03:55 rue_shop3 do you think its a reprogrammable fpga?
04:03 rue_shop3 hmm the fileserver drive should not be at 46c
04:03 rue_shop3 pull it apart tommrow
04:05 deshipu I wonder if there are any lipo charger modules for 2s batteries
04:06 rue_shop3 ok I'm just falling asleep now
05:59 MjrTom any of you guys ever used optical distance sensors along the lines of this one: http://media.digikey.com/pdf/Data%20Sheets/Sharp%20PDFs/GP2Y0A41SK0F_Spec.pdf the ds doesn't really talk about accuracy or resolution; the device simply gives a V output dependant on the distance. I'm wondering if the accuracy I can get out of it is really simply a matter of how accurate I can read it (say
05:59 MjrTom assuming ideal operating conditions for it and all that)
07:25 trows Hi ! does anyone use a firmata-based framework to control an arduino ? I'm trying to implement a PID controller and I need some advices
08:56 mumptai re
08:59 mumptai trows, you might want to rephrase you question into something that can be answered as a closed statement
09:00 deshipu MjrTom: it's more complex, as the voltage graph is not linear
09:01 deshipu MjrTom: also, it will change depending on the lighting conditions and (I think) temperature
09:01 deshipu MjrTom: so the accuracy is not as much depending on how precise a number you get, but more about how consistently you get the same number for the same distance
10:02 MjrTom ok thx deshipu! Since then I've found a real datasheet for it, and I see it's log output (and the fast change area is where I need the most precision instead >.< ), and it's also a little non linear... That's all a bit inconvienient. I suppose I'll need to use some calibration scheme with it. BTW, it does seem to be rather temperature stable, and rather insensitive to lighting. All that
10:02 MjrTom being said; apparently it is somewhat a matter of how well you can sample it (and account for the above, ofc)
10:14 deshipu MjrTom: I think there is a library for it
10:14 deshipu MjrTom: that gives you mm
10:15 MjrTom well... in theory if it can be sampled @ 16 bits, you could get 0.156 mils
10:15 deshipu MjrTom: your results will also vary depending on the kind of surface
10:15 deshipu MjrTom: shiny vs. matte, etc.
10:15 MjrTom what I've read (written in Engrish) seems to suggest the surface is go or no-go
10:15 MjrTom and some no-go surfaces could be corrected
10:17 MjrTom the 0.156 mil figure doesn't really take into account the log output
10:17 MjrTom I think I'll use an antilog amp
10:17 MjrTom *antilog op amp circuit
10:19 deshipu no idea if that will help
10:19 deshipu you will also be amplifying the noise
10:21 MjrTom you mean error? I'm not very concerned about high frequency noise components.
10:22 MjrTom an antilog op amp circuit will make it more convienient to work with the values it outputs
10:23 MrCurious chris_99: https://nurdspace.nl/ESP8266#Translated_datasheet Baud rate at 57600your answer from yesterdays question
10:26 deshipu yeah, error
10:26 deshipu sorry
10:27 MjrTom I think it'll mostly boil down to calibration :|
11:07 rue_shop3 ok so
11:07 rue_shop3 see if today, I can cast the pulley, and, hmm
11:07 rue_shop3 clean dust out of the server
11:07 rue_shop3 and...
11:10 rue_shop3 oh yea, tig weld the sink pipe
11:17 fluffywolf going to make me some light fixtures yet? :P
11:25 uffywolf wants to make led lights that don't
11:26 fluffywolf making retrofit bulbs is a horrible idea. a heavy cast aluminum fixture with emitters screwed to it is a much better idea.
11:26 fluffywolf and you're casting aluminum... :P
11:37 rue_more at the currents they are driving most leds, they dont last more than a few years anyhow
11:38 fluffywolf a solid cast base with good emitters will last ~forever...
11:39 fluffywolf since it can get the heat out of the fixture
11:42 rue_more and I want to do up an even rough ik for the robot arm and see what I win
11:42 fluffywolf ik?
11:43 rue_more inverse kinematics
11:43 fluffywolf ah
11:57 fluffywolf the first step of my robot-related project is to figure out how to machine rzeppa joints with only a 4 axis mill.
11:58 fluffywolf then I can see if the rest of my design might actually work or not. :P
11:58 fluffywolf I could make a mockup with a ball joint...
12:01 fluffywolf actually, first step is to find enough time and money to do anything, including fix the mill...
12:22 rue_more why do you need torque control?
12:23 GargantuaSauce for an arm?
12:24 GargantuaSauce to avoid destructive situations i guess. you can have a human in the workspace if the arm can detect when it collides with them
12:24 GargantuaSauce or a misplaced/unmodeled object etc
12:25 GargantuaSauce for legs i think it's absolutely vital for doing anything fancier than a dumb open-loop gait control
12:25 GargantuaSauce if you want dynamic balance, etc
12:25 rue_more no I mean why does fluffywolf need a rzeppa joint for his printer
12:25 GargantuaSauce oh i couldn't parse the word and thus ignored it :D
12:26 fluffywolf printer?
12:26 uffywolf didn't mention printer
12:26 GargantuaSauce those are fancy
12:27 fluffywolf I need cv joints for the hydraulic motors I want to try building.
12:30 fluffywolf commercial hydraulic motors use cardan (universal) joints, but I want to eliminate an extra degree (of the "of freedom" variety) of force I need to make the crosshead handle...
12:35 GargantuaSauce so you think a big curvy concave force-bearing sleeve that also happens to be smooth enough for bearings to roll across it is nore approachable? :/
12:35 GargantuaSauce more*
12:36 fluffywolf yes. :P
12:36 GargantuaSauce well ah...good luck with that!
12:37 fluffywolf I might try a tripod joint... much easier to make.
12:37 fluffywolf are you familiar with hydraulic motors/pumps at all?
12:38 GargantuaSauce i was just going to suggest that, and you can probably get them from a scrapyard also
12:38 GargantuaSauce not really no
12:38 GargantuaSauce other than in reading way too much about random crap
12:39 fluffywolf well, the short summary then is I'm trying to make some ultra-efficient motors and pumps... with the unusual requirement they need to be able to be stalled for extended periods. to both improve efficiency and reduce wear caused by squishing the oil film out when stopped, I'm avoiding any sliding surfaces... like piston skirts.
12:40 GargantuaSauce stalled as in....applying their maximum torque into something that won't budge?
12:40 fluffywolf yes
12:41 GargantuaSauce sounds like a great way to turn your joints into twisted pieces of scrap :P
12:42 fluffywolf that would mean I need larger ones.
12:48 fluffywolf the real issue isn't strength, it's wear... all the moving parts depend on a film of oil to keep the metal apart... when you stop them, that film is squished out.
12:48 fluffywolf just like the saying that 95% of a car engine's wear is in the first 5 seconds every time you start it.
12:49 ace4016 that's what they say about aircraft engines for turboprops, turbofans, and turbojets
12:50 ace4016 fluid bearing doesn't quite work so well at rest :P
12:50 fluffywolf so, to operate stalled as normal operation, I plan on all ball-bearing construction... including skirtless pistons with crossheads.
12:51 fluffywolf this also helps with the very high efficiencies I hope to see...
12:57 fluffywolf it'll also make my motor have about ten times the parts, but minor details...
13:07 fluffywolf a commercial motor uses several sliding plates, ball joints, and other things that are so nice and simple...
13:16 fluffywolf yay! found the copy of an ebook I was looking for for the last hour... been finding nothing but spam.
13:16 fluffywolf http://en.booksee.org/book/1089734 offers non-spam ebooks
13:19 fluffywolf https://www.google.com/transparencyreport/removals/copyright/domains/booksee.org/ lol! always a good sign. :P
13:21 GargantuaSauce i miss library.nu :(
13:22 fluffywolf I haven't seen this site before, but it seems nice... no ads, direct links...
13:22 fluffywolf too bad the browse interface is only in russian, as is most of everything but basic search and download.
13:24 fluffywolf hrmm, methinks that checking the google takedown reports list for the sites with the most takedown requests would be a good way to find useful sites...
13:25 fluffywolf after all, if they're generating lots of takedown requests, they most likely offer lots of genuine non-spam downloads. :)
13:32 fluffywolf this book has a lot more hard math than I was expecting.
13:33 fluffywolf it makes my calc textbook look fun.
13:41 fluffywolf (which I'd need to dig out from under a few tons of crap in storage to even pretend to work out some of this math...)
15:46 fluffywolf ok. I've decided making an rzeppa joint isn't that hard. making one that doesn't suck, however...
15:49 fluffywolf spherical geometry is easy... but sucks. anything else is complex as fuck. the cage has to be machined to within a few um. case hardening is a must.
15:49 fluffywolf I'm very inclined to go with tripod joints now. heh.
15:51 fluffywolf ... if they can handle the angles.
15:51 fluffywolf and are cv-enough.
16:10 rue_more fluffywolf, why do you need rzeppa joints?
16:10 rue_more fluffywolf, why do you need rzeppa joints?
16:11 fluffywolf to couple the swash plate to the shaft in a hydraulic motor
16:12 fluffywolf commercial motors use flat plates sliding over each other, which doesn't need a joint, or standard u-joints with ball joints to make up for the non-constant velocity.
16:13 rue_more IIRC, THERE Is a high pressure oil hole in the back, that makes everything ok till it plugs
16:24 rue_more if the plate moves freely back and forth, how do the cylinders cause it to rotate?
16:27 fluffywolf the sliding type uses shoes (small plates) that slide on the large plate. the plate stays still, while the cylinder body rotates.
16:28 fluffywolf the other type has a bend in the whole housing rather than just a plate, with a standard universal joint at the bend. this way both the plate and the cylinders rotate, eliminating sliding, but you need ball joints (a form of sliding) to make up for all the funky non-constant movements.
16:31 fluffywolf bent axis (the ones where the whole pump bends) are most efficiency, but not variable displacement. the swashplate pumps with sliding shoes can be variable displacement, but poor efficiency and other issues.
16:33 fluffywolf my design is to fill a very small niche... very high efficiency, able to be stalled without wear, suitable for low speeds and low powers.
16:33 fluffywolf and variable displacement, which rules out bent axis designs.
23:40 kristof What kind of ROM requirements do people find they need for autonomous quadcopters?