#robotics Logs

Apr 25 2013

00:00 GargantuaSauce unless there's something i'm missing here
00:00 Triffid_Hunter Jak_o_Shadows: for best results, post a png
00:03 GargantuaSauce also that 20 ohm resistor had better have a good power rating
00:04 Jak_o_Shadows https://dl.dropboxusercontent.com/u/104809629/Capture.PNG
00:04 GargantuaSauce if you're doing current sensing you should use one a couple orders of magnitude smaller
00:04 Jak_o_Shadows Nah, not doing current sensing
00:04 Triffid_Hunter Jak_o_Shadows: you have the whole right side of the H backwarsd
00:04 Triffid_Hunter Jak_o_Shadows: so D7,D8,Q3,Q4 are backwards
00:05 Triffid_Hunter expect fire if you hook that up ;)
00:05 Triffid_Hunter Jak_o_Shadows: also, ditch R11 completely.. it'll emit fire and make your motor super weak
00:05 Jak_o_Shadows yeah, was skipping that anyway
00:06 Jak_o_Shadows but arg, d7/d8/q3/q4 are soldered in.
00:06 Triffid_Hunter Jak_o_Shadows: I hope you've routed AHO/ALO/BHO/BLO properly, can't tell from your diagram
00:06 Jak_o_Shadows reasonably sure I have.
00:07 Triffid_Hunter the fire will let you know if you have mixed them
00:07 GargantuaSauce q3 and q4 will conduct between the rails
00:08 Jak_o_Shadows https://dl.dropboxusercontent.com/u/104809629/2.PNG
00:08 Jak_o_Shadows fixed?
00:08 Triffid_Hunter Jak_o_Shadows: now you have D12/D13 backwards
00:08 Triffid_Hunter mosfets are good though
00:09 Jak_o_Shadows ah, k
00:09 GargantuaSauce yeah all 4 diodes should be pointing up
00:09 Jak_o_Shadows yeah, that makes some sense.
00:09 GargantuaSauce they are there to prevent the body diodes of the fets from being loaded
00:09 Jak_o_Shadows Now i need to figure out how to change this protoboard.
00:10 GargantuaSauce i find it's usually worth starting over if you have to make big changes to a perfboard
00:10 GargantuaSauce hope you used a socket for the driver!
00:10 Jak_o_Shadows nah, but I have a spare
00:11 Triffid_Hunter naw shouldn't be too hard at all, just swap drain/source for the two mosfets and reverse the two incorrect diodes
00:11 Jak_o_Shadows and I think i can just rotate the two RHS fets 180, move diodes, and then i'm good
00:11 Jak_o_Shadows so I think the chip is good
00:11 Jak_o_Shadows also, what's the easiest way of breaking the lines of track on a protoboard?
00:12 Triffid_Hunter Jak_o_Shadows: why is C7 100uF?
00:12 Triffid_Hunter Jak_o_Shadows: for those two, it's more important that they're fast than that they have a large capacity
00:12 Jak_o_Shadows isn't 100+100 =200?
00:12 Jak_o_Shadows anyway, I don't actually have the capacitors.
00:13 Jak_o_Shadows Unless you can use electrolytic ones.
00:13 Triffid_Hunter Jak_o_Shadows: C6/C7 must be ceramic types, 100nF or more. I used 10uF so I could have the high side mosfets on for a long time without worrying about them running low
00:14 Triffid_Hunter Jak_o_Shadows: basically, every time the low side mosfet is turned on, they get charged up. every time the high side mosfet gets turned on, they basically get shorted to a ~5nF capacitor
00:14 Jak_o_Shadows I don't have any ceramic caps bigger than 47nF, so that's a non-issue till i get it
00:15 GargantuaSauce razor blade Jak_o_Shadows
00:15 Triffid_Hunter Jak_o_Shadows: C8 is the same deal, must be ceramic since it provides the gate charge for the low side mosfets. add an electro in parallel to help hold up the voltage rail
00:15 Jak_o_Shadows thanks GargantuaSauce
00:15 Jak_o_Shadows yeah, will do
00:36 Jak_o_Android Fets ate back in, diode on the side wjer3the3fets weren't moved is flipped around
00:57 rue_bed makeing a driver?
01:03 Jak_o_Shadows yeah.
01:03 Jak_o_Shadows Well, given up for just now, don't have caps, don't have diodes
01:13 Triffid_Hunter Jak_o_Shadows: don't strictly need the diodes, the mosfets have body diodes which do the job
01:14 GargantuaSauce yes with a drop of like 1.5v
01:16 Jak_o_Shadows oh well.
01:46 rue_bed even a little current thru that body diode will cuase a LOT of heat off the fet
01:46 rue_bed esp if its forward breakdown
01:46 yahweh evening Rue
01:46 rue_bed zzzzzzzzzzzzzzzzzz
01:47 rue_bed goin down
01:47 yahweh liar
01:47 yahweh No you aren't
01:47 yahweh 2.5v 11mA blue LED
01:47 yahweh Go
01:47 yahweh Find me one.
01:48 yahweh :p
01:49 yahweh With a resistor at 220 ohms limiting it to 11mA that apparently can be changed.
07:58 GuShH new firefox cellphones are out, RAM backpack sold separately.
08:05 Facepainter Hi! I'd like to seek some thoughts about hydraulics gear motor to be used for building a robot arm as opposed to using dc servomotors
08:07 GuShH Facepainter: step one of your robot girlfriend build?
08:09 Facepainter oh yeah
08:09 Facepainter need me a robot girlfriend
08:27 hackkitten Facepainter: which parts specifically are you thinking about?
08:27 hackkitten hydraulics are powerful, but slow
08:27 hackkitten also fairly complex
08:29 Facepainter Well, in building a robot arm, i initially wanted to use several servo motors to do various functions such as grip, rotator, elbow etc but there's just too much torque needed and it's more expensive to get servomotors for the amount of torque they have
08:29 Facepainter so i am looking into hydraulics gear motors??
08:31 hackkitten if it's torque you need, then hydraulics is a good choice :)
08:33 Facepainter do you know if it's been done?
08:37 Facepainter some people comment that hydraulics is noisy, messy and expensive but i havent really seen anywhere in the interwebz if someone's tried it as i want to know how it turns out.. i assume though that there will be a lot of excess torque so i'm not sure if it's actually cost effective
08:37 Facepainter and also what do you mean by hydraulics are slow? explain pls? haha
08:39 Facepainter and and and what do you know about pneumatics, or anyone else.. what are they like? are they similar to hydraulics? how are they different?
08:39 dunz0r Good hyrdraulics are quite fast actually.
08:39 dunz0r But good and fast hydraulics costs a lot of money
08:40 Facepainter what makes them slow?
08:42 hackkitten pneumatics have a delay and have less torque due to the use of a compressible gas
08:42 hackkitten dunz0r: fast depends on what you define as 'fast' :)
08:43 hackkitten but it's all a matter of trade-offs
08:43 dunz0r hackkitten: Good point. Fast with that amount of torque might be a better wording.
08:43 hackkitten and fast hydraulics also means more maintenance
08:43 hackkitten yup :)
08:45 Facepainter do you guys think using hydraulics to make a robot arm is not cost effective than using servomotors? seeming that there is maintainance on top of the base cost of the hydraulic motors..
08:47 dunz0r Facepainter: It depends on the size/torque.
08:48 dunz0r Servo motors aren't cheap either.
08:48 dunz0r Regular servos is what I would use, but I'm cheap :)
08:48 dunz0r Facepainter: "Real" robotarms (Kuka, ABB etc) all uses hydraulics.
08:48 dunz0r Not all probably, but most I've seen at least.
08:49 Jak_o_Shadows What size/torque capacity are you actually looking at?
08:49 rue_house no not all real robot arms use hydraulics
08:50 Facepainter well im looking to get around 300kgcm
08:50 Facepainter torque
08:50 rue_house some use servomotors some even use steppers
08:50 rue_house Facepainter, at what rpm?
08:51 Facepainter no preference
08:51 Facepainter i dont think there's snail pace rpm anyway haha
08:51 rue_house then you could do it with a pager motor
08:51 dunz0r Haha.
08:51 GuShH hackkitten: hidraulics don't have to be slow
08:51 dunz0r 1000000000000:1 gearing or so :)
08:51 rue_house yup
08:52 rue_house have soemthing that acutally runs slower than a stepper
08:53 rue_house I still need to write the tutorial on how to turn windshield wiper motors into servos
08:53 Facepainter also, can you refer me to arduino interface tutorials with hydraulics?
08:54 rue_house can you actually afford proportionate hydraulic valves?
08:54 rue_house their over $1000ea
08:54 Facepainter i saw some hydraulic motors costing around 600 dollars or so
08:54 rue_house yea, motors and cylinders are cheap
08:54 Facepainter i can afford anything below 1 grand
08:55 rue_house the valves to do anymore than on and off are REALLY expensive
08:55 rue_house proportionate hydraulic valves are over $1000 EACH
08:55 Facepainter can you give me a product example?
08:55 dunz0r Facepainter: Iirc, the valves are solenoids, so a FET should be enough to control them.
08:55 dunz0r Not sure about the fancier ones.
08:55 rue_house not for speed control
08:56 rue_house they are current based inputs
08:56 Facepainter rue i have found a few valves but this one is 55 dollars
08:56 Facepainter http://compare.ebay.com.au/like/251022310460?ltyp=AllFixedPriceItemTypes
08:56 Facepainter wtf
09:03 Facepainter rue_house: did you mean that hydraulic motors are not self contained? ie, plug and play
09:04 rue_house nope, I mean that the valves to control the hydraulic oil cost a lot
09:05 rue_house in a hydraulic system you have an oil tank, a pump, valves, and actuators
09:05 rue_house robot arms do not use on/off valves with the hydraulics, not commercial ones anyhow
09:06 rue_house ABB makes valves iirc
09:06 Facepainter so i dont really need valves then?
09:07 rue_house your not understanding
09:07 Facepainter i understand that its part of a hydraulic system
09:07 rue_house the valves robot arms use have more than on and off, they can varry the amount of fluid flowing
09:07 rue_house which makes the rbot not jerky
09:08 rue_house http://www.boschrexroth-us.com/country_units/america/united_states/sub_websites/brus_brh_i/en/products_ss/08_proportional_servo_valves/index.jsp
09:08 Facepainter thank you for the link im gonna look at it now
09:08 rue_house good thing, the servo valves are what you would want to do it officially
09:10 rue_house why dont you just use electric motors?
09:11 rue_house http://www.ebay.ca/itm/Moog-D651-401-Typ-P-35DBM3AVFA-12-87-Hydraulic-Servo-Valve-Lot-of-2-/170854422855 there is a really good price on a used valve
09:13 rue_house http://www.ebay.ca/itm/MOOG-G631-3705B-631-195-C1-ELECTRO-HYDRAULIC-SERVO-VALVE-NEW-/251062356648 not cheap new
09:16 Facepainter well the arm wswill be tiny as fk if i use normal motors
09:16 rue_house wrong
09:17 rue_house you always gear motors anyhow
09:17 rue_house gearmotors make it easy
09:17 rue_house 300kg/cm?
09:18 Facepainter well it's just that when i calculated before for a 6x4cm robot arm it will need 5 motors with the largest one being 60kgcm torque and the maximum it can lift is only 1.59kg
09:19 rue_house do you want to do that on a shaft or have a linear actuator like a hydraulic cylinder?
09:19 rue_house 6x4cm?
09:20 rue_house http://eds.dyndns.org/~ircjunk/robots/arm5/p1030984.jpg
09:20 rue_house smaller than that?
09:21 Facepainter oh sorry what i meant with 6cm x4 is
09:22 Facepainter there's 4 robot arm parts that are 6 cm each
09:22 Facepainter so the total size is 24 cm
09:22 Facepainter so thats fkin tiny
09:23 Facepainter and it's going to need that much torque.. as i said above
09:23 rue_house even if you use lead the arm wouldn't need motors that big
09:24 rue_house tell me what you think is a reasonable amount of time for a robot to go from straight out (horizontal) to straight up (verticle)
09:24 rue_house 6 seconds?
09:25 Facepainter sure. i have no preference but go on
09:25 rue_house your desgining for 2kg, ok
09:26 rue_house what size do you want? that one in my pic is about 1.3m reach
09:26 Facepainter well thats the thing.. 2kg is too small. i want to expand but it will cost too much buying so many servo
09:26 rue_house so outline what you want
09:26 rue_house I have to leave in a few minutes, I wont back back for 2 days
09:27 Facepainter alright one sec
09:28 Facepainter that size you showed me is actually perfect. so about 1.2m in size is awesome
09:28 Facepainter but i have to ask
09:29 Facepainter whats the whole weight of that arm in the pic if you know
09:29 rue_house 800kg linear actuators from hospital beds
09:30 rue_house that arm weights about 20kg
09:31 rue_house http://www.ebay.ca/itm/New-9-84-250mm-Stroke-Trip-12V-DC-Linear-Actuator-133lbs-60kg-/290710298886
09:31 rue_house maybe I should sell some :)
09:31 Facepainter haha
09:32 rue_house thats for 60kg, mine are 800kg
09:33 Facepainter ok just for a background: i want to build this arm with my friend because we want it to be able to build another arm
09:33 Facepainter so that means this arm has to be able to lift its weight
09:33 Facepainter if it were a 20kg arm i want the minimum amount it can lift to be 20kg
09:36 rue_house hmm
09:36 rue_house stay around for a few days
09:36 Facepainter will do
09:36 Facepainter thanks man
09:37 Facepainter you going now?
09:37 rue_house yep
09:37 Facepainter alright. catch
10:39 theBear hmmm.... if my arm weighed 20kgs, i'd want it to be able to lift a LOT more than that
12:02 SquirrelCZECH hi folks!
12:02 SquirrelCZECH is there anyone with knowledge in hexacopter?
12:03 uirrelCZECH got hexacopter-like project, and is not sure about picking proper propel
12:03 SquirrelCZECH so, resume: it's hexacopter which looks like airship (hindenburg)
12:04 SquirrelCZECH or, it's going to be
12:04 SquirrelCZECH our goal weight is 3kg + load (2kg), I picked 6x250w motors
12:05 SquirrelCZECH (I suppose that if hexacopters with 1.6kg weight and 1.6kg load have 150w motors, this should be enough)
12:05 SquirrelCZECH (mainly because our aim are not acrobatic things)
12:05 SquirrelCZECH and I picked 13x8 propellers... it should be OK?
13:37 theBear hmm, how big is a 150w motor these days ?
13:46 SquirrelCZECH theBear: small
13:46 SquirrelCZECH 250w got diameter around 30mm
13:50 theBear holy crap !
17:46 yahweh Having a hell of a time finding 2.5v blue LEDs. Anyone here know a place?
17:51 ace4016 why 2,5V in particular?
18:15 Tom_itx http://www.lc-led.com/Products/department/5
18:19 Tom_itx http://www.alliedelec.com/search/productdetail.aspx?SKU=70053007
18:20 Tom_itx making a grow light?
18:53 yahweh <ace4016> why 2,5V in particular?
18:53 yahweh Replacing a red LED in a headphone amp.
18:53 Tom_itx so add a resistor to it
18:53 yahweh It's 2.5v and a 220ohm resistor limits it to 11mA
18:54 yahweh How do I calculate what values I need?
18:54 Tom_itx just take the forward voltage from the actual voltage
18:54 Tom_itx and use ohm's law
18:54 Tom_itx for the current
18:55 Tom_itx 2.5v from 5v would be a 2.5v drop across the resistor
18:55 yahweh Okay
18:56 Tom_itx do the same with current
18:56 yahweh So I add the two values?
18:57 yahweh More resistence for voltage plus more resistence for amperage?
18:57 Tom_itx http://led.linear1.org/1led.wiz
18:58 Tom_itx so you need to find the Vf for your led
18:58 yahweh Oh, that looks handy
18:58 Tom_itx forward voltage
18:58 Tom_itx and enter your desired ma draw
18:59 yahweh Yep, okay
18:59 yahweh Thanks man
18:59 yahweh Sorry if I sound dumb :(
18:59 Tom_itx np
21:01 theBear hmmm.... so inductors/coils in general have an inherant phase shift at pass freqs... if i make a 'transformer' with several 'identical' (theoretical) windings, then pulse one of those windings, do the others move in unison with it, or in sympathy with it ? perhaps the load/therefore cutoff freq on 'secondaries' makes a difference ?
21:20 ace4016 each inductor will typically induce a shift; each in the same direction (caps will cause it to go the other way)
21:21 ace4016 i believe they add thier shift...it's been a while
21:39 theBear mmm, makes sense, one shift on the initial coil -> core/field conversion, and same on the way back out of the core/field.... i'm doing a kinda blanket replace/upgrade on this oldish atx that tends to power cycle a little too much (1-infinite times) on this 'new' core2/amt management board at initial powerup, but i was musing about how 'useless' my former measurement attempts were at identifying the exact problem(s)... i figure single multi winding/ta
21:39 theBear p output coil off the main switcher doing at least 3.3, 5, 12 rails, and likely the -'s too... anything overly big/intermittant and small (loadwise) will cause big ugly hiccups on all the lines in the moment before fail/psu shutdown/cycle, but in theory if i probed 2 rails at a time, i would be able to identify the 'problem' load/rail by looking for where the dip occurs first, maybe 90 or 180deg ahead of all the other rails and the protection sensing
21:39 theBear it
21:51 wizrobe maybe i can hook a discussion of this in this channel
21:51 wizrobe http://opencog.org/roadmap/
21:51 wizrobe OUt of all the stuff in this timeline, this one line sticks out at me.
21:52 wizrobe 2015-2016: Advanced Learning and Reasoning
21:52 wizrobe "Humanoid robotic control outside the robot lab within rich environments."
21:52 wizrobe ...
21:52 wizrobe does anyone really think opencog is going to have a robot on two legs getting around outside?
21:52 wizrobe by 2015?
21:52 theBear jeez, 3 years in the future, that IS advanced
21:52 wizrobe I guess im wondering how such a prestigious project like opencog could be so simultaneously ignorant of major problems in robotics
21:52 theBear err, what's a AGI ?
21:53 wizrobe theBear yeah so like you need to assemble a team of guys who know hydraulics and then there's this problem with very low-latency perceptions of cmaeras that shoot like 200 frames a second
21:53 wizrobe THere are completely engineering isues to making a bipedal robot
21:54 theBear and it does only say control in rich environments, and learning/reasoning... for all intents and purposes the "robot" could be a laptop that is already 'obsolete', just imagine what kinda old laptops there's gonna be a few years from now !
21:54 theBear meh, the nips got one or two of 'em
21:54 theBear they got 2 legs
22:04 MrCurious generic macbook pro batteries are $20 ea on ebay NEW, in them are 3 fat lipo cells, a charge/discharge controller with a led level tester. 60Wh
22:04 MrCurious i managed to get 2 for $30 delivered
23:05 wizrobe Yeah I mean.. you can have the most powerful *cognitive" engine known to man.
23:05 wizrobe But there are still serious engineering and dynamics problems to bipedal robots
23:05 wizrobe So given the rest of the opencog timeline, I'm a little worried
23:05 MrCurious if its smart enough, it will work em out on its own
23:05 MrCurious before taking over the world
23:05 wizrobe OH!
23:05 wizrobe This i haven't considered
23:06 wizrobe Yeah I guess theoretically it coudl figure out all of dynamics on its own
23:30 Triffid_Hunter wizrobe: there's already a robot that learns how its legs are arranged from scratch using just an accelerometer
23:30 Triffid_Hunter and wiggling them of course
23:31 theBear hehe, kinda like me when psychadelics are involved and i've been still for too long :)
23:37 MrCurious robo-psychedelics.... that could speed evolution
23:44 theBear heh, could go either way <grin>
23:47 home GUYS
23:47 home what can I do with lasers?
23:47 home Go go
23:50 theBear lol, stick 'em on frickin shark heads
23:53 MrCurious blind yourself, others, and pets most likely
23:54 MrCurious sharks have been done. stuck them on monkey's heads
23:54 MrCurious suppose if you have a 5oz, 5sq mile solar sail, you could drive a space ship
23:55 MrCurious or you could use them to put on really bright shadow puppet shows
23:56 theBear hehe "that monkey only has one ass ! it's of no use to me !"