#robotics Logs

May 06 2013

#robotics Calendar


00:33 MrCurious guess this time of night, all robot making/discussions stop
00:35 rue_bed I object
00:35 rue_bed I merely stopped thinking
00:35 rue_bed I shall probably continue to blather on for hours
00:36 rue_bed I found and wound a core for the 12->24V converter for the mowerbots blade motor
00:37 rue_bed I think the digits for the clock came out nice, I need to make it a colon
00:37 Hukka I just woke
00:38 MrCurious so you are going to arm your robot?
00:38 MrCurious with spinning blades
00:38 MrCurious before establishing trust or expected behavior or any basic guidelines?
00:39 rue_bed the mowerbot is not the wifi bot
00:39 rue_bed the new one may be tho
00:39 MrCurious i was thinking of giving mine a taser
00:39 MrCurious but couldnt think of why it would need one
00:39 MrCurious other than to be cool
00:39 rue_bed whhy move out of the way of obsticles when yo can make them move away from you?
00:42 Curious is pleased with the productive wee
00:59 MrCurious when in doubt.. watch fringe on netflix
01:23 MrCurious my dc to dc converter is cooking. guessing it is not pleased with 8w of pull
01:26 MrCurious perhaps it just needs a heatsink on the coil
01:27 Triffid_Hunter coil is getting hot? might be saturating
01:29 MrCurious go on
01:29 MrCurious pretend like i dont know what you mean
01:30 MrCurious max output current of 3A
01:31 Triffid_Hunter MrCurious: inductor can only hold a certain strength of magnetic field. once that field strength is achieved, increase in current does not cause increase in field
01:31 MrCurious no 8W @5v is like 1.6A
01:31 Triffid_Hunter MrCurious: when that happens, only the resistance of the windings is electrically visible- the inductance vanishes completely
01:31 Triffid_Hunter this is called saturation
01:32 MrCurious http://www.ebay.com/itm/Mini-LM2596-DC-DC4-40V-to-1-5-35V-Step-down-Converter-Power-Supply-Module-/111066266299?pt=LH_DefaultDomain_0&hash=item19dc10babb
01:32 MrCurious that is the module
01:32 MrCurious the part labeled 330 is getting warm
01:32 Triffid_Hunter yep that's the inductor
01:33 Triffid_Hunter wow, price on those has dropped again, they were $1.50/ea when I got my set
01:34 Triffid_Hunter they're probably cheaper than LM2596 @qty 1k now.. seems like someone's running a LOT of grey shifts
01:34 MrCurious whats a gray shift?
01:34 Triffid_Hunter heh yup, $1.80@1k from TI
01:35 Triffid_Hunter MrCurious: grey shift is when a factory is contracted to produce a certain number of items, and when they've made that many simply keep on producing
01:35 Triffid_Hunter then sell the extras themselves
01:35 MrCurious ah nice
01:36 MrCurious whoa just noticed rue has a lot of SLA on his robot
01:36 Triffid_Hunter yeah, old fashioned :)
01:36 MrCurious i have a bigger dc-dc converter i could switch to, but i think i will let this one fail first
01:36 Triffid_Hunter it's all about lipo these days it seems
01:37 Triffid_Hunter MrCurious: as long as it's not higher than 100c it'll probably just keep running like that
01:37 GargantuaSauce letting a ps fail is probably a bad idea
01:37 Triffid_Hunter MrCurious: inductors fail when the enamel insulation on the windings breaks down, and the windings short out
01:37 GargantuaSauce because they tend to take other things out with them!
01:37 Triffid_Hunter but yeah, if you have anything worthy downstream, suggest a crowbar and fuse
01:39 MrCurious whats downstream has value
01:39 MrCurious was hoping to save this dc-dc converter for motors (90W)
01:40 Triffid_Hunter MrCurious: if the inductor fails, the LM2596 may get damaged from the gross overcurrent. silicon frequently fails short circuit
01:45 MrCurious yeah. thinking i may have to order a new one off of ebay
01:45 MrCurious but its no where near its peak current
01:45 MrCurious and its temp while hot, is not so hot i cant hold my finger to it indefinately
01:46 Triffid_Hunter MrCurious: then it's below 45c and is fine
01:46 MrCurious yay!
01:46 Jak_o_Shadows MrCurious: You know, until you pass out from exhaustion anyway
01:46 Triffid_Hunter MrCurious: also, the LM2596 chip itself is rated to 3A. the pcb isn't large enough to dissipate enough heat for it to run much past 2A as you're finding out
01:46 MrCurious actually the chip next to the inductor is hotter
01:47 MrCurious that may be heating up the coil
01:47 Triffid_Hunter the LM2596 itself, or the diode?
01:48 GargantuaSauce it's probably mostly the switcher
01:48 GargantuaSauce also i want to note that core saturation is not an absolute thing
01:49 Triffid_Hunter yeah it follows a curve, not a hard limit
01:49 Triffid_Hunter but easier to explain it as a hard limit at first
01:49 GargantuaSauce one of many sources of delicious nonlinearities
01:51 Triffid_Hunter it's only ~80% efficient at 5v out
01:51 Triffid_Hunter (the LM2596)
01:51 MrCurious that could splain the heat
01:52 Triffid_Hunter still more efficient than a linreg :)
01:52 MrCurious yes.
01:55 MrCurious so the lipo's may burst into flame.
01:55 MrCurious the dc-dc regulator may stop regulating
01:55 MrCurious the computers may stop computing
01:55 Triffid_Hunter heh lipos emitting fire is always a risk
01:56 GargantuaSauce it is not unlikely that any of those things happening will make them all happen
01:56 GargantuaSauce also consider http://www.ebay.com/itm/260755027441
01:56 GargantuaSauce tis what i am using for 3s lipo to 6v for a bunch of servos
01:57 Triffid_Hunter heh I have a hobbyking "5A" ubec, it's just two 3A bucks in parallel
01:57 GargantuaSauce and was the best choice on fleabay 6 months ago when i looked
01:57 GargantuaSauce oh dear
01:57 MrCurious thats why they will be living inside 3mm thick aluminum
01:58 Triffid_Hunter GargantuaSauce: hm I would have skipped that one for 10v min input.. 3S run to or even below 9v before being technically empty
01:58 MrCurious http://www.ebay.com/itm/DC-DC-Buck-Step-Down-Converter-4-5-60V-12v-to-1-25-30V-5v-90W-For-POE-Remote-PSU-/290908896206?pt=LH_DefaultDomain_0&hash=item43bb857bce
01:58 MrCurious wouldnt that be better, and cheaper?
01:58 GargantuaSauce yes
01:59 Jak_o_Shadows oooh, damn, servos need power don't they.
01:59 MrCurious hmmm only 3A max
01:59 GargantuaSauce mm yeah the 90W is a little misleading
02:00 Triffid_Hunter Jak_o_Shadows: yep, don't try and power them from logic power
02:00 GargantuaSauce yeah the damn things peak at well over an amp when loaded
02:01 GargantuaSauce okay sleepytime is long overdue
02:02 Jak_o_Shadows Triffid_Hunter: yeah, hopefully it'll be fine, it's not being used for any other power circuits
02:03 Triffid_Hunter Jak_o_Shadows: those LM2596 modules do the job in a pinch, and they're so cheap you may as well grab a set of 10
02:03 Jak_o_Shadows Yeah, i've got two. One will be for 5V, the other for 12ish
02:04 Triffid_Hunter shouldn't need for 12 unless your Vin is >=20v
02:04 MrCurious that magic smoke tells me that buck converter WAS a boost converter
02:05 Jak_o_Shadows Triffid_Hunter: Just in case. I need to charge the battery and check.
02:05 Triffid_Hunter Jak_o_Shadows: 3S tops out at 12.6 unless your charger is broken
02:05 Jak_o_Shadows 3S?
02:05 GargantuaSauce 3 stupidunits
02:05 Triffid_Hunter lipo pack, 3 cells in series
02:05 GargantuaSauce series
02:06 Jak_o_Shadows hahahahahahaha, using a ni-cd battery.
02:06 GargantuaSauce and don't get me started on "C"
02:06 Jak_o_Shadows Lipos, lol.
02:06 MrCurious nicd? but those dont explode... why waste time on those
02:06 GargantuaSauce they can if you try hard enough<
02:09 MrCurious Triffid_Hunter: what you think of this one? http://www.ebay.com/itm/DC-DC-Buck-Converter-Step-Down-Module-Power-Supply-7V-24-Output-5V-3A-/261210863885?pt=LH_DefaultDomain_0&hash=item3cd161790d
02:11 GargantuaSauce the current continuous 2.5A output peak 4A output Able to provide to 3A of continuous load current
02:11 GargantuaSauce little on the sketchy side
02:11 MrCurious seems beefier than the last one
02:11 GargantuaSauce that monolithic package doesnt look like it can dissipate much heat
02:12 MrCurious yeah. moving on to the next one
02:12 GargantuaSauce at this price point you may as well just get a handful of different ones
02:14 MrCurious yeah
02:14 MrCurious now i am powering the raspberry pi and odroidu2 off of the same dcdc converter
02:14 MrCurious figure i will split em to seperate converters
02:14 MrCurious localize any damage
02:15 GargantuaSauce and fuses and zeners are never a bad idea
02:17 MrCurious wonder if local frys stocks a good dc-dc converter
02:19 MrCurious and radio shack appears to see you coming and lifts your money on teh way in and out
02:32 MrCurious no mounting holes on that dc-dc converter? *PUZZLED*
02:32 Triffid_Hunter mounting holes? who uses mounting holes these days? just 3d print a suitable sized mounting bracket and zip-tie it on
02:36 MrCurious OMG, the kinect pushes this dc-dc converter over the top
02:39 MrCurious if i use 2 of these(dc-dc converters) in parallel, would they balance the load, or would one carry the load while the other coasted
02:40 Triffid_Hunter MrCurious: one will carry most of the load while the other one coasts
02:40 MrCurious figures
02:40 Triffid_Hunter you'd need load-sharing resistors on the output
02:41 Rif has anyone used encoders with a balancing robot?
02:42 Triffid_Hunter Rif: heya :)
02:42 Rif i need some advice or a link on how to incorporate them
02:42 Rif yo
02:42 Triffid_Hunter Rif: the encoders will just plug into your motor PID loop
02:42 Triffid_Hunter derive position or velocity from them
02:43 Rif yes ok, so th start with just count pulses rather than using them for control
02:43 Triffid_Hunter then your gyro+accel gives desired velocity, the encoder gives actual velocity then you feed those two to another PID loop
02:43 Triffid_Hunter for extra points, that loop gives a motor current, and a third loop controls desired vs actual current
02:44 Rif ok i will have a current sense on these drivers, but will leave that for now, it gave me odd readings
02:46 Rif at the moment the gyro and accel are aiming for 0 velocity i guess, will be happy to start by just getting it to balance and stay in one spot, and return to it if bumped
02:47 Triffid_Hunter ah yeah that's where you need the encoders
02:47 Triffid_Hunter to track position
02:47 Triffid_Hunter once you have that working, you can make it drive around simply by changing the target position ;)
02:48 Rif yeah, need to finish these motors brackets, cannot mount them to platform at moment,
02:48 Rif Triffid_Hunter, yep thats the way i was thinking, have each wheel getting its own drive signal now, so will be able to turn etc
02:52 Rif i already have a Time value for the time the loop takes for the gyro, could use it and the encoder counts in that time to get a velocity?
02:53 Triffid_Hunter sure
02:53 Triffid_Hunter assuming you get enough counts that is
02:54 Rif will be 2016 per motor rev, and the motor will be geared 3:1 with timing belt
02:54 Triffid_Hunter here's a really fast smoothing algorithm: have an n-element ringbuffer and an accumulator. at each reading, if the queue is full, pop the oldest reading and subtract from accumulator. add new reading to accumulator and stuff into queue. now, the accumulator holds average * n (buffer size)
02:55 Rif damn forgot to disconnect at work, am gonna annoy myself with the reconnects now :)
02:55 Triffid_Hunter what radius are your wheels?
02:55 Rif 75mm Radius
02:56 Rif made em slightly bigger than what was on the other day, motor was a bit slow
02:56 Rif but has heaps of torque left even at 7.2V instead of 12
02:57 Triffid_Hunter so ~12.8 counts per mm, not bad
02:57 Rif these encoder motor need more voltage though, will run them from 2 7.2V packs in series
02:57 Rif cause they are 24V
02:57 Triffid_Hunter you'll need the smoothing though, else most of the time you'll get zero velocity
02:58 Rif yep that sound like heaps
02:58 Rif ok
02:59 Triffid_Hunter Rif: consider.. you move at 10mm/second but you might sense every 10 milliseconds so you'll get 12 out of 100 readings with a single count advance
02:59 Triffid_Hunter no wait that's at 1mm/s
02:59 Rif oh so i might need a different timer
02:59 Triffid_Hunter but still, if you can boost up to ~50 counts at your desired speed you'll get much smoother response
03:00 Rif a longer interval than the gyro uses
03:00 Rif its 9ms at moment
03:03 Rif Triffid_Hunter, so you reckon i would have the ability to make a working reprap?
03:04 Rif mechanically i think i could, just dunno how much coding would be required that i lack
03:19 Triffid_Hunter Rif: all the coding is done for you already, see marlin and sprinter firmwares
03:20 Rif yep so can do cad designs and then watch the magic, will look up those names
03:21 Triffid_Hunter Rif: github.com/ErikZalm/Marlin from memory
03:28 Rif just calculated 33.48 ticks per mm, seems pretty good
03:31 Rif i must be out somewhere,is way different to yours
03:33 Rif 2016 per motor rev, gearing 2.857:1, wheel circumfirence 172mm
03:34 Rif so about 60mm per rev, or 0.298mm per tick
03:34 Triffid_Hunter Rif: hm possibly, I just did it quickly
03:34 Rif 0.0298*
03:34 Triffid_Hunter if circumference is 172mm, you'll move 172mm per rev
03:35 Rif but there is almost 3:1 gearing after the motor
03:36 Rif yes per rev of the wheel, but i meant 60mm for each rev of the motor, which has the encoder in it
03:37 Rif hrm that seems like alot for these motors, hopefully the motors can handle that, they are just from a printer
03:40 Rif but an old HP, back when printers were still made out of decent parts
06:28 f is in the reprap world
06:29 Rif this encoder to mm per second will have to wait, my brain just can't do it
06:50 Jak_o_Shadows Rif you could make a rerap dead easy I think
10:13 Tom_itx http://www.electronicproducts.com/Passive_Components/Magnetics_Inductors_Transformers/What_Makes_Rare-Earth_Magnets_So_Special_amp_How_are_they_made.aspx
10:15 dunz0r Very interesting Tom_itx, thanks for sharing :)
13:50 MrCurious interesting, so rare earth magnets corrode once the nickel coating is breached
13:50 jumijoze indees
13:50 jumijoze *d
13:53 MrCurious my design was using a rare earth magnet chopped in half
13:53 MrCurious now i wonder how long until it degrades
13:57 MrCurious bingo!
13:57 MrCurious http://www.rare-earth-magnets.com/p-8-nsn0570.aspx
14:08 haryv Anyone familliary with known commerical robotic lawn mowers and have not been satisfied with the performance, getting stuck, and so on?
14:14 theBear nup, but rue got a homemade one that seems to work nicely
14:18 MrCurious the secret of how managers reproduce... http://www.askamanager.org/2008/07/how-to-mentor-someone.html
14:23 haryv Anyone made a robotic lawn mower before?
14:25 haryv Has anyone seen how well the robotic lawn mowers work and if I could make a equvilent with my arduino? I owner a mig welder and could build up a frame and use a existing lawner mower head, and perhaps scotter motors
14:44 haryv MrCurios, thanks for the site
14:45 haryv Crapy managment means lost revenue and low wages
18:41 MrCurious hmmm
18:41 MrCurious http://www.stumbleupon.com/su/2aOogT
19:04 MrCurious : leap motion ship date delayed from may 15 to july 22
19:06 MrCurious going to be hard to top that news today