#robotics Logs

Nov 27 2013

00:48 Jak_o_Shadows I'd like to make this more mathematical so I can analyse it properly to get PID working
00:49 GargantuaSauce i've probably told you this about 30 times now but make sure you have a constant timestep between samples!
00:50 Jak_o_Shadows hahahahaha, it might be.
00:50 Jak_o_Shadows It's on android, not a uC. Besides, that timestep is only for ID as I understand it.
00:51 GargantuaSauce yes
00:51 Jak_o_Shadows What I want to do is grab the system dynamics, and do a proper analysis using matlab, find the root locus and such.
00:52 Jak_o_Shadows Getting the system dynamics is the bit I haven't really done.
00:53 GargantuaSauce that strikes me of limited usefulness but probably an interesting exercise
00:53 GargantuaSauce i am pretty sure the vast majority of practical pid control is done with tuning by hand though
00:53 Jak_o_Shadows well, most people just hand-tune it don't they
00:54 Jak_o_Shadows Probably cause finding a reasonable model of system dynamics is hard. Tuning it isn't so bad
00:54 GargantuaSauce yeah modelling is a losing battle
00:54 Jak_o_Shadows I'm heavily simplifying at the moment: w1 = -w2
00:55 Jak_o_Shadows I don't so much care about moving forward yet.
00:55 GargantuaSauce i believe there are techniques that involve measuring the system's impulse response but i dont know how feasible it is to produce useful control from that
00:56 Jak_o_Shadows As I understand it, that's how you tune it using maths/matlab. You feed it a step, see what comes out.
00:56 GargantuaSauce yeah ok
00:56 Jak_o_Shadows If you feed your real world model a bunch of steps, you may be able to generalize a model out of that I suppose
00:57 GargantuaSauce for one degree of freedom or maaaaaybe two
00:57 Jak_o_Shadows maybe. Can't say I particularly want to run that way.
00:58 GargantuaSauce tis definitely the path of masochism
00:58 Jak_o_Shadows Disappointingly, an upper case theta doesn't look very different.
00:59 GargantuaSauce phi is where it's at
01:00 Jak_o_Shadows Yeah, I just did Φ = ℒ(θ)
01:06 Jak_o_Shadows I can get a transfer function. Don't quite know what to use as the feedback function.
01:10 GargantuaSauce well an idealized version of the compass just gives you an angle right
01:11 Jak_o_Shadows Yeah. I'm presuming that there's no error in it.
01:11 GargantuaSauce so the usual error difference and negative feedback
01:11 Jak_o_Shadows I might just go with unity feedback, cause it goes through a negative summing junction
01:11 GargantuaSauce http://www.youtube.com/watch?v=MqQ2XMrlnj8 alternately, this kind of feedback
01:23 Jak_o_Shadows Also, I'll put in Derivative control first, as the effects of integral control arn't needed as much.
02:22 Jak_o_Shadows according to matlab here, with a quite poor approximation (probably), I want a higher gain than what i'm using
02:27 Faek sin(theta)=theta for non small theta?
09:22 rue_house trying to work out your feedback polarities?
09:22 rue_house do you have the procedure for PID tuning?
09:23 rue_house are you working this all out in s plane?
09:24 rue_house its a mechanical servo?
09:24 rue_house do you have a digital scope to capture the step responce?
09:26 Faek Talking to Jak? I can answer one of those questions - he doesn't have an oscilloscope (either kind)
09:28 rue_house ah, I been working towards an AVR based digital scope for capturing servo responces
09:28 rue_house help to know Wn
09:29 rue_house ^s
14:04 MrCurious https://launch.punchthrough.com makezine.com/2013/11/26/an-early-look-at-the-corta...
16:40 diepchess heh guys
16:41 diepchess question on speed of motors and weight
16:41 diepchess if i have 2 kilo, and a nema17 stepper motor
16:41 diepchess what speed can i start it moving it?
16:49 Tom_itx not all nema17 motors are the same
16:50 Jak_o_Shadows TBH, it really annoys me that reprap and co just say NemaX.
16:51 Tom_itx they probably know nothing else
16:51 Jak_o_Shadows Yeah. Also, you're basically screwed if you're trying to look up proper motor characteristics
16:52 Jak_o_Shadows On one page it says the extruder needs 40 N.cm, slightly below where it says the Z axis needs 13 N.cm
17:13 diepchess what is the calculatio nfor that jak?
17:14 diepchess all those claims without calculations shown always...
17:14 Jak_o_Shadows Yeah.
17:14 diepchess i want to calculate it myself for a 2 kilo setup here :)
17:14 Jak_o_Shadows I'm happy to accept empirical stuff, but at least get it consistent.
17:15 Tom_itx diepchess, alot of them use these: http://www.aliexpress.com/store/product/5pcs-NEMA17-78-Oz-in-CNC-stepper-motor-stepping-motor-1-8A/704350_554160664.html
17:16 diepchess i have right now nema17 tom
17:16 diepchess yet i moved to SBR16UU
17:16 diepchess which appears to be a lot heavier
17:16 diepchess so i have to move 2 kilo now
17:16 diepchess over 2 kilo
17:16 diepchess which is a lot to move at high speeds
17:17 diepchess how to calculate what strength motor i need then?
17:17 Tom_itx what's a SBR16UU?
17:17 diepchess say say 60 mm/s (from standstill)
17:17 Tom_itx there are too many variables to the equasion
17:17 diepchess that's http://reprapworld.com/?products_details&products_id=389&cPath=1655
17:17 Tom_itx the driver voltage will affect it alot
17:18 Jak_o_Shadows You have to fix some stuff.
17:18 Tom_itx i run my steppers at 48v
17:18 Tom_itx wish i could go higher
17:18 diepchess and this http://reprapworld.com/?products_details&products_id=426&cPath=1655
17:18 diepchess i run 'em at 12 volt
17:19 Tom_itx mmm i thought i got away from reprap when i left #reprap
17:48 Jak_o_Shadows Do you find yourself using yours a decent amount?
17:49 Tom_itx who?
17:49 Jak_o_Shadows well, you'll do :P
17:49 Tom_itx i don't have one
17:50 Jak_o_Shadows oh, are you talking about a cnc or mill or else with itx: i run my steppers at 48v
17:51 Tom_itx cnc
17:51 Jak_o_Shadows ah, k
17:51 Tom_itx with gecko drives
17:54 Jak_o_Shadows righto. A cnc would be cool, but I think i'd get more use out of a 3d printer
19:30 Jak_o_Shadows Rue. I know some of the theory behind PID, but i'm a little hazy on actually going between the S plane and the t plane. This is for a differential-drive tricycle. I have a sensor telling me the angle the tricycle is on. My preferred method would be getting a model of the system dynamics/kinematics, taking the laplace transform, and using matlab to tune it.
19:32 Jak_o_Shadows Also, I realise you're probably asleep
19:36 MrCurious include his full name in the messages a few time to wake rue_shop2 rue_house rue_bed up
19:36 Jak_o_Shadows haha
19:44 rue_house hu? what?
19:44 rue_house when? who?
19:46 Tom_itx who's on first?
19:51 Jak_o_Shadows You mentioned PID control yesterday rue
19:53 Tom_L what about it?
20:22 rue_house I did, someone is working on making a pid thing
20:22 rue_house damn, monday I was supposed to start working on the truck
20:31 Tom_L rue_house what sort of conversion is needed between RS232 and 422?
20:33 rue_house 232->ttl->422
20:33 rue_house max232 and max, uh, max...488?
20:34 Tom_itx just ttl level then?
20:34 Tom_itx so a max232 would work?
20:34 rue_house no
20:34 rue_house then ttl to 422
20:34 Tom_itx oh
20:35 Tom_itx 422 is complimentary pair right?
20:35 rue_house uh
20:35 rue_house www.digikey.ca/product-detail/en/MAX3160CAP%2B/MAX3160CAP%2B-ND/1512917
20:35 rue_house hmm
20:36 rue_house iirc, 485 and 422 are the same levels, differential, just one is bidir and one is unidir
20:40 rue_house http://datasheets.maximintegrated.com/en/ds/MAX1487-MAX491.pdf
20:40 rue_house one of those and a max232
20:41 rue_house if you ok half duplex
20:45 Tom_L i'll probably just use my rs422 board to access it
20:45 Tom_L may be simpler
20:55 rue_shop2 for like $3 I got a usb to rs485 stick
20:58 Tom_itx are cat5E cables usually marked as such?
21:00 GargantuaSauce sometimes
21:00 GargantuaSauce you can do gigabit with cat5 just fine in most cases anyway
21:03 MrCurious Good husbands take their wives shopping. The great husbands bring them home.
21:07 Tom_itx which category are you?
21:08 rue_shop2 sounds like a safe electrician, a safe electrician will cap the hot wire, a safer electrician will cap the hot and neutral, a REALLY safe electrician will cap the hot, the neutral, and the ground :)
21:09 GargantuaSauce isnt the ground usually bare copper? <_<
21:09 Tom_itx kinda silly to cap a bare wire ya think?
21:09 rue_shop2 one of our guys said that when he opened a termination box and the ground was the only capped wire
22:49 MrCurious i wish i had a tall drill press
22:50 MrCurious hand drilling end holes on aluminum planks sux
22:50 Tom_itx be inventive
22:50 MrCurious hand drill and cope
22:50 Tom_itx mount it to a table, turn the head sideways and hold the piece from the floor
22:51 MrCurious something tellls me this could use the lathe for this
22:51 MrCurious if i had a head /chuck that could hold bits
22:51 MrCurious should look into tht