#robotics Logs

Oct 21 2015

00:06 anonnumberanon Man that got me thinking about mobility implementations. Two choices come to mind, gasoline powered hybrid with electric actuators, or just batteries and actuators. I think the choice would be clear for such project.
00:07 rue_house the only requirement it to JUST DO IT
00:07 anonnumberanon I'm gonna study robot walking in depth in the next few months.
00:07 onnumberanon shi
00:07 rue_house no, start welding tommorow
00:07 rue_house no measurements, just go for it
00:08 anonnumberanon well i got my 20" robot assembled already (using servos).
00:08 anonnumberanon It's a lot more DOF, 4 in each leg.
00:12 anonnumberanon Did you see it?
00:13 rue_house it was assembled last we saw it
00:13 rue_house did you have a video of it moving?
00:22 anonnumberanon I could but I feel this would be an underachievement. I'd feel safer going crazy with it once I feel it is safe to do so. I'm making a current sensing module right now.
00:26 rue_house what current level?
00:26 rue_house http://ruemohr.org/~ircjunk/robots/buddy_III/buddy-overload.sch
00:26 rue_house its bipolar tho
00:27 rue_house eagle
00:28 rue_house oops, I made the overloads in 2013 and I havn't fixed them yet
00:28 rue_house prolly means the batteries need a topup
00:28 rue_house gnight
00:29 anonnumberanon Only used by hobbyists for RC car so no good data exists.
00:30 anonnumberanon I must have read 1.5A somewhere, at stall torque.
00:30 anonnumberanon To each servo.
00:31 anonnumberanon Planning on multiplexing an ACS712 module.
00:31 anonnumberanon Not sure yet, still researching my options.
05:29 veverak this sucks
05:29 veverak my servos won't leave hong kong :/
05:30 veverak 10 days from acceptance and they are still near hong kong
05:43 joga maybe they ran out of power before they got out :(
06:08 veverak deshipu: got any experience with it?
06:10 veverak http://imgur.com/k5HyqyV
06:10 veverak it's translated to english from chinese
06:11 deshipu veverak: looks like they have a long string of inspections
06:12 deshipu veverak: the box must be looking suspicious ;)
06:12 veverak fu them
06:12 veverak it will propably spend 7 days in Prague also
06:12 veverak ...
06:13 deshipu with all those seals from constant reopening, most likely
06:14 deshipu my guess is that it's large and heavy enough to fit something illegal inside, so they are inspecting it
06:14 veverak nah :/
06:15 veverak second thing
06:16 veverak 4 days in shanghai, should be in plane just now
06:18 deshipu veverak: https://www.youtube.com/watch?v=P9mwELXPGbA
06:18 veverak lol :)
09:28 rue_house anonnumberanon,
09:29 rue_house ever seen my servo tutorial?
09:29 rue_house http://ruemohr.org/~ircjunk/tutorials/elex/hobbyservo/servo101.html
09:29 rue_house look for scope traces
09:30 rue_house er, you know, I should redo that for a stall
09:30 anonnumberanon "like adjusting the gain and force feedback" you got my attention
09:31 rue_house :) I can tell you more later
09:34 anonnumberanon I'll probably want to first fix their problem of "centering", which is something a lot of people complain about, and may be due to a poor quality potentiometer. I haven't even powered one yet to see what I'm looking at. Wouldn't mind actually being able to send back the exact position to the servo controller. That would be extremely useful.
09:34 rue_house did you assume that 1.5ms is ventre?
09:34 rue_house ^c
09:34 anonnumberanon 1520 us apparently is centre according to the net
09:35 rue_house wrong, it varries with ever servo
09:35 anonnumberanon Good to know.
09:35 rue_house oh man, cant speel at all today
09:35 rue_house also, the low and high limit varry with servo
09:38 rue_house the official numbers are 1-2ms
09:39 rue_house but
09:40 rue_house I find the low end is usually around .5ms and the high can be upwards of 2.5
09:40 anonnumberanon What's funny is I'll have to keep the upper body pretty stiff and play only with the legs at the beginning. Making a walking gait will be challenging. I am thinking of a way to control one or two legs at a time with one joystick per joint. If this is successful I think only then, will I be able to add a gyroscope/accelerometer, and make the arms maybe move to balance the robot when it gets pushed or hit.
09:41 rue_house your using digitial servos tho, right?
09:41 anonnumberanon (That is, potentially 4 joysticks to record one leg step).
09:41 anonnumberanon Not digital.
09:41 rue_house ah
09:41 rue_house what is the resolution of the controller?
09:42 anonnumberanon I am designing it.
09:42 rue_house ah
09:42 anonnumberanon I got the frame and the servos because the combo seemed like something that would be dependable, everything else will be my design.
09:42 rue_house I have a m328 serial controller that does 8 serovs with 24000 count resolution 100% stable
09:43 rue_house arduino m328 controller boards are $1.30 from ebay
09:43 rue_house https://www.youtube.com/watch?v=qZF9WJ1Ueew
09:44 rue_house some mechanical slop, but uses a version of the same code
09:44 anonnumberanon brb i need breakfast, will watch that vid
09:45 anonnumberanon .5ms and the high can be upwards of 2.5
09:45 rue_house http://ruemohr.org/~ircjunk/robots/arm8/p1060720.jpg
09:45 anonnumberanon wait what? lol. I guess I'm really used to the ESC's range
09:46 anonnumberanon brb
09:46 rue_house there is a shot on the controller on one of my last arms
09:47 rue_house that sucks, I dont have one up of the realtime tracking control
09:48 rue_house yes I do
09:48 rue_house no, that playback, drat
09:48 rue_house https://www.youtube.com/watch?v=3N7DNfg4p9I
10:10 anonnumberanon what stall torque on the servos from that arm?
10:10 anonnumberanon https://www.youtube.com/watch?v=qZF9WJ1Ueew
10:12 anonnumberanon It's like that robot arm that cracks a smartphone's password by trying all combinations by hitting the screen.
10:12 anonnumberanon Like yours.
10:13 anonnumberanon You must not be drawing a lot of current on that one, straight from the atmel board.. o
10:15 anonnumberanon Unless I'm missing the power supply..
18:02 veverak http://i.imgur.com/Ye2eu3f.png
18:02 veverak "probing algorithm is hardcore they said"
18:03 veverak blue plane: data from probing (generated here)
18:03 veverak line: gcode moves
18:04 [cube] that's cool veverak
18:05 veverak maybe point is that making plate compesation in printers firmware seems bullshity to me
18:05 ace4016 ?
18:06 veverak well, we usually control our printers with something that got much more computing power
18:06 veverak and algorithm to compensate uneven bed if using big matrix of readings are complex
18:06 veverak so, why to bother atmega/weak arm with it when there is no point with it?
18:06 veverak it also means you have to optimize something complex quite a bid
18:06 veverak *bit
18:07 veverak 3d printer should have abillity to "probe on place" and let the thing that controls it modify the gcode
18:07 veverak seems much more effecient to me :)
18:12 Snert_ what is this concept of "micro stepping" a stepper motor? I'm used to 400 steps per revolution, 1.8degrees.
18:12 Snert_ can a stepper motor be 1/2way inbetween 2 steps?
18:12 veverak yep
18:12 Snert_ huh.
18:12 veverak if both pairs of coils are on 50%
18:12 veverak it 'somewhere' in middle between it
18:13 orlock 400*1.8=720
18:13 veverak *it is
18:14 Snert_ more studing to do I guess :)
18:14 Snert_ studying
18:15 ace4016 depending on the motor, you can get more than just the very center iirc (SRMs and the like i believe might have a bit more resolution, but require a lot more feedback and such...though i guess they're technically not steppers per se...they can just act as one)
18:15 veverak well
18:15 veverak with feedback job get's easier :)
18:16 veverak btw: anybody any tips for gearbox for nema17
18:16 veverak that got axis by 90 degrees ?
18:16 Snert_ ok so there may be a varying level of precision that goes along with microstepping.
18:16 veverak Snert_: of course the bigger the microstepping is, the less precise it is
18:17 veverak but it gives you really smooth moves
18:17 veverak so it's somehow worth it anyway :)
18:18 Snert_ in 3d printers I can see it.
18:18 Snert_ in a robotic arm maybe not as much?
18:40 veverak depends
18:41 veverak Snert_: it will make your moves definetly smoother
18:43 SpeedEvil Inertia helps too
18:45 Snert_ I'll keep those things in mind as I get closer.
20:22 rue_house ok, I have made a mcs85 library from my mcs80 library
20:22 rue_house now I can talk to mcs85 chips