Back
[00:15:43] <Wetmelon> Anyone familiar with planetary gearsets?
[00:18:26] <SpeedEvil> I prefer ones from Uranus.
[00:18:33] <DagoRed> What do you want to know about them?
[00:19:39] <Wetmelon> I guess that was a dumb question, should have just asked :P
[00:20:17] * SpeedEvil wishes he had a hobgoblin to make all his gears.
[00:20:28] <Wetmelon> I'm just trying to decide between a belt-drive or a planetary set. Need ~ 4:1 reduction, say 50kW shaft power
[00:20:46] <SpeedEvil> 50kW?
[00:21:04] * Wetmelon nods
[00:21:06] <SpeedEvil> belt drive is I would have thought going to be a lot cheaper unless you can find one from some car or something
[00:21:31] <SpeedEvil> 50kW is in the range of a chain drive for a big bike I'd have thought
[00:21:40] <Wetmelon> That's about right, yep
[00:21:52] <Wetmelon> This is for a small race car ~ 650lb + driver
[00:22:20] <Wetmelon> A lot of (most?) Electric motorcycles I've seen use belts. Not sure why
[00:22:28] <Wetmelon> Could have something to do with compliance
[00:24:34] <Wetmelon> What about weight? Belt still wins, eh?
[00:26:12] <SpeedEvil> Likely, yes.
[00:26:31] <SpeedEvil> Belt is quieter, for one
[00:28:07] <Wetmelon> Other than compactness, does a planetary really have many advantages? Quickly changing gear ratio by changing what gear is locked I suppose?
[00:34:09] <rue_shop3> yea, you can bend them into a mutispeed gearbox
[00:34:11] <rue_shop3> somehow
[00:37:48] <Wetmelon> rue_shop3: Ever seen the planetary set in a Prius? Pretty amazing setup. They used an old Ford design with an expired patent from some 30 years ago though lol.
[00:39:36] <Wetmelon> The ICE runs the planetary carrier, there's a small electric motor on the Sun gear and a bigger one on the Ring. They can play around with the relative power inputs from the ICE and the two electric motors to get effectively infinitely variable gearbox
[00:39:53] <rue_shop3> I'm sure if ford ever had a good idea they would throw it out
[00:41:57] <Wetmelon> lol
[00:43:50] <Jak_o_Shadows> Hmmm. electric cars are gettin gplaces aren't they
[00:44:00] <Jak_o_Shadows> Most new DIY ones are using nic AC motors nowadays
[00:44:28] <SpeedEvil> Or R/C stuff in parallel
[00:44:37] <rue_shop3> electric will be the way to go
[00:45:18] <Jak_o_Shadows> RC stuff in parallel?
[00:46:19] <rue_shop3> no, the cheapest is to use vacuum cleaner motors
[00:46:24] <rue_shop3> I researched it
[00:47:06] <Jak_o_Shadows> Was that including all the supporting stuff to connec them?
[00:51:03] <Wetmelon> I think SpeedEvil means that RC stuff has been improving dramatically, in parallel with electric cars
[00:51:17] <Jak_o_Shadows> ah, yeah.
[00:51:25] <Wetmelon> Maybe not. Maybe he just means that they use a bunch of RC car motors in parallel instead of an AC motor :P
[00:52:29] <SpeedEvil> http://www.hobbyking.com/hobbyking/store/__18128__Turnigy_Aerodrive_SK3_6364_245kv_Brushless_Outrunner_Motor.html - for example
[00:52:31] <SpeedEvil> and yes
[00:52:35] <SpeedEvil> that's an option
[00:53:12] <Wetmelon> I'm a big fan of the Rotomax stuff. I really wanted to use them for our car but they just don't have the torque
[00:53:17] <Wetmelon> (or the power, really)
[00:53:56] <SpeedEvil> http://www.hobbyking.com/hobbyking/store/__25410__Turnigy_RotoMax_80cc_Size_Brushless_Outrunner_Motor_195kv_.html - I considered similar to this.
[00:54:08] <SpeedEvil> Mount three inside each hub, with a planetary gear going directly to the wheel
[00:54:27] <Jak_o_Shadows> What power do you generally need for a car?
[00:54:30] <Jak_o_Shadows> 20+ kW?
[00:54:42] <Wetmelon> This car is limited to 80kW (110hp)
[00:56:00] <rue_shop3> electric HP is WAY more than gas HP
[00:56:02] <Wetmelon> Of course you're traction limited at the low end, so if you're not racing / needing to go 150mph you can get away with less.
[00:56:12] <Wetmelon> :O
[00:56:15] <rue_shop3> casue on a gas engine, you almost never get the rated HP
[00:56:16] * Wetmelon laughs
[00:57:02] <Wetmelon> My professor taught an electric vehicles seminar... and he said something along the lines of "You'll hear people say that 'Electric horsepower is more (or better) than ICE horsepower"
[00:57:14] <Wetmelon> And ofc as engineers we look at each other like "wat?"
[00:57:41] <rue_shop3> its non-engineers perception
[00:57:41] <Wetmelon> But you're effectively right, you don't need the same HP rating, by a significant margin
[00:57:58] <Wetmelon> (under *MOST* driving regimes)
[01:02:11] <Wetmelon> I pulled out my notes. He had it labeled as "flaky rules of thumb" -> "Not so flakey after all!"
[01:05:16] <rue_shop3> have you ever seen the power curve of a gas engine?
[01:05:43] <Wetmelon> Yep, have one right here in front of me :)
[01:05:55] <rue_shop3> its awefull
[01:06:24] <rue_shop3> oh damn, I spent 2 hours making a enclosure for a power supply, and the power supply dosn't fit in it
[01:06:34] <rue_shop3> that would have been a good check
[01:06:59] <Wetmelon> Yeah, they gotta breathe to generate torque (and power). Electric motors are almost the perfect driving power curve. Flat torque curve until they hit the power limit, and then huge torque rise.
[01:07:05] <Wetmelon> D: Oh no!
[01:07:16] * rue_shop3 holds his hands on a hammer for a min and thinks...
[01:07:41] <rue_shop3> I need 12V at 2A
[01:07:49] <rue_shop3> 24w
[01:07:53] <rue_shop3> thats not big
[01:08:07] <Wetmelon> ... What are you planning on whacking? Heatsink?
[01:09:34] <rue_shop3> the case of the power uspply I was gonna put in my case
[01:25:44] * rue_shop3 smashes it open
[01:29:10] <SpeedEvil> Wetmelon: superchargers are a lot flatter
[02:46:53] <rue_house> there is a snag to using dc-dc power adapters for running led lighting
[02:47:00] <rue_house> they have a 2-3 sec startup delay
[02:48:20] <rue_house> I used a laptop supply for my desk light and have the same issue
[02:51:39] <SpeedEvil> varies
[02:51:46] <SpeedEvil> many have no soft start at all
[03:02:50] <Jak_o_Shadows> My (desk) LED lights use an "Asian Power Devices" (presumbly) dc-dc wallwart. Since I use one of the in-line LED strip controllers, I don't have a startup
[03:03:14] <Jak_o_Shadows> But it's not suitable for proper lighting
[03:42:22] <rue_house> I'v seen some of those chinese led drivers that aren't even isolated
[03:42:26] <rue_house> SCARRY
[04:09:18] <SpeedEvil> Isolation doesn't matter if it's in an enclosed housing
[04:10:37] <Jak_o_Shadows> Well, this was a router wallwart possibly? dunno. Those 12V chinese LED drivers are tempting, but they are a bit too dodgy for me to have it sitting on my desk all boring like
[04:21:10] <Jak_o_Shadows> But I do need a desktop power supply
[04:21:49] <Jak_o_Shadows> I'm down to using a 14V, 400mA wallwart with a $2 dc-dc adapter (1A limit). On the other hand, it is a transformer, not a DC-DC adapter
[06:06:14] <SpeedEvil> http://www.nasa.gov/multimedia/nasatv/index.html
[06:06:23] <SpeedEvil> Arm about to grab the ISS resupply craft
[06:10:06] <Jak_o_Shadows> thanks
[06:17:55] <SpeedEvil> And got it
[06:21:36] <Jak_o_Shadows> woo
[08:40:48] <veverak> https://www.parallax.com/product/27808
[08:40:51] <veverak> sweet
[08:52:42] <veverak> deshipu:
https://gist.github.com/SquirrelCZE/03db9e8d3a9e3be50b8a | been thinking about custom controllers for robots
[08:52:50] <veverak> I think I finally got idea that could work
[08:54:58] <deshipu> veverak: arduino micro can simulate a gamepad
[08:55:03] <deshipu> veverak: could be better than keyboard
[08:55:42] <deshipu> veverak: and then use pygame on the pi to read that
[08:55:54] <deshipu> veverak: it can run fullscreen without X
[08:57:09] <deshipu> veverak: I have this thing that I wanted to use for that:
http://www.aliexpress.com/item/New-1pc-4-axis-Joystick-Potentiometer-JH-D400X-R2-R4-5K-10K-4D-with-Button-Joystick/32333754018.html
[08:57:48] <veverak> interesting
[08:59:49] <veverak> deshipu: well, gamepad would lose practical uses
[08:59:55] <veverak> I can du more magic with keyboard
[08:59:56] <veverak> ;)
[09:00:23] <veverak> a tons of tons things can be controlled with keyboard/shortcuts
[09:00:37] <veverak> both I can simulate with arcuno simulating keyboard and harware connected to it
[09:00:43] <veverak> not so much things have gamepad interface
[09:00:57] <veverak> deshipu: and given ROS architecture
[09:01:09] <veverak> you end up using multiple apps anyway
[09:01:10] <deshipu> veverak: you can have both
[09:01:26] <veverak> liek for example dynamic-reconfigure which have nice things etc...
[09:01:32] <deshipu> veverak: it can simulate a composite device
[09:01:32] <veverak> deshipu: that sounds good :)
[09:01:36] <veverak> nice
[09:01:44] <veverak> yeah, than gamepad+keyboard propably makes more sense
[09:02:14] <deshipu> that joystick also has rotation
[09:02:19] <deshipu> as in, you can twist the handle
[09:03:58] <veverak> yeah, I see
[09:04:00] <veverak> nice
[09:04:03] <veverak> :D
[09:24:00] <veverak> deshipu: hmm, I always used controllers in a way that I control joystick only with a thumb
[09:24:06] <veverak> leaving other fingers for buttons mostly
[09:24:15] <veverak> for this 4 axis you can't use just thumb afiak...
[09:24:33] <veverak> question is which solution will be more comfortable, but I suppose only way to figure that out is to try it :D
[09:24:47] <veverak> anyway
[09:24:56] <veverak> deshipu: any tips for "arduino as keyboard+gamepad" solutions?
[09:25:06] <veverak> aaand, detach for a while
[09:30:28] <deshipu> veverak:
https://github.com/NicoHood/HID
[09:31:44] <deshipu> for thumbs I have these:
http://www.aliexpress.com/item/New-Arrive-Left-Right-3D-Button-Analogue-Analog-Joystick-Button-Control-Stick-Repair-Parts-for-PSV/32483090250.html
[10:36:35] <veverak> deshipu: yeah, that I am afraid may be not good enough
[10:36:38] <veverak> but, that needs testing
[10:36:53] <veverak> anyway, I have RC controllers joysticks as backup :)
[12:11:32] <goppo> hi
[12:11:55] <goppo> i'm trying to specify the size of a loose running hole for a 2mm rod
[12:12:06] <goppo> AFAIK, 2.05mm should do. but what should the tolerance be?
[12:12:19] <Wetmelon> See machinery's Handbook
[12:12:21] <goppo> +/- 0.2mm would be too much - right?
[12:22:18] <goppo> but for a 3mm H9/d9 hole, should the shaft be 3m D9/h9?
[12:22:27] <goppo> *3mm
[15:13:40] <robopal> what are you building goppo? got a photo or a drawing?
[17:45:47] <yeoj> does anyone have feedback on best practical solutions now for centimeter-level accuracy for positioning indoors?
[17:46:14] <yeoj> and in a perfect world, would love to learn about what might be on the horizon that could change my potential design
[17:49:46] <deshipu> yeoj: for centimeter accuracy, you pretty much don't have a choice -- use a camera on the ceiling, and markers
[17:50:44] <yeoj> what about millimeter accuracy?
[17:51:28] <yeoj> and what if i have the option to setup a couple beacons, but i'm concerned that 2.4 is already too congested and easy to create interference
[17:51:33] <deshipu> same, only better camera :)
[17:52:16] <deshipu> what are you trying to do?
[17:52:19] <yeoj> www.pozyx.io - what about UWB and something like this?
[17:52:24] <yeoj> is this reliable?
[17:55:14] <deshipu> what are you trying to do?
[17:55:27] <yeoj> highly accurate positioning indoors
[17:55:35] <yeoj> and/or outside, actually
[17:55:46] <yeoj> gps is not sufficiently accurate for my needs
[17:56:09] <deshipu> what are your needs?
[17:56:31] <deshipu> why do you need to know your accurate position?
[17:57:00] <yeoj> if my drone veers off path i'm in trouble
[17:57:34] <deshipu> by a millimeter?
[17:57:38] <ace4016> are your paths soft paths or are you more looking not to bump into things?
[17:57:42] <deshipu> then I have some bad news for you
[17:58:10] <theBear> deshipu, his path is too small ?
[17:58:11] <yeoj> ace4016: both
[17:58:14] <theBear> maybe the whole house ?
[17:58:58] <ace4016> is 10 cm error too much for you? :P
[17:59:19] <yeoj> yeah, 10 cm makes a difference
[17:59:59] <deshipu> well, scratch that idea and think about something that doesn't have such requirements, because this would be prohibitely expensive to do
[18:00:30] <yeoj> what about www.pozyx.io and uwb? are you familiar?
[18:00:47] <deshipu> yeoj: they won't work outdoors
[18:00:56] <ace4016> if you're looking at an indoor GPS, you'll have a hard time. if you're just looking for something that can get you in the ballpark, then use local sensors to avoid bumping into things, and maybe an INS/INU to help with errors, you'll be alright
[18:01:04] <yeoj> deshipu: why not outdoors?
[18:01:10] <deshipu> too small range
[18:01:44] <yeoj> pozyx only needs to receive, right? i can crank up output power on beacon pretty easily
[18:02:10] <deshipu> you can't, legally
[18:02:32] <deshipu> also, you lose accuracy with distance
[18:02:50] <ace4016> why not take a multi-layered approach?
[18:03:04] <deshipu> what you are describing would probably require laser tracking with some very precise instruments
[18:03:24] <yeoj> you're right, those comments all make sense
[18:04:08] <yeoj> is there anything similar to LIDAR Lite that uses an actual laser? i understood that these use LED but are most cost effective
[18:04:13] <deshipu> also, considering wind and such, your drone will probably not be able to stick to that patch anyways
[18:04:16] <yeoj> http://pulsedlight3d.com/
[18:04:45] <yeoj> ace4016: i assume you just mean layer lidar+gps+uwb+camera?
[18:05:02] <theBear> led lasers are basically the same as lasers, for the purposes of lidar kinda stuff, most purposes given appropriate optics
[18:05:11] <yeoj> makes sense, i'm just lazy and was wishfully thinking there was a 'simple' solution ;)
[18:06:03] <deshipu> an external camera tracking a marker on your drone (like a LED blinking in a special pattern) is probably the easiest and cheapest solution
[18:06:33] <deshipu> well, three LEDs, so that you can also guess the distance
[18:07:08] <theBear> oh, forgot context, yeah, even invisible leds work well, and ultrasonic is a traditional choice... you can do delay and phase stuff radar style EASILY at the speeds sound moves in air
[18:07:16] <deshipu> then the precision is a function of distance, camera resolution and the motion of air in between
[18:07:47] <theBear> laser/light nt so much (not talking regarding what deshipu just said btw, that's different, the pattern defines how "fast" you gotta be able to sense and process)
[18:08:00] <deshipu> theBear: doesn't ultrasound vary a lot depending on humidity and temperature?
[18:08:36] <deshipu> not to mention wind
[18:08:41] <yeoj> awesome. thanks for the ideas and feedback, really appreciate it.
[18:08:52] <yeoj> now... what if i wanted to put this drone under water? ;)
[18:08:54] <ace4016> yeoj, yea. multiple GPS and GPS-like positioning (UWB) and multiple onboard outward looking positional systems (lidar, sonar, etc.)
[18:09:40] <ace4016> underwater you'll be limited to LF/VLF waves if you want EM, or sound based systems (sonar and the like)
[18:09:49] <deshipu> yeoj: then I guess the best positioning system would be a very long stick
[18:09:54] <ace4016> light has issues...
[18:10:07] <ace4016> as does any higher frequency EM wave
[18:10:08] <ace4016> in water
[18:10:21] <deshipu> long waves have poor precision
[18:10:53] <ace4016> aye
[18:11:08] <ace4016> so...sound/pressure waves are about it underwater afaik
[18:11:29] <deshipu> but that will vary depending on the temperature and pressure too
[18:11:41] <deshipu> definitely not going to get centimeter precision
[18:12:00] <deshipu> not at long range
[18:12:26] <ace4016> you're going to have to really question why you need cm precision at that point
[18:14:44] <yeoj> i'm just trying to brainstorm really
[18:14:51] <yeoj> this has been super helpful
[18:16:29] <deshipu> it seems to be such a simple thing, naively, that you'd assume it's a solved problem :)
[20:19:15] <deshipu> veverak:
http://augmentedrobotics.com/assets/slikgl.html
[20:28:14] <deshipu> veverak: that uses
http://sci-hub.io/10.1016/j.gmod.2011.05.003#
[20:40:19] <veverak> interesting
[20:40:22] <veverak> :0
[23:55:25] <rue_shop3> hahaha I ahve a smart dummy load :)