#robotics Logs

Jan 10 2013

02:03 Ademan do servos need a flyback diode?
02:04 RifRaf not that i know of
02:12 Triffid_Hunter Ademan: they already have them inside
02:27 Ademan Triffid_Hunter: ah cool
02:37 Triffid_Hunter Ademan: hobby servo is a full h bridge plus pulse extender and monastable oscillator electronically.. when it sees the rising edge of the signal pulse it starts its own osc based on the feedback pot, then takes the difference of the two pulses- ie if the signal is high for longer then it sets the h-bridge to forwards via the pulse extender
02:37 Triffid_Hunter fairly crude but certainly gets the job done. "digital" servo means they put a microcontroller in to sort it instead of the analog processing
02:38 Triffid_Hunter those crazy dynamixel servos that the darwin-op uses has an ARM cortex-M0 in it running some pretty fine grained math.. bit steep at ~$280/ea though
04:16 MaoIsAcat Hi everybody
04:26 MaoIsAcat Do you know if I can use this hub (http://www.robotshop.com/eu/productinfo.aspx?pc=RB-Lyn-99&lang=en-US) with these tires (http://www.robotshop.com/eu/productinfo.aspx?pc=RB-Lyn-16&lang=en-US) ?
04:31 MaoIsAcat or am i in a erroneus channel?
04:43 RifRaf For Lynxmotion neoprene tires NFT-01 through NFT-07
04:43 RifRaf is the tire one of them?
04:45 MaoIsAcat mmmmm nop
04:52 MaoIsAcat but the tires says "Any universal hub" and this hub is a universal hub
05:07 RifRaf they could make it easy and give a hub outside dimension but they don't so not sure
09:06 rue_house you can infer from the grub screw
09:07 rue_house interesting how many people from that competition are comming by here eh?
09:08 rue_house wouldn't it be funny if more than one of them was here long enough to see one of the others
09:08 rue_house hah! if they knew about the log they could snoop on what their compeditors are using
09:23 theBear hehe
09:23 theBear competitors <grin>
14:46 Ademan Triffid_Hunter: wow 280 is definitely steep...
14:47 Ademan For the time being I don't care, but most of the 55g servos I've looked at have been ~1s/60deg which seems pretty slow, but that seems to be the lower bound of what I've found...
14:58 Mickkelo hello
15:26 Ademan hi Mickkelo
21:26 Ademan I have an unlabeled 9g servo I inherited, is it possible for me to measure its "open" resistance?
21:26 Ademan I guess I mean its lowest resistance (across the vcc/gnd)
23:01 Triffid_Hunter Ademan: resistance doesn't apply to dynamic circuits like a servo
23:02 Triffid_Hunter Ademan: I mean sure, you can measure voltage and current at any given instant and calculate one, but it's not useful in any way
23:03 Triffid_Hunter Ademan: I think what you want to measure is stall current. command it to go somewhere, then force it to be somewhere else so it struggles, measure the current. that's gonna be pretty close to worst case. don't do it for long or the motor will burn
23:08 Ademan Triffid_Hunter: I want to make sure I don't fry my MSP430 (and more importantly, whatever computer it's pulling USB power from) so I want to make sure my circuit supplying power to the servo doesn't allow it to draw too much current
23:09 Ademan I have no idea if the servo will even operate with 3v input but I'm gonna try, as I don't have a 5v supply currently
23:11 Triffid_Hunter Ademan: lots will work fine with 3v signal
23:12 Triffid_Hunter Ademan: but you MUST NOT connect the servo power to the same source as your micro
23:12 Triffid_Hunter Ademan: for one, servos can take amperes of current depending on their size and what they're doing, and they're also quite noisy electrically. you'd be best off getting a ubec or similar from hobbyking, or grabbing a switchmode regulator or phone charger or something for it
23:14 Triffid_Hunter Ademan: on my R/C car which I sometimes mount arduinos on, I have the servos connected to a ubec (simply a switchmode buck converter) and the arduino connects direct to the battery and uses its onboard regulator
23:15 Triffid_Hunter since the batteries have astonishingly low ESR, this works great. I've tried it with AAs but the resistance of the contacts is too high and that arrangement causes brown-outs, I had to use a boostbuck set to 8v in front of the arduino
23:17 RifRaf i just dug out my arduino stuff and various others motors, feels good to be making stuff again
23:18 RifRaf just programmed it and got a led blinking :)
23:18 Ademan Triffid_Hunter: Hrm really? I mean it makes sense but also having a shared supply seems convenient, is there a way to sufficiently isolate the two circuits from the same supply?
23:19 RifRaf Ademan trust me you want separate supplies as he says
23:20 Ademan Triffid_Hunter: I'm looking at a supply like this: http://lars.roland.bz/images/power-supply.png where the regulated 3v is for the MSP and the rest is for servos
23:21 Triffid_Hunter Ademan: yeah, each one gets its own voltage regulator, and you add capacitors before the vreg until the ESR is below 100mR or so
23:21 Triffid_Hunter Ademan: if you
23:21 Triffid_Hunter r supply is 5-6v that might work ok but it needs more capacitors
23:23 Triffid_Hunter Ademan: you also want maybe a 1-10r wirewound resistor between the battery and the 3v3 reg with capacitors on both sides- a pi filter if you will
23:23 Triffid_Hunter Ademan: what sort of batteries will you use? if they're R/C grade lipos you can skip some of the filtering because their ESR is astonishingly low
23:24 Triffid_Hunter also you can run your logic from the balance connector and everything else from the main connector, that way they branch off right next to the battery
23:25 Ademan Triffid_Hunter: totally up in the air, atm I'm just exploring my launchpad and using USB power
23:26 Triffid_Hunter Ademan: heh don't even think of running servos from usb
23:26 Ademan haha ok
23:26 Triffid_Hunter Ademan: anything with a DC brushed motor in it should be treated exactly the same as a minature tesla coil in terms of noise and EMR management
23:26 Triffid_Hunter because guess what
23:27 Triffid_Hunter coil + magnetics + imperfect brushes = you guessed it
23:28 Triffid_Hunter and I sure as hell would never connect any sort of tesla coil to my computer ;)
23:28 Ademan In the blog post that power supply circuit it says C1 should be 1uF and C2 0.1uF, in a youtube video describing a regulated power supply the narrator specifically says the 1uF should be an electrolytic cap, any reason for that?
23:28 Triffid_Hunter Ademan: yeah, linear regs will oscillate if the ESR on the output is too low
23:28 Triffid_Hunter electrolytics have a pretty high ESR compared to ceramics
23:29 Triffid_Hunter if you put an 0R05 in series with the output before the capacitor you could use a ceramic
23:29 Triffid_Hunter but it's the input side that you should put big ceramics on, I'd put a 10u multilayer along with a few 0.1 at the input side to help keep motor noise out
23:30 Triffid_Hunter Ademan: remember, all components are imperfect. the question isn't where to get an ideal capacitor, but rather, which capacitor's imperfections best suit the application
23:30 Ademan Also: I see how to place caps to filter noise but I don't really understand the physics or the theory...
23:31 Triffid_Hunter Ademan: basically caps short out AC and block DC, so when you connect them between power and ground, they short out any variance in the voltage without being an actual dead short
23:32 Triffid_Hunter Ademan: just as a demonstration, you can generally pick up DC brushed motors with an AM radio
23:32 Triffid_Hunter but if you put an 0.1u ceramic across the radio's antenna input, the signal will drop profoundly
23:33 Triffid_Hunter Ademan: in terms of general decoupling, even bits of wire or pcb traces aren't perfect. at radio frequencies, they look like strings of weak inductors with tiny capacitors between them. if you suddenly add a load at one end, it takes some time for the current to build up from the other end and top up the voltage.
23:34 Triffid_Hunter Ademan: CMOS ICs may take mere milliamps when running, but they take it in short sharp bursts of higher currents. this means that without decoupling capacitors, the IC's own normal operation can cause it to brown out simply because the current can't get across the PCB fast enough
23:35 Ademan huh
23:35 Triffid_Hunter so the only way to handle this is to make sure that there are small reservoirs all over the board, and especially close to anything taking sharp slurps of current
23:35 Triffid_Hunter municipal water supplies use reservoirs and towers for exactly the same purpose
23:39 Ademan interesting
23:39 Triffid_Hunter imagine if everyone in a certain suburb all turned their taps on at the same time.. the main pressurising pump is maybe 50km away. without local tanks to maintain the pressure, it drops rapidly as the line empties of water, and then comes back up again as the pressure differential wave travels to the main pumps and back
23:40 Triffid_Hunter a local tank maintains the pressure while the wave travels to the main pumps and back, and the suburb hardly notices the much smaller pressure variation
23:40 Triffid_Hunter decoupling capacitors perform exactly the same function but in an electrical sense
23:41 Ademan I think I get it
23:41 Triffid_Hunter Ademan: cool. now imagine a DC motor as a large waterpark next to an industrial district
23:42 Triffid_Hunter and you can imagine why we are adamant that it must be kept as far from sensitive logic as possible
23:42 Triffid_Hunter both physically and electrically
23:43 Ademan Makes sense, so how can I have logic from my microcontroller drive a servo on a different power supply?
23:43 Triffid_Hunter grounds must be connected together
23:45 Ademan So I have USB power for the MSP430, and a 6VDC wall wart, is there any way to safely connect their grounds? that strikes me as an awful idea honestly
23:45 fRaf thinks Triffid_Hunter explains things very
23:46 Triffid_Hunter Ademan: it's necessary unless you want to optoisolate.. generally it works fine if wall wart and computer are on the same power point
23:46 orlok Ademan: What can the wallwart put out?
23:46 Triffid_Hunter thanks RifRaf :)
23:46 Ademan orlok: 1200mA
23:46 orlok Ademan: servo's can chew a fair amount of current at stall
23:46 orlok check the specs
23:46 orlok should be OK for a few
23:47 Triffid_Hunter 5v 2A switchers are common these days, for powered usb hubs and phone chargers and things like that
23:47 Ademan orlok: unfortunately I don't have specs, this is an unmarked hand-me-down for me to play with and break
23:47 Triffid_Hunter 6v 1.2A wart should be quite happy driving servos
23:47 orlok Ademan: ahh, enjoy then!
23:48 Triffid_Hunter Ademan: if it's actually a regular transformer and dc rectifier then it's perfectly safe to connect its ground to your usb
23:48 Triffid_Hunter Ademan: if it's a switcher, it's safer to connect it all up before plugging it in because switchers need capacitors from high side to low side to reduce EMR and that imposes up to twice the mains voltage on the output terminals
23:49 Triffid_Hunter it's not dangerous to us- the current is extremely low, but CMOS chips don't care how low the current is, if you drop 100v into an input it may end up having a bad day
23:49 Triffid_Hunter they do have protection on all their pins and it usually works...
23:49 Ademan Triffid_Hunter: You're recommending connecting the wart before plugging it into the circuit?
23:49 Triffid_Hunter Ademan: if it's a switcher, yes
23:50 Triffid_Hunter usually you can tell from the size and weight
23:50 Ademan Well I don't have a proper adapter for the barrel connector so that's a project for tomorrow heh
23:50 Ademan It's rather heavy so probably a regular transformer I assume
23:50 Triffid_Hunter also, a regular dc supply will put out a volt or two extra with no load, switchers are more tightly regulated
23:51 Triffid_Hunter Ademan: I usually cut them off and put crimp headers on
23:51 Ademan Oh so I should probably put a voltage regulator on it?
23:52 Triffid_Hunter Ademan: servos are usually fine with 7.2v which is what I'd expect it to put out unloaded
23:58 Ademan So when connecting grounds, are there any special precautions? or literally just connect the two?
23:59 Ademan wait, "if it's a switcher[...]that imposes up to twice the mains voltage on the output terminals" can you explain the 2x voltage thing?