#robotics Logs

Mar 07 2014

05:36 anonnumberanon I'm trying to fix this power switch but I don't undersatnd the connections of it.
05:37 Jak_o_Shadows I am alive.
05:39 anonnumberanon http://i.imgur.com/SI5crXN.jpg
05:40 Jak_o_Shadows I would've guessed a bog standard momentary button with a couple more joints purely for mechanical support
05:49 anonnumberanon It's like 5mm big.
05:49 anonnumberanon The power switch for a tablet.
05:53 Jak_o_Shadows are the front two pins connected to any tracks?>
05:54 anonnumberanon I don't believe so.
05:54 Jak_o_Shadows I would say that it's just a normally open momentary button
05:55 anonnumberanon See now the power light is on, but the switch does not turn the screen on and off.
05:55 anonnumberanon Nor does it power down the tablet if you push for a long time.
05:55 Jak_o_Shadows so it's buggered basically.
05:56 anonnumberanon The way I have it setup is I believe I shorted the switch. (At the small white rectangle.)
05:56 anonnumberanon Buggered? What does that mean?
05:56 Jak_o_Shadows haha. Um, in this sense it would be broken.
05:57 anonnumberanon Well it was ripped off the board, so three solders broke and the fourth resisted until the brown plate from the board itself got ripped.
05:58 anonnumberanon wow i actually found it in my soldering iron's wet pad
05:58 Jak_o_Shadows hopefully one of the non-connected ones was ripped?
06:00 anonnumberanon Nope, it was one of the two in the back, connected.
06:00 Jak_o_Shadows oooh. That's less fun to fix
06:02 anonnumberanon There it is. A tiny brown rectangle.
06:02 anonnumberanon About 1.5mm by .75mm.
06:04 anonnumberanon About 1.5mm by .75mm.
06:04 Jak_o_Shadows I don't know how you'd fix that if you've taken up a bit of the tracks.
07:26 anonnumberanon The switch works at least, I'm just gonna have to do a bit of surgery..
09:24 rue_mohr2 unless the iron has been calibrated, dial it up just as much as you need to melt the solder of what your working on, too cold and you will overheat the area trying to get the solder to melt, too hot and you will burn/melt stuff, burning the copper off the circuit board isn't fun
11:05 MrCurious http://www.theverge.com/2014/3/7/5477768/gabor-balogh-circular-smartwatch-concept
11:05 MrCurious i am not a watch person, but that watch could turn me...
11:13 aemquo Nice
19:59 MrCurious totally cool... taser drones....
21:01 rue_more so I ahve to designa smps inductor
21:01 rue_more 100khz?
21:01 rue_more hm
21:01 rue_more 300Khz?
21:02 rue_more at 95%, how many seconds is that
21:02 GargantuaSauce what fets are you using
21:02 rue_more 3us
21:02 rue_more ouch
21:03 GargantuaSauce yeah that's a lot of gate current
21:03 rue_more mayeb irf1404
21:03 rue_more yea
21:03 rue_more about 3nf iirc
21:03 rue_more how about 150Khz
21:03 rue_more 6.666666666666666666666666666666666666666666666666666666666666666666666666666667ns
21:04 rue_more thats 100%
21:05 rue_more 6.3ns for 95%
21:05 e_more reth
21:06 rue_more zippo:/files/temp# ohm -i 5 -p 1
21:06 rue_more Wattage is: 1.000000
21:06 rue_more Current is: 5.000000
21:06 rue_more Voltage is: 0.200000
21:06 rue_more Resistance is : 0.040000
21:06 GargantuaSauce just use google jeez
21:06 rue_more isn't a z44 .34mr?
21:06 rue_more tell me what a person would type into google to find the resistance for 5A with 1w dissipation?
21:07 rue_more 28mR
21:08 rue_more 2nf
21:11 rue_more this chart says I can drive the gate with just 5V for 5A
21:12 GargantuaSauce and what is Vds at that Vgs/Id
21:12 rue_more just working ont hat
21:13 GargantuaSauce probably max dissipation of the package, potentially with heatsink
21:13 rue_more .25ish
21:13 rue_more thats the 4.5V curve
21:13 GargantuaSauce heh i still have a DPAK-shaped indentation in my finger
21:13 rue_more oh next is..
21:14 rue_more .2V
21:14 rue_more thats 5A, so thats 1W
21:14 rue_more woudl save me gate current
21:15 rue_more it IS a 36A fet
21:15 rue_more its only a 5A supply I'm making
21:15 GargantuaSauce also 1/150 kHz is microseconds, not nanoseconds
21:16 GargantuaSauce sooo gate current is actually pretty tiny
21:16 rue_more I just want unreasoanble efficiency
21:16 rue_more oh, I slipped a place? heh
21:16 rue_more hahah i want from milli to nano
21:16 rue_more hahaha
21:16 rue_more thanks
21:16 GargantuaSauce https://www.google.ca/search?q=3+nF+*+12+V+*+150+kHz+%2F+0.95&oq=3+nF+*+12+V+*+150+kHz+%2F+0.95&aqs=chrome..69i57.101391j0j1&sourceid=chrome&ie=UTF-8#q=3+nF+*+12+V+*+150+kHz+%2F+0.05
21:16 GargantuaSauce yeah much more approachable
21:17 rue_more so 30Khz wouldn't be more than 600mA
21:17 rue_more 300
21:17 rue_more .216
21:18 rue_more ? did you parallel the input and output capacitance to get 3nf?
21:18 GargantuaSauce [22:51:46] <rue_more> about 3nf iirc
21:19 rue_more hey,! its 2!
21:19 rue_more https://www.google.ca/search?q=2+nF+*+5+V+*+300+kHz+%2F+0.05&oq=2+nF+*+5+V+*+300+kHz+%2F+0.05&gs_l=serp.3...23410.23410.0.26156.1.1.0.0.0.0.66.66.1.1.0....0...1c.1.37.serp..1.0.0.qvpW106um48
21:19 rue_more sweeeet
21:20 rue_more I can drive that directly with an LM386 then
21:20 rue_more whats the unity gain freq?
21:21 rue_more hmm 300khz bandwidth
21:21 rue_more thats -3db?
21:22 rue_more up to ... oh, thats hz
21:23 rue_more 10db @ 1Mhz
21:25 rue_more hahah its a 1W amp, but the harmonic distortion goes off scale after .2w
21:25 GargantuaSauce yeah well you want it cranking out squares anyway
21:26 rue_more I remember it had sub ns switching time with a pretty big fet
21:26 rue_more zlog,
21:27 rue_more how the hell do I search zlog...
21:27 GargantuaSauce again, the googles
21:27 GargantuaSauce site:
21:27 rue_more tom went to lots of work for google NOT to cache it
21:27 GargantuaSauce this must be recent because i've definitely googled through it
21:27 rue_more yea
21:28 rue_more about 2 months ago
21:28 GargantuaSauce welp, time for wget and grep
21:28 rue_more I better get tobbor back online too
21:32 rue_more haha I forgot hwo to write C code
21:35 rue_more mmm C program stdout piped to bash...
21:36 rue_more damnit, the logs for the rest of the month aren't there
21:40 GargantuaSauce so how are you going to power the opamp
21:40 GargantuaSauce boost converter?
21:40 rue_more bootstrap yea
21:41 rue_more tahts why its only to 95%
21:41 rue_more to lets the boostrap recharge
21:42 rue_more " LM386 has it switched on (worst time) in 4us"
21:42 rue_more <rue_house> the lm741 took 15us to switch the fet
21:43 GargantuaSauce that is a liiiittle slow
21:43 rue_more I think 4us will be ok
21:44 rue_more whats 95% of 300Khz?
21:45 rue_more oh, hmm
21:45 GargantuaSauce yeah.
21:45 rue_more hmm
21:45 rue_more k, fancy fet driver then!
21:45 rue_more I have some AOP605
21:47 rue_more cant drive the main with it, but I can drive the gate
21:48 GargantuaSauce the conclusion i came to after dicking with a more complicated BJT-driven design was to say fuck it and buy monolithic half-bridge drivers
21:48 rue_more http://eds.dyndns.org/~ircjunk/circuits/fetdrive.gif
21:49 rue_more that one is better than 1us
21:49 rue_more er 1ns
21:49 rue_more fast as my scope goes
21:50 GargantuaSauce that is a big inductor
21:51 rue_more I wonder if the value is right
21:51 rue_more I dont usually have marked inductors
21:51 GargantuaSauce well it's an LC low pass filter right
21:51 rue_more yea
21:51 GargantuaSauce you could increase C3 proportional to the decrease in L1
21:51 rue_more I sat and just messed with it, thats was what I came to
21:51 GargantuaSauce changes need to be made to make that the bootstrap cap anyway
21:51 rue_more I couldn't drive it any faster than that
21:52 GargantuaSauce and i am not sure just jamming a diode in front of the inductor will do the trick
21:52 rue_more ? no, you hang a cap off the output
21:53 rue_more I ahve some shottkey diodes
21:53 GargantuaSauce tnftp
21:53 GargantuaSauce o
21:53 rue_more I'd like to do synchronous, but I dont know what happens if the converter goes discontinious with a synchronous converter
21:53 GargantuaSauce fuck
21:54 rue_more hello kitty?
21:54 GargantuaSauce hello irc is not a terminal
21:54 rue_more aah
21:55 GargantuaSauce http://dé.su/bldc.png this is what i tried
21:57 rue_more wanna try that url again?
21:57 GargantuaSauce http://xn--d-bga.su/bldc.png
21:57 GargantuaSauce it's 2014, why does your crap not support punycode :(
21:58 rue_more its not from 2014
21:58 rue_more its the last working copy I had to upgrade to
21:59 rue_more generally looks good, what was wrong?
21:59 rue_more it'll fault at 100% duty of course
21:59 GargantuaSauce Q2 can sink current from the output to ground through Q5 when the output is supposed to be floating
22:00 GargantuaSauce which i believe your design is also susceptible to, assuming you put it on the high side the way i think
22:01 rue_more I'm trying to work that out
22:01 rue_more but I dont really ever do high side drivers, so I wouldn't know
22:02 GargantuaSauce this shit is hard
22:02 GargantuaSauce i don't like power electronic
22:02 rue_more jsut remmebr to test at low power first with a 1A fuse
22:02 GargantuaSauce digital stuff is fun, audio is super interesting but i haven't done anything there yet, but power? blegh
22:02 rue_more and DONT clip your scope leads on current sense resistors
22:02 GargantuaSauce i have an alright bench supply with current limiting anyway
22:03 rue_more mines gonna have usb control :)
22:03 rue_more and monitoring'
22:03 rue_more I dont understand your current path, I'm gonna go try this
22:03 GargantuaSauce remember BJTs are symmetrical
22:04 GargantuaSauce the beta just sucks when it's backwards
22:04 GargantuaSauce symmetrical in topology i should say
22:04 rue_more its pnp, there should be no flow from C to B
22:04 GargantuaSauce switch C and E and it makes sense
22:04 rue_more if you had a pullup/down resistor I could understand
22:05 rue_more yea, you should never use an npn for Q3
22:05 GargantuaSauce this is when Q5 is on remember
22:05 GargantuaSauce the low side works fine
22:05 GargantuaSauce and the high side switching works fine, it's just when both are off that things get fucked
22:05 rue_more its got nowhere to sink from
22:06 GargantuaSauce huh?
22:08 rue_shop3 it shoudl float ok
22:09 GargantuaSauce no it entirely makes sense that it sinks current backwards through Q2 and Q5
22:09 GargantuaSauce the output is sitting at about 24V
22:09 GargantuaSauce both mosfets off
22:09 GargantuaSauce R1 provides a path to ground
22:10 GargantuaSauce a weak one obviously, think i was using 1k
22:10 GargantuaSauce no, 100
22:12 GargantuaSauce anyway i have nice gate driver ICs now, just need to get around to figuring out pcb design and etching
22:12 GargantuaSauce think i have all the materials
22:16 GargantuaSauce in any case this design probably works fine for a dc servo
22:16 GargantuaSauce since you'll always either be driving up or down on both sides i think
22:16 GargantuaSauce my intent was to use it for a brushless motor as the filename implies, where the floating state is necessary
22:21 GargantuaSauce (and maaaaybe the 100 ohms to ground doesn't even make a difference i dunno)
22:32 rue_shop3 redled to redled as transmitter and reciever can respond as fast as 5us
22:33 GargantuaSauce how do you have the receiver configured
22:33 GargantuaSauce i think a reverse bias can make them much faster
22:37 GuShH_ rue_shop3: the screw pump concept worked fine, but there's a problem with the seals on both ends, oil seeps through.. it does pump though, even when it's horizontal.
22:37 GuShH_ cut the screw at 3mm pitch, not full depth since it's delrin and it was bending out...
22:39 GuShH_ it would require over 1000RPM if you wanted a nice constant stream of oil. I haven't coupled it to anything yet, testing it on the lathe.
22:44 GuShH_ I'm not sure how the commercial ones are sealed, I tried a couple o-rings but oil would seep through the bearing anyway. Perhaps the bearings should be fully inside
23:03 rue_shop3 hmm
23:03 rue_shop3 I forgot what I was doing, high side fet driver, right
23:05 GargantuaSauce warning: removing 'gcc-libs' from target list because it conflicts with 'gcc-libs-multilib'
23:05 GargantuaSauce this is gonna be a fun update
23:14 rue_shop3 I been making and taking aprt proects on the same breadboard in the same area of the shop bench for 2 weeks now
23:14 rue_shop3 interestingly enough, the jumpers keep dissapearing
23:14 rue_shop3 each time i do a poject there are less and less jumpers available
23:36 rue_shop3 this gate driver I'm using is a pulldown resistor with a pullup PNP, Ihad to go to 100uF to keep voltage up when switching, and my 330R load isn't hitting ground because of the bootstrap charge current
23:36 rue_shop3 hmmm
23:37 rue_shop3 interestingly enough if my max duty is 95% I could probably use a capacitor for the level converter
23:41 GargantuaSauce i am quite sure that you have to have a low-impedance drive both up and down on the gate in order to get satisfactory performance
23:42 GargantuaSauce a pulldown resistor won't cut it
23:43 GuShH_ rue_shop3: ever heard about "CMT" tools?
23:43 GuShH_ http://www.cmttool.com/products_detail/&productId=ab22b62f-68d4-4f92-be5b-ac387f23f099.html
23:43 GuShH_ of course it's just TiN coated, but they say "titanium!"
23:44 GuShH_ it's an 80 bucks set in the US... of course I'm paying twice as much, which makes me wonder if they're worth it at all
23:52 rue_shop3 yea, dunno where
23:52 rue_shop3 hmm
23:52 rue_shop3 so the problem with the bootstrap, is that for really small loads, the min drive votlage is pullup up by the bootstrap system
23:53 rue_shop3 I have a 330R load and its offset up by about 3V
23:53 rue_shop3 so...
23:54 rue_shop3 what if there were a 'output' that the bootstrap system used, that wasn't an output?
23:54 rue_shop3 or what if my power dosn't come from the bootstrap system
23:55 rue_shop3 the idea of making a little power bord with an array of isolated outputs gains more interest
23:57 rue_shop3 I wonder how long the average little toroid can take at 1V/turn