#robotics Logs

Mar 14 2013

00:03 Rif and was gonna stop at jaycar
01:59 Guest8892 theBear: being a drunk just got more accessible http://www.ebay.com/itm/Whiskey-Pocket-Hip-Flask-Wine-Liquor-Alcohol-Wedding-Party-Bar-Drink-Bottle-Cap-/300770344369?pt=LH_DefaultDomain_15&var=&hash=item46074f3db1
02:09 RifRaf never seen hip flasks? i cannot see anything new about it
02:17 Guest8892 RifRaf: just the low price
02:17 Guest8892 4 bucks and it comes from china
02:18 Guest8892 how come if I want something from downtown it costs me 40 USD on shipping alone?
02:18 Guest8892 and the courier is just a dude on a bike
02:18 Guest8892 ybr125 most likely.
02:19 RifRaf ah ok then, yeah is amazing how they ship everything to us for free these days
02:21 RifRaf gush i have a measured vref os 50mV, i can measure 7mV on adc pin, have set code to use external ref, but all adc outputs are 1023 no matter what i plug in or out
02:22 RifRaf will try 200mV vref
02:22 RifRaf is 1023 without any voltage to the vref pin
02:23 RifRaf so that won't help, must have something wrong as usual
02:23 Guest8892 could be all a software issue
02:23 Guest8892 try to use proven code
02:24 RifRaf like arduino?
02:24 est8892 bl
02:24 Guest8892 like, someone else's example to test the adc
02:24 Guest8892 arduino lib is proven to be bloated and not entirely useful for real world applications.
02:24 RifRaf have found nothing for small external vref, they all just use 5V generally
02:27 RifRaf is maybe the prescaler, will check datasheet again
02:30 RifRaf if i give the vref a higher voltage gives out other values
02:30 RifRaf lets see now
02:35 RifRaf going back to internal 1.1
02:39 RifRaf and getting proper results
02:49 rue_bed you can use 5V with the amp circuit I'll get ya
02:50 rue_bed RifRaf, it helps if you can find a 79L05, -5V regulator, a 78L12 would help too
02:50 rue_bed maybe less so
03:03 RifRaf coulda said so earlier, now i have to try dig one out
03:10 RifRaf rue_bed do you know why i could not use a simple diode to offset ground by a bit?
03:10 RifRaf what theBear was trying to help me with
03:17 RifRaf found a few tl072 opamps
03:23 RifRaf max233, lm358, 50 x mega168, man i need to find some stuff to use these in
03:23 RifRaf no 7905 yet but giving up for now
03:23 RifRaf i am trying to avoid a real negative supply rue
03:24 RifRaf ok maybe i should try with a newer opamp
03:56 RifRaf we have success
03:56 RifRaf 7mV shows up as 340mV now, opamp set to 49X gain
03:56 RifRaf using lm358 with 5V single supply
03:56 RifRaf was using a dumb opamp
03:56 Jak_o_Shadows1 as opposed to one that finished high school?
03:58 RifRaf yep like me
03:59 RifRaf if you are implyed people are dumb because they did not finish school
03:59 RifRaf mine was done at 14 years but never stopped learning
04:00 RifRaf this thing should work ok down to a 2mV input according to datasheet
04:03 Jak_o_Shadows1 nah.
04:05 RifRaf is cool, at 3.7mV gain seems higher than it should be, will go lower
04:13 RifRaf 2mv in 116mV out, can live with that, 99mV would be perfect but am getting perfect after about 5mV
04:15 RifRaf lm324 needed about 5mV to start working and 20mV to start working near accurate
04:20 zap0 not all skools are places of learning
04:25 RifRaf wow, with 2.5mA through led (dim) , 0.7mV across shunt, 51mV output still,, 0.3mV, 35mV out, at this point my meters are giving false readings too i imagine, is all good enough for wat i need
04:27 RifRaf sooo wish that i found this opamp before the other one, would not have lost 4 days
04:27 RifRaf but learn as usual
04:28 RifRaf can go back to 0.01R shunt now
04:42 Jak_o_Shadows1 Motor driver circuit! http://jak-o-shadows.users.sourceforge.net/electronics/random/motorDriver.PNG Hopefully final!.
04:48 Jak_o_Shadows1 Only thing i'm worried about, my half-baked simulation software says that the power through the mosfets is too big for em
04:57 RifRaf when you push both switches at the same time?
05:28 Jak_o_Shadows1 possibly
05:28 Jak_o_Shadows1 I should try to avoid that I think.
05:28 RifRaf that would cause a direct short
05:28 RifRaf at all costs yep
05:28 theBear don't cause a direct short ! it doesn't get much worse than that in this game !
05:29 Jak_o_Shadows1 I'll probably need to heatsink the fets?
05:29 RifRaf yep, for any decent current
05:29 RifRaf and even a fan as well for big ones
05:30 RifRaf theBear i am happy, and drunk, and things still working
05:30 Jak_o_Shadows1 If the fan isn't directly attached to the HS, but a bit further away, just blowing over it, is that still effective?
05:31 RifRaf yes as long as its moving the hot ait away from them
05:31 RifRaf air*
05:31 Jak_o_Shadows1 that shouldn't be too much problem.
05:31 RifRaf so they get new cool air to heat up, and they do heat up
05:33 Jak_o_Shadows1 remember, no kill like overkill
05:36 RifRaf something weird going on here, the more i test the less gain i am using and the more i wonder why i am even using the opamp
05:37 RifRaf down to 10X gain now and i want to goto 5x
05:37 RifRaf i think i should sleep at this point before i screw it all up again
05:38 RifRaf maybe 5x was all i needed
05:38 RifRaf stay tuned for the next episode :)
05:38 Jak_o_Shadows1 ha
05:38 Jak_o_Shadows1 well, night then,
05:42 RifRaf just had some good wasabi, woke me up again, i do recall wanting to use a range from 0 to 200mV from the 0.01R sense resistor
05:43 theBear hooray ! i'm surprisingly happy tho not comfortable, but i am getting drunk pretty quick
05:43 RifRaf and now with the 1.1vref would be just over a 5x gain required to use it fully, so am kinda happy that matches up
05:46 RifRaf theBear if you ever wanna use lm324, don't, try lm358 instead, and use 2 of them if you need the 4 opamps
05:47 RifRaf working great on 5V single rail supply
05:47 theBear hmmm, very interesting....
05:48 RifRaf found a bunch of other opamps while trying to find a lm7905 for rues next experiment
05:48 RifRaf and after datasheeting a few seemed like that was the one for 5v
05:48 theBear at the same time 324 is DIRT cheap and easy to find, and always works fine for audio kinda stuff.... actually, i gotta glance at that analog pid-like thing i made a few months (a year?) ago again, it acted all crazy when i should been comfortable on maybe 0-12v rails and not TOO close to the edges, i suspect it was a 324
05:49 theBear heh, i coulda told you that, my opamps have always lived one or two compartments away from the regs
05:49 RifRaf the specs say it works from 2mV input but i still had results down to 0.3mV, just extra gain
05:50 RifRaf i found the wrongest one first, but learnt alot in the process so all good
05:50 RifRaf i am wondering if your diode thing would work on this but not game to try now
05:51 theBear that's err, quite damned impressive design, i been looking at a LOT of opamps theory wise recently (not specific models, more general designs/implementations) for various reasons, i've pretty much settled on the conclusion that an opamp is a TINY TINY pushpull/a-b class poweramp, pretty much 100% the classic design with longtail pair/mirror on the input and splitters and drivers for the output stage a little feedback here and there.... and well, howe
05:51 theBear ver you look at it the whole thing is basically transistors and maybe a little resistance, getting that close to the rails must take some sneakiness
05:51 theBear screw it, it was always an ugly hack, just the easiest under the conditions... also it made the dropout on yer v+ reg a little tight
05:52 RifRaf yes i was reading about the first opamps this evening, big things with 2 nice tubes
05:53 theBear heh, you don't wanna think about that, regular 'a class' tube stages are messy enough, it's a theoretical nightmare by the time you got them opamp-ing
05:54 RifRaf yes i have never studied tubes at all, but i knew you would understand
05:54 RifRaf this was but in the 60's
05:54 RifRaf back*
05:55 RifRaf the inventors of the opamp. i went way back to find the answers
05:56 RifRaf page 13 http://www.stanford.edu/class/ee122/Handouts/2-Op-Amp_Concepts.pdf
05:57 RifRaf thats a real odd document to read btw, not sure if intentional or if someone altered it still
05:58 home hey
05:58 RifRaf hey homie
06:00 home hi
06:01 RifRaf watchaupto?
06:02 home me? I am just checking git
06:02 home to see if my changes were pushed or not
06:02 home and talking to people in other channels to see what they are up to with their projects
06:03 RifRaf oh theBear one more thing, using external vref for adc was not much good blow 0.8V and useless below .5V
06:04 RifRaf home cool, i got my adc working and so life is a little bit better now, can rest in peace tonight
06:05 home RifRaf: gj !! :D
06:05 home I never got any ADC's, since they come prebuilt with the stuff I usually have..so yeah
06:05 RifRaf i am using the ones in the avr
06:05 RifRaf but was trying to measure low voltages
06:06 Jak_o_Shadows Ultrasonic sensors connect to an ADC, unless you get a fancy one that connects via i2C?
06:07 RifRaf the ones i have just use echo and trigger pins that can connect to any i/o pins
06:08 RifRaf like sr-04\
06:09 RifRaf gonna add one to this light meter soon to see distance from light source
06:17 home Jak_o_Shadows: fancy? I reckon not
06:20 RifRaf i2c is pretty easy once you get it going. stilll learning it but
06:50 home i2c is weird
14:12 RifRaf yay have volts amps and watts displayed correctly on avr's lcd
14:22 home RifRaf: pic
14:30 RifRaf nope avr
14:31 RifRaf distance colour and illuminence combined as well, just need a few buttons now
15:41 jeet i have an issue, when i turn my motors on, the xbee radio link seems to mess up
15:41 jeet i.e. i can't process the new motor signal
15:43 jeet one other issue i had was: trying to power a wi-fi camera (2W peak power), the adapter it came with steps 100-240AC to 5V, 1.2A.. we were trying to power it with 4xAA = 6V into a 5V, 1A regulator... didn't really work? thoughts? use 2 x 4AA in parallel for more current draw?
15:53 theBear if it came with a 1.2a adapter, it's VERY unlikely it only uses 2w, or they would have used a MUCH cheaper .5a or even smaller one
15:53 theBear and what kind of regulator, usually 1v isn't enough 'room' for most regulators to work
15:54 theBear and finally, depending on how the motors are powered and controlled, you probably just need a decent sized cap somewhere to help with surges under high load (like starting) and switching or brushing noise in general
16:43 jennie hello
16:47 jennie Hi
17:01 Burninate Hello
17:01 Burninate If anyone's about that does computer vision, I'm trying to find out if the PS Eye has been surpassed by something with a larger sensor size / resolution
17:02 Burninate and/or where in hell I can find a decent listing of webcam sensor sizes
17:14 theBear erm, from memory the ps cam is on par with current technology electronic-eye-replacement implants, ok, i exaggerate, but only by a factor of maybe 4 or so... these days even a REALLY cheap noname webcam does 640x480, and usually MUCH higher
17:15 theBear and remember, unless you're doing crazy fringe-stretching photography, you don't care about the size of the sensor, just the rez
17:30 Burninate theBear: What?!
17:30 Burninate Of course you care about the sensor
17:31 theBear but not the SIZE of it, physically
17:31 Burninate That's the dominant factor in the quality of cameras
17:31 theBear unless you are trying to capture super-low-light situations or do crazy lens tricks for infinity focus or something
17:32 theBear not really, optics and resolution are the dominant factors in well lit situations where you don't care about depth of field/focus much, and you DID say webcams, which aren't exactly 'quality' at any level
17:33 Burninate photometric S:N at a given light level and exposure time scales directly with sensor size, or alternately, for a given S:N and given light level, exposure time scales inversely with sensor size, if you maintain expectations of an F/ratio in a certain range
17:34 Burninate super-low-light situations will expose the level setting for what it is, but moderate light levels will be a lot worse quality than high light levels as well
17:35 theBear indeed, but with a webcam you can't possibly have any expectations for that kind of thing, at best you get something that can be focused well somewhere in the 2-5foot region, and focused over a range of just a few inches, however lumpy your average head is... that's what webcams are about, they aint cameras
17:37 Burninate the difference between a 1/7" webcam and a 1/3" webcam is a light-gathering factor of 5x
17:38 Burninate or, equivalently, an S:N 5x higher
17:39 Burninate or, equivalently, a framerate for the same S:N 5x higher
17:40 Burninate I'm not aware of any webcams that can't focus at infinity
17:40 Burninate and they all appear to have rather deep depth of fields
17:42 Burninate I guess if you've got hundreds of watts of light to throw at the problem, it's less of a big dela
17:42 theBear i'm familiar with the math, and within arms reach i have both a webcam with fairly flat/infinite focus, AND another which can't focus well over more than about 9-12"
17:43 Burninate what I'm interested in is computer vision for robotics, where you don't always get a sunlit day
17:44 Burninate Can I take it your advice would be to abandon all hope, ye who enter here, and adopt a $1000 machine vision camera?
17:44 GargantuaSauce webcams are kind of terrible
17:45 theBear i'd say go to a 'big' computer store, and just spend 15 seconds trying each one, check how flat it focuses and at what point the auto-brightness struggles to get any kind of decent contrast/noise-free image
17:47 Burninate the PS Eye is $16, 1/4", 480p60 (USB2.0-constrained) raw signal
17:48 Burninate I'm amenable to spending ten times as much for a better camera, but not thirty
17:49 theBear 480p, so err, roughly ntsc i guess, that's what i meant at the start when i said it's shitty
17:49 theBear a $5 noname webcam will beat that EASILY, tho probly won't break 24/30fps
17:49 Burninate That's rather naive
17:50 Burninate A $100 Logitech or Microsoft webcam won't do 30fps HD
17:50 GargantuaSauce yeah as soon as you turn up the resolution they get sloooow
17:50 Burninate they lie, do 10fps or 15fps and frame-double
17:50 GargantuaSauce and that increased resolution is mostly interpolated anyway
17:50 Burninate or they 'interpolate' the nonexistent pixels on a lowres sensor
17:51 theBear 480p isn't hd !
17:51 Burninate the resolution effectively goes to double digits when there's any motion
17:51 theBear and it's not naive, cos i've got a $5 noname that my buddy picked up in china maybe 10 years ago now, that DOES do that
17:51 Burninate because they alll use heavy inferframe compression
17:52 Burninate *interframe
17:52 theBear err, now that's naive, many of them use little or no compression, and they blur like hell cos they got shitty dirt-cheap sensors and often gain the hell out of them or visually saturate them
17:54 theBear and for reference, logitech and ms have never been at the forefront of webcam tech, and generally charge literally 20x what a 'noname' brand does for EXACTLY the same product in a different case (no, i don't mean the same pcb, tho it wouldn't surprise me)
17:54 Burninate who has been at the forefront?
17:54 theBear chinese nonames, webcams just aren't that hard technically, it's all about cheap and easy to manufacture/design
17:55 Burninate alright, so what's the next step up?
17:55 theBear for 30 bucks you should find one that does 1024x768 up to at least 15fps
17:55 Burninate I don't want "1024x768" out of a sensor without that many pixels, and I don't want 15fps
17:56 theBear next step up might be a VERY high quality security cam, but that'll be pal output, no digital, then you get into 'real' digital cams, somewhere between a few hundred bucks aimed at stuff like endoscopes and microscopes/lab work, maybe image capture, then up to high frame rate and stupid high quality ones for crazy money
17:57 GargantuaSauce i've had good results using a DV cam over firewire
17:57 Burninate it just feels like there's this huge gulf between a PS Eye and a Point Grey
17:57 theBear i don't care what you want, i'm just making you aware of the facts about webcams.... and you asked who is at the forefront, i answered in 2 questions
17:57 theBear wtf is a point grey ?
17:57 Burninate http://ww2.ptgrey.com/USB3
17:58 theBear if you think a playstation cam is anything but a steaming heap of shit, you really don't know much about webcams... they're not built to be special as i alraedy explained early on while you threw basic lens math at me, but the playstation ones are SHIT
17:59 GargantuaSauce you are doing it again
17:59 theBear yeah, medical, industrial, etc etc, those are 'real' cams, like a mid-highgrade dv-cam without the handle and recorder attached
17:59 rninate apologises for coming on a bit rudely in an attempt to cla
18:00 theBear webcams are webcams
18:00 Burninate GargantuaSauce, is there a preferred Firewire DVCam?
18:01 theBear you just pick like you would shopping for one to shoot video with, firewire video is a VERY solid/universal standard
18:01 GargantuaSauce I don't know, I just used what was available...a pretty low-end mini sony dvcam
18:01 GargantuaSauce wasn't ridiculously noisy and had a steady framerate
18:02 theBear time for a rest, gnight then
18:05 GargantuaSauce it sure would be nice if there were more consumer-grade imagers designed for cv usage
18:06 theBear there would be if consumers had any idea what computer vision was or how to achieve it... this time for sure
18:07 GargantuaSauce that is a pretty strange criterion
18:07 GargantuaSauce considering that the two main ones on the market for game consoles don't require the user to understand its function...
18:13 ace4016 as it should be really
18:13 ace4016 most consumers don't care about how it works; they just want it to work
18:35 SorcererX Burninate: yes, there is a huge gap pricewise.
19:19 rue_house RifRaf, you didn't ask the last design question
19:39 RifRaf rue but i solved the problem already
20:15 Tom_itx that doesn't count
20:15 Tom_itx you still gotta play the game
21:04 RifRaf well my question is then how many days does it take to get a parts lists rue? if you gimme a few a day might have the list by next weekend
21:06 Tom_itx he would just dig it outta a box ya know
21:13 RifRaf yes i tryed that night, my boxes don't have as much junk as his though
21:13 Tom_itx nor mine as much as yours
22:51 RifRaf how much current can we source from an avr pin?
22:57 RifRaf 50mA?
23:02 rue_house no
23:03 rue_house more like 15 or 20
23:05 RifRaf thanks
23:05 RifRaf any advances on 20? like 25
23:06 RifRaf rue what part of the puzzle will you give me today?
23:06 RifRaf have to head off in an hour
23:06 rue_house I just woke up, the last question was, do yo uwant a range of +-3A
23:06 RifRaf yes
23:07 RifRaf current i am egtting from 0 to 5A, in about 25mA steps
23:07 RifRaf currently
23:10 RifRaf rue_house do you think the lm358 makes a decent opamp for 5V single supply
23:11 rue_house nope
23:11 rue_house the lm324 is close
23:11 rue_house RifRaf, you have to use a - bias supply to do it right
23:11 RifRaf lm324 is horrible, well it worked fine apart from low inputs, the lm358 was great
23:11 rue_house so will -3A to +3A sense work for you
23:11 RifRaf it works right now
23:11 rue_house if you dont want help then just say :)
23:12 RifRaf i am happy to try something else
23:12 RifRaf to learn
23:12 rue_house the circuit for buddy's current sense is close to what your making
23:13 rue_house but its sense range is -3 to +3A
23:13 rue_house I also have a circuit for high side monitoring
23:14 rue_house I'm trying to remember where I put the schematic
23:15 rue_house !assist buddy_III
23:15 rue_house reaaaaaaaaaaaaaaaaaaaaly?
23:17 RifRaf and your substitute memory is on strike it seems
23:17 rue_house http://eds.dyndns.org/~ircjunk/circuits/olsch.png
23:17 rue_house yea, something is odd, need to work it out
23:18 rue_house oh ho, good show, no values
23:19 rue_house oh, maybe I didn't use lm324....
23:19 rue_house I'll find some values for that in a bit
23:19 RifRaf have some tl072s but they seemed to want a dual rail supply
23:29 rue_house to do this right you have t use dual rail supply
23:29 rue_house thats why I asked if you have a max232
23:29 rue_house it'll make +-10V for you
23:30 rue_house Tom_itx, theBear how big is the main shaft in your washing machine?
23:30 RifRaf about 25mm on my f&p ones
23:30 rue_house hmm
23:30 RifRaf oh sorry i am not theBear or Tom_itx
23:30 rue_house thats descent size for making lathe things from
23:31 RifRaf but i have made a few windmills with washing machines
23:31 rue_house didn't think you been into a machine RifRaf
23:31 rue_house ah yes
23:31 RifRaf der, i get go and vernier one for you within 20seconds
23:31 rue_house no, just wondering if they have descent material in them
23:32 rue_house I dont have any round stock over 1/2"
23:32 RifRaf yes decent, have had em in some big winds
23:34 RifRaf rue_house you remember oxsta?
23:34 RifRaf maple
23:37 rue_house hmm
23:40 RifRaf goin up for his 40th, so gone all weekend
23:45 RifRaf i should give you a schematic for a more simple current meter, though has no protection for hooking it up the wrong way yet
23:47 rue_house was that his name
23:47 RifRaf jamie
23:47 rue_house ah
23:48 rue_house jamie of australlia,
23:48 RifRaf yes
23:48 rue_house more popular than that pope guy
23:48 RifRaf he decided your name was maple remember, i thought that would ring a bell
23:48 rue_house RifRaf, my current sense had to do negitive current too
23:48 rue_house yea
23:48 RifRaf yes wat is negative current?
23:49 rue_house reverse
23:49 RifRaf ah oh yep
23:49 rue_house buddyIII uses + and - current on the motors
23:50 RifRaf with 2 batteries in servies , using where they join for ground?
23:50 rue_house yes, that way I only need 2 transistors for controlling each motor
23:50 rue_house and I get bidir pwm
23:51 RifRaf interesting
23:52 rue_house http://eds.dyndns.org/~ircjunk/circuits/motordriver.png
23:52 rue_house there are so many motors on buddy its worth the extra power trouble
23:53 RifRaf pretty old school
23:53 RifRaf has he got lots of legs or something?
23:53 rue_house thats the 18 axis hexapod
23:53 RifRaf is that the big hexapod
23:53 RifRaf ah yes
23:53 rue_house yes
23:54 RifRaf is it walking yet?
23:54 rue_house overload boards are finished
23:54 rue_house It will take about 24A peak running, and if I put that much power in without single channel overload protection I'll likley smoke something
23:54 rue_house hmm, overload boards are not finished
23:55 rue_house is what I meant to say
23:55 rue_house I'v etched them, still assembling
23:56 rue_house it'll be fun to have it all working
23:56 rue_house the overload boards can trip out each leg on a per-motor overload sense
23:56 RifRaf sure will
23:56 rue_house each motor gets 1.5A max, otherwise it trips a 555 that operates a cutout relay
23:57 rue_house the computer can read the state and reset
23:57 RifRaf not bad for such a big robot
23:57 rue_house also, the motors can strip the gearboxes
23:57 rue_house which is the other reason for the current limiting