IRC log for #edev on 20100816

00:01.18*** join/#edev Crofton (~balister@207.47.37.7.static.nextweb.net)
00:25.27*** join/#edev pytey (~pytey@howler.blatero.com)
01:19.40*** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil)
01:50.54*** join/#edev Crofton (~balister@207.47.37.22.static.nextweb.net)
01:51.16*** join/#edev balister_ (~balister@207.47.37.22.static.nextweb.net)
03:15.33*** join/#edev cbx (~cbx___@unaffiliated/cbx)
04:04.52*** join/#edev mbuf (~user@118.102.130.6)
05:00.11*** join/#edev kergoth_ (~kergoth@ip24-251-170-95.ph.ph.cox.net)
05:04.30*** join/#edev balister_ (~balister@207.47.37.22.static.nextweb.net)
05:11.36bsdfox_http://www.woot.com/ $65 roku wifi/hd device
05:11.43bsdfox_probably good for hacking
05:14.39*** join/#edev cbx (~cbx___@unaffiliated/cbx)
06:34.09*** join/#edev yukach (~yukach@195.222.85.229)
08:17.44*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
08:42.53*** join/#edev cbx_ (~cbx___@unaffiliated/cbx)
08:46.29*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
08:57.41*** join/#edev florian_kc (~fuchs@Maemo/community/contributor/florian)
09:06.29*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
09:30.06*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
09:53.22*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
10:04.00*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
10:13.13*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
10:18.07*** join/#edev JoeLlama (~snork@unaffiliated/joellama)
10:40.16*** join/#edev roxfan (dunno@91.180.128.210)
10:41.22*** join/#edev devilsadvocate (~devilsadv@202.3.77.203)
10:47.13*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
10:51.14*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
10:59.45*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
11:04.12*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
11:07.45*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
11:10.07*** join/#edev devilsadvocate (~devilsadv@202.3.77.203)
11:14.04*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
11:17.21*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
11:24.49*** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho)
11:31.09*** join/#edev methril_work (~methril@201.35.65.90)
11:34.14*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
11:40.55*** join/#edev yukach (~yukach@195.222.85.229)
11:47.45*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
12:25.33*** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil)
12:27.28*** join/#edev calculus (~calculus@gentoo/user/calculus)
12:27.56*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
12:44.53*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
13:02.55*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
13:12.47*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
13:27.52*** join/#edev devilsadvocate (~devilsadv@202.3.77.41)
13:31.13*** join/#edev sjhill (~sjhill@home.bethel-hill.org)
13:31.13*** mode/#edev [+o sjhill] by ChanServ
13:37.29*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
13:56.29*** join/#edev Crofton (~balister@207.47.37.8.static.nextweb.net)
14:01.35*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
14:03.25melchiori'm writing an application for an embedded system which will be sold on the market sooner or later. I 'd like to use the readline library which is in GPL but am I right to use it or the license don't allow me to use this lib for my embedded application ?
14:06.22SpeedEvilIt depends.
14:06.38SpeedEvilIf you use a GPL library in your program, you need to GPL the entire program
14:06.51SpeedEvilAnd supply source on request, or along with it
14:07.18SpeedEvilAlso source for any other GPL elements.
14:08.57melchiorEven if we don't sell the program but a product on which the program works ?
14:09.19florianonly if you distribute the binary
14:09.44SpeedEvilIf you don't distribute the binary, then no, you don't need to.
14:11.29SpeedEvilIf you are writing anapplicationfor sale, then when you sell it, you will need to provide source if you link against GPL.
14:11.49melchiorno the binary is a simple command line interface on which the user operate upon connection to the device
14:13.48SpeedEvilIf the user owns the device, your program runs on the device, then that's a clear violation of the licence on readline, if you don't provide source to your app. Assuming readline is licenced GPL, not lGPL
14:14.28*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
14:15.19sjhillSpeedEvil: it's GPL which is why we don't use it in any of our products
14:15.30SpeedEvilK
14:15.49SpeedEvillooks at two devices running busybox with no source. :/
14:24.20*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
14:30.20*** join/#edev prpplague (~danders@nat/ti/x-ofoccqzpzzriwzbx)
14:35.11*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
14:35.57*** part/#edev melchior (~melchior@AStrasbourg-552-1-23-123.w90-13.abo.wanadoo.fr)
14:36.22*** join/#edev melchior (~melchior@AStrasbourg-552-1-23-123.w90-13.abo.wanadoo.fr)
14:44.22*** join/#edev adyer (~adyer@173-161-64-17-Illinois.hfc.comcastbusiness.net)
15:04.49*** join/#edev sndcrb (~sndcrb@88.194.0.144)
15:13.57*** join/#edev penguinece (~penguinec@2610:48:100:8da:213:20ff:fe0a:7c8f)
15:14.16rektideanyone heard of the DASH7 --- err, protocol, I guess?  or alliance?
15:21.13*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
15:21.47prpplaguerektide: only Blakes7
15:21.56prpplaguejokes with rektide
15:23.16chouimatmorning
15:23.40prpplaguechouimat: hey bud
15:24.38kergoth_morning guys
15:31.29prpplaguekergoth_: hey bud
15:31.33prpplaguekergoth_: hows work?
15:31.45kergoth_good, you?
15:34.57rektidei ordered a couple econotag's... they're cheap freescale arm units with 802.15.4 onboard
15:35.38rektidenext i have to figure out how to make a MOSFET driver with them
15:36.10*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
15:38.17rektidehttps://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IRFP4468PBF these mosfets
15:38.29rektidedont even know what kind of gate current i need
15:39.32rektidebut i'm looking for suggestions on how to make a voltage controller good for 4-20 V
15:40.19rektidehopefully i can find a DC-DC converter than can step down 60-90V to a stable reference, then i'm not sure what i need to do.  frankly i dont really know how DAC's work or what's available.
15:41.12rektideDAC's probably wont be able to sink enough current anyways, i suppose i'll need some op amps to drive or some such to be the actual gate driver
15:42.30prpplaguekergoth_: not too bad, just swamped with work
15:43.07prpplaguekergoth_: TI's procedures are somewhat difficult to grasp
15:43.50kergoth_heh, well, that's true of most big companies
15:43.55rektideshit, R(G) is .8 ohm; gate driving is gonna be a bigger project than i would've guessed
15:44.23kergoth_when I was there it took a bit of getting used to the whole configuration management process, design docs, etc, for their processes
15:44.47chouimatkergoth_: sound extremelly "fun" </sarcasm>
15:45.48sjhillprpplague: howdy
15:45.50sjhillhey kergoth_
15:45.59sjhilllo' chouimat
15:46.22sjhillsmacks rektide
15:46.59prpplaguesjhill: hey bud
15:47.09rektideouch
15:48.32sjhillrektide: no reason, just to remind everyone that the universe is random and unfair
15:52.24chouimat:)
15:52.40kergoth_like we needed reminding :)
15:56.14*** join/#edev sassan (~vadmin@cvs.mpc-ogw.co.uk)
15:58.25rektidecan someone check out that mosfet sheet, and tell me what i'm missing
15:59.31rektidea mosfet whose _gate_ sinks 20V * 25A does not seem correct.  but at 20V max, with a .8 ohm resistance, that's what'd it be.
16:06.31*** join/#edev Crofton (~balister@adsl-76-201-143-217.dsl.pltn13.sbcglobal.net)
16:11.19*** join/#edev JoeLlama (~snork@unaffiliated/joellama)
16:27.16*** join/#edev mbuf (~mbuf@124.123.187.59)
16:34.24SpeedEvilA MOSFETs gate sinking any current has assploded.
16:35.02SpeedEvilYou generally need ~1uA or so to keep a MOSFET on or off - even at 125C.
16:35.07SpeedEvilThe leakage is small.
16:35.39SpeedEvilThe reason you need appreciable gate current is because you want to switch it on and off fast in most cases, to limit the time it spends in the in-between state.
16:36.07SpeedEvilIt may dissipate .1W when fully on - but 100W when the resistance is - say - 50 ohms and equal to the load.
16:36.52SpeedEvilIf it's turned on and off once a minute, then you can without too much hastle - if 100W peak is fine - turn it on and off over a millisecond using a couple of milliamps to charge and discharge it.
16:37.20SpeedEvilIf you're turning it on and off at 100KHz, you need considerably more beefy drivers to move the gate charge out in 10-50ns.
16:40.32*** join/#edev davidc__ (~davidc__@S0106000f662a10a4.vc.shawcable.net)
16:40.34rektideyou'r talking about leakage
16:41.24rektidei think i'm starting to follow
16:42.06rektidethe R(gate) is used to calculate what kind of current you'll experience when you are chargin the gate?
16:42.55SpeedEvilyes
16:43.20rektidedo i need to drain the gate as well, for powering down?  what figures do i need to look at to know how rapidly the charge will dissipate ?
16:43.44SpeedEvilIf you apply 5V, you'd get instantaneously the difference between the gate voltage and 5V * resistance current flowing till it charges.
16:43.49SpeedEvilIt won't dissipate.
16:43.55SpeedEvilIt is a capacitor.
16:44.04rektidesans whatever small leakage there is
16:44.07SpeedEvilyeah
16:44.13rektideok so i need to drive, and drain it
16:44.33SpeedEvilbut leakage can be low enough that it's hours or days
16:44.36rektidei'm thinking a high current op amp will suit here
16:44.44SpeedEvilwhat are you switching?
16:45.01rektidehttps://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IRFP4468PBF these mosfets
16:45.52SpeedEvilwhats the load, and how often
16:45.53rektidei'd really like to do more than just switch the gate
16:47.34*** join/#edev gratz|home (~gratz@cpc3-brig15-2-0-cust237.brig.cable.virginmedia.com)
16:47.38rektideits for a motor, http://endless-sphere.com/forums/viewtopic.php?f=30&t=16728
16:47.53SpeedEvilAre you PWMing it?
16:47.59rektidethey want to
16:48.06rektidewe will to some degree for sure
16:48.15SpeedEvilyou do know this is a brushless motor/
16:48.24rektidei'd also like analog gate control
16:49.16rektideyes
16:49.21SpeedEvilwhy?
16:49.23rektideit is brushless
16:49.40rektidewell, i have this stupid notion that it'd help us drive the damned beast
16:50.13SpeedEvilYou really don't want the transistors in analog mode at any time, unless you know just what you're doing.
16:50.15rektidestart at 18V to beat down the induction in the motor with huge current, then drop to a nominal 8V for most of the cycle
16:50.25SpeedEvilUmm - no.
16:50.32SpeedEvilYou do that with PWM.
16:50.39SpeedEvilNot by varying the gate drive.
16:51.08SpeedEvilThere are also modular ESCs that will 'just work' for a motor of that size.
16:51.10rektideswitching adds heat, i figure if we can avoid switching
16:51.28rektideand abuse the peak current draw of this mosfet
16:51.29SpeedEvilYou still need to do the switching somewhere.
16:51.35SpeedEvilCan't do that.
16:51.41SpeedEvilLook at the safe operating area.
16:51.42rektideyes absolutely, switching is still the primary goal
16:51.50SpeedEvilThe FET might be - say - 100A/50V.
16:52.03SpeedEvilIt however cannot dissipate more than 100W without exploding.
16:52.28rektidethis is a 100V / 200A fet...
16:52.33SpeedEvilIf you try to simply linearly regulate to 100A@25V, you lose 2500W in the transistors, which rapidly explode.
16:52.35rektidewhich dissipates 520W
16:52.49rektidehowever it has a peak of 1120A drain
16:53.22SpeedEvil##electronics is more appropriate for this discussion anyway.
16:53.43SpeedEvilYou will lose _way_ less than 520W just PWMing it.
16:55.06SpeedEvilThere are also brushless motor control chips that may be of use. But I don't have any part numbers.
16:55.06rektidei'm trying to use the analog gate drive to adjust current, not voltage
16:55.15SpeedEvilThat is utter fail.
16:55.15rektidei'm well past the triode region
16:55.22rektidei'm in to saturation
16:55.27rektideits a question of how hard i'm driving saturation
16:55.33SpeedEvilIt varies _extremely_ part-part, as well as with temperature.
16:55.35rektidethats entirely different from what you are talking about
16:55.39SpeedEvilI know.
16:56.16SpeedEvilIf you want an efficient drive, varying the motor current by analog regulation is insanity.
16:56.25rektidemostly, i believe i can safely very very briefly (within single PWM) pulse the gate voltage higher then drop it down fast
16:56.29SpeedEvilYou use PWM, and regulate the current that way.
16:56.39rektideyeah since there's a huge inductor
16:56.42SpeedEvilYou're missingthepoint.
16:56.54SpeedEvilGate current has no relation at all to drain current.
16:57.07rektidei suppose just driving 15V-- way past the MOSFET's power dissipation-- then derating regular ops by PWM'ing it, is the way to go
16:57.26SpeedEvilDrain current is set by the transfer charachteristic of the FET, the Vds, and the gate voltage.
16:57.42SpeedEvilGate current isn't anywhere in this.
16:58.13SpeedEvilGate current only sets how fast it charges the gate capacitance, and how fast it goes on and off.
16:58.15rektideum i was reading my chart headers wrong
16:58.32rektideV(gs) does indeed affect I(ds)
16:58.36rektideFigure 3
16:58.51rektidehttp://endless-sphere.com/forums/viewtopic.php?f=30&t=16728
16:58.55rektideshit wrong link
16:58.56rektidesorry
16:59.02rektidehttps://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IRFP4468PBF
16:59.14rektide"Fig 3."
16:59.40rektidethe graph only goes up to 6.5V
16:59.46SpeedEvilYou're reading it wrong.
16:59.47rektideV(gs)
16:59.57SpeedEvilI can't look at the graph at the moment.
17:00.10SpeedEvilDoes it say next to it *(Vds 10v)
17:00.14SpeedEvilOr similar.
17:00.45rektideno, its showing the triode region.  but past the triode region, there's still some correlation.
17:01.03JoeLlamaoOo 2.6 mOhms :)
17:01.05rektideits for Vds of 25V
17:01.11rektideyeah baby
17:01.12SpeedEvilrektide: Right.
17:01.25JoeLlama100 volts 290 amps and 2.6mOhm
17:01.29JoeLlamaneat (:
17:01.32SpeedEvilrektide: That is holding 25V across the drain and source.
17:01.50SpeedEvilAnd trying to turn it on with the given gate voltage. Which will have to be pulsed.
17:01.59rektideJoeLlama: endless sphere has a guy named Luke / LiveForPhysics whose helping coordinate a to-be open sourced motor controller using
17:02.00SpeedEvilThis is _NOT_ the mode you want to operate in.
17:02.30JoeLlamarektide nice part )
17:02.37rektideSpeedEvil: indeed not.  the graph only goes to 6.5V V(Gs), but it shows correlation past that region.  and that makes sense.  the bigger the channel (more V) the bigger the path for current.
17:02.40SpeedEvilrektide: You're trying to make a car, and using instead of the throttle, slipping the clutch.
17:02.41rektideJoeLlama: i was blown away
17:02.50rektideJoeLlama: now i just have to figure out how the hell to drive the damned thing
17:02.51SpeedEvilrektide: yes - but that's quite irrelevant.
17:02.57JoeLlamayeah of course that RS(on) is at only 10 volts :)
17:03.18SpeedEvilrektide: You are way over your head here.
17:03.28rektideSpeedEvil: i was indeed trying to be fancy
17:03.29SpeedEvilI suggest wandering over to ##electronics.
17:03.37SpeedEvilThere are people there that have built brushless controllers.
17:03.39rektideSpeedEvil: and PWM is the path i'll look at instead
17:03.49SpeedEvilYou can do it analog.
17:04.09SpeedEvilBut you will need huge heatsinks, perhaps 20* the number of transistors, and bigger batteries.
17:04.15rektidei got laughed out by someone for mentioning the R(gate) is 0.8 ohms, he said gate's dont have current and told me to go back to school. :(
17:04.26rektideso much noise in ##electronics, apologies if i'm upsetting anyone's tranquility in here
17:04.55chouimatrektide: don't worry I stopped reading a long time ago :)
17:04.59SpeedEvilIn general, gate resistance is something that you don't care about until you get to really advanced stuf.
17:05.11SpeedEvilAnyway - AFK.
17:05.18SpeedEvilMost datasheets don't even quote it.
17:05.22rektideSpeedEvil: again, there is some I(ds) correlation based on V(gs), and i thought it'd be cute to exploit that small area.
17:05.31rektidebut i'm beginning to agree
17:05.41rektideits too complex and too neglidgible a gain
17:05.47rektideto be worth implementing
17:06.03rektidethis is a pretty big gate that needs to be driven fast, so gate drive is actually important.
17:06.16rektidedriving it to all sorts of crazy voltages, less useful
17:06.22rektideand not for this project, chiefly
17:07.04rektidethank you so much for bouncing this off me.  i just dusted off my very dusty digital electronics book on MOSFET's a month ago, and this is the first sanity check i've had
17:09.22SpeedEvilSearch on the various makers - ST, IR, .. for 'brushless motor'
17:09.51SpeedEvilyes - that can be exploited in some apps. The problem is that trying to do it in a power app means huge wasted power, and ...
17:24.09*** join/#edev devilsadvocate (~devilsadv@202.3.77.41)
17:49.18*** join/#edev NullMoogleCable (~fake@cpe-69-205-123-203.rochester.res.rr.com)
18:42.03*** join/#edev davidc__ (~davidc__@S0106000f662a10a4.vc.shawcable.net)
19:12.36*** join/#edev florian (~fuchs@Maemo/community/contributor/florian)
19:25.06*** join/#edev kergoth_ (~kergoth@ip24-251-170-95.ph.ph.cox.net)
19:26.38adyerR(G) definitely exists, but it's "connected" to the total gate capacitance, so the two combined effectively limit how fast you can slew the gate assuming a perfect driver.  A bunch of people make mosfet gate driver chips.  I would recommend using something off the shelf.
19:44.42*** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil)
20:04.18*** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil)
20:07.36*** join/#edev methril_work (~methril@201.35.65.90)
20:14.12*** join/#edev JoeLlama (~snork@unaffiliated/joellama)
20:23.51*** join/#edev miknix (~miknix@gentoo/developer/miknix)
20:29.47*** join/#edev methril_work (~methril@201.35.65.90)
20:31.14rektideyeah i'll PWM the gate with a driver.  that makes this whole exercise a lot less interesting.  now i've got to focus on the hard work
20:31.25rektideof figuring out what a brushless controller actually needs
20:54.25*** join/#edev maluta (~maluta@189.107.31.6)
21:04.12*** join/#edev JoeLlama (~snork@unaffiliated/joellama)
21:08.57MirellIf only Starbuck's wireless was faster than dial-up.
21:49.23*** join/#edev SpeedEvil (~user@tor/regular/SpeedEvil)
22:02.59*** join/#edev bsdfox (~bsdfox@unaffiliated/bsdfox)
22:13.04*** join/#edev bgamari (~ben@pool-96-240-200-109.spfdma.east.verizon.net)
22:57.10*** join/#edev gustavoz_mile (~gustavoz_@host199.190-136-153.telecom.net.ar)
23:25.06*** join/#edev SpeedEvil (~user@tor/regular/SpeedEvil)
23:46.10*** join/#edev rue_mohr (~rue@24.207.119.38)
23:46.20*** join/#edev krushia (~krushia@h69-131-92-179.cntcnh.dsl.dynamic.tds.net)
23:46.33rue_mohrI want to write my own blackberry apps, anyone know if the blackberry has a jtag port?

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.