highaltitude.log.20091104

[00:13] Luiswu (n=Luiswu@cpe-098-121-125-085.triad.res.rr.com) joined #highaltitude.
[00:14] <Luiswu> Evening
[00:14] <natrium42> hi
[00:16] <Luiswu> Anyone got any experience with pressure switches? Looking for something to activate a beeper once payload gets below 2k feet
[00:18] <natrium42> hmm
[00:18] <natrium42> you could use an absolute pressure sensor and a microcontroller
[00:18] <fergusnoble_> Randomskk: that bit isnt working yet
[00:18] <Luiswu> I was hoping for something simpler
[00:19] <Luiswu> Mainly because this will be in a payload container several feet from the main payload
[00:21] <fergusnoble_> Randomskk: the python bit is being rewritten by rich
[00:22] <natrium42> Luiswu: maybe an analog pressure sensor connected to a biased MOSFET
[00:23] <natrium42> small mcu sounds more reliable imo
[00:25] <Luiswu> natrium42: I was thinking somethink like that. If I am remembering correctly, I've seen pressure switches about 1"x1"x.25" that were either NO or NC... I was wondering if anyone knew about them or where to find them
[00:25] <natrium42> never seen/used those, sorry
[00:26] <Luiswu> If I can find them, and get them to work, I'll post info here
[00:27] <natrium42> cool
[00:27] <Luiswu> KISS seems to be a good idea for projects on the low end as mine
[00:29] <Luiswu> Although my partner wants to use an IMU and servos to stabilze the cameras.
[00:37] natrium42 (n=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left #highaltitude.
[01:00] fergusnoble_ (n=fergusno@fn217.quns.cam.ac.uk) left irc:
[01:04] juxta (i=fourtytw@122-49-175-48.ip.adam.com.au) joined #highaltitude.
[03:08] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[03:09] GeekShad__ (n=Antoine@48.167.204-77.rev.gaoland.net) left irc: Read error: 104 (Connection reset by peer)
[03:54] SpikeUK (n=chatzill@host86-136-203-64.range86-136.btcentralplus.com) joined #highaltitude.
[03:57] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: "The cake is a lie !"
[04:02] SpikeUK (n=chatzill@host86-136-203-64.range86-136.btcentralplus.com) left irc: "ChatZilla 0.9.85 [Firefox 3.0.15/2009101601]"
[04:21] bittwist (n=a@unaffiliated/bittwist) left irc: "( www.nnscript.de :: NoNameScript 4.1 :: www.regroup-esports.com )"
[04:56] bittwist (n=a@unaffiliated/bittwist) joined #highaltitude.
[05:24] bittwist (n=a@unaffiliated/bittwist) left irc: "( www.nnscript.de :: NoNameScript 4.1 :: www.regroup-esports.com )"
[05:28] bittwist (n=a@unaffiliated/bittwist) joined #highaltitude.
[05:29] jasonb (n=jasonb@dsl027-180-244.sfo1.dsl.speakeasy.net) left irc: Read error: 110 (Connection timed out)
[05:44] borism (n=boris@195.50.215.148) got netsplit.
[05:44] Tiger^ (i=tygrys@217.149.240.132) got netsplit.
[05:44] borism (n=boris@195.50.215.148) returned to #highaltitude.
[05:44] Tiger^ (i=tygrys@217.149.240.132) returned to #highaltitude.
[05:52] jasonb (n=jasonb@m4f0536d0.tmodns.net) joined #highaltitude.
[05:54] bittwist (n=a@unaffiliated/bittwist) left irc: "( www.nnscript.de :: NoNameScript 4.1 :: www.regroup-esports.com )"
[05:57] bittwist (n=a@unaffiliated/bittwist) joined #highaltitude.
[06:08] jasonb_ (n=jasonb@m330536d0.tmodns.net) joined #highaltitude.
[06:14] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[06:24] jasonb_ (n=jasonb@m330536d0.tmodns.net) left irc: Read error: 60 (Operation timed out)
[06:26] jasonb (n=jasonb@m4f0536d0.tmodns.net) left irc: Read error: 110 (Connection timed out)
[06:34] natrium42 (n=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[06:42] bittwist (n=a@unaffiliated/bittwist) left irc: "( www.nnscript.de :: NoNameScript 4.1 :: www.regroup-esports.com )"
[06:46] bittwist (n=a@unaffiliated/bittwist) joined #highaltitude.
[06:49] Nick change: bittwist -> bt42
[06:55] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[07:09] gamers600 (n=gamers60@ip68-99-233-6.ph.ph.cox.net) joined #highaltitude.
[07:09] gamers600 (n=gamers60@ip68-99-233-6.ph.ph.cox.net) left irc: Client Quit
[07:33] jasonb_ (n=jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) joined #highaltitude.
[08:24] rjharrison (n=rharriso@gateway.hgf.com) joined #highaltitude.
[08:24] <rjharrison> Morning all
[08:46] Paul2 (n=paul@2001:ba8:1f1:f0c0:216:5eff:fe00:13d) left #highaltitude.
[08:57] SpeedEvil1 (n=user@tor/regular/SpeedEvil) joined #highaltitude.
[08:57] SpeedEvil1 (n=user@tor/regular/SpeedEvil) left #highaltitude.
[09:03] RocketBoy (n=Steve@217.47.75.8) joined #highaltitude.
[09:04] <RocketBoy> morning!
[09:04] <SpeedEvil> Yes, it is.
[09:04] Action: SpeedEvil is probably up.
[09:05] <RocketBoy> rjharrison: Tomorrow is now out - Saturday or Sunday are looking good ATM
[09:10] <RocketBoy> If I had a £1 for every time ran the CUSF predictor or wunderground I'd be richer than Abramovich
[09:57] borism_ (n=boris@195-50-212-148-dsl.krw.estpak.ee) joined #highaltitude.
[10:02] borism (n=boris@195.50.215.148) left irc: Read error: 145 (Connection timed out)
[10:40] Nick change: jasonb_ -> jasonb
[10:58] junderwood (n=chatzill@adsl.jcu.me.uk) joined #highaltitude.
[11:18] <MikeMc> Morning
[11:30] EI5GTB (n=Paul@78.16.7.183) left irc: Read error: 104 (Connection reset by peer)
[11:31] EI5GTB (n=Paul@213-202-176-49.bas504.dsl.esat.net) joined #highaltitude.
[11:39] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[11:44] <RocketBoy> An interesting pump failure - this had been immersed in Methanol for a couple of weeks then the methanol was left to evaporate - the pump being covered in the vapor - http://imagebin.org/70457
[11:45] <RocketBoy> the connecting wire has lost all its flexibility
[11:46] <MikeMc> why was it left in methanol?
[11:47] <RocketBoy> testing of methanol ballast pumping
[11:49] <MikeMc> why methoinal as ballast?
[11:50] <RocketBoy> its been outside for a few weeks - and I only juts got round to looking at it
[11:50] <RocketBoy> low freezing point
[11:50] <RocketBoy> I'm at a bit of a loss as I didn't expect that sort of failure
[11:55] <SpeedEvil> PVC leaching
[11:56] <SpeedEvil> if the wire is PVC - it has plasticisers in that make it flexy
[11:56] <SpeedEvil> and not solid like hard PVC
[11:57] <RocketBoy> yeah - that sounds right
[11:58] <RocketBoy> BBL
[11:58] RocketBoy (n=Steve@217.47.75.8) left irc: "Leaving"
[12:02] <MikeMc> does it need to be flexible once it's in-situ ?
[12:36] <juxta> which flight prediction models have people had success with?
[12:37] <juxta> I'm using the CUSF one, which seems very cool, but I get rather variable results with it - one day it might estimate 150km of deflection to the east, the next it might say 150km to the west
[12:46] rjmunro (n=chatzill@87.127.91.51) joined #highaltitude.
[12:50] <SpeedEvil> that's - largely - weather forecasting
[12:53] <juxta> hmm
[12:53] <juxta> how accurate to the models tend to be SpeedEvil?
[12:54] <SpeedEvil> As accurate as any forecast
[12:54] <juxta> might pose a bit of an issue - to the east is fine, but I live on the coast, even a small deviation to the west will put me in the ocean :)
[12:55] <SpeedEvil> Generally it's a good idea to try launching +-1 hour from your target date, and +-20km or so from your launch point
[12:56] <SpeedEvil> If all of the points go to the same placeish, it indicates that you're in a locally smooth bit of the weather
[12:56] <juxta> ah, good idea
[12:56] <SpeedEvil> If however they all go to different corners of the earth, you probably should avoid.
[12:57] <SpeedEvil> And of course, be prepared to call the launch off if the latest forecast says no.
[12:57] <juxta> hehe
[12:57] <juxta> yeah
[12:57] <juxta> the alternative is I drive far enough out so that it doesn't matter where it blows
[12:58] <juxta> are there other models I should try besides the CUSF one?
[12:58] <SpeedEvil> Don't know.
[13:00] <juxta> fair enough, cheers for the suggestions :)
[13:01] <SpeedEvil> you might see if you can get information on released metereological balloons, and comapre the paths with the model to get an idea
[13:01] <rjharrison> Use the CUSF one
[13:02] <rjharrison> IT's better for quite a few reasons
[13:02] <rjharrison> But not well tested
[13:02] <rjharrison> SpeedEvil good idea
[13:02] <rjharrison> Your local met office might be able to help juxta
[13:04] <juxta> good idea, I'll check it out :)
[13:09] <juxta> I just read the page re in flight landing spot prediction on one of the pegasus payloads, looks pretty cool
[13:16] rjharrison (n=rharriso@gateway.hgf.com) left irc:
[14:15] GeekShado_ (n=Antoine@94.184.204-77.rev.gaoland.net) joined #highaltitude.
[14:33] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: Read error: 110 (Connection timed out)
[15:13] rjmunro (n=chatzill@87.127.91.51) left irc: "ChatZilla 0.9.85 [Firefox 3.5.4/20091016081620]"
[15:36] natrium42 (n=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left irc:
[16:01] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[16:06] sbasuita (n=sbasuita@unaffiliated/drebellion) joined #highaltitude.
[16:19] jcoxon (n=jcoxon@212.183.140.17) joined #highaltitude.
[16:19] <jcoxon> afternoon all
[16:21] Matt_APEX (n=chatzill@152.78.175.54) joined #highaltitude.
[16:45] juxta (i=fourtytw@122-49-175-48.ip.adam.com.au) left irc: Read error: 104 (Connection reset by peer)
[16:47] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[16:53] jcoxon (n=jcoxon@212.183.140.17) left irc: Read error: 110 (Connection timed out)
[17:14] RocketBoy (n=Steve@217.47.75.8) joined #highaltitude.
[17:14] jiffe97 (n=jiffe2@209.159.247.189) left irc: Read error: 110 (Connection timed out)
[17:18] jasonb (n=jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) left irc: Read error: 110 (Connection timed out)
[17:33] icez (n=icez@unaffiliated/icez) joined #highaltitude.
[17:55] Laurenceb (n=Laurence@137.205.17.64) joined #highaltitude.
[17:55] <Laurenceb> hi
[18:11] <Laurenceb> http://www.youtube.com/watch?v=EF3-U9Lb12k
[18:12] <SpeedEvil> eAlso - yes - if it's an avr32
[18:13] <SpeedEvil> I think that's also using a 'smart' LCD
[18:14] <Laurenceb> apparently not
[18:14] <Laurenceb> its making use of sprites ect on the SD
[18:15] <Laurenceb> but yeah to do it properly youd want to use a copy of beagleboard or ngw100
[18:16] jasonb (n=jasonb@m430536d0.tmodns.net) joined #highaltitude.
[18:35] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Remote closed the connection
[18:36] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[18:40] <Laurenceb> Randomskk: about?
[18:47] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Remote closed the connection
[18:48] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[18:48] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Client Quit
[18:59] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[19:12] natrium42 (n=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[19:13] rjharrison (n=rharriso@80.176.172.227) joined #highaltitude.
[19:13] <rjharrison> PING RocketBoy
[19:13] <natrium42> hi
[19:14] <natrium42> sup rjharrison ?
[19:14] <rjharrison> Hey natrium42
[19:14] <rjharrison> Well the UK is being crap for launch weather
[19:14] <rjharrison> There might be a window of opportunity this w.e
[19:14] <natrium42> :S
[19:14] <rjharrison> but not really perfect for an altitude attempt
[19:19] jasonb (n=jasonb@m430536d0.tmodns.net) left irc: Read error: 110 (Connection timed out)
[19:19] <Laurenceb> you can come to SEDS instead
[19:20] sbasuita (n=sbasuita@unaffiliated/drebellion) left irc: Read error: 110 (Connection timed out)
[19:24] <natrium42> try edwards airforce base
[19:28] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Remote closed the connection
[19:30] <natrium42> or white sands
[19:30] <rjharrison> Hum
[19:30] <rjharrison> I could
[19:33] <RocketBoy> rjharrison: - yo
[19:35] <rjharrison> Hi RocketBoy
[19:35] <rjharrison> It's still looking dicy for the w/e
[19:35] <rjharrison> dicey
[19:35] jasonb (n=jasonb@dsl027-180-244.sfo1.dsl.speakeasy.net) joined #highaltitude.
[19:35] <RocketBoy> I guess
[19:35] <rjharrison> You have any thoughts
[19:35] <RocketBoy> just see how it pans out
[19:36] <rjharrison> The following week doesn't look much better
[19:36] <rjharrison> Is this typical for the time of year?
[19:36] <RocketBoy> yeah - but I think there is so much variance in the predictions not to worry bout them unless its 2 or 3 days out
[19:37] <rjharrison> Ok
[19:37] <RocketBoy> what may seem like a bad patch a week out now might look good in a few days
[19:37] <rjharrison> I'll watch the sat/sun predictions ofer the next few days
[19:38] <RocketBoy> yeah
[19:42] <RocketBoy> I get the impression that someone here is playing with the CUSF predictor code?
[19:45] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[19:47] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: Remote closed the connection
[20:00] bittwist (n=a@unaffiliated/bittwist) joined #highaltitude.
[20:01] <Randomskk> Laurenceb: hi
[20:01] <Randomskk> brb one sec but I am around now
[20:05] bt42 (n=a@unaffiliated/bittwist) left irc: Read error: 60 (Operation timed out)
[20:07] jasonb (n=jasonb@dsl027-180-244.sfo1.dsl.speakeasy.net) left irc: Read error: 110 (Connection timed out)
[20:08] <Laurenceb> Randomskk: hows the quadcopter cooming on?
[20:08] <Laurenceb> are you working on the ESCs ?
[20:12] <Randomskk> yea
[20:12] <Randomskk> got a motor spinning
[20:12] <Randomskk> have had approximately zero time to work on it though
[20:12] <Randomskk> oh my god examples papers
[20:13] <Randomskk> so all I've got is a motor spinning on my breadboarded design
[20:13] <Randomskk> and not very well, the code could do with significant improvement
[20:13] <Randomskk> current stage is designing the PCB, then I'll send off for that and get them made, at which point I should be able to work on the code in anger
[20:13] <Laurenceb> cool
[20:14] <Laurenceb> ESC pcb?
[20:14] <Randomskk> yea
[20:14] <Laurenceb> what are you doing for the main pcb?
[20:14] <Randomskk> to start with, the stm32 breakout I made, with jumper wires to my sensors
[20:14] <Laurenceb> ah cool
[20:14] <Laurenceb> sounds nice and simple
[20:14] <Randomskk> once I know which sensors I want to use and have some level of control, I will make a pcb with the arm and the sensors all onboard
[20:14] <Randomskk> hopefully from there I should be able to easily and relatively cheaply make multiple quads
[20:14] <Randomskk> going to work on the chasis at some point
[20:14] <Laurenceb> I'll be happy to help with the guidance code
[20:15] <Randomskk> thanks
[20:15] <Laurenceb> you need a ublox5 onboard :P
[20:15] <Randomskk> will let you know when I get to that point :P give it a few terms/years/...
[20:15] <Randomskk> considering it
[20:15] <Randomskk> atm just got LS23002 or whatever
[20:15] <Randomskk> 5hz nmea
[20:15] <Laurenceb> I hear ublox is better
[20:15] <Randomskk> but again that's why I'm prototyping first and then doing it integrated
[20:15] <Randomskk> yea, so do I
[20:16] <Laurenceb> its 4Hz ubx
[20:16] <Randomskk> I think cusf have/are getting some
[20:16] <Laurenceb> my ublox receivers are certainly very nice
[20:16] <Laurenceb> I put some UBX code up on the wiki
[20:16] <Randomskk> will check it out
[20:16] <Randomskk> current timeline is something like this
[20:16] <Randomskk> get ESC spinning a motor fast enough to generate lift
[20:17] <Randomskk> get main ARM controlling ESCs so all four run at constant speed
[20:17] <Randomskk> get main ARM to respond directly to gyros to roughly level the thing
[20:17] <Randomskk> run gyros through a PID controller
[20:17] <Randomskk> add accelerometers
[20:17] <Randomskk> integrate a kalman filter
[20:17] GeekShado_ (n=Antoine@94.184.204-77.rev.gaoland.net) left irc: "The cake is a lie !"
[20:17] <Randomskk> worry about magneto and baro for yaw/altitude
[20:17] <Laurenceb> you need a kalman filter really for all the last stages
[20:17] <Randomskk> then gps for position hold
[20:17] <Randomskk> also at some point radio control and telemetry
[20:18] <Randomskk> and ground station software
[20:18] <Laurenceb> if it was my I'd get a attitude filter running
[20:18] <Randomskk> the first steps are really so I can get something actually flying soonish
[20:18] <Randomskk> for certain values of flying
[20:18] <Laurenceb> as soon as you have the ARM board running
[20:18] <Laurenceb> hmm you weant to be careful
[20:18] <Randomskk> the ARM board itself is running, but writing the kalman filter is tricky and takes time
[20:18] <Laurenceb> yes, but its important for it to work
[20:18] <Randomskk> vs going "here run all the motors at 10% speed and at least it will move"
[20:19] <Laurenceb> I guess
[20:19] <Randomskk> but I agree, the kalman filter is definitely important
[20:19] <Randomskk> right now time is just the big constraint
[20:19] <Randomskk> way way more than when I was at school or on holiday
[20:19] <Randomskk> despite having less scheduled activities
[20:20] <Randomskk> I now actually have work that has to be done :P
[20:20] <Randomskk> for certain values of has, etc
[20:22] <Laurenceb> I'd get the gyros ect talking to the ARM, then dump to data to a terminal
[20:22] <Randomskk> ideally I wanna get it to the point where it can keep itself level and gps position hold, then all the radio commands need to be is "go to this position" and then maybe stuff to do with payload, e.g. take photo/rotate camera/whatever
[20:22] <Laurenceb> then you can test in matlab
[20:23] <Randomskk> I really need to get playing with matlab.
[20:23] <Laurenceb> yeah that sounds perfect
[20:23] <Randomskk> I've got the gyros talking to the arm, so to speak - analogue gyros so it's just reading an ADC
[20:23] <Laurenceb> I put some quick code on the wiki
[20:23] <Randomskk> and sending them out over serial is easy, I want to send them out over USB for speed etc
[20:23] <Randomskk> oh, I think I saw that a while back
[20:23] <Laurenceb> thats for just three gyros and an accel
[20:23] <Randomskk> I've not actually touched matlab before, although I think we do "cover it" to some level at the course here
[20:23] <SpeedEvil> serial may be a better idea
[20:23] <Randomskk> ofc we also cover c++ and that was a very very short introduction, really more of a get you started kinda thing
[20:24] <Randomskk> SpeedEvil: it's simpler but slower
[20:24] <SpeedEvil> slower?
[20:24] <Randomskk> 115200 baud seriel over usb-serial
[20:24] <SpeedEvil> What are you doing that 115200baud would be too slow?
[20:24] <Randomskk> high speed adc readings
[20:24] <Randomskk> maybe it's not too slow, I've not tried running any code on the computer side.
[20:24] <Laurenceb> you dont need that many readings
[20:25] <Randomskk> the more the better when you are integrating to get attitude though
[20:25] <SpeedEvil> you don't do it that way though
[20:25] <Laurenceb> low pass filter first
[20:25] <SpeedEvil> Analogue filter with a 20hz or so corner
[20:26] <Randomskk> the arm can sample over a variable time period as it were which may suffice
[20:27] <Randomskk> so you get the average voltage over your sample time period
[20:27] <SpeedEvil> you low-pass filter, and then sample faster than nyquist
[20:28] <Randomskk> okay
[20:28] <Randomskk> but ultimately you still need to get a relatively high sample rate
[20:28] <Laurenceb> what
[20:29] <Laurenceb> CTCP/PING request from fluppie : 3480605192
[20:29] <SpeedEvil> No, you need 50Hz or so.
[20:29] <Randomskk> unix time probably
[20:29] <Laurenceb> you get that?
[20:29] <Randomskk> that's a timestamp they sent back, or so.
[20:29] <SpeedEvil> 50Hz * 3 gyros = 150Hz
[20:29] <Randomskk> it's not a unix timestamp though
[20:29] <Randomskk> SpeedEvil: 50Hz sample rate?
[20:29] <SpeedEvil> Randomskk: yes - with a 20Hz filter first
[20:29] <Randomskk> I've seen 500Hz as the slowest you want your entire control loop running
[20:29] <Randomskk> going up to 1kHz
[20:29] <Randomskk> including sampling
[20:30] <SpeedEvil> Not really directly relevant.
[20:30] <SpeedEvil> The control loop speed isn't relevant to the sampling.
[20:30] <Laurenceb> you only want enough data to test in matlab
[20:30] <Randomskk> pointless for the control loop to run faster than the sample rate
[20:30] <Randomskk> it wouldn't have any new data
[20:30] <SpeedEvil> you can sample the gyros at 50Hz - and with a 20Hz filter on the frontend - completely reconstruct the input signal.
[20:31] <Randomskk> completely reconstruct the 20Hz-filtered signal, sure
[20:31] <Randomskk> but not the original signal out of the gyro
[20:31] <SpeedEvil> At worst you get a small phase offset.
[20:31] <Randomskk> I'm saying the gyro output updates faster than 20Hz
[20:31] <SpeedEvil> Which means you are - perhaps 0.05 degrees off for 1/10th of the second.
[20:32] <SpeedEvil> The actual craft cannot stop or start rotating significantly to exceed 20Hz
[20:32] <SpeedEvil> to approach even
[20:32] <SpeedEvil> unless it hits something
[20:32] <Randomskk> the motors can certainly change speed a lot faster than that
[20:32] <SpeedEvil> No, they can't.
[20:32] <Randomskk> relatively sure they can, depending on the momentum of the load I guess but still
[20:32] <SpeedEvil> The delta speed in 1/20th of a second is really quite small.
[20:33] <SpeedEvil> And that makes the delta thrust really small too.
[20:33] <SpeedEvil> And the torque effects on the vehicle.
[20:33] <Randomskk> I can change the speed of the motor even within a single commutation phase, and there are six of those per revolution, and it is rotating at 10k RPM
[20:33] <SpeedEvil> Hence maximum rotational accelleration is small.
[20:33] <SpeedEvil> No, you can't.
[20:33] <SpeedEvil> The propellor mass is _significant_.
[20:34] <Laurenceb> you mean moment of inertia
[20:34] <SpeedEvil> You are limited in the torque you can apply.
[20:34] <SpeedEvil> yes.
[20:34] <SpeedEvil> I've done way too much work on lights and stuff today, and have a headache.
[20:35] <SpeedEvil> Work out hte rotational inertia of the propellor + motor. And the maximum torque at x000RPM
[20:35] <Randomskk> assuming 20hz is the maximum system response rate, that would imply there's no cause to run the control loop significantly faster?
[20:36] <SpeedEvil> I don't see a good reason for it.
[20:36] <SpeedEvil> Perhaps trhere's something I'm missing.
[20:36] <Randomskk> then I am at a loss as to why all the functioning quads I've seen run at 500Hz and above
[20:36] <SpeedEvil> But if your inputs are all changing at under 20Hz, your outputs cannot, ...
[20:36] <Randomskk> that's the assumption I'm less sure about
[20:36] <Randomskk> I don't see why the gyros can't change significantly faster than 20hz
[20:37] <SpeedEvil> If they're not prefiltering the inputs, you do need high rates to properly integrate.
[20:37] <SpeedEvil> Because the gyro is bolted onto a heavy thing.
[20:37] <SpeedEvil> It has a limited amount of thrust.
[20:37] <Randomskk> okay, point taken there
[20:37] <SpeedEvil> How fast can it accellerate rotationally - given +100% of thrust on two sides, and 0% on the other.
[20:38] <Randomskk> the quad as a whole has a high enough moment of intertia that rapid rotational accelerations are unlikely
[20:38] <SpeedEvil> It's fast - as humans percieve time - but not really absolutely fast.
[20:38] <SpeedEvil> If you don't filter the inputs, then you do need to sample fast, or vibrations may screw you
[20:38] <Laurenceb> I'd say 50Hz control loop
[20:39] <Randomskk> that all seems correct, but in that case I don't see why everyone else has such significantly faster control loops
[20:39] <Laurenceb> as they dont understand properly?
[20:39] <Randomskk> even failing that, the person I've spoken to tried a range of speeds and found the higher he could go, the more stable it was - 500hz being his hardware limit
[20:40] <SpeedEvil> Many people don't really understand the maths.
[20:40] <Laurenceb> it may be his filter wasnt very good
[20:40] <SpeedEvil> And hence try tweaking stuff, to improve performance, and improve it without really working out why.
[20:40] <Randomskk> there's also that group of MIT engineers who made a quad, and theirs was a 1kHz loop
[20:40] <Randomskk> I'd hope they got the maths
[20:41] <Laurenceb> e.g. it became unstable with larger angle changes between iterations
[20:41] <Randomskk> that said, thinking about it, I believe their kalman was at 50Hz or so on a ground computer
[20:41] <Randomskk> but the onboard control loop was definitely at 1kHz
[20:42] <Laurenceb> I dont see how that worked
[20:43] <Laurenceb> unless they put the gyro data through a direction cosine matrix or something
[20:44] <Randomskk> that was another thing I was looking at doing, but in their case...
[20:44] <Laurenceb> I think it'll pretty much be picking up vibrations
[20:45] <Randomskk> iarc.angel-strike.com/2009SymposiumPapers/2009MIT.pdf
[20:45] <Randomskk> and theirs is incredibly stable
[20:46] <Randomskk> obviously a cluster of ground stations is cheating massively but still
[20:47] <Randomskk> http://www.youtube.com/watch?v=5qQJwLJ857s
[20:49] <SpeedEvil> AIUI, the absolute bare minimum of speed you need your position filter to run at would be the time at which the cross-coupling errors start to become non-nelgigable.
[20:49] <Laurenceb> yeah Ive seen it before
[20:49] <SpeedEvil> for example when rotation couples into change in position.
[20:49] <Laurenceb> hmm you looks like they have PD control at 1KHz
[20:50] <Laurenceb> so the D term is easy - just pass through the gyro data
[20:50] <Laurenceb> P might not actually be at 1KHz, uits unclear
[20:50] <Randomskk> yea, I think so. PD direct from their IMU at 1KHz, then the kalman for their position estimate on the ground station used for nagivational decisions rather than vehicle stability?
[20:51] <Randomskk> oh, planner at 0.3Hz, using SLAM
[20:51] <Laurenceb> alternatively you could just do linear integration of the gyros
[20:51] <Laurenceb> atarting from the last kalman filter output
[20:51] <Laurenceb> for the P term
[20:52] <Laurenceb> *starting
[20:52] <Randomskk> if the kalman is running fast enough, couldn't you just take its output d theta and theta?
[20:52] <Randomskk> rather than integrate from its last output
[20:53] jasonb (n=jasonb@dsl027-180-244.sfo1.dsl.speakeasy.net) joined #highaltitude.
[20:55] <Laurenceb> yes
[20:55] <Laurenceb> IMO you dont need 1KHz
[20:55] <Randomskk> seems like that'd be the better way to go to get smoother gyro data
[20:55] <Randomskk> yea, I've seen perfectly stable 500hz
[20:56] <Laurenceb> I've seen stable stuff with RC rate gyro unitsd
[20:56] <Laurenceb> and flown by hand
[20:56] <Randomskk> but I don't wanna fly by hand :P
[20:56] <Randomskk> point taken though
[20:56] <Randomskk> will have to have a play with that
[20:56] <Randomskk> maybe even put all this new mechanics knowledge to use
[20:57] <Randomskk> though we've not covered rotational inertia yet. lol
[20:57] <Laurenceb> wikipedia is your friend
[20:57] <Randomskk> yea
[20:58] <Randomskk> I'm sure I can get it figured out
[20:58] <Laurenceb> I'd use the motor rpm feedback as a control vector input
[20:58] <Randomskk> like I said though, the current phase is "design esc pcb"
[20:58] <Randomskk> hmm really?
[20:58] <Laurenceb> yes, its useful
[20:58] <Randomskk> well, that's easy enough to do
[20:58] <Randomskk> and I guess rpm is easier to relate to the motion than the arbitrary value throttle level would be
[21:00] <Randomskk> given diameter, pitch of propellor plus whatever weird aerodynamics you need
[21:00] <Laurenceb> yes, as you have to consider actualy speed
[21:00] <Randomskk> so work out some formula for rpm->force
[21:01] <Randomskk> and from there you could even run the calculations to work out acceleration etc but I guess it'd be easier to just try and respond
[21:01] <Randomskk> ha, integrating force from rpm over time to work out attitude, altitude, position would be fun
[21:02] <Randomskk> pity about the wind
[21:02] <Laurenceb> thats exactly what you do do
[21:02] <Randomskk> but without wind it could be done - sensorless navigation :P
[21:02] <Laurenceb> you dont have a control vector as such
[21:02] <Laurenceb> just rpm as a data inoput effectively
[21:02] <Randomskk> to be fair, ground vehicles do just that, rotary encoder on the wheel and you know your position, speed, etc
[21:03] <Laurenceb> you need to gyros and accel ect for it to work
[21:03] <Randomskk> in bitter unpredictable reality, yes
[21:03] <Laurenceb> motor rpm is only going to get you a little bit more performance
[21:03] <Randomskk> still, every little counts
[21:04] <Randomskk> eurgh, I have enough mechanics examples papers to do as it is :P
[21:06] <SpeedEvil> For ideal performance, you want to step as fast as possible to the desired speed
[21:06] <SpeedEvil> not wander there as you might with a normal control algorithm
[21:07] <Randomskk> SpeedEvil: the wandering comes from making small adjustments to speed until the output state is reached?
[21:07] <Laurenceb> ideally you want to solve the equations required to get you to the desired position
[21:07] <Randomskk> vs making a single adjustment that will take you all the way?
[21:07] <Laurenceb> rather than PID dontrol
[21:07] <Randomskk> I see what you mean
[21:07] <Randomskk> is that feasible?
[21:08] <Laurenceb> I think so
[21:08] <SpeedEvil> And by step, I mean optimally control the motor speed to implement the effects on the vehicle that youwant in the most optimal way - which won't always actually be a step
[21:08] <Laurenceb> its complicated by the fact theres usually no solution that completes in one timestep
[21:08] <Laurenceb> also you have errors
[21:08] <Randomskk> though hopefully the kalman helps with the errors
[21:08] <SpeedEvil> Rather than going 'no, still need a bit more, turn it on faster' 'oops - overshot' ...
[21:08] <Randomskk> as much as possible
[21:08] <Laurenceb> but you can find an optimal throttle value to each esc at each timestep
[21:09] <Randomskk> SpeedEvil: though the PID should hopefully not overshoot if the gains are correct?
[21:09] <Laurenceb> yes, its very handy
[21:09] <Laurenceb> the kalman gives you measures of uncertainty
[21:09] <SpeedEvil> Gains may vary for example if you're in ground effect
[21:09] <Laurenceb> you you can arrive at optimal values
[21:09] <Randomskk> Laurenceb: I see what you mean, so while the PID is more of a trail and error approach, you can work out the actual optimal values
[21:09] <Laurenceb> yes
[21:09] <SpeedEvil> yes
[21:09] <Randomskk> trial* even
[21:09] <Laurenceb> PID is a bodgey way
[21:10] <Randomskk> yea.
[21:10] <Laurenceb> you something this simple you can work out optimal solutions
[21:10] <Laurenceb> as its really quite well behaves and easy to model
[21:10] <Randomskk> if you say so :P
[21:11] <Laurenceb> http://en.wikipedia.org/wiki/Optimal_control
[21:11] <Laurenceb> you have prop thrusts and moments of inertia
[21:12] <Laurenceb> and thats about it
[21:12] <Randomskk> yup, I guess so.
[21:12] <SpeedEvil> In a windless environment
[21:13] <Randomskk> will worry about wind once it's stable enough to be taken out of my room :P
[21:13] <Randomskk> though admittedly getting it stable enough to be in my room may require going outside where it is less likely to destroy my stuff
[21:13] <Laurenceb> basically at each iteration you have some erranous attitude and attitude rate
[21:13] <SpeedEvil> one of those kiddy trampolines with guard mesh.
[21:14] <Randomskk> SpeedEvil: that'd be perfect
[21:14] <Laurenceb> the question is how do you correct as soon as poss
[21:14] <Randomskk> so long as it doesn't cut the mesh up, of course :P
[21:14] <Randomskk> and a bit of a roof
[21:14] <SpeedEvil> (with shrouded props)
[21:14] <Randomskk> shrouded props is something I really need to work on, heh
[21:14] <Laurenceb> that would make time the cost function
[21:14] <Randomskk> well the whole chasis design really
[21:14] <Laurenceb> you may make cost some function of time and energy consumption
[21:14] <Randomskk> might redesign the central bit and get it cut out of some sheet metal
[21:15] <Randomskk> Laurenceb: depending on if you wanted to optimise for stability or optimise for battery life though I guess?
[21:15] <Randomskk> I guess there's one solution that uses the least additional energy from the motors, and another that uses the most
[21:15] <Laurenceb> e.g. you throttle up to 100% then have to throttle to -100%
[21:15] <Laurenceb> that uses a lot of energy
[21:15] <Randomskk> I guess I'd have to worry about air resistance too
[21:15] <Laurenceb> but gets you to the desired state in minimal time
[21:16] <Randomskk> yea. so you can optimise for battery life
[21:16] <Randomskk> heck, you could have a slider on the control software
[21:16] <Randomskk> just like the cpu scaling on a laptop
[21:16] <Laurenceb> its not going to make a huge difference unless its in wind
[21:16] <Laurenceb> you may be able to make it auto tune
[21:17] <Laurenceb> so it increases the reposiveness if it senses that its starting to lose control
[21:17] <Randomskk> so optimise for minimal battery consumption but automatically increase as needed to maintain a minimum level of control
[21:17] <Laurenceb> yes
[21:17] <Laurenceb> so in your room it would go to pretty much optimal battery life
[21:17] <Randomskk> and above that a sliding scale the user can set depending on whether stability or battery life is more important
[21:18] <Laurenceb> outside it would start to be more aggressive
[21:18] <Randomskk> right, due to wind.
[21:18] <Laurenceb> yes
[21:18] <Randomskk> but equally I could have it be very aggressive in my room to keep it as stable as possible, where I can easily recharge
[21:18] <Randomskk> though admittedly I don't like the sound of a very aggressive flying thing in my room ;P
[21:19] Action: Laurenceb contemplets wireless recharging
[21:19] <Randomskk> would there be any significant reason to increase energy consumption if a good-enough solution exists for minimum energy?
[21:19] <Randomskk> I guess it depends on the difference in the time taken between the two solutions
[21:19] <Laurenceb> probably not
[21:19] <Laurenceb> its only going to make a few % difference
[21:19] <Randomskk> so just generally optimise for time and battery life
[21:19] <Laurenceb> yeah
[21:20] <Laurenceb> but you may want to start with pid
[21:20] <Randomskk> so the important thing is more to always be powerful enough to stabalise, which may require lots more power in certain conditions, therefore giving the same approximate stability even in very different conditions
[21:20] <Randomskk> would make predicting battery life a total bitch
[21:20] <Laurenceb> I dunno, guess you could sim the entire thing pretty well in matlab
[21:20] <Randomskk> :P
[21:21] <Laurenceb> and test pretty much the entire algorythm
[21:21] <Randomskk> I definitely need to get some experience in with matlab
[21:21] <Randomskk> hopefully by the end of term, or failing that at least by the end of the winter vacation I can get the hardware pretty well working
[21:21] <Randomskk> then the control stuff comes into play properly
[21:23] <Laurenceb> can you keep it open?
[21:23] <Randomskk> in what sense?
[21:25] <Laurenceb> open source
[21:25] <Randomskk> of course
[21:25] <Randomskk> entirely
[21:25] <Laurenceb> cool
[21:30] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[22:02] edmoore (i=836f0142@gateway/web/freenode/x-85bddfcca4183a32) joined #highaltitude.
[22:03] <edmoore> yo. rjharrison - what's latest?
[22:03] <Randomskk> hey edmoore, I ran the cusf svn through a visualiser tool and it made a cool video of every file edit
[22:03] <edmoore> examples papers going well, then?
[22:03] <Randomskk> you can tell? :P
[22:04] <Randomskk> not atrocious tbh, my run of supervisions is over for this week and for the one today I had done all the questions we could cover so that was good
[22:04] Action: MikeMc is not happy
[22:04] <Randomskk> nothing until monday when I have maths and I've even done most of that. but then tuesday onwards is not good
[22:04] <Randomskk> still. spent all day up to dinner and then supervision today doing examples papers
[22:05] <Randomskk> anyway the video shows how utterly crazy your svn gets pretty quickly
[22:05] <Randomskk> but it's like 100mb and I forgot to stop it as I was at dinner so it runs until 2012 and as you can imagine not much happens on your svn then.
[22:06] <edmoore> uhuh
[22:06] <edmoore> any news on launches forthcoming?
[22:06] <edmoore> am only doing a flying visit here. am still in dept, working on hobble pcbs
[22:06] DanielRichman (n=daniel@unaffiliated/danielrichman) joined #highaltitude.
[22:07] <edmoore> whether or not i do an all-nighter depends on whether or not i have to track balloons tomorrow!
[22:07] <Randomskk> is hobble your 4th year project?
[22:08] <Randomskk> 09:05:22 <RocketBoy> rjharrison: Tomorrow is now out - Saturday or Sunday are
[22:08] <Laurenceb> edmoore: going to SEDS?
[22:08] <Randomskk> looking good ATM
[22:08] <Randomskk> 19:14:16 <rjharrison> Well the UK is being crap for launch weather
[22:08] <Randomskk> 19:14:31 <rjharrison> There might be a window of opportunity this w.e
[22:08] <edmoore> Laurenceb: we were going to talk there, and then all of a sudden we dissappeared off the schedule a few days ago to be replaced by something about careers in ESA (who would do that to themselves!?) and they didn't even tell us. so sod them.
[22:09] Action: MikeMc is not happy he missed James' talk at Hackspace
[22:09] <Randomskk> edmoore: so your talk was actually cancelled and they also didn't tell you?
[22:09] <edmoore> yes
[22:09] <Randomskk> ouch
[22:09] <edmoore> lucky iain checked or we'd have driven down
[22:09] <Laurenceb> bah
[22:10] <Laurenceb> Ive paid £14 to get in
[22:10] <edmoore> right, need to get head down so I'll be off. remember to fling emails around if there's likely to be a launch
[22:10] <Laurenceb> sucks more for you
[22:10] <Randomskk> seeya
[22:10] edmoore (i=836f0142@gateway/web/freenode/x-85bddfcca4183a32) left irc: "Page closed"
[22:17] <MikeMc> I wonder how James got on
[22:36] <SpeedEvil> I wonder when the DEA will start shooting down drug smuggling UAVs.
[22:37] <Randomskk> it does seem like a pretty good idea
[22:37] <Randomskk> tie your parcel of crack to it
[22:37] <Randomskk> autopilot using gps
[22:37] <Randomskk> drop off the crack
[22:37] <Randomskk> autopilot into the sea
[22:37] <Randomskk> the perfect crime?
[22:38] <DanielRichman> Well it does spend 5 hours of flight broadcasting
[22:39] <DanielRichman> and when you say drug-smuggling I guess you mean from another country into england, which requires some nice weather, some complicated legal stuff (? does it?) if you want to try to remain inconspicuous and some pretty pro ballooning
[22:39] <MikeMc> Speedeveil ar eyou in the US ?
[22:40] <SpeedEvil> No.
[22:40] <Laurenceb> http://www.sparkfun.com/tutorial/AdTutorials/SFE03-0008-PrintAd05-Vanessa-BIG.jpg
[22:40] <Laurenceb> ^ lulwut
[22:41] <SpeedEvil> I just used DEA as an acronym, as I couldn't recall the UK agency
[22:41] <MikeMc> lol
[22:41] <MikeMc> ok
[22:41] <MikeMc> Laurenceb: wtf
[22:41] <SpeedEvil> I've also pondered undersea electric people smuggling subs, which also seem quite possible.
[22:42] <Laurenceb> cheaper prostitutes
[22:44] junderwood (n=chatzill@adsl.jcu.me.uk) left irc: "ChatZilla 0.9.85 [Firefox 3.0.15/2009101601]"
[22:45] <Laurenceb> and crack
[22:45] <Laurenceb> we just need to get it ready by 2012
[22:45] <Randomskk> hehe
[22:47] GeekShadow (n=Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[22:49] fuzzylugnuts (n=hush@c-68-34-212-42.hsd1.tn.comcast.net) joined #highaltitude.
[22:53] Matt_APEX (n=chatzill@152.78.175.54) left irc: Read error: 110 (Connection timed out)
[22:53] DanielRichman (n=daniel@unaffiliated/danielrichman) left irc: "Leaving"
[22:56] <fuzzylugnuts> Hi
[22:58] Action: Laurenceb has been trying nitrous/kerosene in his rocket sim code
[22:58] <SpeedEvil> Did it light?
[22:58] <Laurenceb> I think the nebula pn-prize project would work with just two stages
[22:58] <MikeMc> ugh?
[22:59] <Laurenceb> wel.. using a CF tank
[23:00] <Laurenceb> I'm not sure what the entire design is, but its using an annular aerospike for the first stage engine
[23:01] <Laurenceb> and ground launch
[23:01] <Laurenceb> I'm not convinced that plumbing parts will take the pressure
[23:03] <Laurenceb> but hes using high pressure aluminium helium cylinders for the tests - not propane tanks
[23:05] <Laurenceb> you could use them for the entire rocket but it soon needs a lot of stages
[23:05] <Laurenceb> and subsequently gets very heavy
[23:08] <Laurenceb> still the use of off the shelf parts for pretty much everything is very impressive
[23:11] RocketBoy (n=Steve@217.47.75.8) left irc: "Leaving"
[23:23] natrium42 (n=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left #highaltitude.
[23:25] borism_ (n=boris@195-50-212-148-dsl.krw.estpak.ee) got netsplit.
[23:25] Tiger^ (i=tygrys@217.149.240.132) got netsplit.
[23:25] Tiger^ (i=tygrys@moo.pl) joined #highaltitude.
[23:35] GeekShadow (n=Antoine@reactos/tester/GeekShadow) left irc: Connection timed out
[23:36] borism_ (n=boris@195-50-212-148-dsl.krw.estpak.ee) got lost in the net-split.
[23:39] fuzzylugnuts (n=hush@c-68-34-212-42.hsd1.tn.comcast.net) left irc: "oswald cobblepot"
[00:00] --- Thu Nov 5 2009