highaltitude.log.20070321

[00:03] LaurenceB (n=laurence@host86-133-69-166.range86-133.btcentralplus.com) left irc:
[00:39] edwardmoore_ (n=edwardmo@88.212.167.121) left irc: "toodle-pip"
[01:03] icez (n=icez@ip68-3-56-121.ph.ph.cox.net) joined #highaltitude.
[06:02] VA3NEE (n=Snoopl3s@209.105.206.171) left irc: Read error: 110 (Connection timed out)
[06:54] icez (n=icez@ip68-3-56-121.ph.ph.cox.net) left irc: Remote closed the connection
[09:34] VA3NEE (n=Snoopl3s@209.105.206.171) joined #highaltitude.
[10:30] defy (n=defy@60-234-174-220.bitstream.orcon.net.nz) left irc: Read error: 110 (Connection timed out)
[14:00] icez (n=icez@ip68-3-56-121.ph.ph.cox.net) joined #highaltitude.
[14:05] edwardmoore (n=edwardmo@88.212.167.121) joined #highaltitude.
[14:06] edwardmoore (n=edwardmo@88.212.167.121) left irc: Client Quit
[14:54] icez (n=icez@ip68-3-56-121.ph.ph.cox.net) left irc: Remote closed the connection
[16:16] LaurenceB (n=laurence@host86-145-126-177.range86-145.btcentralplus.com) joined #highaltitude.
[16:22] edwardmoore (n=edwardmo@88.212.167.121) joined #highaltitude.
[16:25] flowolf (n=flowolf@87.18.219.24) joined #highaltitude.
[16:37] <LaurenceB> Hi all
[16:37] <LaurenceB> Can't stay for long... Not much wind today=glider tests
[16:38] <edwardmoore> nice
[16:38] <edwardmoore> it's too cold though
[16:38] <LaurenceB> tried last night and smashed the tail off :-(
[16:38] <LaurenceB> have fixed it with a byro tube
[16:38] <edwardmoore> lol
[16:39] <LaurenceB> carbon fiber is not that strong... or at least the stuff in cheap planes
[16:40] <LaurenceB> I've been thinking, really need to know more about glider behaviour...
[16:40] <LaurenceB> thinking of an unguided rogallo test using the electronics off MiHAB3
[16:41] flowolf (n=flowolf@87.18.219.24) left irc: "Leaving"
[16:41] <LaurenceB> could aim for a target of 2 1/2 weeks to "flight readyness"
[16:41] <edwardmoore> that would be useful probably
[16:41] <edwardmoore> once you can characterise the response you're laughing
[16:42] <LaurenceB> would you be interested in a flight some time starting in 2 1/2 weeks?
[16:42] <LaurenceB> i.e. weekend after next
[16:43] <edwardmoore> that's the weekend after easter weekend isn't it?
[16:43] <LaurenceB> other payload(s) could be put on above the glider
[16:43] <LaurenceB> erm doh when is easter
[16:43] <edwardmoore> quite possibly, but that's really very far ahead!
[16:44] <LaurenceB> yes, I'll have to have tim to get something good built
[16:44] <LaurenceB> what it have an SD card with FAT16 formatting
[16:44] <edwardmoore> ?
[16:44] <LaurenceB> -to have
[16:45] <edwardmoore> talking to the avr?
[16:45] <LaurenceB> I've got libraries to taking care of everything, basically theres a print to file command
[16:46] <LaurenceB> well avr writes onto card
[16:46] <edwardmoore> oh cool, as long as u have the libraries
[16:46] <LaurenceB> will also get a UV filter for the camera
[16:46] <edwardmoore> otherwise it'd be a thankless task
[16:47] <LaurenceB> yes you can download them I think, looking for site.....
[16:47] <LaurenceB> http://members.aon.at/voegel/
[16:48] <LaurenceB> ooh it supports FAT32 as well didnt know that
[16:49] <edwardmoore> very smart
[16:49] <LaurenceB> video would be very useful for examining the response as well, probably have 12 second clips
[16:51] <LaurenceB> The main code will take a little bit of work, all the functions are a copy and paste job, but I'll have it work on an altitude, not time basis
[16:52] <LaurenceB> ie with every 100M ascent, it will take a photo or a video clip
[16:52] <LaurenceB> and do all the other stuff i.e. radio, txt, write to flight log
[16:55] <edwardmoore> i think just measuring position and attitude at a fairly high sample rate through a turning manouvre would be useful
[16:55] <LaurenceB> yes, I'll probably have to use the 1hz gps, my hz sems a bit glitchy
[16:56] <LaurenceB> -4hz
[16:57] <LaurenceB> might set it to do that in bursts of 20 seconds or so
[16:59] <LaurenceB> do you know when pegasus 5 launch is planned?
[17:11] <LaurenceB> I'll take that as a no... Anyway I'd better be going
[17:12] LaurenceB (n=laurence@host86-145-126-177.range86-145.btcentralplus.com) left irc:
[17:21] jcoxon (n=jcoxon@212.44.26.183) joined #highaltitude.
[17:52] jcoxon (n=jcoxon@212.44.26.183) left irc: Operation timed out
[18:23] flowolf (n=flowolf@87.18.219.24) joined #highaltitude.
[18:26] flowolf (n=flowolf@87.18.219.24) left irc: Client Quit
[18:28] icez (n=icez@ip68-3-56-121.ph.ph.cox.net) joined #highaltitude.
[18:30] flowolf (n=flowolf@host81-187-static.13-87-r.retail.telecomitalia.it) joined #highaltitude.
[18:51] flowolf (n=flowolf@host81-187-static.13-87-r.retail.telecomitalia.it) left irc: "Leaving"
[19:03] jcoxon (n=jcoxon@212.44.26.183) joined #highaltitude.
[19:03] <edwardmoore> hi jcoxon
[19:08] <jcoxon> hey
[19:12] <jcoxon> did rocketboy speak to you about friday instead of sat?
[19:22] whiteknight (n=jcoxon@212.44.26.183) joined #highaltitude.
[19:39] jcoxon (n=jcoxon@212.44.26.183) left irc: Read error: 110 (Connection timed out)
[19:44] whiteknight (n=jcoxon@212.44.26.183) left irc: Read error: 110 (Connection timed out)
[19:46] rocketboy (n=steve@host81-154-222-228.range81-154.btcentralplus.com) joined #highaltitude.
[19:49] jcoxon (n=jcoxon@212.44.26.183) joined #highaltitude.
[19:49] <jcoxon> stupid library internet
[19:51] <rocketboy> K?
[19:53] <jcoxon> keeps disconnecting me
[20:04] jcoxon (n=jcoxon@212.44.26.183) left irc: "Leaving"
[20:28] LaurenceB (n=laurence@host86-145-126-177.range86-145.btcentralplus.com) joined #highaltitude.
[20:28] <LaurenceB> Hi all
[20:28] <LaurenceB> been testing my glider, flew well, locked onto target
[20:29] <LaurenceB> but oscillated quite a bit, looking at some nice graphs now...
[20:32] <LaurenceB> thinking it may need a derivative term in the feedback loop
[21:07] <rocketboy> Suggest you implement the whole PID thing and fiddle with the parametrs until its fast enought without oscillation
[21:14] <LaurenceB> hmmm yes the graphs are extremely interesting
[21:14] <LaurenceB> when I reduced the rudder size, the period of oscillation increases
[21:16] <rocketboy> humm - it takes longer to get the same turning effect with a smaller rudder
[21:17] <LaurenceB> but the amplitude of the oscillations still increases with each one
[21:17] <LaurenceB> by a factor of around 2 to 2.5
[21:18] flowolf (n=flowolf@host81-187-static.13-87-r.retail.telecomitalia.it) joined #highaltitude.
[21:18] <rocketboy> so thats quite an unstable system
[21:18] <LaurenceB> your not kidding!
[21:19] <rocketboy> and thats without any I or D ?
[21:20] <LaurenceB> with I and P
[21:20] <LaurenceB> it limited how long the flights could be as it doubled back on itself into the hillside
[21:23] <LaurenceB> lost sight of the glider on one test, thought I had lost it over a housing estate
[21:23] <LaurenceB> found it in the top of a tree :P
[21:23] <rocketboy> Did you try taking out the I altogether
[21:23] <LaurenceB> I was small
[21:24] <LaurenceB> and there were limits on it, it could only shift servo by +-10 degrees
[21:25] <LaurenceB> p=0.7 for servo angle as function of offset from heading
[21:25] <LaurenceB> brb food..
[21:25] <rocketboy> ok - BBL
[21:38] <LaurenceB> I think that a D term may help solve the problem, especially as there is arond a 1 second latency with the gps
[21:39] <LaurenceB> an entire oscillation has a period of 12 seconds with the smaller rudder
[21:39] <LaurenceB> and about 10 with the larger one
[21:54] <LaurenceB> okay graphs up on wiki page
[22:07] jcoxon (n=jcoxon@jac208.caths.cam.ac.uk) joined #highaltitude.
[22:10] <LaurenceB> Hi
[22:10] <jcoxon> hey LaurenceB
[22:17] <LaurenceB> check the wiki page for my glider:)
[22:19] <jcoxon> cool
[22:19] <jcoxon> bbl (sorry but am being called away
[22:19] jcoxon (n=jcoxon@jac208.caths.cam.ac.uk) left irc: "Leaving"
[22:30] <LaurenceB> hmh33 - lol your electrical connectors guide is so true!
[22:35] <rocketboy> thats a fine sine wave oscillator you have developed laurence
[22:41] <rocketboy> I would say that the I + the lag of control and the body system must be adding up to the 10/12 seconds
[22:43] <rocketboy> although that seems a lot to me
[22:44] <LaurenceB> hmm looking on wikipedia at the Ziegler-Nichols method
[22:45] <LaurenceB> apparently there should be a critical value of P below which it wont oscillate
[22:45] <rocketboy> sure P = gain
[22:46] <LaurenceB> but thats not quite what I observed, when P was effectivley reduced by reducing rudder size
[22:46] <rocketboy> have to have at least unity gain to oscillate
[22:46] <LaurenceB> there period increased
[22:46] <LaurenceB> and the stability seemed unchanged
[22:46] <rocketboy> runnder size is part of the overall gain
[22:48] <rocketboy> its posible that the rudder size changed the mechanical system dynamics - e.g. its weight ... moment of inertia
[22:48] <LaurenceB> yes, think I'll increase the rudder size and follow the Ziegler-Nichols method
[22:48] <LaurenceB> dont think so, it was just a small piece of plastic
[22:49] <rocketboy> humm - it shold only affect the gain then
[22:50] <rocketboy> - what about its increased size giving the servo more load - and therefore taking longer to get to its proper position?
[22:50] <LaurenceB> well looking at the Ziegler-Nichols equations and putting in some rough figures from my tests it seems credible
[22:50] <LaurenceB> good point, the servo is a cheap 4gram one
[22:51] <LaurenceB> glued straight onto the tail
[22:51] <rocketboy> really there needs to be a change in lag to explain the frequency differences
[22:51] <LaurenceB> but I dont know the critical value of P for the Ziegler-Nichols
[22:52] <LaurenceB> hmm the gps dehaviour is quite complex to say the least
[22:52] <LaurenceB> -behaviour
[22:52] <LaurenceB> http://en.wikipedia.org/wiki/PID_controller
[22:53] flowolf (n=flowolf@host81-187-static.13-87-r.retail.telecomitalia.it) left irc: "Leaving"
[22:56] <rocketboy> Humm - in my day it was all about measuring step responce and comming up with its inverse IIR
[22:56] <LaurenceB> sorry thats over my head lol
[22:57] <LaurenceB> oh you feed it with a step function?
[22:58] <edwardmoore> stick a gyro in laurence :p
[22:58] <rocketboy> yep and measure it - you can get the gain and various time constants from the measurements
[22:59] <LaurenceB> hmm yes it would be good for tests
[23:00] <LaurenceB> but gyros wander all over the place with temperature so for a high altitude one you would need to correct it with the gps
[23:00] <LaurenceB> which is a right pain to say the least
[23:01] <rocketboy> Then I assume you put P and the I and D time constants into the pid and bingo an optimised response
[23:02] <LaurenceB> yes according to the Ziegler-Nichols method then if youve got the oscillation period and critical P you can derive them
[23:03] <LaurenceB> but i'm not even sure if there is a critical P
[23:04] <LaurenceB> and if there is its going to be small and the tests will be tricky due to limited glide time
[23:04] <rocketboy> control systesm was one of the more complex engineering subjects in my uni days - but it seems that PIDs and the various rule based tuning methods make it much simpler now (you lucky guys)
[23:05] <LaurenceB> yes the tuning techniques on wikipedia dont seem too hard
[23:06] <LaurenceB> the rudder size was reduced two times, firstly halved then down to approx 66% of the previous
[23:07] <rocketboy> it would be interesting to see if compensating with P whould return the system to the same response for different rudder sizes
[23:08] <LaurenceB> but only run F was done at the intermediate rudder size, and it was not thrown very well so is not vey useful
[23:08] <LaurenceB> yes good point
[23:09] <LaurenceB> it could be that at the larger rudder size, the torque on the rudder effected the response
[23:10] <rocketboy> yep - i think whay I proposed would prove that true/false
[23:10] <LaurenceB> but the effects is the wrong way round, as larger size = shorter period
[23:10] <rocketboy> way = what
[23:10] <rocketboy> whay = what
[23:10] <LaurenceB> but looking at a simple equation for a spring, if the spring is stronger it oscillates faster
[23:11] <LaurenceB> the P is like the spring constant
[23:11] <rocketboy> no don't think so
[23:12] <LaurenceB> which begs the question what is the period you use in the Ziegler-Nichols method
[23:12] <rocketboy> you meed a mass ans a spring to oscillate
[23:12] <LaurenceB> mass= moment of intertia spring = rudder
[23:14] <LaurenceB> so a larger tail would effect things by making the moment of inertia less important
[23:14] <rocketboy> - the spring stiffness ("stength" in your words) anmd mass (iniertia) sets the oscillation period
[23:14] <edwardmoore> I can never think of a good analogy for the I term when you're talking about mass spring systems
[23:16] <LaurenceB> the I term was 0.015 per second so insignificant over 10 second or so period for values for 45 degrees or so
[23:16] <LaurenceB> for=of
[23:17] <LaurenceB> having said that not completely insignificant
[23:18] <rocketboy> hey ho - I'll dig out my old control systems notes if I can find them
[23:20] <edwardmoore> is it silly of me to suggest you whack in a gyro, set up a PID, and tune it the old fashioned way?
[23:21] <LaurenceB> I'll need a gyro
[23:22] <edwardmoore> aye, that you will
[23:22] <edwardmoore> but you'll have a muchly better dynamic response
[23:22] <LaurenceB> I reacon I can get it working with a bit more data
[23:23] <edwardmoore> say.... angular velocity? :p
[23:23] <rocketboy> I you want to borrow a gyro chip I have one (MC gave it to me)
[23:23] <LaurenceB> that would be good thanx
[23:24] <rocketboy> OK will look it out tomorrow - I have to go now - CU guys
[23:24] <LaurenceB> cu
[23:25] <LaurenceB> I'd better be off as well, hopefully do some more tests tomorrow
[23:25] <rocketboy> cu
[23:25] rocketboy (n=steve@host81-154-222-228.range81-154.btcentralplus.com) left irc: "Leaving"
[23:25] LaurenceB (n=laurence@host86-145-126-177.range86-145.btcentralplus.com) left irc:
[23:25] edwardmoore (n=edwardmo@88.212.167.121) left irc: "toodle-pip"
[00:00] --- Thu Mar 22 2007