highaltitude.log.20090125

[00:32] <Laurenceb> hmm theres loads of noise from my tft
[00:33] <Laurenceb> moving windows around chucks out tons of stuff around 434MHz
[00:35] <SpeedEvil> wireless mouse?
[00:35] <Laurenceb> no its definately the mointor
[00:35] <Laurenceb> the traces must be acting as antennas
[00:36] <SpeedEvil> the monitor will be repainting at 60Hz anyway
[00:36] <SpeedEvil> if it's an LCD or CRT
[00:36] <SpeedEvil> sure it's not the CPU?
[00:36] <Laurenceb> yeah, but the freq distribution changes
[00:36] <Laurenceb> depending where the windows are
[00:36] <Laurenceb> hmmm
[00:36] Action: Laurenceb thinks remote tft snooper
[00:36] <SpeedEvil> been done
[00:37] <Laurenceb> for crt
[00:37] <SpeedEvil> LVDS radiates quite a lot
[00:37] <Laurenceb> tft as well?
[00:37] <SpeedEvil> yes
[00:37] <Laurenceb> nice
[00:37] <Laurenceb> yeah, moving my hand in front of the screen blocks some of it
[00:38] <hallam> hey
[00:38] <Laurenceb> hi hallam
[00:38] <hallam> Laurenceb: does your C C/A code generator definitely work?
[00:38] <Laurenceb> hmm
[00:38] <Laurenceb> I think I got acquisition working
[00:38] <hallam> we are having issues
[00:38] <Laurenceb> check the acquisition first
[00:39] <hallam> doing acq in matlab
[00:39] <Laurenceb> yeah, compile the c acquisition code and try
[00:39] <hallam> but we just compared the result of your C/A code generator to the matlab one
[00:39] <hallam> and they're different
[00:39] <Laurenceb> oh yeah
[00:39] <Laurenceb> sorry
[00:39] <Laurenceb> woops I should have told you
[00:39] <Laurenceb> sorry if you wasted ages - for some reason the outputs are off
[00:40] <hallam> haven't wasted too long
[00:40] <Laurenceb> so for prn 2 you need to tell it prn 3
[00:40] <Laurenceb> dont know why
[00:40] <hallam> it's just +1?
[00:40] <Laurenceb> no
[00:40] <Laurenceb> I already fixed the change in array index
[00:40] <Laurenceb> from matlab to c
[00:40] <hallam> right but
[00:40] <hallam> for sv 11 do I ask it for sv 12?
[00:40] <Laurenceb> so I doint know why its doing that
[00:40] <Laurenceb> yeah
[00:40] <hallam> ok I'll try that
[00:41] <Laurenceb> I'm just setting up a test of the radio board v2
[00:41] <Laurenceb> but without an audio cable I cant get fldigi working on it
[00:41] <hallam> still doesn't seem right
[00:42] <Laurenceb> how so?
[00:42] <hallam> e.g.
[00:43] <hallam> for SV 11
[00:43] <hallam> I call the C code with parameter 12
[00:43] <hallam> it outputs 11111101......
[00:43] <hallam> matlab gives 1 1 1 0 1 0 0 0
[00:44] <Laurenceb> hmm
[00:44] <Laurenceb> maybe try 10
[00:45] <hallam> ok, that did the trick
[00:45] <hallam> weird eh
[00:47] <Laurenceb> hmf - I forgot the pinout of my usb - spi,i2c,serial ect board
[00:47] <Laurenceb> and the manufacturer has a new version now :-/
[00:51] <SpeedEvil> mail them?
[00:52] <Laurenceb> think I have it store on my windows partition
[00:52] <Laurenceb> I can work out the serial lines from the breadboard setup it was on
[00:52] <Laurenceb> got it working now
[00:52] <Laurenceb> tune up pulses :P
[00:52] <SpeedEvil> :)
[00:52] <Laurenceb> loud and clear
[00:52] <Laurenceb> well its like 10cm from the icom
[00:53] <Laurenceb> really need an audio cable to get any further
[00:54] <Laurenceb> hallam: whats your antenna setup?
[00:54] <Laurenceb> hmm theres a drunken fight outside my room
[00:54] <Laurenceb> the joys of student life
[00:55] <SpeedEvil> As long as they don't pull out AK47s.
[00:55] <Laurenceb> beeeeeeeeeebaaaaaaaaaababadbdbdbdbadbadbadbabdbbdbadbbaabbadbdag hisssssssssssss
[01:00] <Laurenceb> grig is working very nicely
[01:00] <Laurenceb> maybe I just need to learn to set it up, but I've had a lot more sucess with grig than fldigi for controlling the icom
[01:06] <Laurenceb> I've forgotten the name of the serial terminal in ubuntu
[01:06] <Laurenceb> anyone?
[01:07] <gordonjcp> gtkterm
[01:07] <Laurenceb> mincom
[01:07] <Laurenceb> I remember
[01:08] <gordonjcp> gtkterm is a nice gui app
[01:09] <gordonjcp> and its configuration is a lot less fail
[01:10] <Laurenceb> ok yeah
[01:11] <Laurenceb> it was a bit annoying, I'm installing gtkterm
[01:11] <gordonjcp> gtkterm is not without its problems
[01:11] <gordonjcp> but much as I am fond of minicom, it's only slightly better fun than hantavirus
[01:13] <Laurenceb> hmm better double check the pinouts here
[01:17] <Laurenceb> working
[01:18] <Laurenceb> sounds like rtty :P
[01:18] <gordonjcp> good show
[01:18] <Laurenceb> I can relay irc over 434 MHz :P
[01:18] <Laurenceb> have to wait for an audio cable to check it properly with fldigi
[01:24] <Laurenceb> this is very cool
[01:24] <gordonjcp> nice
[01:24] <Laurenceb> grig works fine
[01:25] <Laurenceb> is it ok to turn off the prc1000 with rs232 still going in?
[01:26] <SpeedEvil> Generally, all RS232 devices are designed to be pretty bulletproof
[01:27] <Laurenceb> well I unplugged rs232 first
[01:47] <Laurenceb> gordonjcp: can I pick up satellite comms with the pcr-1000 ?
[01:50] <gordonjcp> depends
[01:50] <gordonjcp> most of the downlinks are on 70cm, so if you can tune that you should be okay
[02:07] jiffe20 (n=jiffe@209.159.247.189) joined #highaltitude.
[02:09] <Laurenceb> hallam: http://www.kaztek.com/datasheets/dsmSTSL.pdf
[05:06] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-cef973153232afd3) left irc: "http://www.mibbit.com ajax IRC Client"
[05:07] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-f55956a521b1d1b2) joined #highaltitude.
[05:17] iNatrium (i=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[05:24] iNatrium (i=natrium4@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left irc: "Rooms " iPhone IRC Client " http://rooms.derflash.de"
[07:11] akawaka (n=akawaka@66-215-97-198.dhcp.malb.ca.charter.com) left irc: Read error: 110 (Connection timed out)
[07:38] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-f55956a521b1d1b2) left irc: "http://www.mibbit.com ajax IRC Client"
[09:12] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[10:15] rjharrison (n=rharriso@80.176.172.227) joined #highaltitude.
[12:30] <EI5GTB> :O "{}" anything but {}
[12:43] rjharrison (n=rharriso@80.176.172.227) left irc:
[14:51] <Laurenceb> natrium42: used fldigi?
[14:55] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: Read error: 104 (Connection reset by peer)
[14:56] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[15:12] <Laurenceb> nvm, got it working :P
[15:12] <Laurenceb> unfortunately my sound hardware is screwy
[15:12] <Laurenceb> my laptop has onboard sound, and I think it uses an RC oscillator, as the RX offset has to be -20000ppm
[15:16] <SpeedEvil> Or another random clock that's almost right from somewhere onboard
[15:23] <Laurenceb> it appears to change
[15:23] <Laurenceb> I had the same problem with the v1 radio board
[15:23] <Laurenceb> but I thought it may be related to the radio/scanner
[15:24] <Laurenceb> guess I'd better get a usb soundcard
[15:25] <Laurenceb> the tracking loops in fldigi rely on the clock being accurate to a few hundered ppm
[15:25] <Laurenceb> I've tried to tune it by hand, but theres still occasional glitches that shouldnt be there
[15:26] <Laurenceb> maybe those are actual sound card clock glitches
[15:26] <Laurenceb> I'm shoving large text files through the link
[15:33] jcoxon_ (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) joined #highaltitude.
[15:33] <SpeedEvil> dealextream - $3.49
[15:34] Action: SpeedEvil is planning on ordering a few
[15:34] <SpeedEvil> extreme
[15:37] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[15:40] <Laurenceb> SpeedEvil: how long for delivery?
[15:42] <SpeedEvil> 2-3 weeks typ
[15:43] <SpeedEvil> however, ebuyer have them cheap too - not as cheap
[15:43] <SpeedEvil> and next day
[15:43] <SpeedEvil> though not next day today of course
[15:44] <SpeedEvil> DX can be dangerous though.
[15:44] <SpeedEvil> 'Oooh - solder paste - only $3
[15:44] <jcoxon_> Laurenceb, i use a usb soundcard on my old ibook
[15:44] <SpeedEvil> Oooh - nice LED torch - only $10
[15:44] <SpeedEvil> Oooh - comedy chicken monitor surround, only $4
[15:44] <Laurenceb> lol
[15:44] <SpeedEvil> Pretty soon you're up to $10000
[15:44] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc:
[15:49] <Laurenceb> http://www.ebuyer.com/product/106540
[15:49] <Laurenceb> slight overkill
[15:54] Xenion (n=robert@p579FC18C.dip.t-dialin.net) joined #highaltitude.
[15:54] <Laurenceb> http://www.currys.co.uk/martprd/store/cur_page.jsp?BV_SessionID=@@@@0314129103.1232898682@@@@&BV_EngineID=ccceadegfhfdhehcflgceggdhhmdfhm.0&page=Product&fm=1&sm=2&tm=undefined&sku=488071&category_oid=-35758#productInformationSection
[15:55] <Xenion> Guten Abend / Good evening :-)
[15:59] <Laurenceb> hi
[15:59] <Laurenceb> theres one with a SSD for £175
[16:05] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[16:08] <jcoxon_> hey edmoore
[16:08] <edmoore> hi
[16:14] <jcoxon_> hows tricks?
[16:15] <edmoore> just talking to fergusnoble on the repeater
[16:16] <jcoxon_> :-)
[16:19] <Laurenceb> http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=290285889019&cguid=038ee03311d0a0e203641a84fffc749b
[16:19] <Laurenceb> ^ is something like that going to work with ubuntu?
[16:23] <Xenion> (Mac version has 2 channel playback only) @ Laurenceb
[16:23] <Xenion> i guess the same limitation imply for unix/linux as well
[16:23] <Laurenceb> I see
[16:38] <Laurenceb> http://pastebin.com/m30734f83
[16:38] <Laurenceb> thats some random c code I sent through the link
[16:39] <Laurenceb> seems to work very well, but I'm not happy about using the onboard sound
[16:54] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-a0e6a0fbb47ae0eb) joined #highaltitude.
[16:54] <fergusnoble> hallam: hello
[16:54] <fergusnoble> hallam: any progress?
[16:56] <fergusnoble> saw you put the code up on the SVN
[16:56] <Laurenceb> hey hallam
[16:56] <Laurenceb> good "morning"
[16:57] <Laurenceb> hallam: I thought of an issue with the blackfin approach - we cant dump data samples in the same way
[16:58] <Laurenceb> as they are grouped together into 32 bit words
[16:58] <Laurenceb> so you have to have 16 seperate PRN tables, with differing phase
[16:59] <Laurenceb> then the DLL swaps between lookups, if it wants to go from lookup 0 to 15 or 15 to 0 then you do the dump business
[16:59] <Laurenceb> its 32KB of lookups in total
[17:00] <fergusnoble> brb
[17:03] <Laurenceb> fergusnoble: can you recommend an easy way to create a GUI with python?
[17:05] <fergusnoble> Laurenceb: wxpython is good
[17:05] <fergusnoble> Laurenceb: i have a script that wraps it to make it even easier too
[17:06] <Laurenceb> what about tkinter?
[17:07] <fergusnoble> not used it for anything proper
[17:08] <fergusnoble> i think its quite basic and doesnt use native widgets
[17:08] <fergusnoble> and more and more people are moving to wx
[17:08] <fergusnoble> not exactly objective but:
[17:08] <fergusnoble> http://wiki.wxpython.org/Choosing%20wxPython%20over%20Tkinter
[17:09] <Laurenceb> I see
[17:10] <gordonjcp> heh
[17:10] <gordonjcp> that's not really a great decision
[17:10] <fergusnoble> so i made a wrapper for wxpython where you give it a resource file and it draws to
[17:10] <fergusnoble> sorry, takes care of everything
[17:11] <gordonjcp> that's like "Choosing a Morris Minor over an Aveling-Barford traction engine"
[17:11] <fergusnoble> and lets you acces objects on your windows using a more simple syntax
[17:11] <fergusnoble> gordonjcp: enlighten us, whats the better choice?
[17:11] <gordonjcp> pygtk
[17:11] <fergusnoble> gordonjcp: not so portable
[17:12] <gordonjcp> neither is wx
[17:12] <gordonjcp> wx is fundamentally broken in different ways on OSX and Linux, Windows I haven't tried
[17:12] <Laurenceb> lol
[17:12] <Laurenceb> flamewar
[17:13] <Laurenceb> so, I've never played with guis apart from with visual studio
[17:13] <Laurenceb> and I want to make some simple stuff with python on linux
[17:13] <Laurenceb> wheres the best place to start?
[17:13] <gordonjcp> pygtk and glade ;-)
[17:14] <gordonjcp> well - if you're using a GTKish environment
[17:14] <fergusnoble> Laurenceb: ive not done much gui stuff either, can only tell you about my experience learning wx, which was pretty smooth
[17:14] <Laurenceb> ok
[17:14] <fergusnoble> Laurenceb: picked it up in a couple of hours
[17:14] <Laurenceb> know of a good tutorial?
[17:15] <fergusnoble> cant remember what i used, but there is good stuff out there
[17:17] <fergusnoble> Laurenceb: http://pastebin.com/m4e6f90c3
[17:19] <fergusnoble> just make a class for your app that inherits SimpleApp and pass it an XRC file of your window (can be made with editors like XRCed
[17:20] <edmoore> just heard ISS repeater
[17:20] <fergusnoble> edmoore: what freq?
[17:20] <edmoore> 145.800
[17:20] jcoxon_ (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) left irc: "This computer has gone to sleep"
[17:21] <edmoore> gone now
[17:21] <Laurenceb> what ant are you using?
[17:21] <fergusnoble> Laurenceb: then you can access widgets on the window like self.controls.nameofcontrol.stuff
[17:21] <Laurenceb> right
[17:22] <Laurenceb> fraid I dont 100% follow, but thanks for the code
[17:22] <Laurenceb> I'll read through a tutorial that goes through step by step
[17:22] <fergusnoble> e.g. self.controls.label1.SetLabel("some text")
[17:23] <fergusnoble> Laurenceb: i do recommend using XRC files rather than defining the look of the window in code
[17:23] <Laurenceb> ok
[17:24] <Laurenceb> where do I find XRCed?
[17:24] <fergusnoble> also you can add event handlers like self.ctonrolls.button.Bind(wx.EVT_BUTTON, function)
[17:24] <fergusnoble> and that makes it run "function" when the button is pressed
[17:24] <Laurenceb> nice
[17:25] <hallam> hey all
[17:25] <Laurenceb> I cant find XRCed in the repository
[17:25] <Laurenceb> hi hallam
[17:25] <hallam> Laurenceb: Fergus and I noticed that problem with them being in chunks of 16 samples
[17:25] <hallam> but he solved it with two shift operations
[17:26] <Laurenceb> interesting
[17:26] <Laurenceb> you have a one word buffer?
[17:26] <hallam> yeah
[17:26] <hallam> http://www.mibbit.com/pb/fQnhYZ
[17:26] <Laurenceb> ok
[17:26] <fergusnoble> edmoore: do you think this might be a point to introduce ourselves on the repeater?
[17:26] <Laurenceb> hallam: nice
[17:27] <Laurenceb> so are you now at the point of having a c struct that simulates the behaviour on the blackfin?
[17:27] <hallam> getting there
[17:27] <Laurenceb> cool
[17:27] <hallam> about to implement the DLL and PLL
[17:27] <edmoore> fergusnoble: possibly
[17:27] <edmoore> am with sam currently
[17:27] <fergusnoble> Laurenceb: XRCed should be included in the wxwidgets package
[17:27] <Laurenceb> yeah I was thinking the lookup would save time
[17:28] <Laurenceb> but its ver minimal compared to that operation
[17:28] <Laurenceb> you only have to setup the lookup table once per PRN code iteration
[17:29] <fergusnoble> hallam: did yo get the interleaved carrier mixing working?
[17:29] <Laurenceb> but yeah % wise a few shifts add little to the overall cpu clocks
[17:30] <fergusnoble> Laurenceb: the blackfin has a nice one cycle shift by n operation
[17:30] <fergusnoble> which helps
[17:30] <Laurenceb> this is very cool - my idea of an fpga and soft core with imu and gps could all be done on a blackfin
[17:30] <fergusnoble> hopefully
[17:31] <Laurenceb> its a lot of work, but looks doable
[17:31] <Laurenceb> at least we have working matlab code to go on
[17:32] <fergusnoble> Laurenceb: actually, i forgot but with the wrapper i sent you its even easier to add event handlers
[17:32] <fergusnoble> you can just do self.controls.button.OnClick = some_function
[17:32] <Laurenceb> nice
[17:33] <Laurenceb> thats just the code you linked on pastebin?
[17:33] <fergusnoble> but currently only OnClick and OnPaint are implemented that way
[17:33] <fergusnoble> Laurenceb: yes
[17:33] <Laurenceb> so how do I use that code?
[17:33] <fergusnoble> Laurenceb: one sec, ill knock up a demo app
[17:33] <Laurenceb> thanks, sorry to be slow
[17:34] <hallam> fergusnoble: yes
[17:34] <hallam> fergusnoble: at least I think so
[17:35] <Laurenceb> nice
[17:35] <hallam> still generating the carrier the slow way
[17:35] <Laurenceb> you have bins for real and imaginary filling up ?
[17:35] <hallam> yeah
[17:35] <hallam> and they look vaguely plausible, though it's very hard to tell without the loop filters working
[17:35] <Laurenceb> did you see what I was saying yesterday?
[17:36] <Laurenceb> take a u16 for lo phase
[17:36] <hallam> I like the way Fergus suggested
[17:36] <Laurenceb> then use the 2msb to get a 32 bits words from lookup tables
[17:36] <hallam> which is to have a counter for "
[17:36] <Laurenceb> well 4 values
[17:37] <hallam> "samples remaining until transition"
[17:37] <hallam> I think it works better than the inverse at low frequencies
[17:37] <hallam> and on the rocket we'll have actual 0 IF
[17:37] <Laurenceb> hmmmm
[17:37] <Laurenceb> if you have 0IF
[17:37] <Laurenceb> you need to be able to have a -ive carrier
[17:38] <hallam> hm, that's true
[17:38] <Laurenceb> I like th eu16 idea
[17:38] <hallam> which is hard to do with samples_remaining
[17:38] <Laurenceb> erm u16
[17:38] <Laurenceb> you can add a u16 and s16 in a single cycle
[17:38] <Laurenceb> then shift to get the 2msb
[17:39] <Laurenceb> and get the u32 to xor with the data
[17:39] <hallam> so your carrier would never include a transition within a chunk?
[17:39] <Laurenceb> add 4 and get the seconds u32
[17:39] <Laurenceb> no
[17:39] <Laurenceb> transitions at chunk boundaries
[17:39] <hallam> yeah
[17:39] <hallam> I don't know if that's losing too much
[17:40] <Laurenceb> its hardly significant I think
[17:40] <hallam> ok
[17:40] <Laurenceb> well... theres one way to find out
[17:40] <Laurenceb> it will add some noise
[17:40] <hallam> btw I did a little check and the matlab tracker is keeping phase lock to within about 10 degrees of the carrier wave
[17:40] <Laurenceb> like a sort of time quantisation noise
[17:40] <Laurenceb> but it should be small
[17:40] <hallam> i.e. 6mm
[17:40] <Laurenceb> neat
[17:41] <Laurenceb> attitude time
[17:41] <Laurenceb> 4 front ends :P
[17:41] <hallam> 2's enough
[17:41] <Laurenceb> yeah, for a rocket
[17:41] <fergusnoble> Laurenceb: http://pastebin.com/m70117d4d
[17:42] <Laurenceb> thanks
[17:43] <fergusnoble> Laurenceb: also if you want to add more event handlers in the style of the .OnClick = function
[17:43] <Laurenceb> hmm
[17:43] <Laurenceb> right how do I get this running?
[17:44] <fergusnoble> its quite easy, in wxMeta.py, look at the function __setattr__ and add to the list event_handlers
[17:44] <fergusnoble> Laurenceb: hehe, i didnt test the code so it may have a bug
[17:44] <fergusnoble> make sure the first file i sent is called wxMeta.py and is in the same folder
[17:45] <Laurenceb> right
[17:45] <fergusnoble> to the event_handlers list you just need to add the name you want to use e.g. OnSomething and then the wxpython constant for that event handler, e.g. wx.EVT_SOMETHING
[17:46] <Laurenceb> for some reason whe I c&p from pastebin
[17:46] <Laurenceb> I get the line numbers and about 4 \n
[17:47] <Laurenceb> any chance you could drop it in another pastebin thingy?
[17:47] <fergusnoble> copy from the window below?
[17:47] <Laurenceb> ooh
[17:47] <fergusnoble> that has no line numbers
[17:47] <Laurenceb> durrr
[17:48] <Laurenceb> I always used to just select the text - dont know why its stopped working
[17:52] <Laurenceb> fergus: are there more graphicsy things like lines and canvases ?
[17:53] <Laurenceb> didnt you make an altitude indicator?
[17:55] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: Read error: 104 (Connection reset by peer)
[17:56] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[17:59] <edmoore> fergusnoble: ping
[18:00] <SpeedEvil> GPS gets lots easier if you don't care much about lock times, or accept uploaded orbits.
[18:06] Nick change: edmoore -> edmoore|away
[18:08] <fergusnoble> edmoore|away: hello
[18:09] akawaka (n=akawaka@66-215-97-198.dhcp.malb.ca.charter.com) joined #highaltitude.
[18:09] <fergusnoble> Laurenceb: yes, you can do all sorts
[18:10] <fergusnoble> to draw on a thing use the onpaint event handler
[18:10] <Laurenceb> I see
[18:10] <Laurenceb> I'm installing all the gumf at the moment
[18:10] <fergusnoble> put a wxPanel on your window i think is the best way, which is just a blank box
[18:11] <fergusnoble> and then add an onpaint handler to it
[18:11] <fergusnoble> ill add it to my example
[18:14] <fergusnoble> Laurenceb: thats an example of drawing http://pastebin.com/m5d0f084
[18:25] <Laurenceb> ok
[18:25] <Laurenceb> do I need an XRC file?
[18:27] <Laurenceb> fergusnoble: ping
[18:27] <fergusnoble> hi
[18:27] <fergusnoble> Laurenceb: yeah, im just making one to see if that code works :)
[18:27] rjharrison (n=rharriso@80.176.172.227) joined #highaltitude.
[18:27] <Laurenceb> what do I do for the XRC file?
[18:28] <fergusnoble> use an editor i would
[18:28] <Laurenceb> thanks, I cant seem to find XRCed
[18:28] <fergusnoble> is wxwidgets installed?
[18:28] <Laurenceb> I think so
[18:28] <fergusnoble> did you say you were on linux?
[18:28] <Laurenceb> that doesnt exist
[18:28] <Laurenceb> but I think its part of something else
[18:29] <Laurenceb> I'll have another search
[18:44] <Laurenceb> I dont believe it
[18:44] <Laurenceb> just spent all this time searching for the right libraries
[18:44] <Laurenceb> still no XRCed
[18:46] <fergusnoble> are you on linux?
[18:46] <Laurenceb> yes
[18:46] <Laurenceb> ubuntu
[18:47] <fergusnoble> ok, it looks like it actually comes with wxpython
[18:47] <fergusnoble> http://wiki.wxpython.org/index.cgi/UsingXmlResources
[18:47] <Laurenceb> I'm sure that was the first one I installed
[18:47] <fergusnoble> ok, odd
[18:48] <fergusnoble> for me on ubuntu i can just type xrced at the command line
[18:48] <Laurenceb> hmm
[18:48] <fergusnoble> oh, maybe you need something like wxpython-dev
[18:48] <Laurenceb> ok its not wxpython
[18:48] <fergusnoble> or wxwidgets-dev
[18:48] <Laurenceb> wx_python ?
[18:48] <Laurenceb> yeah I have wxwidgets-dev
[18:49] <Laurenceb> and wxpython-dev I think...
[18:49] <Laurenceb> no theres no such thing
[18:51] <fergusnoble> http://wiki.wxpython.org/InstallingOnUbuntuOrDebian
[18:51] <fergusnoble> looks like you need python-wxgtk2.8, python-wxtools, python-wxaddons, wx2.8-i18n
[18:52] <Laurenceb> I dont have the last one
[18:54] <fergusnoble> whould have thoght that xrced would be in python-wxtools
[18:54] <hallam> Having trouble getting the loop filters to work
[18:54] <hallam> I think it's the carrier
[18:55] <fergusnoble> Laurenceb: there is another xrc editor called wxglade which is supposed to be good, its proper drag and drop graphical style
[18:55] <fergusnoble> but ive not used it
[18:56] <Laurenceb> k
[18:56] <Laurenceb> hallam: remember 0->-1, 0.5->0, 1->1
[18:56] <Laurenceb> so the gains all need changing
[18:56] <Laurenceb> and the discriminator functions
[18:56] <hallam> well, you subtract something from all the correlations
[18:57] <hallam> then change the gain
[18:57] <hallam> is there more to it than that?
[18:57] <Laurenceb> yeah
[18:58] <fergusnoble> Laurenceb: i found some errors in my code, ill just post the updated versions
[18:59] <Laurenceb> still no XRCed
[19:00] <Laurenceb> omg
[19:00] <Laurenceb> xrced
[19:00] <Laurenceb> it works
[19:00] <Laurenceb> that was a waste of time :/
[19:02] <hallam> Laurenceb: any idea what else to change?
[19:02] <hallam> I've taken the DLL out of the equation by just precalculating an appropriate chuck rate with matlab
[19:03] <Laurenceb> hmmm
[19:03] <Laurenceb> can you setup a fairly accurate doppler to start with
[19:03] <fergusnoble> Laurenceb: new versions, and an xrc file that works with it http://www.pastebin.com/m2fbac11e
[19:03] <Laurenceb> then plot the output bins?
[19:04] <fergusnoble> http://pastebin.com/m2fbac11e
[19:05] <fergusnoble> Laurenceb: actually saw another error, i suck http://pastebin.com/m73806a29
[19:05] <fergusnoble> Laurenceb: note the syntax for adding event handlers, i was wrong about that before
[19:11] <hallam> http://i39.tinypic.com/fepfdv.png should be this
[19:12] <hallam> http://i39.tinypic.com/zus50j.png actually this
[19:12] <hallam> ignore the bottom right plot
[19:12] <hallam> brb
[19:13] <fergusnoble> bbl
[19:13] <fergusnoble> Laurenceb: hope that code helps
[19:13] <Laurenceb> hmm
[19:14] <Laurenceb> thanks, its screwing up, but I dont think I updated everything
[19:15] <Laurenceb> hallam: whats red and blue?
[19:16] <hallam> red=real(prompt correlation), blue = phase
[19:16] <SpeedEvil> Laurenceb: how does L2 code compare with L1?
[19:16] <hallam> actually blue = phase(i * prompt correlation)
[19:16] <SpeedEvil> Laurenceb: is it basically similar on different freq, or radically different?
[19:16] <hallam> SpeedEvil: it's longer
[19:16] <Laurenceb> SpeedEvil: its P code
[19:16] <hallam> 1 week long instead of 1ms
[19:16] <SpeedEvil> oh
[19:16] <hallam> and 10 times faster
[19:16] <Laurenceb> yeah
[19:17] <hallam> then it's encrypted
[19:17] <hallam> but the encryption is only applied at 500kHz
[19:17] <Laurenceb> you can track it to an extent using that feature
[19:18] <SpeedEvil> So you can lock using power detectors and the 20 long sequence?
[19:19] <hallam> there's a word in the C/A nav message that gives you an index into the P code
[19:19] <hallam> I think the P code is publically know, or maybe even published, just not the encryption
[19:20] Action: SpeedEvil idly wonders about SAW filters micromachined to pop out code-phase directly.
[19:20] <hallam> there's also P code on L1
[19:21] <hallam> you and your idle wonderings
[19:21] <SpeedEvil> You have a crystal, with micromachined bits that reflect inverted or non-inverted and then sum them.
[19:21] <Laurenceb> http://pastebin.com/m2b966419
[19:22] <SpeedEvil> _really_ low power GPS.
[19:22] <Laurenceb> fergusnoble ^
[19:22] <Laurenceb> SpeedEvil: still need a vco for the carrier
[19:23] <SpeedEvil> Laurenceb: I was meaning you get a 1us 1KHz pulse of 1.2GHz per satellite
[19:24] <SpeedEvil> Laurenceb: I was meaning you get a .5us 1KHz pulse of 1.5GHz per satellite
[19:24] <SpeedEvil> Instead of having to do the correlation - it's done in the crystal
[19:25] <SpeedEvil> per PRN
[19:25] <hallam> one crystal per prn?
[19:26] <SpeedEvil> Imagine the incoming RF as a water wave
[19:26] <SpeedEvil> Going from left-right
[19:26] <SpeedEvil> you have 1023 steps which either reflect part of this wave inverted, or non-inverted
[19:26] <SpeedEvil> back from right-left
[19:27] Hiena (n=Hiena@81.93.195.181.datatrans.hu) joined #highaltitude.
[19:27] <SpeedEvil> actually - this wouldn't work - it'd have to reflect them in another direction - so that they all come together in a crest
[19:28] <SpeedEvil> With 32 (or whatever) of these, so you have one micromachines SAW crystal per sat.
[19:29] <SpeedEvil> I was trying to come up with 'mechanical' ways of recieving GPS, to see if it gave any insight.
[19:29] <Laurenceb> yeah there are RFID tags like that
[19:30] <SpeedEvil> There are?
[19:30] <Laurenceb> fergusnoble: any ideas?
[19:30] <Laurenceb> yes, they have a unique prn
[19:30] <Laurenceb> laser etched on
[19:31] <SpeedEvil> neat
[19:32] <hallam> that's pretty cool
[19:32] <hallam> Laurenceb: any ideas? :)
[19:33] <Laurenceb> on the bugs
[19:33] <Laurenceb> in the gui code
[19:33] <hallam> I was asking you if you have any ideas on my PLL
[19:33] <hallam> I've fiddled with the gain quite a bit already
[19:33] <Laurenceb> I dont see the problem
[19:33] <Laurenceb> what happens?
[19:34] <hallam> well it just doesn't seem to lock
[19:34] Action: EI5GTB hands hallam a padlock
[19:35] <hallam> thanks :p
[19:35] <Laurenceb> what were those graphs of?
[19:35] <EI5GTB> np
[19:35] <hallam> http://i39.tinypic.com/fepfdv.png this is matlab
[19:35] <hallam> http://i39.tinypic.com/zus50j.png this is C
[19:35] <hallam> same sample data
[19:36] <hallam> DLL deactivated and replaced with a steady chuck rate
[19:36] <hallam> PLL gain adjusted so I think it's the same as the matlab
[19:37] <Laurenceb> theres something wrong
[19:37] jcoxon_ (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) joined #highaltitude.
[19:38] <Laurenceb> the PLL is changing its output way to fast
[19:38] <jcoxon_> evening all
[19:38] <Laurenceb> in the C
[19:38] <Laurenceb> hi jcoxon
[19:38] <hallam> Hi James
[19:38] <hallam> okay
[19:39] <hallam> Laurenceb: so that's not just a matter of the gain?
[19:39] <Laurenceb> and the initialisation of code phase is way different
[19:39] <jcoxon_> hey guys
[19:39] <jcoxon_> hows it going?
[19:39] <Laurenceb> hallam: yeah maybe
[19:39] <hallam> yeah ignore the code phase graph, I didn't transfer the data properly for it
[19:39] <Laurenceb> ok
[19:40] <Laurenceb> look at the functions in th ematlab code
[19:40] <Laurenceb> there is a gain coeficient in there
[19:40] <hallam> yeah, it translate directly to tau1
[19:40] <Laurenceb> I dont see why it should me a difference to scaling the discriminator
[19:40] <Laurenceb> but...
[19:40] <Laurenceb> *make
[19:41] <Laurenceb> fergusnoble: ping
[19:41] <Laurenceb> also check the code of course :P
[19:42] <Laurenceb> do you have seperate xor words
[19:42] <jcoxon_> hallam, henry, any ideas of your next launch?
[19:42] <jcoxon_> thats a when question really
[19:42] <Laurenceb> for the imaginary and complex bins?
[19:42] <Laurenceb> when the gps is ready :P
[19:43] <hallam> jcoxon_: not really I'm afraid. Hopefully not too long
[19:43] <hallam> Laurenceb: yes I think so
[19:43] <Laurenceb> and you inverted the phase of I
[19:43] <Laurenceb> but pi
[19:44] <jcoxon_> hallam, okay
[19:44] <Laurenceb> so i*i = -1
[19:44] <hallam> http://pastebin.com/d164f581b
[19:45] <jcoxon_> we should backup the last few days of logs concerning the development of gps units :-p
[19:45] <Laurenceb> grooo thats a bit convoluted
[19:45] <hallam> correlation I bin = data I xor carrier I + data Q xor carrier notQ
[19:45] <Laurenceb> hallam: I'm a bit knackered
[19:45] <Laurenceb> erm try it on paper
[19:45] <Laurenceb> :P
[19:45] <hallam> Q bin = data I xor carrier Q + data Q xor carrier I
[19:45] <hallam> I did
[19:46] <hallam> that's how I worked out how to make the carrier
[19:46] <Laurenceb> yes
[19:46] <Laurenceb> that looks spot on
[19:49] <jcoxon_> bbiab
[19:49] jcoxon_ (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) left irc: "Leaving"
[19:49] <Laurenceb> carrier_i = ( fmod(chan->carrier_phase + 1,4) < 2)?1:0;
[19:49] <Laurenceb> why the +1 ?
[19:49] <hallam> 4 use to be 2*PI
[19:49] <hallam> and 1 used to be PI/2
[19:49] <hallam> and 2 used to be PI
[19:50] <Laurenceb> ok yeah
[19:53] <edmoore|away> hallam: we need to think of a title for this talk
[19:53] <edmoore|away> and a 2 sentence description
[19:55] <Laurenceb> hmm fmod has an argument of 4
[19:55] <Laurenceb> is that correct?
[19:55] <hallam> yes
[19:55] <hallam> I think so
[19:56] <hallam> edmoore|away: High altitude ballooning with CU Spaceflight
[19:56] <hallam> sorry, can't think of anything better atm
[19:56] <edmoore|away> i guess it's a fallback
[19:56] <edmoore|away> :)
[19:56] Nick change: edmoore|away -> edmoore
[19:56] <Laurenceb> guess so
[19:56] <Laurenceb> why not normal modulus?
[19:56] <hallam> doesn't work with floats
[19:56] <Laurenceb> ah
[19:57] <hallam> I think the carrier generation is okay
[19:57] <hallam> I'll check though
[19:58] <Laurenceb> carrier_notq = carrier_q?0:1;
[19:58] <Laurenceb> carrier_notq = !carrier_q;
[19:58] <Laurenceb> cant you do that?
[20:00] <hallam> maybe, I'm not sure if true in C is guaranteed to be 1 or if it's just guaranteed to be nonzero
[20:00] <hallam> oh I have an idea
[20:01] <hallam> 1 ms isn't an exact multiple of chunks
[20:01] <hallam> there are actually 4091 samples in a millisecond, we pad the rest of the PRN with magic zeros
[20:01] <hallam> but I'm generating the carrier for all 4096
[20:08] <hallam> well piss, that didn't fix it
[20:08] <Laurenceb> hmm
[20:09] <Laurenceb> it works in matlab
[20:09] <hallam> yeah, it's not absolutely totally the same though
[20:10] <Laurenceb> no
[20:10] <Laurenceb> oh did you do the 01010101 business?
[20:10] <hallam> the magic zeros?
[20:10] <hallam> you only need to do them in PRNearlylate, right?
[20:10] <Laurenceb> 111111101010101010000000
[20:10] <Laurenceb> yes
[20:10] <hallam> I don't think there's anywhere else where there are zeros
[20:11] <hallam> and yes I did them there
[20:11] <Laurenceb> hmm
[20:11] <hallam> but PRNearlylate should be totally irrelevant because the DLL is not used
[20:11] <Laurenceb> maybe sim it in matlab
[20:11] <Laurenceb> yes
[20:11] <hallam> I turned the DLL off in matlab too (just have the constant code rate correction)
[20:12] <Laurenceb> ok
[20:22] <Laurenceb> edmoore: are you familiar with python?
[20:22] <edmoore> not really, and espcially not wxpython
[20:22] <Laurenceb> ok
[20:22] <Laurenceb> nvm
[20:22] <Laurenceb> SpeedEvil? rjharrison?
[20:37] <edmoore> rjharrison: http://garydion.com/projects/whereavr/
[20:37] <edmoore> may be of interest if you've not bumped into it yet
[21:04] Hiena (n=Hiena@81.93.195.181.datatrans.hu) left irc: "-=Halt! Hammerzeit!=-"
[21:23] <hallam> http://showcase.netins.net/web/wallio/GPSrcvrsvs60kft.htm
[21:25] Action: Laurenceb is working on a mini rogallo ground control application
[21:25] <Laurenceb> or rather ground data display
[21:27] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-a0e6a0fbb47ae0eb) left irc: "http://www.mibbit.com ajax IRC Client"
[21:31] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[21:35] <Xenion> http://www.googlelunarxprize.org/files/images/lunar/teams/artwork/micro-space.jpg
[21:35] <Xenion> wrong channel
[21:35] <Xenion> sorry
[21:40] jcoxon_ (n=jcoxon@86.158.31.172) joined #highaltitude.
[21:41] <jcoxon_> hey all
[21:42] <EI5GTB> ohai
[21:43] <jcoxon_> hey EI5GTB
[21:43] <EI5GTB> howdy
[21:43] <jcoxon_> hows things?
[21:43] <EI5GTB> not too bad
[21:43] <EI5GTB> just getting finished up with work here..
[21:43] <EI5GTB> been a long weekend!
[21:44] <jcoxon_> :-)
[21:44] <jcoxon_> i'm just starting work again on Pegasus VI
[21:44] <EI5GTB> cool
[21:45] <jcoxon_> but really want to find a flex connector for the 24pin socket on the gumstix
[21:45] <jcoxon_> seem to be impossible to find
[21:56] <Xenion> Gute Nacht jungs ! / Good Night folks :-)
[21:56] <jcoxon_> night Xenion
[22:03] <edmoore> jcoxon_: will you be around later?
[22:03] <edmoore> 11.30ish
[22:03] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-0466d4ce456314d0) joined #highaltitude.
[22:03] <hallam> Laurenceb: I've reimplemented the C version in matlab
[22:03] <hallam> and it shows the same behaviour
[22:03] <hallam> that is, I'm running it at 4Msamp/s
[22:04] <hallam> I didn't do the interleaving stuff
[22:04] <Laurenceb> hmmm
[22:04] <hallam> or the xors
[22:04] <Laurenceb> :-/
[22:04] <hallam> just +/-
[22:04] <hallam> so the problem is not in any of those
[22:04] <Laurenceb> ok
[22:04] <Laurenceb> so what exactly iss it doing?
[22:04] <Laurenceb> +- with a carrier?
[22:05] <Laurenceb> so... how is that different from before?
[22:05] <Laurenceb> with rounded carrier
[22:05] <hallam> http://uploads.mibbit.com/up/1ok403KK.png
[22:05] <Laurenceb> i see
[22:05] <edmoore> hallam: 'Supersonics, RF-Harmonics, Digital Electronics and Teddy Bears. An talk by CU Spaceflight'. Is that too silly?
[22:05] <hallam> I tried with a +1 / -1 carrier and with an analog complex exponential one, no difference
[22:06] <hallam> edmoore: sounds great
[22:06] <edmoore> sure?
[22:06] <hallam> yes
[22:06] <Laurenceb> hmm so you are doing the real multiplication
[22:06] <hallam> except
[22:06] <hallam> should be "a" not "an
[22:06] <hallam> "
[22:06] <Laurenceb> then imaginary and real bins
[22:06] <hallam> Laurenceb: yes
[22:06] <edmoore> oh sure yeah
[22:06] <edmoore> sorry
[22:06] <Laurenceb> hmm
[22:06] <hallam> Laurenceb: doing it in complex form
[22:06] <hallam> just like in the old matlab
[22:07] <Laurenceb> it works?
[22:07] <hallam> the only real change is going from 8MS/s decimated, to 4MS/s
[22:07] <hallam> no
[22:07] <hallam> see the picture
[22:07] <Laurenceb> oh
[22:07] <hallam> I can't see how that change ought to screw it, but it does seem to
[22:07] <Laurenceb> so the difference is its 4MS/s
[22:07] <Laurenceb> ?
[22:07] <hallam> yeah
[22:07] <edmoore> fergusnoble: for when you get back: ''Supersonics, RF-Harmonics, Digital Electronics and Teddy Bears. A talk by CU Spaceflight' Do you think that's ok as a title for our talk.
[22:07] <Laurenceb> right
[22:08] <hallam> it worked okay at a decimated 8MS/s
[22:08] <hallam> so it's not like there's not enough information
[22:08] <Laurenceb> hallam: less gain
[22:08] <edmoore> adjustment: "Supersonics, RF-Harmonics, Digital Electronics and Teddy Bears: a talk by CU Spaceflight"
[22:08] <Laurenceb> oh hang on
[22:08] <edmoore> (colon)
[22:08] <Laurenceb> no if you measure angle for discriminator
[22:09] <Laurenceb> then no
[22:09] <hallam> I've tried varying the gain a bunch, with no luck
[22:09] <hallam> but as you say, with that discriminator I can't see how it could make any difference
[22:09] <Laurenceb> are you sure you have scaled the local osc frequency
[22:09] <Laurenceb> and the prn
[22:09] <hallam> yeah
[22:10] <Laurenceb> hmf
[22:10] <hallam> I'll check a 4th time
[22:10] <Laurenceb> I thought we'd already tried it at 4MS/s
[22:10] <Laurenceb> ?
[22:10] <hallam> nah, we'd tried it at 8MS/s but with every other sample set to 0
[22:10] <hallam> but look at the correlation from the first millisecond
[22:10] <hallam> it's high
[22:10] <hallam> and if you change the input frequency or anything else, it's not high
[22:10] <hallam> so the correlation's working
[22:12] <jcoxon_> hey guys, what pcb people have you used
[22:12] <jcoxon_> only want to make a simple breakout board
[22:12] <Laurenceb> batchpcb
[22:12] <hallam> Cambridge Engineering if you want it fast and don't mind no solder mask / plated-through holes
[22:12] <Laurenceb> ooh
[22:13] <hallam> olimex if it's small and not too fancy
[22:13] <hallam> goldphoenix for the nice stuff but it takes nearly 2 weeks unless you want to pay a lot, and the minimum panel size is quite large
[22:13] <Laurenceb> batchpcb is low but highe spec than olimex
[22:13] <Laurenceb> the use gold phoenix 2~3 weeks
[22:13] <Laurenceb> *they
[22:14] <hallam> yeah batchpcb uses goldphoenix, it's slow because it has to go from china to colorado to the UK
[22:14] <jcoxon_> basically i want to have a 24 pin 0.5mm FFC connector
[22:14] <Laurenceb> yes
[22:14] <hallam> how many of those pins do you want to use?
[22:14] <jcoxon_> which breaks out to normal thru holes
[22:14] <jcoxon_> i'd like to break them all out to the standard 0.1 inch header
[22:15] <jcoxon_> like this:
[22:15] <jcoxon_> http://www.toby.co.uk/content/catalogue/products.asp?series=FFC5-xx-R-L-Tx
[22:19] <hallam> a CUED pcb could probably do that
[22:19] <hallam> if you give me the artwork in PDF format I'll get one made for you
[22:19] <hallam> couple of days
[22:20] <jcoxon_> okay
[22:21] <jcoxon_> could i do it in eagle?
[22:21] <hallam> yeah
[22:22] <hallam> http://uploads.mibbit.com/up/ppxALmeE.png 8MS/s
[22:22] <hallam> http://uploads.mibbit.com/up/08QGaNau.png 4MS/s
[22:24] <Laurenceb> hmmm
[22:24] <Laurenceb> something is wrong
[22:24] <Laurenceb> hopefully not too serious
[22:24] <Laurenceb> it shoulld be able to track
[22:24] <Laurenceb> you compacted the data down?
[22:25] <jcoxon_> hallam, okay this is the connctor i'm going for: http://uk.farnell.com/hrs-hirose/fh12a-24s-0-5sh-55/socket-ffc-fpc-zif-0-5mm-24way/dp/1324565
[22:25] <Laurenceb> its not got zeros in it
[22:25] <hallam> yeah
[22:25] <hallam> I just plotted the carriers for both methods for the first 2 ms
[22:25] <hallam> they look the same
[22:26] <hallam> jcoxon_: it'll be kind of a bitch to solder, but possible
[22:27] <jcoxon_> hallam, that was my concern
[22:27] <Laurenceb> hmm if should track
[22:27] <Laurenceb> *it
[22:27] <hallam> that's what I keep telling it, but does it listen?
[22:27] <hallam> track_you_bastard=1;
[22:27] <jcoxon_> hallam, is there any way of making it easier for me?
[22:28] <hallam> extend the pads out past the footprint of the connector
[22:28] <hallam> or have traces running up to them
[22:28] <Laurenceb> hallam: you are using the angle of prompt ?
[22:28] <hallam> I'll get the PCB guys to tin it
[22:28] <hallam> and you can use solder paste and maybe a hotplate
[22:28] <hallam> Laurenceb: for the discriminator, yes
[22:28] <Laurenceb> hmmm
[22:28] <hallam> I didn't change that from the 8MS matlab
[22:29] <Laurenceb> 11 is the strogest?
[22:29] <hallam> yeah
[22:29] <Laurenceb> this isnt right
[22:29] <hallam> would you like to see the whole m-file?
[22:29] <Laurenceb> its impossible for it not to be able to track it
[22:29] <Laurenceb> ok
[22:29] <hallam> http://pastebin.com/d20b058e6
[22:30] <Laurenceb> are you sure your taking the right samples?
[22:30] <hallam> pretty sure
[22:30] <hallam> as I said, it correlates fine on the first couple of milliseconds
[22:31] <Laurenceb> hmmm
[22:32] Xenion (n=robert@p579FC18C.dip.t-dialin.net) left irc: "Verlassend"
[22:32] <fergusnoble> Laurenceb: whats up?
[22:32] <hallam> and that's true even if I start it an exact no. of ms later
[22:32] <Laurenceb> hi fergus
[22:33] <hallam> hey Ferg
[22:33] <Laurenceb> fergus : I got the gui working
[22:33] <Laurenceb> but I'm trying to write a data display
[22:33] <Laurenceb> how can I have a while 1: that doesnt interrupt the gui?
[22:35] <Laurenceb> hallam: looks to me like the filter is unstable
[22:35] <fergusnoble> Laurenceb: yeah, i grappled with that one for a while
[22:35] <hallam> right
[22:36] <hallam> how do I make it less unstable?
[22:36] <Laurenceb> are you sure you did I,Q,dump,dump,I,Q ?
[22:36] <Laurenceb> hallam: have to work out why its unstable in the first place
[22:36] <Laurenceb> it can track 32
[22:36] <Laurenceb> so I dont see how it can be noise
[22:36] <Laurenceb> if it cant even track 11 now
[22:37] <fergusnoble> Laurenceb: one way is to bind a function to the main window's wx.EVT_IDLE event
[22:37] <fergusnoble> thats called all the time the gui is not doing something
[22:37] <Laurenceb> ok
[22:37] <fergusnoble> but it can't be blocking
[22:37] <Laurenceb> its going to run a lot !
[22:37] <fergusnoble> in mine i had something that waited for a line from stdin
[22:38] <Laurenceb> ah
[22:38] <fergusnoble> and that would cause the gui to only update when it recieved a line
[22:38] <Laurenceb> hmmm then you need to read a logfile line by line
[22:38] <hallam> Laurenceb: I just checked again and the data is definitely in the right order
[22:38] <fergusnoble> so i changed it to use tail to always grab the last line of the log
[22:38] <Laurenceb> ah ok
[22:39] <Laurenceb> so it graps the last lines and looks to see if it has changed?
[22:39] <fergusnoble> Laurenceb: the proper way to do it is really to have another thread for processing, but thats a bit annoying
[22:39] <fergusnoble> Laurenceb: yeah
[22:39] <Laurenceb> yeah, thats what I was thinking
[22:39] <Laurenceb> and its annoying :P
[22:39] <fergusnoble> shall i send you the code for the cusf tracker?
[22:39] <Laurenceb> yeah, thanks
[22:40] <Laurenceb> I'm going to want to put on some heading dials and stuff
[22:40] <Laurenceb> windspeed, temperature
[22:40] <fergusnoble> http://pastebin.com/m23a6c650
[22:40] <fergusnoble> ok, that has a heading dial
[22:40] <Laurenceb> neat
[22:40] <fergusnoble> its not great code quality as i only ever tack chanes on in a rush just before launches
[22:41] <fergusnoble> *changes
[22:41] <fergusnoble> (or during launches)
[22:41] <Laurenceb> hehe
[22:42] <hallam> Laurenceb: is there anything I should do except keep fiddling the filter coefficients?
[22:42] <hallam> the 8MS/s version is stable for a pretty wide range of them
[22:43] <Laurenceb> tango gps ?
[22:43] <Laurenceb> hallam : the time interval is still the same
[22:43] <hallam> yeah it really shouldn't change, should it
[22:44] <hallam> but I don't know what else to do
[22:44] <Laurenceb> no
[22:44] <fergusnoble> Laurenceb: its mapping program i modified to display the track and predictions
[22:44] <Laurenceb> cool, free?
[22:45] <fergusnoble> yeah
[22:45] <Laurenceb> this is very awsome
[22:45] <fergusnoble> the good thing is there is a way to use it to grab huge chunks of google maps data
[22:45] <fergusnoble> and store it for offline use
[22:45] <Laurenceb> any chance you could pastebin the xml?
[22:45] <fergusnoble> sure
[22:46] <Laurenceb> it nicks google maps data?
[22:46] <fergusnoble> oops, shouldnt have said that ;)
[22:47] <Laurenceb> hehe
[22:47] <Laurenceb> hallam: I'm confused by this tracking issues
[22:47] <Laurenceb> looks like the pll
[22:47] <Laurenceb> but I really cant see it
[22:47] <fergusnoble> http://pastebin.com/m36a6d6bf
[22:49] <fergusnoble> in the main part where it doecodes the telem string note its inside a try block
[22:49] <fergusnoble> so it just assumes the data is ok and makes whatever assumptions
[22:49] <fergusnoble> and if the data isnt ok, python throws an exception and the data is ignored
[22:50] <Laurenceb> I see
[22:50] <Laurenceb> hallam: maybe its the dll
[22:51] <Laurenceb> did you adjust the number of samples to chuck?
[22:54] <hallam> yes
[22:54] <Laurenceb> hmf
[22:54] <hallam> but I also disabled the DLL in both for testing purposes
[22:54] <Laurenceb> oh yeah
[22:54] <hallam> or rather, gave it a fixed discriminator
[22:54] <Laurenceb> ok, try it with no pll
[22:55] <hallam> hm, on both? that's worth a try
[22:55] <Laurenceb> yeah, they should both drift out of lock about the same time
[22:56] <hallam> should it be *exactly* the same?
[22:56] <Laurenceb> not quite
[22:56] <hallam> it's not, but I think it should be
[22:56] <Laurenceb> as theres noise
[22:56] <hallam> where does the difference come from?
[22:56] <Laurenceb> but practically the same
[22:56] <hallam> aren't they operating on the same data?
[22:56] <Laurenceb> theres extra somples
[22:56] <Laurenceb> they are a bit noisy
[22:57] <hallam> remember the 8MS/s is decimated
[22:57] <hallam> so it's effectively 4
[22:57] <Laurenceb> but as we are averaging and they are interleaved
[22:57] <Laurenceb> yeah
[22:57] <Laurenceb> so very small difference
[22:57] <hallam> the nonzero samples that the 8MS/s version gets are identical to the samples that the 4MS/s version gets
[22:57] <Laurenceb> shouldnt really be visable in the noise
[22:57] <Laurenceb> yeah
[22:58] <hallam> I made the 8MS/s version chuck 0, 2 or 4 instead of 0, 1 or 2
[22:58] <hallam> I'm trying to make them exactly the same
[22:58] <Laurenceb> so you should have a noisy correlation going out of lock with both
[22:58] <Laurenceb> but the noise will differ
[22:58] <hallam> what noise?
[22:58] <hallam> it's the same noise in each
[22:58] <Laurenceb> no
[22:58] <Laurenceb> theres device noise
[22:58] <hallam> I don't understand
[22:58] <Laurenceb> and noise that gets through the filter
[22:58] <hallam> they're using the same digital data
[22:59] <Laurenceb> eh
[22:59] <Laurenceb> you chuck 1/2 the data
[22:59] <Laurenceb> for the 4Msps
[22:59] <hallam> yeah
[22:59] <hallam> and in the 8Msps I make 1/2 the data zero
[22:59] <hallam> the same half
[22:59] <Laurenceb> ah
[22:59] <Laurenceb> ok sorry
[22:59] <hallam> hence decimation
[22:59] <Laurenceb> identical in that case
[22:59] <hallam> (vs downsampling)
[23:00] <Laurenceb> but even if you didnt
[23:00] <hallam> right, even if not, you'd expect them to be close
[23:00] <hallam> but they're not identical
[23:00] <Laurenceb> if should be practically indistinguishable
[23:00] <hallam> so I guess I'll try to track down why they aren't identical
[23:00] <Laurenceb> yeah
[23:00] <Laurenceb> and I'm heading off
[23:00] <Laurenceb> back in half an hour
[23:00] <Laurenceb> good luck
[23:01] <hallam> thanks
[23:01] <hallam> fergusnoble: I'm about to make spaghetti carbonara if you're peckish
[23:04] <fergusnoble> hallam: i just got back from having dinner
[23:04] <fergusnoble> thanks though
[23:04] <fergusnoble> hows the gps going?
[23:04] <hallam> egh, got this weird issue, the PLL isn't working
[23:04] <hallam> I've replicated the effect in matlab
[23:04] <hallam> have to track it down
[23:05] <hallam> it works fine at 8Msamples/second, with half of the samples set to 0
[23:05] <fergusnoble> think it might be the carrier?
[23:05] <hallam> yes I think so
[23:05] <hallam> not the carrier generation
[23:05] <hallam> it doesn't work at all at 4Msamples/sec
[23:05] <hallam> which should be identical to the decimated 8MS/s
[23:05] <fergusnoble> yeah
[23:05] <fergusnoble> very odd
[23:06] <edmoore> fergusnoble: you were away when hallam and i had a quick talk chat
[23:06] <edmoore> hope you think the title is ok
[23:06] <fergusnoble> when you change the matlab to use 4Ms/s does it give the same correlation as the C?
[23:06] <fergusnoble> edmoore: yeah, looks good
[23:08] <hallam> fergusnoble: think so
[23:08] <fergusnoble> ok, thats good to know
[23:08] <hallam> ok I'm starving
[23:08] <hallam> back in a bit
[23:08] <fergusnoble> any youve tried it with the sine carrier?
[23:08] <fergusnoble> ok, bye
[23:13] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-0466d4ce456314d0) left irc: "http://www.mibbit.com ajax IRC Client"
[23:17] <edmoore> fergusnoble: did you have a good meal?
[23:17] <fergusnoble> we went to the eagle in the end
[23:18] <fergusnoble> so it wasnt as good as the castle would have been, but it was very pleasent
[23:18] <edmoore> cool
[23:18] <edmoore> you an ou?
[23:18] <fergusnoble> ou?
[23:18] <edmoore> who
[23:19] <fergusnoble> chris and ximin
[23:19] <edmoore> ximin sounds like a parameter in matlab
[23:19] <edmoore> I'm assuming that's not the case though
[23:19] <fergusnoble> i think you met chris, not sure about ximin
[23:19] <fergusnoble> no
[23:19] <edmoore> chris - another physicist?
[23:19] <edmoore> slightly northern?
[23:19] <fergusnoble> compsci and from manchester
[23:20] <edmoore> close
[23:20] <edmoore> but i think I know the guy
[23:20] <fergusnoble> cool
[23:20] <fergusnoble> anything good on the airwaves?
[23:20] <edmoore> not since ISS pass
[23:21] <edmoore> though there's a southafrican guy called laurence who seems quite technical
[23:21] <edmoore> 2E0JYK is on a lot
[23:25] <fergusnoble> yeah, worth getting the radio out?
[23:25] <jcoxon_> hmmm got an eagle question for someone
[23:25] <edmoore> jcoxon_: fire
[23:25] <edmoore> fergusnoble: not really - i'm about to grab a bath
[23:29] <jcoxon_> making a breakout board, found the two parts and put wires between them on teh schematic
[23:29] <edmoore> not been any activity for 15 mins
[23:29] <jcoxon_> then when i do switch to board and generate it not all the wires are there
[23:29] Action: SpeedEvil gets into (his own) bath.
[23:29] <SpeedEvil> splashies
[23:29] <edmoore> jcoxon_: on a matter of technical specificity, are we talking air-wires or actual tracks?
[23:29] <jcoxon_> hmmm i guess i just drew lines
[23:29] <edmoore> on the schematic though
[23:29] <jcoxon_> yeah
[23:29] <edmoore> sure. so on the layout view, you should get faint yellow 'air wires'
[23:29] <edmoore> they just show that there needs to be a connection. you then draw the physical track between them
[23:29] <jcoxon_> yeah
[23:29] <jcoxon_> okay
[23:29] <jcoxon_> i'll have a play
[23:29] <edmoore> it should have all the air-wires on it corresponding to wires you drew on the schematic - if not, double check you actually did make a connection on the schematic between the two places
[23:29] <edmoore> right, going for bath
[23:29] Nick change: edmoore -> edmoore|away
[23:36] <natrium42> edmoore|away: finally, i was about to mention it
[23:37] <jcoxon_> sorted that bit
[23:37] <jcoxon_> hadn't made proper connections - had slightly gone over
[23:38] <natrium42> jcoxon_: have you ever tried using mosfets to power electronics on/off?
[23:39] <natrium42> i have a power converter with inhibit pin, but it still draws 1mA
[23:41] <natrium42> so i am considering just using a mosfet at the Vin line
[23:41] <natrium42> err, voltage converter, not power converter
[23:41] <natrium42> lol
[23:41] <jcoxon_> natrium42, i haven't, only for cutdowns - not very good with my electronics
[23:41] <Laurenceb> fergusnoble: what goes in tracker_settings ?
[23:44] <fergusnoble> very little, just the callsign and some parameters used by the landing prediction
[23:44] <natrium42> jcoxon_: k, fair enough
[23:44] <fergusnoble> oh, and the lat,lon window for a coord to be valid
[23:44] <jcoxon_> fergusnoble might no
[23:44] <fergusnoble> and launch location
[23:44] <jcoxon_> know*
[23:44] <fergusnoble> natrium42: not done it myself, but have heard its ok
[23:44] <natrium42> it should be less power draw, right?
[23:44] <natrium42> leakage current is quite small
[23:44] <fergusnoble> oh yeah, not much will get through
[23:44] <Laurenceb> fergusnoble: could you pastebin it?
[23:44] <fergusnoble> just thinking if its on the high side you might have to do something clever with interfacing the base to the microcontroller
[23:44] <fergusnoble> http://pastebin.com/m36aae375
[23:44] <Laurenceb> thanks
[23:44] <fergusnoble> depending on the type of fet
[23:44] <Laurenceb> high side what current?
[23:44] <Laurenceb> or voltage?
[23:44] Action: Laurenceb doesnt follow
[23:44] <fergusnoble> Laurenceb: ill put it all on the svn before too long, just want to make it a bit neater before the world can see it
[23:45] <Laurenceb> :P
[23:45] <Laurenceb> I have a national semiconductor switched mode ic on the rogallo
[23:45] <Laurenceb> with an inhibit pin
[23:45] <Laurenceb> so you can turn off the servo and rc receiver
[23:45] <Laurenceb> thats just on a gpio
[23:48] Nick change: edmoore|away -> edmoore
[23:48] <jcoxon_> if olimex says: Minimum tracks and spaces: 0,203 mm
[23:48] <fergusnoble> natrium42: this is probably more than you really want to know :)
[23:48] <fergusnoble> http://www.industrialcontroldesignline.com/howto/199204060
[23:49] <jcoxon_> how do i set up eagle autorouter?
[23:49] <edmoore> the best approach to high side driving on mosfets is to buy a logic level one, which has it built in :p
[23:49] <edmoore> jcoxon_: no
[23:49] <edmoore> don't ask questions like that
[23:50] <edmoore> I won't help you, as a matter or principle
[23:50] <jcoxon_> ?
[23:50] <Laurenceb> hehe
[23:50] <natrium42> srsly jcoxon_ ...
[23:50] <Laurenceb> its not very good
[23:50] <edmoore> the autorouter is the devil
[23:50] <edmoore> unless you're designing a mobo with a very high end bit of kit, it's immoral/poor
[23:51] <Laurenceb> surely thats the new us president
[23:51] <edmoore> a high end design package*
[23:51] <jcoxon_> of course it is, but when you are just playing i don't see why i can't ask
[23:51] <edmoore> mobo isn't the best example...
[23:51] <edmoore> ok, well there are loads of guides on the web, but I've said my bit. It's not so much th dark side, just the shameful one
[23:51] <Laurenceb> mojo
[23:51] <fergusnoble> jcoxon_: the problem is you have two things you can change, where the components are (and their orientation) and how the tracks are routed
[23:52] <fergusnoble> the autorouter cant do the first part
[23:52] <fergusnoble> and so will fail to find the best solution
[23:52] <fergusnoble> and generally will route your board like it just smoked a bunch of crack
[23:53] <jcoxon_> wow i feel like i'm being told off
[23:53] <edmoore> routing is a tough one to solve, algorithmically. You do have to dive in at one end, start working through, then probably go back and change the stuff you started with to make latter stuff work.
[23:53] <edmoore> i'd rather fly spears of death again than use the eagle autorouter
[23:54] <edmoore> ...(I don't actually stand by that)
[23:55] <fergusnoble> jcoxon_: actually its probably a good idea to have a play with it to see what tis like
[23:55] <edmoore> like a toddler sticking its fingers into a plug
[23:55] <edmoore> I'm sort of enjoying this :)
[23:56] <jcoxon_> edmoore, i'm not
[23:56] <edmoore> have a play, but it is pretty crappy.
[23:57] <edmoore> http://www.instructables.com/id/EHQB31NZQUES9J6DAT/
[23:58] <natrium42> jcoxon_: try TopoR --> http://www.freestyleteam.com/index.php?topic=topor&lang=en
[23:59] <natrium42> it does some crazy routing
[23:59] <natrium42> Laurenceb-esque
[23:59] <edmoore> didn't laurence try that?
[23:59] <natrium42> oh?
[23:59] <natrium42> http://www.freestyleteam.com/index.php?topic=toportour&lang=en
[23:59] <edmoore> i'm not sure
[00:00] --- Mon Jan 26 2009