highaltitude.log.20090123

[00:01] <fergusnoble> how do you change between reversed and non-reversed rtty on fldigi?
[00:04] <edmoore> dude has a point
[00:05] <edmoore> hmm, I really do want to try some HF now
[00:05] <edmoore> only so much to hear on UHF
[00:06] <fergusnoble> edmoore: ive got the badger trasmitting exactly 425 shift now, which helps on fldigi and other programs which dont allow non standard shifts
[00:07] <edmoore> at the bottom i think fergusnoble
[00:07] <fergusnoble> its now non-reversed on LSB too
[00:07] <edmoore> 'Rv'
[00:07] <edmoore> oh that probably is just the USB switch
[00:09] <edmoore> fergusnoble: ok that's cool
[00:09] <fergusnoble> edmoore: looks like the VFO knob on the icom changes rtty width when in rtty mode
[00:10] <fergusnoble> might want to check that though
[00:11] <edmoore> i don't have a vfo knob
[00:11] <fergusnoble> hehe ok
[00:14] <hallam> fergusnoble: how do you ensure exactly 425 shift?
[00:14] <fergusnoble> well, on the bench, and i shouldnt really say exactly
[00:14] <Laurenceb> omg omg
[00:14] Action: Laurenceb fixed sdrgnss
[00:15] <SpeedEvil> :)
[00:15] <hallam> nice!
[00:15] <Laurenceb> time to run it over all the data
[00:15] <fergusnoble> but as close as i can get by tweaking the dac parameters
[00:15] Action: SpeedEvil passes Laurenceb a tropy.
[00:15] <hallam> how did you fix it?
[00:15] <SpeedEvil> (It's a singed oven-fried chip)
[00:15] <Laurenceb> just rewrote the acquisition file
[00:15] <hallam> HenryGNSS has locked on to the nav message, now to interpret it...
[00:15] <Laurenceb> lol HenryGNSS
[00:15] <fergusnoble> hallam: awesome
[00:15] <edmoore> very awesomeness
[00:17] <edmoore> fergusnoble: what's the lingo for an all-freq tx mod?
[00:17] <edmoore> I've forgotton
[00:17] <fergusnoble> dunno
[00:17] <fergusnoble> out of band something
[00:18] <Laurenceb> this may take a while
[00:18] <Laurenceb> luckly its not my machine
[00:18] <Laurenceb> ssh ftw
[00:18] Action: Laurenceb is building up a file full of ssh details
[00:19] <Laurenceb> it can do each sat at 1/3 real time
[00:20] <Laurenceb> but theres a gui and stuff to drive
[00:21] <Laurenceb> half way now
[00:21] <Laurenceb> not sure how long a nav solution takes
[00:21] <edmoore> fergusnoble: mods.dk, just for ref
[00:21] <fergusnoble> my coordinates are di
[00:21] <fergusnoble> configring the mainframe
[00:21] <edmoore> right, am going to kip
[00:22] <edmoore> down with 9ams
[00:22] <fergusnoble> don't let the bed bugs bite
[00:22] Bluenarf (n=Paul@apollo.paulsnet.org) joined #highaltitude.
[00:22] <fergusnoble> blue naf!
[00:25] <fergusnoble> ... and thats what happens when you let arts students near a computer
[00:26] <edmoore> alex?
[00:26] <fergusnoble> yes
[00:26] Nick change: Bluenarf -> EI5GTB
[00:27] <edmoore> friendly
[00:27] <Laurenceb> hi
[00:27] <Laurenceb> I hope there is enough nav data
[00:32] <Laurenceb> http://maps.google.co.uk/maps?f=q&source=s_q&hl=en&geocode=&q=%4052.202106,0.118178&sll=53.800651,-4.064941&sspn=12.635315,39.550781&ie=UTF8&t=h&z=16&iwloc=addr
[00:32] <Laurenceb> I win
[00:33] <edmoore> that's impressive
[00:33] <hallam> nice work
[00:33] <Laurenceb> I started to lose 32
[00:33] <Laurenceb> how far out is it?
[00:34] <hallam> yeah I lost 32 at about 25s
[00:34] <hallam> 10m max
[00:34] <Laurenceb> nice
[00:34] <hallam> how much data was that working from?
[00:35] <Laurenceb> 500ms
[00:35] <hallam> cool
[00:35] <hallam> did you preload the ephemeris?
[00:35] <Laurenceb> yeah
[00:35] <Laurenceb> but not from a server
[00:35] <Laurenceb> from the 40s sample
[00:35] <hallam> what from?
[00:35] <hallam> oh, ok
[00:36] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc:
[00:36] <fergusnoble> Laurenceb: very very nice work
[00:36] <fergusnoble> Laurenceb: was that from your code?
[00:36] <Laurenceb> yeah
[00:37] <Laurenceb> did write it all myself
[00:37] <Laurenceb> if based on Borre's code
[00:37] <fergusnoble> that must be about the coolest thing ive seen in a while
[00:37] <fergusnoble> hopefully hallam is hot on your heels
[00:37] <fergusnoble> :)
[00:37] <Laurenceb> I had to rewrite some files to get it to work in a low signal environment
[00:38] <hallam> how long can you track satellite 32 for, Laurenceb ?
[00:38] <Laurenceb> just a sec, looking at the plot
[00:38] <fergusnoble> hallam: so it looks like we can do this, i guess we need to look at getting the data into the blackfin!
[00:39] <fergusnoble> i say we, i really mean you/Laurenceb
[00:40] <Laurenceb> its tracking 32 until the end
[00:40] <Laurenceb> but the real component of prompt is decreasing
[00:41] <Laurenceb> theres quite a few noise spikes crossing 0 near the end
[00:42] <Laurenceb> but with an averaging based 1ksps/50bps conversion its trackable I think
[00:42] <Laurenceb> I didnt write that file, I'll take a lok
[00:43] <SpeedEvil> you can track it quite a way after you've lost the data of course
[00:44] <fergusnoble> hallam: shaved a third of the ram usage off the badger code
[00:44] <Laurenceb> whats the total use?
[00:46] <Laurenceb> yeah, Borres code averages each group of 20 samples
[00:46] <Laurenceb> then rounds to 0 or 1
[00:46] <Laurenceb> or rather +-1 then converts to binary
[00:50] <fergusnoble> Laurenceb: no 16.5K
[00:50] <fergusnoble> *now
[00:51] <Laurenceb> thats quite a lot
[00:51] <Laurenceb> guess it does fat
[00:51] <Laurenceb> so not bad
[00:51] <Laurenceb> the min rogallo is 214 bytes :P
[00:52] <Laurenceb> and the radio is about 500 bytes of flash
[00:53] <Laurenceb> hmm interesting
[00:54] <Laurenceb> Borre correlates with the packet header
[00:54] <SpeedEvil> yeah - I was thinking of that
[00:54] <SpeedEvil> for low signals
[00:54] <Laurenceb> i.e. he doesnt downconvert to 50bps until the header if found
[00:54] <SpeedEvil> but I don't think it buys you that much over simply looking at the absolute power.
[00:54] <SpeedEvil> Oh.
[00:54] <SpeedEvil> That's nifty
[00:54] <Laurenceb> I was thinking of just finding a timing such that you sink with the largest changes in real(prompt)
[00:55] <SpeedEvil> so you use the header as an extra 3 bits or so of PRN
[00:55] <SpeedEvil> getting you a little deeper.
[00:55] <Laurenceb> well
[00:55] <Laurenceb> from what I can see the DLL and PLL screw up before the 50bps is lost
[00:55] <SpeedEvil> Though at that noise level, you won't be picking up the message
[00:56] <Laurenceb> but I guess it makes it extremely noise tolerant
[00:56] <SpeedEvil> I was pretty much assuming for low noise you'd be _vastly_ reducing the filter bandwidth, and largely relying on your position being static, with appropriate corrections going back from the upper levels and satellites you have in lock
[00:56] <Laurenceb> and also tolerant of glitches in the 1KHz loop v 20ms data bit syncronisation
[00:56] <SpeedEvil> for high noise
[00:57] <SpeedEvil> The aim isn't to use it for a good position solution, but to keep it locked at all costs, so you can use it rapidly if it gets better
[00:58] <SpeedEvil> I was also wondering about for known bitsequences, backpropagating those too to the lower levels.
[00:58] <Laurenceb> I'll try rerunning with tighter filters
[00:58] <Laurenceb> yeah, there are techniques to do that
[00:59] <SpeedEvil> So you have a known frame - say part of the long orbital data - you tell the lower level code what the bits are
[01:00] EI5GTB (n=Paul@apollo.paulsnet.org) left irc: Read error: 104 (Connection reset by peer)
[01:04] <hallam> fergusnoble, nice work
[01:07] <fergusnoble> hallam: hows it going with the nav data?
[01:08] <hallam> slow work interpreting it
[01:08] <hallam> going to crib a lot from the textbook I think
[01:08] <SpeedEvil> Nav bits are easy
[01:08] <fergusnoble> cool
[01:08] <fergusnoble> when does it arrive?
[01:08] <SpeedEvil> interpreting them isn't very.
[01:09] <SpeedEvil> I note that you can simply throw 4 satellites 50bps streams over a com-link, along with doppler data, and do it on the ground though
[01:10] <Laurenceb> just use borres code
[01:10] <Laurenceb> I posted a zip aerlier
[01:10] <fergusnoble> hallam: is there anything i can help with on the blackfin?
[01:11] <fergusnoble> hallam: or maybe porting the sd code to your board?
[01:11] <fergusnoble> it does use about 18k of flash for the fat stuff though
[01:12] <fergusnoble> maybe can be trimmed down if that doesnt fit on the pic18
[01:12] <Laurenceb> theres gcc for the blackfin right?
[01:12] <fergusnoble> Laurenceb: yup i think so
[01:12] <Laurenceb> wonder if the math library makes use of the dsp stuff
[01:13] <SpeedEvil> Making a big file - the maximal file size on the card - lets you just write into the whole file without problems
[01:14] <fergusnoble> dunno, even with out it its got rediculous power
[01:14] <SpeedEvil> at least on FAT
[01:14] <SpeedEvil> AIUI
[01:14] <fergusnoble> SpeedEvil: yeah thats one solution
[01:14] <Laurenceb> I'm re running with tighter loops
[01:14] <fergusnoble> maybe the best one for the gps graber
[01:15] <fergusnoble> but flexibility is good, esp since the blackfin is also running the camera
[01:15] <Laurenceb> for the GPS SAR they are planning on DDR2
[01:15] <fergusnoble> would be nice for it to makes new files for each image
[01:15] <SpeedEvil> how much is a blackfin these days
[01:15] <fergusnoble> dunno, hallam?
[01:16] <Laurenceb> shark+spraypaint
[01:17] <hallam> chips ~$10
[01:17] <hallam> board you can use ~$150
[01:17] <SpeedEvil> hallam: and no external ROM/RAM?
[01:17] <SpeedEvil> internal
[01:17] <hallam> no internal flash, some internal ram
[01:18] <hallam> 256kB or thereabouts
[01:18] <hallam> can boot off an SPI flash chip
[01:18] <SpeedEvil> k
[01:18] Action: SpeedEvil is currently playin with stm32s, which are also fun.
[01:21] <Laurenceb> hmf
[01:21] <Laurenceb> almost exactly the same place
[01:21] <fergusnoble> down to 11k
[01:21] <Laurenceb> think thats as good as its going to get
[01:24] <Laurenceb> I've lowered the DLL bandwidth, and made the chip spacing 0.35
[01:24] <fergusnoble> hallam: also, food for thought, it looks like the lpc serial peripheral is fast enough to read the gps data
[01:25] <Laurenceb> lpc?
[01:25] <fergusnoble> arm7
[01:25] <Laurenceb> oh course
[01:26] <fergusnoble> anyway, bed for me i think
[01:26] <fergusnoble> night
[01:26] <SpeedEvil> Night.
[01:26] <Laurenceb> bye
[01:26] Action: Laurenceb goes to get some food
[01:40] <Laurenceb> back
[01:40] <Laurenceb> ooh a bit different
[01:43] <Laurenceb> http://maps.google.co.uk/maps?utm_campaign=en&utm_source=en-ha-emea-uk-goog-gm&utm_medium=ha&utm_term=google%20maps
[01:43] <Laurenceb> hallam ^ nearer?
[01:46] <hallam> try a link with some coords in it :P
[01:47] <Laurenceb> http://maps.google.co.uk/maps?f=q&source=s_q&hl=en&geocode=&q=%4052.20211072,0.1181422778&sll=52.202118,0.118271&sspn=0.0002,0.000603&ie=UTF8&ll=52.202089,0.118238&spn=0.0002,0.000603&t=h&z=21&iwloc=addr
[01:47] <Laurenceb> woops
[01:47] <hallam> yes fergusnoble, it seems most decent micros (16-bit and up) can eat the data fast enough, the trouble is doing something with it
[01:47] <hallam> there's not enough ram to store very much
[01:47] <fergusnoble> 32k - thats a few ms?
[01:48] <hallam> 32ms
[01:48] <fergusnoble> but i agree the blackfin is better soln
[01:48] <hallam> at the slow data rate
[01:48] <fergusnoble> ok, night
[01:48] <hallam> it's kind of a bitch to recover position from that
[01:48] <hallam> could be done, but it's a bitch
[01:48] <hallam> goodnight
[01:48] <Laurenceb> you actuallt sleep?!
[01:48] Tigga (n=chatzill@pc-232-235-103.magd.cam.ac.uk) left irc: Read error: 110 (Connection timed out)
[01:48] <Laurenceb> hallam: where was the ant?
[01:49] <Laurenceb> nvm night
[01:49] <hallam> no, saying night to fergus
[01:49] <hallam> not sleeping yet
[01:49] <Laurenceb> lol
[01:49] <hallam> not until I have SOMETHING from this nav msg
[01:49] <Laurenceb> how long have you been up?
[01:49] <hallam> long
[01:50] <Laurenceb> anyway, whereabout was the ant?
[01:50] <hallam> hang on I'll ask google maps
[01:50] <Laurenceb> thanks
[01:51] <hallam> 52.2021, 0.1184
[01:52] <Laurenceb> cool
[01:53] <Laurenceb> would help if google maps wasnt at an angle
[01:53] <Laurenceb> I'll try the os site
[01:55] <SpeedEvil> at an angle?
[01:55] <Laurenceb> the photos arent from zenith
[01:56] <Laurenceb> hard to say how far out it is
[01:56] <SpeedEvil> they should be rectified though.
[01:58] <Laurenceb> its probably several meters over the height of that building
[01:59] <Laurenceb> hallam: window onto penbroke street?
[02:00] <hallam> yup
[02:01] <hallam> what do you get for the altitude?
[02:01] <Laurenceb> -83m
[02:01] <Laurenceb> not sure what geoid that is
[02:03] <Laurenceb> hmm yeah 32 was probably fading out of view
[02:03] <Laurenceb> maybe even over the 40s
[02:03] <Laurenceb> thats a lead roof
[02:04] <Laurenceb> hmm 32 is ok for the first 15 seconds
[02:04] <Laurenceb> then starts to fade out
[02:04] <Laurenceb> by about 15% after 40s
[02:12] <Laurenceb> wgs84
[02:12] <Laurenceb> http://earth-info.nga.mil/GandG/wgs84/gravitymod/egm96/intpt.html
[02:16] <Laurenceb> I think theres issues with multipath
[02:16] <Laurenceb> with all those lead rooves ect
[02:17] <Laurenceb> I'd say try it in a park or somewhere next
[02:17] <Laurenceb> but its working pretty well given the conditions
[02:20] akawaka (n=akawaka@66.77.144.8) left irc: Remote closed the connection
[02:20] <Laurenceb> I just plotted all the points in 3D
[02:20] <Laurenceb> they are spread out in a thin disk
[02:20] <hallam> cool
[02:21] <hallam> what sort of spread?
[02:21] <Laurenceb> 80m radius
[02:21] <Laurenceb> ~15m thick
[02:22] <Laurenceb> the axis is pointing down at 45 degrees to the south east
[02:22] <hallam> nice
[02:22] <hallam> argh wtf matlab
[02:22] <Laurenceb> doesnt look quite right to me
[02:23] <Laurenceb> looks like you are receiving reflections off stuff below you
[02:23] <hallam> usually there's more uncertainty in alt than position
[02:23] <Laurenceb> yeah, but its such a perfect disk
[02:24] <Laurenceb> guess 32 was weak, so all your sats were in that direction more or less
[02:26] <Laurenceb> no
[02:26] <Laurenceb> hmm have I mixed up the aces...
[02:29] <Laurenceb> whats that big thing on the other side of trumpington street?
[02:35] <Laurenceb> hmm guess the shape is all an effect of the sat alignments
[02:35] <Laurenceb> but its hardly ideal conditions
[02:35] <Laurenceb> need to test in a park Id say
[02:36] <SpeedEvil> how many sats
[02:36] <SpeedEvil> ?
[02:37] <Laurenceb> 4
[02:38] <Laurenceb> I'm prettyimpressed by being able to track prn32
[02:38] <Laurenceb> theres a lead roof in the way
[02:41] <SpeedEvil> multipath, or reflections
[02:41] <SpeedEvil> maybe
[02:43] <Laurenceb> data bits are earliy read
[02:43] <Laurenceb> *easily
[02:50] <SpeedEvil> yeah - well multipath bouncing off hte ground
[03:02] <hallam> woohoo got time-of-week
[03:05] <SpeedEvil> :)
[03:18] <hallam> argh SO SLEEPY
[03:18] <hallam> must..get..pseudorange
[03:18] <Laurenceb> http://imagebin.org/36566
[03:18] <Laurenceb> hallam ^ my desktop
[03:19] <hallam> impressive
[03:20] <Laurenceb> well done with time of week
[03:20] <hallam> I'm glad to have mostly built this one myself (of course, with a ton of help from you guys) rather than just using SoftGNSS though
[03:20] <Laurenceb> did you average the 20 samples
[03:20] <hallam> it's much more satisfying and I udnerstand it
[03:20] <hallam> 20 samples per bit?
[03:20] <hallam> yeah
[03:21] <hallam> that's very solid
[03:21] <hallam> no problem decoding nav bits unless the carrier PLL loses it
[03:21] <hallam> (I assume it's the carrier PLL that kills sv32, not the DLL)
[03:22] <Laurenceb> yeah probably
[03:22] <Laurenceb> I get nav bits for 32 all the way through
[03:22] <Laurenceb> how are you finding headers?
[03:22] <Laurenceb> correlating at the adc sample rate?
[03:22] <Laurenceb> erm
[03:22] <Laurenceb> 1KHz
[03:25] <Laurenceb> hallam: you you correlate with the 1KHz prompt dumps to find the headers?
[03:26] <hallam> headers?
[03:26] <Laurenceb> erm
[03:26] <hallam> the preamble?
[03:26] <Laurenceb> HOW
[03:26] <Laurenceb> yeah
[03:26] <hallam> well first I turn everything into nice bits
[03:26] <Laurenceb> ah ok
[03:27] <Laurenceb> softgnss does it differently
[03:27] <Laurenceb> it correlates the 1KHz samples to find a preamble
[03:27] <hallam> yeah I don't know why they did that
[03:27] <Laurenceb> guess it doesnt mind DLL glitches
[03:27] <hallam> I already know the code phase of the start of each bit
[03:28] <Laurenceb> yeah, but you may end up off by some integer ms
[03:28] <Laurenceb> after some glitches
[03:28] <hallam> don't think so
[03:28] <Laurenceb> and the easiest way to match up is to correlate withthe preamble
[03:28] <Laurenceb> yeah, but its more robust
[03:28] <hallam> I know the time of each edge to better than half a microsecond
[03:29] <Laurenceb> yeah
[03:29] <hallam> I track continuously in code phase
[03:29] <hallam> no glitches afaik
[03:29] <Laurenceb> oh
[03:29] <Laurenceb> they track in constant numbers of samples
[03:29] <Laurenceb> then stretch the prn to match
[03:29] <hallam> suckers
[03:30] <Laurenceb> so the data bits slip
[03:30] <hallam> I like this way better
[03:30] <Laurenceb> we solved that problem at the start
[03:30] <hallam> right
[03:30] <Laurenceb> and its much faster
[03:30] <Laurenceb> :P
[03:30] <hallam> how slow is theirs/
[03:30] <hallam> ?
[03:30] <Laurenceb> 1/3 real time
[03:30] <Laurenceb> for one sat
[03:30] <hallam> oh, about the same then
[03:30] <Laurenceb> not sure what its running on tho
[03:30] <hallam> mine's about half real time
[03:30] <hallam> on my decent-ish laptop
[03:30] <Laurenceb> some random machine I found on my travels
[03:31] <Laurenceb> seems to be running debian
[03:31] <Laurenceb> I sshed in from a gateway machine, then setup X forwarding
[03:32] <hallam> I think it's bed for me
[03:32] <hallam> ttyl
[03:32] <Laurenceb> me too
[03:32] Laurenceb (i=83e3dd23@gateway/web/ajax/mibbit.com/x-f0a8cab886fd7b84) left irc: "http://www.mibbit.com ajax IRC Client"
[03:53] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-3abcd75ce62bed10) left irc: "http://www.mibbit.com ajax IRC Client"
[06:25] Simon-MPFH (n=simon@phantom.mpfh.co.uk) joined #highaltitude.
[07:04] Simon-MPFH (n=simon@phantom.mpfh.co.uk) left irc: "Leaving"
[08:07] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[08:18] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) left irc:
[08:57] Ebola (i=ebola@unaffiliated/ebola) got netsplit.
[08:57] fergusnoble (n=fergusno@fn217.quns.cam.ac.uk) got netsplit.
[08:59] fergusnoble (n=fergusno@fn217.quns.cam.ac.uk) returned to #highaltitude.
[08:59] Ebola (i=ebola@unaffiliated/ebola) returned to #highaltitude.
[09:19] M0TEK (i=836f0142@gateway/web/ajax/mibbit.com/x-3bae6152768f968f) joined #highaltitude.
[09:33] M0TEK (i=836f0142@gateway/web/ajax/mibbit.com/x-3bae6152768f968f) left irc: "http://www.mibbit.com ajax IRC Client"
[10:13] rharrison (n=rharriso@gateway.hgf.com) joined #highaltitude.
[10:14] <rharrison> Any radio heads out there. Can some one explain why I need to send a square wave to the tx port on the radiometrix to get a tone on fm.
[10:15] <rharrison> eg 1kHz
[10:15] <rharrison> It uses 5V CMOS logic
[10:15] <gordonjcp> rharrison: what would it transmit if you just fed it DC?
[10:16] <fergusnoble> rharrison: you should get a tone if you just feed it dc
[10:16] <fergusnoble> on ssb that is
[10:17] <rharrison> Yep I do
[10:17] <fergusnoble> the radiometrix uses the voltage input to shift the carrier frequency
[10:17] <rharrison> But unfortunalty i'm into FM this is for the challenge
[10:18] <fergusnoble> on fm the reciever outputs the amount of shift of the carrier
[10:18] <fergusnoble> so to get a tone you have to shift the carrier by an oscillating amount
[10:18] <fergusnoble> to get a tone out
[10:19] <rharrison> What I don't really understand is how you send data in FM. Steve said send it a square wave for a tone. Which I'm sure is right. I just don't understand how it works. If I want to do rtty so I send one freq. followed by another for the 1 and 0
[10:19] <gordonjcp> rharrison: you'd need to send two tone
[10:19] <fergusnoble> i.e. when your reciever is in fm you get something out roughly like what you put into the radiometrix
[10:20] <gordonjcp> rharrison: have you ever owned a ZX Spectrum>
[10:20] <rharrison> Ok I see so the sqr wave is occilating causing the deviation hence the tone
[10:20] <fergusnoble> rharrison: exactly, you use one frequency for the mark and a different one for the space (1 and 0)
[10:21] <rharrison> gordonjcp funny enough I did have a zx81
[10:21] <fergusnoble> and they should be in the middle of the audio range so your radio reciever doesnt filter them out, as its designed for audio
[10:21] <fergusnoble> and spaced apart by one of the standard shifts used in rtty like 425Hz
[10:21] <rharrison> Ok cool and if i read the data sheet I should get some idea of what occilation to use
[10:22] <gordonjcp> rharrison: yup
[10:22] <gordonjcp> rharrison: tbh squarewaves are fine, but pass it through a lowpass filter to knock the edges off a bit
[10:22] <rharrison> So the occilation should be spaced by the shift. 1Khz and 1.425khz
[10:22] <gordonjcp> rharrison: that would do nicely
[10:22] <fergusnoble> so something like 1000Hz and 1425Hz would be ideal
[10:22] <fergusnoble> hehe
[10:23] <rharrison> hehe
[10:23] <fergusnoble> beat me to it
[10:23] <rharrison> cool I understand a lot better now
[10:23] <rharrison> I think the RM has a LPF on it anyhow so I may get away with out having to filer
[10:23] <fergusnoble> rharrison: if you can use a sine wave instead you will get more performance out of the rtty decoder
[10:24] <fergusnoble> rharrison: yeah, the radiometrix should lpf the worst of it away
[10:24] <rharrison> Ahh and generate the sin wave at the different freq.
[10:24] <rharrison> Cool
[10:24] <fergusnoble> but a sine wave is quite easy if you use a lookup table
[10:24] <gordonjcp> overkill
[10:24] <rharrison> Surely there must be a little IC for sin wave generation
[10:25] <rharrison> and use that to sent to the RM
[10:25] <rharrison> RM = Radiometrx
[10:25] <gordonjcp> you can actually buy modem ICs
[10:26] <fergusnoble> the decoder basically finds the difference in how much youve got of each frequency, if you use a square wave there is a chance that some of the harmonics of the square wave will appear to the decoder as the other frequency than the one your trying to transmit, and so reduce the signal to noise
[10:26] <fergusnoble> its called intersymbol interference
[10:27] <fergusnoble> but like gordonjcp says im sure its not really much of an issue in this case
[10:27] <fergusnoble> probably best to just give it a go
[10:27] <rharrison> Ok great I'll test and I really apprecite the 5 min on radio practical
[10:27] <rharrison> Alot clearer understanding now
[10:28] <fergusnoble> np
[10:29] <fergusnoble> rharrison: another advantage of fm is your frequency wont drift as the temperature changes hardly at all
[10:30] <gordonjcp> fergusnoble: uhm
[10:30] <gordonjcp> how would a harmonic of 1kHz affect AFSK with tones of 1kHz and 1.45kHz?
[10:30] <gordonjcp> what would the second harmonic be?
[10:32] <fergusnoble> the first harmonic would be 3khz :)
[10:32] <fergusnoble> if its a perfect square
[10:33] <fergusnoble> but you can see on the waterfall on the square edges of our radio transitions a wide broadband blip
[10:34] <fergusnoble> we only change on the bit edges so its not so much of an issus, if you were doing that 1000 times a second you might be in more trouble
[10:38] <gordonjcp> fergusnoble: the first harmonic would, by definition, be 1kHz
[10:38] <fergusnoble> i would call that the fundamental
[10:38] <gordonjcp> 2nd harmonic would be 2kHz and 3dB down, 3rd harmonic would be 3kHz and 6dB down
[10:38] <gordonjcp> fergusnoble: yes
[10:39] <gordonjcp> most people do
[10:39] <gordonjcp> no-one uses the term "first harmonic" because it's misleading ;-)
[10:40] <fergusnoble> ok, sorry
[10:40] <gordonjcp> as it happens a square wave hasn't got much 2nd harmonic anyway, so you're really only going to get the third and fifth harmonics
[10:40] <gordonjcp> all of which will be pretty damn close to being off the edge of the world anyway
[10:40] <fergusnoble> yeah, fourier series of a square wave is just odd multiples of the fundamental
[10:40] <gordonjcp> precisely
[10:41] <gordonjcp> so you pick non-harmonically-related frequencies for your tones
[10:41] <gordonjcp> it's interesting to note that if you use SSB you can receive the FSK at any frequency you like
[10:41] <gordonjcp> and harmonically-related pairs *are* harder to decode, for the reasons you outlined
[10:42] <fergusnoble> gordonjcp: ok, so maybe its not the harmonics but when you put a square edge through the radiometrix you get a broadband blip
[10:42] <gordonjcp> fergusnoble: hence knocking the edges off with a lowpass filter
[10:42] <gordonjcp> in an ideal world you *would* use a proper sinewave, but you've got to ask if the tradeoff is worth it
[10:43] <fergusnoble> for me at least i prefer software changes to hardware ones, so i would see generating a sine in software as win win
[10:43] <gordonjcp> you gain a small fraction of a dB better SNR, but you've drastically increased the brainpower needed to generate the tone
[10:43] <gordonjcp> now you need a faster MCU, with higher power requirements
[10:43] <gordonjcp> *now* you need a bigger battery
[10:44] <fergusnoble> gordonjcp: i get the point
[10:44] <gordonjcp> is half a dB worth that, compared to adding a resistor and a capacitor?
[10:44] <fergusnoble> gordonjcp: but it really doesnt take much grunt to look up an array index
[10:44] <gordonjcp> no, absolutely
[10:44] <gordonjcp> if you're already using something capable of it then go to it and good luck
[10:44] <fergusnoble> and that can lead in to pulse shaping or more advanced encodings like MFSK
[10:45] <gordonjcp> yup
[10:45] <fergusnoble> or phase based encodings
[10:45] <gordonjcp> or you could use a single-chip GMSK modem
[10:45] <fergusnoble> meh, i think its really a matter of personal preference rather than a clear cut engineering choice
[10:46] <fergusnoble> i prefer to have one bit of hardware that gives a lot of flexibility through software changes
[10:46] <gordonjcp> fergusnoble: broadly speaking, my point is that you shouldn't overthink these things if you haven't got *anything* working at all
[10:46] <gordonjcp> ie. if you're just starting from scratch, try the simplest thing that could work
[10:47] <fergusnoble> indeed, and thats what i suggested doing
[10:47] <gordonjcp> feeding a squarewave from the counter/timer through a little lowpass filter to the TX is just about as simple as you can get
[10:48] <fergusnoble> ok, got to dash to a lecture
[10:48] <gordonjcp> of course I'm coming at this from the POV of using a very low power single-chip MCU - if you're using a proper board then you've got the power to use proper sinewaves
[10:48] <gordonjcp> okay, take care
[10:48] <gordonjcp> ttyl
[10:48] <fergusnoble> but gordonjcp you are right, its obv better to get something working first
[10:49] <fergusnoble> where you go from there, there are many choices
[10:55] <rharrison> Not wanting to look too thick here but I can't see which frequency I should be starting from ideally. Here is the link to the datasheet http://www.radiometrix.co.uk/dsheets/ent1enr1.pdf
[10:56] <gordonjcp> rharrison: frequency as in RF output, or frequency as in AF modulation?
[10:56] <rharrison> frequency as in entry to the tx pin on the transmitter
[10:56] <rharrison> ie square wave
[10:57] <gordonjcp> okay
[10:57] <gordonjcp> by "starting from", do you mean "frequency when I'm not sending data"?
[10:58] <rharrison> Yep I guess so
[10:58] <rharrison> Call that a 1 I think you send high in rtty when doing nothing
[10:58] <gordonjcp> yup
[10:59] <gordonjcp> what baud rate were you planning?
[10:59] <rharrison> 50 baud
[10:59] <gordonjcp> okay generally the "space" tone is higher
[10:59] <rharrison> 425Hz shift
[10:59] <gordonjcp> so pick two frequencies that you can set your receiver to and let it rip
[11:01] <gordonjcp> as you and fergusnoble discussed earlier, 1kHz and 1.425kHz would be fine
[11:01] <rharrison> So if I send 1kHz into the tx to I get a signal 1k off 169.406
[11:01] <rharrison> Is that sort of how it works
[11:01] <gordonjcp> is the TX transmitting on 169.406?
[11:01] <rharrison> Yep
[11:01] <gordonjcp> ah, no
[11:02] <gordonjcp> FM modulates the frequency of the carrier, hence the name
[11:02] <gordonjcp> so you'd find that the output frequency is wobbling up and down around 169.406
[11:03] <rharrison> and that wobble is changed by the freq. I pump in the tx end
[11:03] <gordonjcp> rharrison: if you had a slow sinewave and the transmitter was DC coupled, you could measure the frequency and watch it slide up and down a few kHz
[11:03] <gordonjcp> bingo
[11:03] <gordonjcp> stick in 1Hz, it wobbles up and down at 1Hz
[11:03] <rharrison> And that wobble get converted to sound by the radio
[11:03] <gordonjcp> yup
[11:03] <rharrison> Right I'm there
[11:03] <rharrison> Sorry I know bugger all about radio in practice
[11:04] <gordonjcp> an easy way to visualise it is to think about the way that people used to detect FM, with slope detection
[11:04] <rharrison> The clouds are slowly lifting though
[11:04] <gordonjcp> you know how an AM radio hasn't got a sharp "peak" when it's tuned in, but has sloping sides?
[11:04] <rharrison> yep
[11:04] <gordonjcp> right
[11:04] <rharrison> Sort of
[11:05] <gordonjcp> well tune an AM radio a little bit above an FM signal so it's not *quite* tuned in properly
[11:05] <gordonjcp> as the F is M'ed up a bit, it "rides up the slope" as it were
[11:05] <gordonjcp> and you get really really shitty audio
[11:05] <rharrison> Ok i see that
[11:05] <rharrison> I can do that on my radio and see that
[11:05] <gordonjcp> we do it better ways now
[11:06] <gordonjcp> yup
[11:06] <gordonjcp> local FM broadcast on 96.3MHz, tune on AM to 96.305MHz and you'll hear crappy distorted but recognisable sound
[11:06] <gordonjcp> rharrison: the other thing is deviation
[11:06] <rharrison> the 1k deviation then going back a bit would be an occilation +- .5k either side of the carrier?
[11:07] <gordonjcp> which is not nearly as fun as it sounds
[11:07] <gordonjcp> exactly
[11:07] <gordonjcp> the stronger the modulation, the deeper the deviation
[11:07] <rharrison> And the more space you need between channels?
[11:08] <gordonjcp> yup
[11:09] <gordonjcp> but the better the quality of the signal
[11:09] <gordonjcp> you know how "broadband" gives you a faster internet connection?
[11:09] <gordonjcp> because you pack more information into a particular moment in time?
[11:09] <gordonjcp> same idea
[11:10] <rharrison> hence the name broad band
[11:10] <rharrison> Cool Right well thanks very much gordonjcp: I have a better idea of what I'm doing now
[11:10] <gordonjcp> ayup
[11:11] <gordonjcp> rharrison: if you look at various PSK modes in a waterfall then it actually shows up as a band
[11:11] <gordonjcp> and the faster modes are quite, well, broad
[11:17] <rharrison> Humm should I be using PWM to generate the frequency on the avr
[11:18] <gordonjcp> that would work
[11:18] <gordonjcp> simple software fix, too ;-)
[11:18] <gordonjcp> rharrison: you might find you don't need to be too accurate with your sinewave - even a couple of steps would do
[11:19] <rharrison> OK cool
[11:20] <gordonjcp> I have a modem that approximates a sine wave with two squarewaves
[11:21] <gordonjcp> or something like that
[11:22] <gordonjcp> anyway imagine a squarewave, with a little square box on the top of the middle of each cycle
[11:22] Action: SpeedEvil ponders pirate DAB radio.
[11:42] edmoore (n=edmoore@pomegranate.chu.cam.ac.uk) joined #highaltitude.
[12:52] <rharrison> Any one got anything excitng planned for the w/e
[12:53] Action: rharrison thinks launches...
[13:13] gregHome (n=gleblanc@75.108.33.75) left irc: Read error: 113 (No route to host)
[13:39] <fergusnoble> rharrison: are the wids good?
[13:41] <fergusnoble> edmoore: going to HoM?
[13:42] <fergusnoble> edmoore: i took another overnight stress test of the badger loogin and no corrption
[13:42] <fergusnoble> so its looking good
[13:43] Action: SpeedEvil plans on taking over the WORLD!
[13:43] <SpeedEvil> Exactly how is something I'm still brainstorming.
[13:44] <SpeedEvil> But you'll know by monday.
[13:44] <fergusnoble> SpeedEvil: isnt that the same thing yo do every night?
[13:44] <SpeedEvil> Pretty much.
[13:45] Action: SpeedEvil puts an ad in craigslist for unwanted minions.
[13:48] Laurenceb (i=83e34f19@gateway/web/ajax/mibbit.com/x-40fd48d5906cbcdd) joined #highaltitude.
[13:48] <Laurenceb> hello
[13:48] <SpeedEvil> Hello!
[13:49] Action: SpeedEvil wonders if Laurenceb reads craigslist.
[13:49] <Laurenceb> dont even know what it is
[13:49] <Laurenceb> well I have a vague idea
[13:49] Action: SpeedEvil just being silly, nvm.
[13:49] <gordonjcp> anyone following the JAXA launches?
[13:50] <SpeedEvil> Nope.
[13:50] <SpeedEvil> Anything fun?
[13:51] <gordonjcp> just heard KKS-1, very loud
[13:51] <SpeedEvil> :)
[13:51] <SpeedEvil> What's it, a packet relay?
[13:56] <Laurenceb> where is the balloon?
[13:57] <gordonjcp> SpeedEvil: don't know
[14:03] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-69b7fc7a11e98948) joined #highaltitude.
[14:04] <fergusnoble> hallam: hello
[14:04] <Laurenceb> hi hallam
[14:05] <Laurenceb> slept well?
[14:21] <Laurenceb> wow this is odd
[14:21] <Laurenceb> ! ran the newton raphson estimator at 20Hz over all the data then averaged
[14:22] <Laurenceb> if you plot it in 3D, theres voids in the spread of poiints
[14:22] <SpeedEvil> 20Hz?
[14:22] <Laurenceb> almost like bubbles
[14:22] <Laurenceb> yeah
[14:22] <Laurenceb> so averga epseudorange over 50ms
[14:22] <Laurenceb> then solve
[14:22] <SpeedEvil> How many points are tehre?
[14:23] <Laurenceb> 800
[14:23] <SpeedEvil> it'll just be a random walk won't it?
[14:23] <Laurenceb> no
[14:23] <SpeedEvil> Unless you have a lot of datapoints, you're gonna get bubbles
[14:23] <Laurenceb> its reaches a valid solution after about 4 iterations
[14:23] <SpeedEvil> well - not random - random constrained by axes where there are no noise
[14:24] <SpeedEvil> Oh - right.
[14:24] <SpeedEvil> you're getting multiple consecutive estimates?
[14:24] <SpeedEvil> like you might if you had it outputting a position at 20Hz
[14:24] <Laurenceb> yes
[14:25] <SpeedEvil> What's the error like if you do that - and if you do it over ~1s?
[14:25] <Laurenceb> noy sure, due to the multipath
[14:25] <edmoore> fergusnoble: sorry was away
[14:25] <edmoore> ok cool, that's good news
[14:25] <SpeedEvil> apparent error, inc multipath
[14:25] <Laurenceb> brobaby 5m or so due to the noise I'd say
[14:25] <SpeedEvil> nice
[14:25] <Laurenceb> oh about 10 to 15 overall
[14:25] <fergusnoble> edmoore: abot to implement some domino
[14:26] <Laurenceb> but henrys sats are pretty bad
[14:26] <fergusnoble> edmoore: looks easy enogh if you have the fldigi source :)
[14:26] <edmoore> can we try some vnc action?
[14:26] <edmoore> haha
[14:26] <Laurenceb> they are all in a line for starters
[14:26] <edmoore> ok
[14:26] <edmoore> wow
[14:26] <fergusnoble> iuw
[14:26] <fergusnoble> how do i set it up my end?
[14:26] <fergusnoble> actually, pm me
[14:26] <fergusnoble> or repeater, im tuned in
[14:26] <edmoore> well if it's just trimming someone elses c++ into c... though actually it'd be good to try
[14:26] <edmoore> are you on your eee?
[14:27] <fergusnoble> no, hackintosh
[14:28] <edmoore> can you hear me on the repeater?
[14:29] <fergusnoble> no, try again
[14:29] <Laurenceb> theres a tropospheric correction
[14:29] <SpeedEvil> and ionospheric too
[14:29] <Laurenceb> it integrates over an atmospheric model
[14:29] <Laurenceb> yeah
[14:29] <Laurenceb> thats simpler
[14:29] <Laurenceb> as you can use ephemeris data
[14:30] <edmoore> oh i think i fiddles earlier
[14:30] <edmoore> let me try something
[14:30] Action: SpeedEvil forgets.
[14:30] <SpeedEvil> L2 is broadcast in a different band?
[14:30] <edmoore> fergusnoble: how about just then?
[14:31] <fergusnoble> heard you, but i dont seem to have much tx power
[14:31] <edmoore> moving my antenna away from the window frame has taken to to S9
[14:32] <Laurenceb> SpeedEvil: yeah
[14:35] <Laurenceb> hmm my cart2geo function uses ED50
[14:35] <Laurenceb> not WGS84
[14:35] <Laurenceb> I'm guessing google maps is WGS84
[14:35] <Laurenceb> it may exoplain the shift
[14:39] <edmoore> natrium42: yo
[14:39] <Laurenceb> http://maps.google.co.uk/maps?f=q&source=s_q&hl=en&geocode=&q=%4052.2020729722222,0.118206&sll=52.202147,0.118181&sspn=0.000291,0.000603&ie=UTF8&ll=52.202167,0.11861&spn=0.000582,0.001207&t=h&z=20
[14:40] <Laurenceb> works slightly better if I average pseudrange for 1 second then do newton raphson
[14:40] <Laurenceb> thats about 10m total error
[15:02] <Laurenceb> I think it maybe just suffering from multipath, but the height is a bit off
[15:03] <Laurenceb> It converts from X,Y,Z corords to UTM, then to WGS84
[15:03] <Laurenceb> but I think the WGS84 conversion may miss the altitude off
[15:13] <Laurenceb> damnd geoids
[15:48] <Laurenceb> hmm they seem to have grabbed the code out of a 1950s text book
[15:57] fergusnoble (n=fergusno@fn217.quns.cam.ac.uk) left #highaltitude.
[16:00] Bluenarf (n=Paul@apollo.paulsnet.org) joined #highaltitude.
[16:00] Nick change: Bluenarf -> EI5GTB
[16:03] <Laurenceb> hi Ei5GTB
[16:03] <EI5GTB> hai
[16:04] <EI5GTB> hows things now?
[16:05] <Laurenceb> not sop bad
[16:05] <EI5GTB> gdgd
[16:05] <Laurenceb> trying to improve this gps code
[16:05] <EI5GTB> i see, whats it doingf?.
[16:05] <Laurenceb> its about 10m off atm
[16:06] <Laurenceb> finding where hallam lives
[16:11] <rharrison> EI5GTB: How is it going
[16:11] <EI5GTB> 'ello
[16:11] <EI5GTB> not too bad
[16:11] <rharrison> Hoe to get a doc together on how to progress challenge
[16:11] <EI5GTB> Laurenceb, what does it do? parse nmea> or work out the actual position from raw data?
[16:11] <rharrison> I have some modules from radiometrix and with a bit of luck we should be doing a range check in feb
[16:12] <EI5GTB> cool
[16:12] <Laurenceb> EI5GTB: it works out the position from ADC samples
[16:12] <EI5GTB> oh, nice
[16:12] <rharrison> EI5GTB: Raw data... Those buggers are mad
[16:12] <EI5GTB> lol
[16:12] <EI5GTB> whats wrong with nmea?
[16:12] <Laurenceb> its limited
[16:12] <rharrison> You should read the logs from the last 48 hours
[16:12] <Laurenceb> you cant stick it on rockets
[16:14] <EI5GTB> i see
[16:14] <Laurenceb> also you cant combine it with a mems IMU
[16:17] <EI5GTB> i see
[16:17] <Laurenceb> well, not optimally
[16:19] <SpeedEvil> You need to be able to input the accelerometer, gyro inputs back into the stage where the position is determined really, for optimal results. With NMEA, you just get 'cooked' data.
[16:21] <Laurenceb> well not just that
[16:21] <rharrison> EI5GTB PM
[16:21] <Laurenceb> something I hadnt realised before is you have to loosen up the tracking loops for it to work
[16:21] <Laurenceb> in high dynamic environments
[16:22] <Laurenceb> hopefully a combined IMU/GPS will get round that problem
[16:22] <SpeedEvil> I mean - if you can push the loops directly with the accels, then you don't need to loosen up the loops
[16:22] <Laurenceb> yeah
[16:22] <Laurenceb> or better lose them completely
[16:22] <SpeedEvil> yeah
[16:22] <Laurenceb> you dont need the PLL locked properly
[16:22] <Laurenceb> if you use code rate for velocity correction
[16:23] <Laurenceb> and a noncoherent code phase measurement
[16:23] <SpeedEvil> That implies FFT or similar doesn't it?
[16:23] <Laurenceb> no
[16:23] <SpeedEvil> Which implies large hardware.
[16:23] <Laurenceb> nope, just abs(prompt)
[16:24] Action: SpeedEvil needs food.
[16:24] <SpeedEvil> ah - yes.
[16:24] <Laurenceb> rather than real(prompt)
[16:24] <SpeedEvil> Things in many ways get lots easier if you don't care about low signal environments
[16:24] <SpeedEvil> With the gains and sensitivities of modern chipsets.
[16:25] <SpeedEvil> especially with decent antennas
[16:25] Action: SpeedEvil ponders the use of multiple digitising chipsets with multiple antennas.
[16:25] <Laurenceb> accels are very good as well
[16:25] <Laurenceb> been done
[16:25] <SpeedEvil> I assume it has.
[16:26] <SpeedEvil> X - -X antenna, ...
[16:26] <Laurenceb> http://www.sstl.co.uk/assets/Downloads/SGR-20.pdf
[16:27] <SpeedEvil> As an ideal omni antenna that doesn't suffer from hairy ball.
[16:28] <rharrison> I couldn't see the price on the flyer
[16:28] <Laurenceb> a lot
[16:28] <rharrison> hehe
[16:28] <Laurenceb> more than your house
[16:32] <Laurenceb> hmm where to by a streo to stereo jack...
[16:32] <Laurenceb> need to connect my icom to my laptop
[16:37] <SpeedEvil> got any broken headphones?
[16:38] <Laurenceb> nope :/
[16:42] <hallam> Laurenceb: is it stereo or mono? stereo cables don't work well with my icom (ic-r20)
[16:42] <Laurenceb> oh
[16:43] <Laurenceb> any idea why?
[16:43] <Laurenceb> they dont make contact correctly?
[16:43] <Laurenceb> it should just be mono yeah
[16:43] <hallam> yeah, I'm not quite sure what it is but the contacts are in the wrong place
[16:43] <Laurenceb> ok... I've just run the nav filter at 1KHz
[16:44] <Laurenceb> hmm
[16:44] <hallam> stereo sockets can use mono cables but not the other way round
[16:44] <Laurenceb> I'm might wander over to curries/homebase ect
[16:44] <hallam> question before you go
[16:44] <hallam> were you trying to code the tracking in C at one point?
[16:44] <Laurenceb> no I'm not going now
[16:44] <Laurenceb> yes
[16:44] <hallam> how did that go?
[16:44] <Laurenceb> I've got c tracking code
[16:44] <Laurenceb> works fine
[16:45] <hallam> how much faster than matlab?
[16:45] <Laurenceb> I can pastebin it if you want
[16:45] <Laurenceb> about an order of magnitude
[16:45] <hallam> sure I'd love a copy of that
[16:45] Action: SpeedEvil wishes 'countones' was a common instruction.
[16:46] <Laurenceb> http://www.mibbit.com/pb/k9jKY4
[16:46] <hallam> thanks!
[16:46] <SpeedEvil> (read in lots of samples, xor, countones)
[16:46] <Laurenceb> http://www.mibbit.com/pb/k8xo0h
[16:47] <hallam> SpeedEvil: you mean dot product?
[16:47] <hallam> Laurenceb: that one does acquisition as well?
[16:48] <Laurenceb> erm the second one is just ca code gen
[16:48] <hallam> qait
[16:48] <Laurenceb> first one is aquisition and code gen
[16:48] <hallam> first one looks like it does acquisition
[16:48] <hallam> ok
[16:48] <Laurenceb> you need fftw3
[16:48] <hallam> yeah I'll get that
[16:49] <SpeedEvil> hallam: instead of doing readsignalbit,MAC with local PRN. You do read 32 signal bits, XOR with local copy, countones.
[16:49] <hallam> I was planning to do the acq in matlab and then pass off to C for tracking
[16:49] <SpeedEvil> hallam: for a many, many times speedup.
[16:49] <Laurenceb> he
[16:50] <hallam> SpeedEvil: sure, but if I were implementing such an instruction I'd combine the xor and countones, and call it a dot product
[16:51] <SpeedEvil> hallam: sure
[16:51] <SpeedEvil> hallam: I was just meaning that some processors have had this instruction - the transputer forex.
[16:51] <hallam> heh
[16:51] <SpeedEvil> Elektor reported on a GPS using one, back in the day.
[16:51] <hallam> how about a GPS on a Propellor>
[16:51] <SpeedEvil> ~1988?
[16:51] <hallam> ?
[16:51] <Laurenceb> yeah
[16:52] <Laurenceb> theres one here
[16:52] <Laurenceb> based on that design
[16:52] Action: SpeedEvil hasn't really looked at propellor much.
[16:52] <SpeedEvil> I probably should.
[16:52] <Laurenceb> hallam: where are you?
[16:52] <SpeedEvil> But my todo stack is overflowing.
[16:52] <hallam> home
[16:52] <Laurenceb> as in where was the ant?
[16:52] <hallam> 2 pem st
[16:52] <hallam> you want the lat and lon again?
[16:53] <Laurenceb> can you give me a point accurate to a meter or so?
[16:53] <hallam> I'm not sure, it's hard to say from google maps
[16:53] <hallam> I'll try live earth
[16:53] <SpeedEvil> Don't assume google images is accurate
[16:53] <hallam> yeah it's not really, at the meter level
[16:53] <SpeedEvil> In at least some places, it's metres off.
[16:53] <Laurenceb> its about 20m out atm
[16:54] <SpeedEvil> You might check with GPS logs from openstreetmap in the area
[16:54] <Laurenceb> odd as I ran it at 1KHz
[16:54] <Laurenceb> its 10m out at 1Hz
[16:54] <hallam> Laurenceb: what's the distribution like?
[16:54] <hallam> the ant was completely stationary
[16:54] <SpeedEvil> Is it 4 sats?
[16:54] <hallam> yeah
[16:54] <hallam> and one of them is noisy as hell
[16:55] <SpeedEvil> trying dropping a sat out of the solution might be entertaining.
[16:55] <hallam> btw, I did manage to track sv32, had to increase the DLL loop gain
[16:55] <hallam> can you do it with 3? have to assume altitude=0?
[16:55] <SpeedEvil> yes
[16:55] <SpeedEvil> well - no
[16:55] <Laurenceb> hmm
[16:55] <Laurenceb> yeah
[16:55] <SpeedEvil> you assume altitude = known
[16:55] <hallam> right
[16:55] <Laurenceb> its not a very good plan
[16:55] <hallam> this is Cambridge, it's 0
[16:55] <Laurenceb> as I'm not sure about this geoid code
[16:56] Action: SpeedEvil watches the waves wash over hallam.
[16:56] <Laurenceb> its oput by about 100m
[16:56] <SpeedEvil> Or you can fix lat and lon, and solve for alt
[16:56] <Laurenceb> the distridution is a disc
[16:56] <Laurenceb> about 80m radius
[16:56] <hallam> Laurenceb: is it WGS84 or somethign else?
[16:56] <Laurenceb> wel...
[16:56] <Laurenceb> I think it is for lat and long
[16:56] <Laurenceb> as google maps is so close
[16:57] <Laurenceb> but for altitude I think itsd international geoid 1924
[16:57] <Laurenceb> looking at the comments in the code
[16:57] <Laurenceb> its taken froma 1951 book
[16:58] <hallam> edmoore: are you there? and do you have one of those xmos dev things?
[16:58] <Laurenceb> how do I link to a lat, long in open streetmap?
[16:58] <SpeedEvil> umm
[16:58] <SpeedEvil> I think you can search for it
[16:58] <SpeedEvil> I just use JOSM
[16:59] <SpeedEvil> (Java Open Streetmap EditorM)
[16:59] <Laurenceb> JOSM?
[16:59] <Laurenceb> no java on here
[16:59] <SpeedEvil> click 'permalink' and then edit
[17:01] <Laurenceb> http://www.openstreetmap.org/?lat=52.20209575&lon=0.1181225&zoom=18&layers=B000FTF
[17:01] <Laurenceb> hard to see where the center is
[17:01] <hallam> I think the XMOS would be well-suited to GPS tracking
[17:01] <Laurenceb> is that a multicore?
[17:01] <SpeedEvil> Now click 'edit' and then 'show GPS tracks'
[17:02] <hallam> yeah, 4-cores at 400MIPS each
[17:02] <SpeedEvil> you will need to make an account
[17:02] <hallam> in a chip
[17:02] <hallam> each core can run 8 hardware threads
[17:02] <Laurenceb> wow
[17:03] <Laurenceb> they certainly put emphasis on ease of use
[17:03] <Laurenceb> the pdf tutorial make it look like piss
[17:03] <SpeedEvil> What tutorial?
[17:04] <Laurenceb> on the xmos site
[17:05] <hallam> of course, an FPGA would also work
[17:05] <Laurenceb> hallam: are you facing onto pembroke st?
[17:05] <SpeedEvil> ah
[17:05] <hallam> yes
[17:05] <Laurenceb> the place with the lead roof?
[17:06] rharrison (n=rharriso@gateway.hgf.com) left irc:
[17:07] <Laurenceb> hmm the positions are in a region 20mx100x80
[17:07] borism (n=boris@195-50-211-250-dsl.krw.estpak.ee) joined #highaltitude.
[17:09] <Laurenceb> its probably worth looking for some better lat/long/alt calculation code
[17:09] <SpeedEvil> Or just get another sample in a better location, or with a known GPS.
[17:09] <Laurenceb> the distribution of the positions is fine
[17:09] <Laurenceb> yeah
[17:10] <Laurenceb> but the average point just seems toi be wrong
[17:10] <Laurenceb> I remeember when I first started playing about with gps, google maops was about 100m out
[17:10] <Laurenceb> but they seem to have fixed that
[17:10] jcoxon (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) joined #highaltitude.
[17:10] <Laurenceb> hallam: http://maps.google.co.uk/maps?f=q&source=s_q&hl=en&geocode=&q=%4052.20209575,0.1181225&sll=52.202105,0.118362&sspn=0.00028,0.000603&ie=UTF8&ll=52.202096,0.118348&spn=0.000291,0.000603&t=h&z=21&iwloc=addr
[17:10] <Laurenceb> hi jocoxon
[17:11] <hallam> 0.1181022789285464, 52.20197972383436
[17:11] <hallam> precise enough for you? :P
[17:11] <Laurenceb> with 1Hz pseudorange averaging, its closer
[17:11] <jcoxon> hey Laurenceb
[17:11] <Laurenceb> but only a bit
[17:11] <Laurenceb> and theoretically 1KHz should be a bit better
[17:12] <Laurenceb> I think the hacked up XYZ to lat long alt functions in softsngg are to blame
[17:13] <Laurenceb> altitude is -77m
[17:13] <hallam> Lauyrence, if you're typing SoftSngg you should probably get some sleep
[17:13] <Laurenceb> should be about 55meters
[17:13] <Laurenceb> lol
[17:14] <Laurenceb> WGS84 is about 45m over cambridge
[17:14] <Laurenceb> oh crap
[17:14] <Laurenceb> no.. arg
[17:15] <Laurenceb> yeah, whuich makes sea level 45m
[17:15] <Laurenceb> so you should be about 55m
[17:15] <Laurenceb> oh on the corner wow
[17:16] <Laurenceb> in that caase 1KHz works very well :D
[17:17] <Laurenceb> about 8meters out
[17:17] <SpeedEvil> You mean you've got the alt right?
[17:17] <Laurenceb> no
[17:17] <SpeedEvil> That's damn fine.
[17:17] <SpeedEvil> If it's 8m out
[17:17] <Laurenceb> horizontal error=8 meters
[17:17] <SpeedEvil> ah
[17:17] <Laurenceb> altitude is erm -77m
[17:17] <Laurenceb> which is kind of wrong
[17:17] <Laurenceb> so it looks like its just a case iof the correct geoid
[17:17] <SpeedEvil> you haven't got the sign wrong anywhere?
[17:18] <Laurenceb> I'm a bit confused to be honest
[17:18] <SpeedEvil> Though actually, 77m is ratehr too high for that building
[17:18] <Laurenceb> my lassen iq says 96m for my room here
[17:18] <Laurenceb> and the geoid is 45m
[17:18] <Laurenceb> 96-45=51m
[17:18] <Laurenceb> and ordinance survey+height iof windowsill
[17:19] <Laurenceb> =51m +-1meter or so
[17:19] <Laurenceb> so I was going off that
[17:19] <Laurenceb> but - lassen iq uses WSG84
[17:19] <SpeedEvil> Err
[17:19] <Laurenceb> this altitude code seems to use international 1924
[17:19] Action: SpeedEvil vaguely recalls some GPSs may output WGS84 height too
[17:20] <Laurenceb> I looked it up
[17:21] <Laurenceb> now, it I look up Internation geoid 1924...
[17:24] <hallam> Laurenceb: the code you pastebinned does acq, but did you do any that does tracking?
[17:24] <Laurenceb> hallam: nope fraid not
[17:24] <hallam> ah ok no probs
[17:24] <hallam> I'll write some
[17:24] <Laurenceb> hmm
[17:25] <Laurenceb> its about6.5 meters if you assume google hasnt accounted for the fact the photos arent from zenith
[17:25] <hallam> I think that's pretty good, considering ionosphere errors
[17:26] <Laurenceb> yeah its pretty much the limit
[17:26] <Laurenceb> especially considering the very poor signals
[17:26] <Laurenceb> and bad alignment
[17:26] <hallam> Laurenceb: if you want something to try, have a go at downsampling it to 4MHz
[17:26] <Laurenceb> ok, but not right now :P
[17:26] <hallam> hehe
[17:27] <hallam> do you think it's ok to do that just by setting every other sample to 0 I and Q, and using the same code without modification?
[17:27] <Laurenceb> I'm investingating geoid seperation
[17:27] <Laurenceb> yeah
[17:27] <hallam> or do I have to actually downsample it and recode the matlab to work at 4MHz
[17:27] <hallam> cool, I'll try it
[17:27] <Laurenceb> as you cant lower the lowpass filters on I and Q
[17:28] <Laurenceb> and most of the noise isnt from the ADC
[17:28] <Laurenceb> so you probably cant get much better than just taking every other sample
[17:28] <Laurenceb> also, thats a valid approximation of running the SE4120 in the 4MHz mode
[17:29] <Laurenceb> as all that happens is the PLL for the sample clock is reconfigured
[17:30] <hallam> yeah that's what I was trying to simulate
[17:30] <hallam> hm, acq isn't nearly as good
[17:30] <hallam> no longer finds 13 and 32
[17:30] <SpeedEvil> Important navigation note - http://xkcd.com/
[17:31] <hallam> I'll bear that in mind
[17:32] <hallam> Laurenceb: after decimation like that, should I adjust the acq threshold?
[17:32] <jcoxon> hehe cost of skynet 999999999
[17:32] <Laurenceb> hallam: yeah
[17:32] <Laurenceb> halve it
[17:32] <hallam> ok
[17:32] <Laurenceb> https://www.og.berr.gov.uk/regulation/guidance/co_systems/co_sys_06.htm
[17:33] <Laurenceb> ED50 is lower
[17:33] <Laurenceb> erm... thats not good
[17:34] <Laurenceb> unless its using WGS84
[17:34] <Laurenceb> in which case ialtitude is <30m out
[17:34] <Laurenceb> but thats not what the comments in the code say
[17:35] <Laurenceb> http://www.mibbit.com/pb/3Uo1I6
[17:38] jcoxon (n=jcoxon@host86-158-31-172.range86-158.btcentralplus.com) left irc: "Leaving"
[17:39] <hallam> Laurenceb: do you have your blue stick with you?
[17:40] EI5GTB (n=Paul@apollo.paulsnet.org) left irc: "Leaving"
[17:42] <hallam> http://i43.tinypic.com/dd13jm.png how many sats am I searching for, and how many did I find?
[17:43] <Laurenceb> not atm
[17:43] <hallam> after dinner I'll go out and get some more data from a few different points
[17:43] <Laurenceb> hallam: 32
[17:44] <Laurenceb> yeah, good plan
[17:44] <Laurenceb> ooh dual core
[17:44] <hallam> no, not 32
[17:44] <Laurenceb> fancy
[17:44] <hallam> yeah I built this laptop from barebones and I rigged it for number crunching
[17:44] <hallam> 4GB ram
[17:45] <hallam> look at the memory usage graph
[17:46] <Laurenceb> oh lol
[17:46] <Laurenceb> 17?
[17:46] <Laurenceb> and you found 5 ?
[17:46] <hallam> heh, not quite
[17:47] <hallam> searched for 5 (five big peaks from the coarse acquisition)
[17:47] <hallam> found 4 (four smaller ones from the fine acq)
[17:47] <Laurenceb> usign the c code?
[17:47] <hallam> nah that's still matlab
[17:47] <hallam> I think I'll actually leave the acquisition in matlab for now
[17:48] <hallam> it's the tracking I'm really interested in
[17:48] <Laurenceb> oh hang on
[17:49] <Laurenceb> the ellipsoid height on that page is the height of the test point
[17:49] <Laurenceb> so... if it goes down 46m from WGS to ED, ED is higher
[17:49] <Laurenceb> WGS is 45m
[17:50] <Laurenceb> 45+46=91
[17:50] <Laurenceb> 91-77=
[17:50] <Laurenceb> erm 14m
[17:52] <Laurenceb> ok, get a map says 10m altitude
[17:52] <Laurenceb> MSL - your windowsill is 4 m up
[17:53] <hallam> not bad
[17:55] <Laurenceb> yeah only a few meters out at the most
[17:58] <Laurenceb> http://www.responsivespace.com/psearch.asp?cat=L
[17:58] Last message repeated 1 time(s).
[17:58] <Laurenceb> erm
[17:58] <Laurenceb> Paper Number RS1-2003-6005 sorry doublepost
[18:00] <Laurenceb> its almost good enough to see what side of the road you're on
[18:08] <Laurenceb> ok, working it out as best I can, I get 6.4 meters from pembroke street to your windowsill
[18:12] <Laurenceb> seems a pretty sensible number
[18:13] Bluenarf (n=Paul@apollo.paulsnet.org) joined #highaltitude.
[18:14] Nick change: Bluenarf -> EI5GTB
[18:16] EI5GTB (n=Paul@apollo.paulsnet.org) left irc: Remote closed the connection
[18:16] <hallam> the decimated data seems to work fine except that I even with the higher-gain DLL I can't get SV32 to stay locked more for than about 20 seconds
[18:17] <hallam> but I can probably live with that
[18:17] <Laurenceb> I get it ok
[18:17] <hallam> even with decimated samples?
[18:17] <Laurenceb> hallam: have you got the softgnss code ?
[18:17] <Laurenceb> havent tried
[18:17] <hallam> yeah I have
[18:18] <hallam> I just haven't been using it except to decode the ephemeris
[18:18] <Laurenceb> I'll pastbin you my acquisition file
[18:18] <Laurenceb> http://www.mibbit.com/pb/zsZHRu
[18:19] <hallam> ok, this is because theirs sucks?
[18:20] <Laurenceb> indeed
[18:20] <Laurenceb> also their postnavigation code sucks
[18:20] <Laurenceb> it doesnt average the pseudoranges
[18:21] <Laurenceb> I think thats one reason why a 1KHz position filter works better
[18:21] <Laurenceb> of course what you could do is find dopplers and get velocity
[18:22] <Laurenceb> then kalman fiilter each 1Kz position solution with the velocity solution
[18:22] <Laurenceb> but I dont think you could get the position down to less than 2 meters or so error - due to sat timing errors and ionosphere
[18:23] <hallam> I think there's quite a lot of error in the doppler
[18:23] <Laurenceb> yeah
[18:23] <hallam> is it going to be roughly the same as the noise bandwidth of the PLL?
[18:23] <Laurenceb> but kalman filtering may help
[18:23] <Laurenceb> erm shouldnt be
[18:23] <hallam> ok
[18:24] <Laurenceb> I think if it gets as large as the noise bandwidth, it starts to lose lock
[18:24] <hallam> I don't really understand what all those coefficients mean
[18:24] <hallam> ok
[18:24] <Laurenceb> so a larger noise bandwidth makes it more responsive
[18:25] <Laurenceb> the xmos thing is so cool
[18:25] <Laurenceb> its like propellor only without the goofyness
[18:26] <hallam> it still has some goofyness
[18:26] <hallam> but it does look quite nice
[18:27] <hallam> man I want to build a correlator on an FPGA now
[18:28] <Laurenceb> http://www.datasheetcatalog.org/datasheet/zarlinksemiconductor/zarlink_GP2021_JAN_01.pdf
[18:28] <Laurenceb> ^ is the place to start
[18:31] <hallam> it would be nice to have everything except the front end on a single FPGA
[18:32] <Laurenceb> yeah, soft core, fpu, and tracking units
[18:32] <Laurenceb> then front end, mems, magnetometer and sun sensor/thermos externally
[18:41] <Laurenceb> right I'm off, hopefully I can find a cable
[18:41] <Laurenceb> then icom fun awaits :P
[18:42] rjharrison (n=rharriso@80.176.172.227) joined #highaltitude.
[18:42] <Laurenceb> cya all
[18:43] Laurenceb (i=83e34f19@gateway/web/ajax/mibbit.com/x-40fd48d5906cbcdd) left irc: "http://www.mibbit.com ajax IRC Client"
[18:47] <hallam> http://www.weblab.dlr.de/rbrt/pdf/ISSFD_XVI15.pdf
[18:49] <rjharrison> hallam: wow, that looks like another comlicated rad
[18:49] <rjharrison> read
[18:49] <rjharrison> I'm going to settle for getting on with the competition
[18:51] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-69b7fc7a11e98948) left irc: "http://www.mibbit.com ajax IRC Client"
[19:42] fergusnoble (n=fergusno@fn217.quns.cam.ac.uk) joined #highaltitude.
[19:47] <edmoore> fergusnoble: did you fire up the dominoEX at about 6.30ish?
[19:47] <fergusnoble> erm, no
[19:47] <fergusnoble> why do you ask?
[19:47] <edmoore> hearing things in white noise
[19:47] <edmoore> standard.
[19:48] <fergusnoble> ive only just got my laptop out of its bag again
[19:48] <fergusnoble> not drinking then?
[19:48] <fergusnoble> ive just had the marvelous descovery that i have another computing assignment in
[19:49] <fergusnoble> and ive already has a few ales
[19:52] <natrium42> ginger ale?
[19:52] <natrium42> :D
[19:52] <fergusnoble> indeed, whisky and ginger ale
[19:52] <fergusnoble> a fine drink
[19:53] <natrium42> XD
[19:53] <fergusnoble> this is digressing
[19:53] <fergusnoble> edmoore: are you paving or staying in tonight?
[19:54] <edmoore> probably paving
[19:54] <edmoore> or similar
[19:54] <natrium42> "paving" ?
[19:54] <natrium42> brit speak
[19:54] <fergusnoble> PAV is a horrendous night at churchill college
[19:54] <natrium42> aah
[19:55] <fergusnoble> edmoore: does it stand for anything?
[19:55] <fergusnoble> pulling and vomiting maybe?
[19:56] <edmoore> Pavilion
[19:56] <rjharrison> hah
[19:56] <edmoore> where it used to be
[19:56] <edmoore> but pulling and vomiting is firly accurate
[19:57] <natrium42> lol
[19:59] <fergusnoble> edmoore: do you have the email for the uni communications person?
[19:59] <fergusnoble> if so could you forward it to me?
[19:59] <fergusnoble> ive still yet to return her call
[19:59] Bluenarf (n=Paul@apollo.paulsnet.org) joined #highaltitude.
[20:00] Nick change: Bluenarf -> EI5GTB
[20:02] <edmoore> fergusnoble: lisa?
[20:02] <edmoore> hang you on
[20:02] <edmoore> it'll be here somewhere
[20:02] <rjharrison> natrium42 did you get any where on the tracker?
[20:02] hallam (i=836fc8c8@gateway/web/ajax/mibbit.com/x-2719d4232682d91a) joined #highaltitude.
[20:02] <fergusnoble> hallam: hello
[20:02] <hallam> Hi
[20:02] <natrium42> unfortunately not
[20:02] <hallam> How's it going?
[20:02] <edmoore> lc460@admin fergusnoble
[20:02] <hallam> SpeedEvil: I solved the mystery of the 7km/s range rate error
[20:03] <hallam> mostly, anyway
[20:03] <fergusnoble> hallam: good, are you up to much tonight?
[20:03] <fergusnoble> hallam: i need an excuse not to go to the bop
[20:03] <hallam> more GPSing was the plan
[20:03] <hallam> want to come and see how it works?
[20:03] <fergusnoble> oh deffo
[20:03] <hallam> cool beans
[20:04] <fergusnoble> i need to code a bit for degree first, and eating in about half an hour
[20:04] <hallam> I've just roasted a chicken
[20:04] <fergusnoble> oh, wow
[20:04] <hallam> ok
[20:04] <hallam> and ate about 10% of it, so there's plenty left
[20:04] <fergusnoble> thats hard to refuse but alex has just started making something
[20:04] <hallam> ok
[20:04] <hallam> we can snack on it later
[20:04] <fergusnoble> awesome, i will bring some coding to do
[20:05] <fergusnoble> hallam: whats the latest on the gps?
[20:07] <hallam> just maths for the position, but I'm taking a sidetrack to rewrite the tracking code in C
[20:08] <hallam> I figure if I can make it work at 4x realtime on the PC, I should have a chance at it doing the same on the blackfin
[20:08] <hallam> er, at running in real time on the BF
[20:08] <hallam> 500MHZ vs 2GHz, and the BF is more efficient per MHz
[20:09] <edmoore> what about FP?
[20:09] <hallam> you only need that in the nav solution, not the tracking
[20:09] <edmoore> that makes total sense
[20:09] <edmoore> cool
[20:09] <hallam> so FP performance just limits the nav update rate
[20:09] <edmoore> ye
[20:09] <edmoore> p
[20:10] <hallam> acquisition takes either lots of CPU cycles or lots of memory
[20:10] <fergusnoble> whats to do to get the data onto the blackfin?
[20:10] <hallam> tracking takes lots of CPU cycles
[20:10] <hallam> nav doesn't take much at all
[20:10] <hallam> acq and track are both improved by parallelization
[20:10] <edmoore> hence hardware
[20:11] <edmoore> I guess that's the next project :)
[20:11] <hallam> acq isn't as important because you can have it doing that before it's left the balloon, or preload an ephemeris on the ground
[20:12] <hallam> edmoore: as soon as I learn some verilog/vhdl I feel up to implementing trackers in FPGA, but I'm trying to see if maybe it can be done on the blackfin in software
[20:13] <hallam> if nothing else, it would be nice to not have to store 1MB/sec of data
[20:13] <hallam> (I did some tests earlier today, and it can work fine at the 1MB/s rate rather than 2MB/s)
[20:14] <fergusnoble> hallam: im sure it can be done on the blackfin
[20:14] <fergusnoble> hallam: the only question is if that leaves room for the camera capture
[20:14] <hallam> well, I back-of-the-enveloped it, and it could only just be done
[20:14] <hallam> tracking that is, not storage
[20:14] <hallam> storage is no problem
[20:15] <fergusnoble> im mean the camera must use up some cpu
[20:15] <hallam> depends
[20:15] <hallam> it doesn't use any CPU to get the pictures into RAM
[20:15] <fergusnoble> how many correlators are you planning on runnin?
[20:15] <hallam> for tracking you only need as many as you have satellites
[20:15] <hallam> so maybe 6
[20:15] <fergusnoble> ok
[20:16] <hallam> JPEG compression takes some CPU
[20:16] <fergusnoble> does the camera not compress onboard?
[20:16] <hallam> no
[20:16] <edmoore> it's 15fps isn't it? that's be a feat
[20:16] <fergusnoble> thats a shame
[20:16] <hallam> not too much I think, I'd guess maybe 25% at 640x480 15fps
[20:16] <fergusnoble> ok
[20:16] <hallam> ed, there are cameras that do that
[20:16] <hallam> I have one
[20:16] <hallam> it's tiny and the optics are shite
[20:16] <edmoore> how much do you pay for them?
[20:16] <hallam> $10
[20:17] <edmoore> oh nice. but i guess the optics are the driver
[20:17] <fergusnoble> edmoore: the sparkfun cameras do jpeg onboard
[20:17] <hallam> http://www.sparkfun.com/commerce/product_info.php?products_id=8668
[20:17] <fergusnoble> its not unusual
[20:17] <hallam> that particular one sucks though
[20:18] <hallam> last I checked, nobody understands the register set
[20:18] <edmoore> 'This camera is also unique in that it offers on-board JPEG compression' cf 'it's not unusual' :P
[20:18] <hallam> so it gives weird colours etc
[20:18] <hallam> sparkfun are full of shit, there are plenty with onboard jpeg
[20:18] <hallam> a lot of the omnivision ones
[20:18] <hallam> just not the OV9655 that I have
[20:18] <edmoore> someone should tell them
[20:18] <hallam> it'd be a full-time job to tell sparkfun every time they're full of shit
[20:18] <fergusnoble> unique amoing the sparkfun range maybe
[20:19] <fergusnoble> hallam: sparkfun are good!
[20:19] <hallam> heh yeah, their range of 3, all out of stock
[20:19] <hallam> I'm just being crotchety
[20:19] <hallam> they are cool
[20:19] <hallam> they sell in small quantities, which redeems all their faults
[20:19] <edmoore> they're not as bad as maplin
[20:19] <hallam> lots better than maplin
[20:20] <hallam> but they're still full of shit from time to time
[20:20] <fergusnoble> they are good at making new tech acessible to people who dont do smt
[20:21] <fergusnoble> and are a good cross section of whats available but they are by no means the be all and end all
[20:21] <hallam> sure
[20:21] <hallam> I think what annoys me about them is that they don't seem to mind selling things without any access to documentation or reference designs, some of which nobody (on the forums) have ever been able to make to work
[20:22] <hallam> e.g. some of the cameras and some of the GPS chips
[20:22] <edmoore> they seem to be going the way of lots of outfits like that - becoming a shop window for lots of disparate sellers items
[20:22] <fergusnoble> yeah, at least with the camera the had a competition to get it working
[20:22] <hallam> and if they advertised them as such it wouldn't be a bad idea
[20:22] <edmoore> which I guess doesn't help with consistant standards of documentation or quality
[20:23] <hallam> that competition camera was good, because they said they had no docs
[20:23] <hallam> look at that one I linked though
[20:23] <hallam> "The nice thing is that we have a complete datasheet on this camera along with a good supplier."
[20:23] <hallam> now look at the datasheet
[20:23] <hallam> you can't even read the dimensions of the drawing to make a PCB layout
[20:23] <hallam> and there's nothing about how to set the registers
[20:25] <hallam> okay it's a new revision compared to when I looked at it last, but still little information
[20:25] <hallam> at least you can read the numbers on the drawing now
[20:26] <hallam> tens of people asking on the forum how to read out the picture over I2C
[20:26] <hallam> which of course is impossible, you have to use the data bus
[20:26] <hallam> but they should make a note on the product page about that
[20:26] <hallam> they have this image of being prototype friendly, but some of their products really aren't
[20:26] <hallam> anyway, enough griping
[20:27] <hallam> more C
[20:27] <fergusnoble> "standard data+i2c interface"
[20:27] <fergusnoble> i dont think any camera has an interface you could call standard
[20:27] <fergusnoble> but yes, i have nothing against them, i think they do some good work
[20:27] <hallam> well it's usually 8 (or 10) data pins and a clock out, there's not much to it
[20:27] <hallam> sure, on the whole I think they're good
[20:29] Xenion (n=robert@p579FCF5D.dip.t-dialin.net) joined #highaltitude.
[20:30] <fergusnoble> ok, so for this assesment i need to solve some equation with newton-raphson or bisection that isnt a square root
[20:30] <fergusnoble> any interesting contenders?
[20:31] <hallam> do you get marks for interesting?
[20:31] <Xenion> Guten Abend alle miteinander / Good evening everyone :-)
[20:32] <hallam> Hallo Xenion, wie gehts?
[20:32] <Xenion> geschäftig ;-)
[20:33] <hallam> ich auch
[20:33] <Xenion> d.h. mir geht es gut
[20:33] <Xenion> :-)
[20:34] <edmoore> fergusnoble: drinks settling in so this could be wonderfully unhelpful
[20:34] <edmoore> so you want to use N-R to find stuff other than square roots?
[20:35] <hallam> cube root?
[20:35] <hallam> SpeedEvil: http://www.aggregate.org/MAGIC/#Population%20Count%20(Ones%20Count)
[20:36] <edmoore> surely you can use NR to find the roots of any fuctions with continuous derivatives
[20:37] <edmoore> so like, say (in latexy sintax) x_{n+1} = x_{n} - g(x_{n})/g'(x_{n})
[20:37] <edmoore> i think that's right
[20:37] <hallam> yeah edmoore, he just wants an interesting function to solve for his examples paper
[20:38] <edmoore> oh right
[20:38] <edmoore> yeah some cubic
[20:38] <edmoore> but right it in the style or Cardano
[20:38] <edmoore> or Ferrara
[20:38] <edmoore> or ferdi
[20:38] <edmoore> or targtallia (sp?)
[20:38] <edmoore> write it*
[20:39] <edmoore> of Cardano*
[20:39] <hallam> fergusnoble: I can find an astrodynamics equation that is suitable
[20:39] <hallam> SpeedEvil: the Blackfin has a single-cycle ones count :D this might just make it possible to do real-time on it
[20:39] <edmoore> just say 'if newton and/or raphson predated Cardano, he could have solves his problem presented in section poo.bum or Ars Magna with the following iterative method:
[20:40] <edmoore> and solve it
[20:40] <edmoore> whats his face poo bum Mackay may like historical context
[20:40] <edmoore> he's a hippie afterall
[20:47] <hallam> fergusnoble: Kepler's equation (easy) or Barker's equation (fairly easy)
[20:55] <fergusnoble> hallam: im not allowed to use the math libs to compute sin or cos
[20:55] <fergusnoble> so that rules out kepler i think?
[20:56] <fergusnoble> could use an approx for sin but i have to get the answer to 5dp
[21:02] <edmoore> fergusnoble: do a cordic
[21:02] <edmoore> in software
[21:02] <edmoore> it's the only thing I've ever written in VHDL
[21:02] <edmoore> very simple
[21:03] <fergusnoble> meh
[21:03] <fergusnoble> i think maybe just another polynomial would suffce
[21:03] <edmoore> but a cordic is bisecty
[21:03] <fergusnoble> just a more interesting one than c-x^2=0
[21:03] <edmoore> it's in the spirit of the exercise
[21:04] <edmoore> bisecty is a word, yes
[21:04] <fergusnoble> edmoore: ok, ill look into it
[21:05] <edmoore> that's well clever that is
[21:05] <edmoore> right, time to commit
[21:05] <edmoore> bbl
[21:06] <fergusnoble> edmoore: not usre if i like the lookup taleyness
[21:07] <fergusnoble> *tableyness
[21:10] <SpeedEvil> hallam: nifty!
[21:33] Laurenceb (n=laurence@dyres221-35.surrey.ac.uk) joined #highaltitude.
[21:34] <Laurenceb> hello
[21:37] <Laurenceb> no luck getting a cable for my icom
[21:37] <Laurenceb> wheres the best place to get bnc connector?
[21:37] <Laurenceb> rapid or similar?
[21:47] <Laurenceb> hmm X forwarding work much more smoothly on ubuntu
[21:47] <Laurenceb> you can hardly tell the difference
[21:56] <SpeedEvil> doesn't cambridge have a Maplins store?
[21:57] <SpeedEvil> That'd be quickest
[21:57] <SpeedEvil> rapid, or farnell, or even maplin work OK mailorder
[21:57] <SpeedEvil> hallam: re: countones
[22:05] <rjharrison> Quick question if I sent rtty ofver fm should I be able to decode it in any rtty program?
[22:05] <rjharrison> or does it need to be usb
[22:10] <Laurenceb> yeah, prob a rapid/rs/farnell jobby
[22:10] <Laurenceb> need to get a yagi from somewhere ideally
[22:10] <Xenion> Gute Nacht / Good Night folks ! :-)
[22:11] Xenion (n=robert@p579FCF5D.dip.t-dialin.net) left irc: "Verlassend"
[22:11] <Laurenceb> whats the best coax for rigging up yagis and things?
[22:25] <Laurenceb> can anyone help me setup my icom with fldigi?
[22:29] <rjharrison> have u got the interface?
[22:29] <rjharrison> oh Laurenceb you have the computer type icom
[22:29] <Laurenceb> yeah
[22:29] <rjharrison> Do you want to have rig control or just decode rtty
[22:30] <Laurenceb> both
[22:30] <rjharrison> shall we start with rtty decode
[22:30] <rjharrison> do you have any local rtty you can pick up to decode?
[22:31] Action: rjharrison thinks payload
[22:31] <Laurenceb> yeah
[22:31] <Laurenceb> hmm
[22:31] <Laurenceb> fldigi hangs when I initialise
[22:31] <rjharrison> hmm that's not good
[22:31] <rjharrison> what are you running it on?
[22:31] <rjharrison> Ubuntu
[22:31] <Laurenceb> yeah
[22:31] <rjharrison> windows
[22:31] <rjharrison> ok
[22:32] <rjharrison> for the eee
[22:32] <rjharrison> ?
[22:32] <Laurenceb> what baud rate?
[22:32] <rjharrison> 50
[22:32] <Laurenceb> no
[22:32] <Laurenceb> for comms to the icom
[22:32] <rjharrison> for rtty you only need that audio in
[22:32] <rjharrison> I would skip the cat for a sec
[22:33] <rjharrison> get rtty working firts
[22:33] <rjharrison> first
[22:33] <Laurenceb> well I cant get my radio working atmn
[22:33] <rjharrison> ahh no front pannel?
[22:33] <Laurenceb> nope
[22:33] <rjharrison> OK now I get the picture
[22:34] <rjharrison> flgigi will only do cat if you have hamlib installed
[22:35] <rjharrison> The manual will probably give you the deafult baud settings
[22:35] <rjharrison> 9600 ususally for yeasu
[22:36] <rjharrison> U might be better off initially in windoze if it's an option
[22:36] <Laurenceb> yeah already tried
[22:36] <Laurenceb> it works fine there
[22:37] <rjharrison> baud rate there will be the same in linux
[22:37] <Laurenceb> oh that was using the stupid icom widget
[22:37] <rjharrison> what version of fldigi are you using?
[22:38] <Laurenceb> 2.08
[22:38] <rjharrison> 3.03 is the latest
[22:38] <Laurenceb> ok
[22:38] <rjharrison> You can grab that from the web site as a precomiled binary
[22:38] <Laurenceb> how can I find the latest version on the repo?
[22:38] <rjharrison> just copy over and replace your current
[22:38] <Laurenceb> as I'm not sure if its opdated very often
[22:39] <rjharrison> yum upgrade fldigi
[22:39] <Laurenceb> I had to build 2.08 from source
[22:39] <rjharrison> update even
[22:39] <rjharrison> apt-get infact
[22:40] <rjharrison> personally I would just grab the binary off the fldigi site
[22:40] <rjharrison> recompile when you have a spare evening
[22:40] <rjharrison> You are going to be better off straight away with the latest version
[22:41] <Laurenceb> I fear this is going to take ages to get working :-/
[22:43] <Laurenceb> guess I may as well update ubuntu
[22:43] <Laurenceb> my icom smells of joss sticks
[22:44] <rjharrison> fraid so I have got fldigi doing cat fine with yaesu ft 897D
[22:48] <Laurenceb> bbl
[23:10] laurence_ (n=laurence@dyres221-140.surrey.ac.uk) joined #highaltitude.
[23:12] laurence_ (n=laurence@dyres221-140.surrey.ac.uk) left irc: Remote closed the connection
[23:12] Laurenceb (n=laurence@dyres221-35.surrey.ac.uk) left irc: Read error: 110 (Connection timed out)
[23:18] Laurenceb (n=laurence@dyres221-140.surrey.ac.uk) joined #highaltitude.
[23:19] <SpeedEvil> Hmm.
[23:19] <SpeedEvil> http://www.dealextreme.com/details.dx/sku.19634
[23:19] <SpeedEvil> Ok - it doesn't take serial.
[23:19] <SpeedEvil> But it's 15 quid.
[23:19] Action: SpeedEvil sighs.
[23:19] <SpeedEvil> 20 quid
[23:19] Action: SpeedEvil stabs exchange rate.
[23:19] <SpeedEvil> (canned GSM 'bug')
[23:21] <SpeedEvil> I wonder how fast you can pass FSK through GSM. 10baud?
[23:22] <SpeedEvil> Laurenceb: http://www.dealextreme.com/details.dx/sku.19634
[23:22] <Laurenceb> yeah
[23:23] <Laurenceb> interesting
[23:23] <hallam> open it, you'll find one of those GSM modules
[23:23] <natrium42> hmm, is it hackable?
[23:23] <hallam> (probably)
[23:23] <hallam> which will probably take serial
[23:23] <SpeedEvil> hallam: almost certainly
[23:23] <SpeedEvil> hallam: I'm just wondering about a minimal cost/effort solution.
[23:23] <hallam> it's something that will have been produced in small enough quantities to still use GSM modules rather than a totally integrated thing, but large enough quantities to be cheap
[23:23] <SpeedEvil> hallam: if you can get 5 baud even, that's plenty to spit out a location.
[23:24] <hallam> waiting for a location at 5 baud is unpleasant
[23:24] <SpeedEvil> Even assuming binary
[23:24] <SpeedEvil> I'm assuming this would be for recovery
[23:24] <hallam> I guess it's better in binary
[23:24] <SpeedEvil> not tracking
[23:24] <Laurenceb> hmm fldigi is still 2.08
[23:24] <hallam> than ascii
[23:24] <hallam> hm, guess so
[23:24] <Laurenceb> but I have a new kernel
[23:24] <hallam> why not have a speech synthesizer?
[23:24] <SpeedEvil> Even ascii it's only a minute or two
[23:24] <SpeedEvil> hallam: code one on an attiny for me!
[23:25] <SpeedEvil> (assuming you send a whole NMEA sentance, which'd be silly)
[23:26] <hallam> you can get these nice chips which do it for you
[23:26] <SpeedEvil> Yeah.
[23:26] <SpeedEvil> This is a _MINIMAL_ cost thingy though
[23:26] <hallam> fair enough
[23:27] <hallam> why do you think FSK can only go as slow as 10 baud?
[23:27] <SpeedEvil> 20 quid for above GSM thingy. Fiver for MCU + buzzer to talk to GSM, 15 quid camera.
[23:27] <SpeedEvil> hallam: because GSM makes an active effort to throw away anything that's not voice
[23:27] <SpeedEvil> pure tones sort-of get through OK.
[23:27] <hallam> ah
[23:27] <SpeedEvil> Anything else basically is shredded.
[23:27] <hallam> maybe faster would be better then\
[23:28] <SpeedEvil> No - it has a model of the vocal tract
[23:28] <hallam> those bastards
[23:28] <SpeedEvil> In short - if you can sing it - you can transmit it.
[23:28] <hallam> speech synth is the way to go then
[23:28] <SpeedEvil> And it's designed so that errors produce similar sounding output
[23:28] <SpeedEvil> This results in - if you send in non-speech - a one-way hash - especially after two loads of air-side errors
[23:29] <SpeedEvil> Heroic efforts - data-driven vocal tract simulator - have been able to push 1300bps over it. But I can't find that paper.
[23:30] <Laurenceb> http://www.dealextreme.com/details.dx/sku.13301
[23:32] Action: SpeedEvil is looking for a suitable MP3 player to hack into that waterproof DAB radio I bought a couple of days ago
[23:33] <Laurenceb> one for the ladies
[23:35] <rjharrison> I think so too
[23:35] <hallam> oh boy, free shipping
[23:35] <rjharrison> right anyone know anything about audio
[23:36] Action: SpeedEvil has ears.
[23:36] <rjharrison> hhe
[23:36] <SpeedEvil> Compression waves in gaseous, liquid, and plasma mediums.
[23:36] <SpeedEvil> Compression waves in gaseous, liquid, solid and plasma mediums.
[23:37] <SpeedEvil> Damn.
[23:37] <rjharrison> I have AF out of DC bias 0.75V
[23:37] <SpeedEvil> A hell of a lot of the stuffs price on DX has _CRASHED_
[23:37] <SpeedEvil> http://www.dealextreme.com/details.dx/sku.19818
[23:38] <SpeedEvil> I guess that must more reflect the real price they can sell it for.
[23:38] <rjharrison> Is that low or high compared to normal Audio
[23:38] <SpeedEvil> 'out of DC' ?
[23:38] <SpeedEvil> your typical 'line' level output is around a volt AC
[23:38] <rjharrison> SOrry AF out with a DC bias of 0.75V
[23:39] <SpeedEvil> what's the DC bias got to do with it? (as Tina Turner diddn't sing)
[23:39] <rjharrison> External load should be >10Kohms /// <100pF
[23:40] <SpeedEvil> What's the context of this
[23:40] <rjharrison> This is out of the radiometrix
[23:40] <SpeedEvil> and what are you trying to work out?
[23:41] <rjharrison> I have to turn the volume up full on my amp to get a signal on the pc and I thought this was be cause the output on the AF is very low compared to noraml
[23:41] <rjharrison> normal
[23:41] <SpeedEvil> what is the AC output voltage?
[23:41] <SpeedEvil> most amps will load with 1K typucally, which is <10K
[23:41] <SpeedEvil> you might want a buffer
[23:42] <SpeedEvil> Got any random opamps around?
[23:44] <rjharrison> no
[23:46] <rjharrison> Should it have an AC output on the audio
[23:46] <SpeedEvil> ?
[23:47] <rjharrison> see above what is the AC output V
[23:47] <SpeedEvil> You specified only DC bias
[23:47] <SpeedEvil> that doesn't say if the AC level is 75mV, or 75Kv
[23:47] <rjharrison> OK I don't have a clue what I'm doing
[23:47] <rjharrison> :)
[23:48] <hallam> if it's a pure tone coming out
[23:48] <hallam> the voltage on the scope will look like a sine wave
[23:48] <SpeedEvil> rjharrison: do you have a scope?
[23:48] <hallam> DC bias = how far it is away from the center line
[23:48] <SpeedEvil> And why are you connecting it through your amp?
[23:48] <hallam> AC level = how big the peaks are
[23:50] <rjharrison> Right if i connect it to my computer speakers I can hear it nicely
[23:50] <rjharrison> If I connect to my head phones lovely
[23:50] <rjharrison> If i plug it in the sound card the comp barly hears it
[23:50] <SpeedEvil> try playing with the mixer levels on the PC
[23:51] <Laurenceb> back
[23:51] <rjharrison> I have on both and if I really push them up I get a faint sound
[23:51] <Laurenceb> ok... I'm not brilliantly hot with linux
[23:51] <Laurenceb> I have the fldig tarball
[23:51] <SpeedEvil> rjharrison: is this windows?
[23:51] <Laurenceb> how do I install it?
[23:51] <rjharrison> AF level is 400mVp-p
[23:51] <rjharrison> Both linux and windows
[23:51] <SpeedEvil> Laurenceb: tar -xzvf tarball
[23:51] <SpeedEvil> to extract it
[23:52] <Laurenceb> where to?
[23:52] <rjharrison> Laurenceb is this fldigi or ubuntu
[23:52] <Laurenceb> fldigi
[23:52] <Laurenceb> on ubuntu
[23:52] <rjharrison> tar -xzvf fldigi...tgz
[23:52] <Laurenceb> wut
[23:53] <Laurenceb> where does that extract it to?
[23:53] <rjharrison> this shoudl give you the executable file
[23:53] <SpeedEvil> current dir
[23:53] <rjharrison> yep
[23:53] <rjharrison> ./fldigi
[23:53] <Laurenceb> which is a temporary files location
[23:53] <rjharrison> and see what happens
[23:53] <Laurenceb> hmm ok
[23:53] <Laurenceb> I'll dump it in there then set to work
[23:54] <rjharrison> BTW SpeedEvil AF level is 400mVp-p
[23:54] <hallam> rjharrison: there's a difference between mic input and line input ports on the PC
[23:54] <hallam> one has much weaker amplification than the other
[23:54] <rjharrison> ahh true
[23:54] <hallam> I don't know the details
[23:54] <rjharrison> I only have line in on the eee
[23:54] <SpeedEvil> 400mV should be just fine however
[23:56] <rjharrison> Laurenceb any joy
[23:56] <rjharrison> did it load
[23:58] <fergusnoble> hallam: ok, just done Kepler
[23:58] <fergusnoble> hallam: was a good choice actually, hope they like the fact its relating the math to a physical problem
[23:59] <fergusnoble> am i correct in saying it doesnt have an analytical solution?
[23:59] <Laurenceb> where the hell does firefox download to?
[23:59] <SpeedEvil> Nobodies found it yet
[23:59] <SpeedEvil> Laurenceb: ~/Desktop
[23:59] <hallam> fergusnoble: yes
[23:59] <SpeedEvil> often
[23:59] <Laurenceb> no luck there
[23:59] <Laurenceb> I know it should
[23:59] <hallam> fergusnoble: if I remember rightly, anyway
[23:59] <fergusnoble> hallam: ok, cool
[23:59] <Laurenceb> but its dissapeared
[23:59] <fergusnoble> hallam: still want me to come oveR?
[00:00] --- Sat Jan 24 2009