resolve hax

Carrier delay in 12.2(53) SE1 broken

2011-11-09T17:21:00.000+10:00

The "carrier delay" setting on an interface is broken before 12.2(55). You have been warned.

I set up a stack of 3550s for a proof-of-concept to demonstrate how EIGRP could use feasible successor routes to provide a fast failover between two Ethernet segments. Due to some design constraints, SVIs were used for the Layer 3. All well and good, but when we actually tested it, we saw around 950ms of actual outage when we pulled out a cable or shut down a port. Why?

I did some investigation and googling and found a post on Packetlife detailing carrier delays with SVI convergence. It seemed to describe my problem but setting the carrier-delay had no impact. I tried the whole thing again on 3560s, and again saw the same result. I got in touch with Lincoln Dale from Cisco, who for starters described this method of testing as flawed because of its reliance on logged times not actual outage times, but also managed to get some Cisco engineers to check the source code for the carrier-delay functionality and recommend a course of action. It was definitely in there. After a few hours of packet captures, tests and various things, I noticed a discrepancy in the logs - one switch logged everything in a different order, and from the packet captures it seemed that it was converging correctly when it lost a port, and the other switch wasn't, so the pings went out but never came back. Why? I thought their IOS versions were the same but the one that was working was at 12.2(55) SE3. I upgraded them all to 12.2(55) and everything worked exactly as expected.

This sucks for 3550 owners who probably already know that 3550 IOSs stopped at 12.2(44), but hell, it's legacy equipment and it's just for testing anyway.

DHCP client broken in c7301-jk9s-mz.124-25d.bin

2011-06-01T11:08:00.001+10:00

The DHCP client on c7301-boot-mz.124-17.bin does not work. You have been warned.

I was trying to run up a router on a LAN segment that was notorious for having layer 2 issues and kept not being able to get an IP. Eventually after using a known-good cable on a known-good port where DHCP worked (the one my PC was plugged into minutes earlier) I noticed the packet counters were still slowly incrementing on the interface. I hard-set the IP and it worked.

File this under "Information I could have used 20 minutes ago".

Google DNS - 8.8.8.8

2011-04-20T17:18:00.000+10:00

Recently at home I've been having weird international speed issues, but only to certain locations. Vimeo was unusably slow, the App store was barely usable. My ISP (Internode, go buy from them) reported no issues. Swapped routers, no change.. Speedtests and line sync were fine, some downloads great, others crawled.

I finally nailed it down today - in a stroke of laziness I'd set 8.8.8.8 for the DHCP pool's DNS server. Turns out, 8.8.8.8 doesn't seem to exist in Australia anymore. I swear this used to be in Sydney, it was around 60ms away. Now, from my home, every box I've got around Au and every Traceroute site I can find, it's showing up in California.

or-10mf.bne#traceroute 8.8.8.8

Type escape sequence to abort.

Tracing the route to 8.8.8.8

1 150.101.180.24 28 msec 28 msec 28 msec

2 150.101.180.50 24 msec 28 msec 28 msec

3 203.16.212.13 44 msec 44 msec 44 msec

4 72.14.223.1 48 msec 44 msec 44 msec

5 66.249.95.226 44 msec 44 msec

66.249.95.224 44 msec

6 209.85.249.52 148 msec 144 msec

66.249.95.166 196 msec

7 209.85.255.34 160 msec 148 msec 144 msec

8 209.85.255.39 180 msec

209.85.255.217 176 msec 176 msec

9 209.85.250.101 176 msec 176 msec 180 msec

10 209.85.241.154 176 msec

209.85.241.162 184 msec

209.85.241.154 192 msec

11 8.8.8.8 176 msec 176 msec 176 msec

A geolocation of the last hop before it finds its target: http://www.ip2location.com/209.85.241.154 - definitely in the US. So Google's anycast DNS from the US is returning US-local mirror addresses for things instead of using Au CDNs. And it's causing weird, per-destination slowness. So that's the reason my 'net has been sporadically terrible this month.

Lesson: Use 8.8.8.8 for testing purposes only. Set your DNS to your own or your ISP's own DNS servers otherwise.

Fixing problems with Steam hanging on login attempt on Win7 x64

2011-03-31T16:44:00.000+10:00

Having problems with your Steam install taking forever to log in, updating incredibly slowly (5kb/s), or just refusing to let you do either? I've had this problem on two computers now and the reason for it just became apparent. They're both machines that had a single hard drive until a small SSD was installed for the OS, meaning that Steam had to be moved to the secondary drive.

Open up Regedit and search for "x86)\Steam", then carefully delete or update every entry. I'm not sure why exactly, but this hanging seems to be caused by old registry entries pointing to where Steam no longer exists. Removing them causes Steam to update normally and let you log in. I suspect that the default behaviour when it can't find config or cache files is to try to contact default servers, which aren't nearby, which would account for the slowness, although I can't imagine it would cause failure to be able to log in. Who knows.

Either way - try that for a fix. I've done this on two machines now and it's fixed both instantly.

Advice for building things

2011-03-22T11:49:00.000+10:00

I'm a big fan of Autospeed, a car enthusiast web site by Julian Edgar. He writes a lot of interesting articles about vehicles and driving, but also about modification, repair and just general building things. I haven't read it in a while, but here are a number of articles of his that you should definitely read if you are thinking about repairing or building anything mechanical.

1. How to use hand tools
2. Making things
3. Advice to ignore

These three posts contain a number of valuable lessons that everyone will much prefer to learn from a web site than through trial and error. Happy repairing!

Goodbye to a faithful server

2010-11-10T15:25:00.000+10:00

Back in 2004, I installed a server of my own at PIPE, a 1RU white box called Refused.

I was fresh out of uni and it was a massive learning experience actually having to figure out how things really worked. This little workhorse is ex-webcentral. I paid $20 for it, and it was manufactured in 2001. It had no disk drive covers or drive caddies, so I made my own with tape and the packaging from a phone headset. Nearly ten years later, it has never crashed, except for a drive failure in 2005. I finally finished migrating off it, and decommissioned it today. It’s the most stable machine I’ve ever owned, and I have it to thank for my Linux proficiency today.

Goodbye old buddy. I’ll miss you.

iPad serial port

2010-11-02T23:13:00.006+10:00

I kind of just assumed that it would work with an iPad, but I didn't actually think to test it until someone asked me to. So, with thanks to Jay from NetSIP for letting me borrow his iPad, I present to you the serial port working just as well as it did on my phone.

Version 2 of the prototype should be ready this weekend, with the production model ready shortly after that. I'm looking forward to seeing them shipped off to people! To be honest, I'm happy with the outcome already; I set about building this because I wanted one, and was sick of being stuck without one. Whatever happens from here on in is just gravy.

On a related note, it's been about a week since I built the thing and it got its first useful run in the field today. I'd installed a new router downstairs at PIPE and when I got back to my desk, I couldn't reach it or any servers past it. I groaned and reached for my laptop, then remembered the serial port, so I headed off to the datacentre with my phone in hand. After getting down there and logging into the console (and realising I had ip domain lookup enabled), I checked the IP addressing, and found that when I burnt the config out to the router I'd not unshut Fa0.

It was vastly easier than using a laptop because you can hold it easily while standing up and still retain full usage. I'm looking forward to having the miniaturised version in my hand so I don't have to lug this Jaycar project box about.

iPhone serial port

2010-10-26T17:39:00.041+10:00

A while back I found out that iPhones, iPods and the like have a serial port in the bottom. A company called RedPark announced one back in February but it never eventuated, and when I contacted them in August, the last the guy had heard was that Apple had said no. So, I set about seeing if I could make my own plug-in module.

After a bit of research and the wonderful pinouts.ru, I discovered that the iPhone serial port is in fact a TTL device - super low voltage. I looked at what I'd need for the dock connector too, and quickly realised that I'd have to be a LOT better at soldering than I am. Not wanting to spend time learning, instead I bought a breakout board. I'd been meaning to buy one of these for a long time for the purposes of building a line-in from a mixing desk, but eventually bought one for this project. Later I'm going to hook up some of the other pinouts on the thing, but that's for later - check them out.

Step 1: The breakout board

This particular dock connector breakout cost $35USD because it was fully labelled, but you can buy cheaper ones. They also get cheaper if you don't want them pre-soldered.

Step 2: Power

The voltages in regular RS232 serial would fry a TTL port quick smart, so I needed something to convert between the two. Maxim makes a chip called a MAX3232 that will do the job. You can buy one of these pre-soldered to a DB9 connector and PCB, or you can buy the raw chip and wire it up yourself. The raw chip is cheaper but again, more work. You'll need five 0.1uF capacitors and the knowledge of how to wire them up. This information is pretty readily available, but the datasheet for the MAX3232 is confusing, so here's a schematic on what this raw wiring should look like. There are plenty of prebuilt ones on eBay, but they all have female DB9s, so I had to make up a custom headshell.

The MAX3232 runs on 3.3v, which is convenient because the iPhone has a 3.3v power output in the dock connector. It's really got everything in there. Connect to Pin 18 for Vcc and Pin 1 for ground, and your chip should power up.

Kernel hackers be warned: the 3.3v isn't available until the phone is finished booting. If you want to watch the boot process and interfere with it, you'll need to provide external 3.3v power.

Step 3. Wire it up

Now, this part got confusing because Cisco devices don't use standard RJ45 serial pins for communication, and all of these devices are labelled vaguely as to which direction the RX and TX are relative to. Pin 12 is the PHONE's TX. Pin 13 is the PHONE'S RX. Connect Pin 12 to the TX-IN on the TTL converter, and Pin 13 to the RX-IN. This saves you 30 minutes of pain right there. While you would assume that the converter labels its RX/TX relative to itself, this assumption is pain-in-the-ass-fully wrong.

You'll also need to connect a 470kohm resistor between pin 1 (GND) and 21 (Accessory detect) to enable the UART.

Now grab a male DB9 head, and a rollover cable. Cut the head off the rollover, then solder it to the DB9. Wire Pin 3 of the RJ45 to Pin 2 of the DB9, Pin 6 of the RJ45 to pin 3 of the DB9, and Pins 4 and 5 of the RJ45 to Pin 5 of the DB9. Of course, this isn't necessary if you bought the correct gender DB9 in the first place, and you should just wire them up normally straight off the chip, then plug your regular rollover cable into it.

It's also super easy to make this an RJ45 port as this was earlier in the prototype process by doing the same in reverse with an RJ45 socket instead of the DB9. Again, Cisco rollover console is Pin 3 TX, Pin 6 RX, and Pins 4 & 5 are Ground.

Again, the schematic is here.

Now that all this is wired up, you probably want to put it in a case to protect it. I got a basic project case for $5 from Jaycar.

Step 4: Success

Jailbreak your iPhone, Install Terminal, OpenSSH server and Minicom. Change the root password using the passwd command (assuming your iOS is a version in which passwd works) before enabling the SSH server. The default password is alpine, and it's commonly known, so you really need to change it.

Once you've installed minicom, open up your terminal or SSH in, start minicom and change the settings - same as you would on a linux box. The device is /dev/tty.iap, speed 9600 8N1, no hardware or software flow control. Remove all the device initialisation, dialling and hangup strings, and save the new configuration as something, preferrably cisco. Save and exit.

Restart minicom with your new config by typing 'minicom cisco' and you're done. The real benefit in all of this is that there are so many console packages for iPhone in Cydia now that you can have a fully functional computer, as useful as a linux box, but without carrying around a laptop. SSH into it from a windows machine on-site or via an iPad and you're set. 3G iPhones with smashed screens are cheap, you can just leave one in your bag with the serial port, and you never have to worry about being locked out of a device.

There you have it! iPhone serial port for consoling into switches/routers/allthingsnetworkrelated. All parts were acquired from eBay, Ridax and Jaycar. By the time I bought everything it probably cost around $80, but you can probably build it cheaper.

FAQ:

1. No, I won't build you one. I was going to set up a manufacturing run but it turns out I'm a network engineer not a phone salesman and have little interest in such things.

2. You DO need to jailbreak your phone to do this. You can't access the iPhone serial port without jailbreaking.

3. That crap all over the screen is actually dead pixels. The glass itself is undamaged but the phone took a particularly hard beating, rattling around inside the pouch on my bike after it escaped its pocket and bounced around with screwdrivers and tools for 10km.

4. I'm not an electrical engineer, so I probably can't help you if you're having problems with your own attempt. I'll give it a shot though.

5. I am however a network engineer, so I can fix your network or set you up with a commercial out-of-band management solution if you want to be able to remote serial console into your equipment. Book it in through www.ionetworks.com.au/contact-us and I'll get back to you as soon as possible.

The last day of preparations

2010-09-11T18:16:00.001+10:00

After resealing the duck and getting it all set up again, I put it in the sink and turned it on. A lot of noise, no movement. I could've cried. Thank god I didn't change the water after Geordie washed his hands in it, because pushing it down further, I noticed a churning below the body. A thought I had briefly last night that'd made me sick with worry was true. The impeller was spinning the wrong way.

After freaking out briefly, then remembering I had an electric engine, I reversed the polarity and tried again.

The water jet kicked so hard that it blew most of the water out of the sink.

It turns out that the replacement gearbox kit was made to be mounted end-to-end over the drive train, not raised over it. It was designed to be pointing the other way, thus, was spinning in the wrong direction. The thought of the driveshaft pointing out longer on one end than the other had sparked me as odd when I got it, but I thought nothing of it at the time. So, I took it down to the pool to test.

The same problem as last year, although less pronounced - it wanted to tip over forwards. Also, the impeller wasn't biting enough, and Archimedes displacement principle wasn't enough to get a proper rear-leaning. I was going to buy sinkers, until Hayley pointed out that I had three massive jars of change that, per dollar per kilo, were cheaper to tape to a duck than to buy sinkers. So after fitting two rolls of them to the rear, we were in business.

Faster, much faster. But, we couldn't get past about 50% throttle without the nose diving. So, I set about doing something stupid and building a ski for the front. My theory was that by extending the base of the hull in a straight line, it would produce extended planing and cause the front of the duck to rotate around a longer axis, meaning it wouldn't reach the dip point.

Wrong.

It was nowhere near long enough and barely sat out of the water. The angle of naturality didn't make any difference, so it acted as a scoop and pulled it down instead. I had a quick thought, and went to get a drill and a piece of string. Drilling two quick holes near the top and making a halter-neck for the duck gave it a curvature that'd make the leading edge stay out of the water, which it mostly did, except for that the water then rushing behind it slammed into the front of the duck, leaving us in an even worse position. It never tipped over but 50% throttle and 100% were identical speed-wise, the latter just churned more water. So, back to the drawing board. It needed a bow.

So that's what it's getting. I've run out of light, it's dark now, so pool testing will be difficult, I'll just have to build it all up and hope for the best tomorrow morning. 18 hours remain.

archimedes buoyancy principle

2010-09-10T10:40:00.002+10:00

Ah, Archimedes you sweet bastard, I could kiss you.

The weight distribution of the duck is about even as its dimensions go. I thought this was going to cause problems with the tail not sitting far enough down in the water, but it probably isn't. Because I, dear readers, had forgotten about Archimedes' buoyancy principle. The tail of the duck is thin and elongated. The front is fat and wide. The rear will displace noteably less water than the front, causing it to have less buoyancy, so if the weight distribution is even, the rear should sit down in the water. How much remains to be seen, but some basic calculations seem to at least imply that it will be about as much weight as I wanted to add to the tail anyway.

You sweet buoyant bastard.

fast duck friday

2010-09-10T09:24:00.000+10:00

Woke up early this morning to do the rest of the sealing and make the hatch. My sikaflex experience today goes as follows.

10 PRINT Hey, sikaflex is pretty neat. This stuff is good to use.
20 PRINT Oh shit I touched it. Now it's on me. Now it's on something I touched.
30 DO CLEANUP_MESS
40 GOTO 10

It's all sealed now, cut the hatch to size. No more work today as the sealant needs to dry, but tomorrow morning when I wake up, time to take this thing to the pool.

at last, some hope

2010-09-09T23:28:00.002+10:00

After a quick detour to the jam rooms to practise for Friday night's show, it was time for the steering and finishing the hull. After realising that I hadn't made a mess of the control arm holes, and that I could bend the arms to work on an angle, everything worked out.

After this, it was a matter of cutting, cutting and more cutting. Cut out the duck body, cut out the hull, cut out the extraneous clips and edges from the platform. Cut cut cut. Finally it was all done.

And then, inserted - like loading a clip.

Ok so it's pretty much the same as the last photo, but you see the fitment now. As it stands, the duck is now upside down with the glue holding these two parts setting - tomorrow morning when it's done, I wake up and sikaflex seal the tail end for water-tightness, and we're ready to water test.

I'm nervous, for two reasons. One, we put the platform in my bathtub earlier, and weren't able to get it to move the water at all.

I was scared that I'd messed up somewhere and seized the drivetrain or untapped the impeller. Problem was, it just wasn't deep enough. I'm pretty sure the first time we put this thing in the water, it's not going anywhere. It's going to need to be weighed down in order to get it low enough in the water for the impeller to bite, I just hope we can actually do that.

Two and a half days left.

why do my posts always start with 'bloody'?

2010-09-09T15:16:00.002+10:00

http://www.flickr.com/photos/neverhadachoice/4973428643/Thursday bloody thursday.

Three days to go. This morning I woke up early to glue the suction tube and impeller pipe into hull. It took a lot of glue, and I've been using this two-part epoxy resin type caper that works underwater, sets hard, is very durable and is generally awesome. The only thing is .. well. What's the best way to say this.

It smells like jizz. No word of a lie.

I took a quick trip home at lunch after the glue had set to finish the job as there was a small part I couldn't reach, and to set up the servos. As my lunch break and my parking time expired, I had the reversing flap set up and working using one of the old helicopter servos. Fortunately I learned enough during this to be reasonably sure I'm going to be able to get the steering working pretty easily. The only thing is that I've held the reversing flap servo on with double-sided tape - very strong double-sided tape, but double-sided tape nonetheless. I doubt the steering servo will fare as well with such weak restraint, so I'm not sure what I'm going to do about that, but that's a problem to solve after I've figured out where it needs to sit. It might be as easy as drilling into the hull, but it might require zip ties. I kind hope it requires zip ties.

When I get home tonight, it's installing the steering servo, then attempting to seal all of this and glue it into the duck. In a perfect world I'd be able to seal the jug body and test in a sink, but when has anything ever gone according to plan?

wednesday, bloody wednesday

2010-09-09T00:39:00.001+10:00

I've said before on this blog that even good engineers make mistakes, but they tend to happen at a much more alarming rate when you don't know what you're doing, and your suppliers don't know anything about anything.

Pictures accompanying this post can be found on flickr as usual.

Tuesday at lunch I went in search a replacement belt, again. This time, I contacted John Darke of Darke Horse Models. I'd seen him listed on the Graupner site as an Australian partner, but with no working email address, no web page, a disconnected 1300 number and a phone number that rang out, I wasn't hopeful. However, as luck would have it, I managed to trawl through enough forums to find a PDF with his phone number on it. I drove out to Loganlea (evidently I really wanted it) with Hayley to visit the guy as he runs his business out of his house. I wondered what kind of guy was into RC stuff but wasn't any good at internets, and the answer was apparent - a retired mechanical engineer who has built more boats than I've ever seen, and good ones. He didn't care about profit, and certainly didn't want to deal with dumbass customers. He just wanted to help good people build good boats. Fortunately, he mistook me for one of those people.

It turned out he had a single unit left, a 700BB turbo motor with the belt and the gearbox, but not just any gearbox - a heavy duty one. He complained at length about the gearbox I'd got and showed me why they fail all the time. The belt was identical but he convinced me it was worth using this gearbox instead. We left, and he had $105 of my human dollars but I had the gearbox! And a spare motor, which to be honest runs at a lower voltage so is better for this application - a trade-off of RPM for torque.

So back home to build. What followed was an absolute clusterfuck. The drive shaft on the new motor, and the new gearbox were both slightly different sizes to my old ones, and I HAD to use my old drive-shaft - the new one wasn't even close to long enough. I was able to get my shaft tube into place by removing the extra bearings, and flipping the rear cover, but I realised all this time I'd had the tube in backwards and when set up the right way, it was relying on its original bearing to hold it in place. In addition to that, how was I supposed to mount it all in the wood? Matt cut the duck up so we could see where everything fitted but it was getting out of hand. The waterline would be too high, the duck would nosedive under any pressure from the rear. We looked at building a hull for the rear and so many other options but it just wasn't going to be worth it. I went to bed at 12:30am, nerves and spirit shattered. My design was in tatters, I couldn't get the new gearbox working reliably, I was facing the possibility of having to go in and say "sorry I spent all our money and have nothing to show for it".

Before going to bed, I sat down and wrote out a basic plan for tomorrow. Stop troubleshooting problems that hadn't arisen yet, and absolute step one HAD to be, get a working product. Once it's working, then it can be improved.

Today was more productive - discussions at work lead to a Big W expedition for a drink container as a prebuilt housing for the motor and jet. We also realised our water-line problems didn't exist as the duck needed to plane, and could have its balance adjusted with weights. I arrived home with jugs in hand (lol) and got to it.

Task list:
1. Fix drivetrain issues. Resolved by cutting a part I thought I'd destroyed into shape as a bearing case from the original gearbox.
2. Get jet set up in container

It turned out the drive just didn't fit fully enclosed, so I had to cut the container to make way for it. And oh god, the problems. Turns out I got a vital step wrong yesterday and I was supposed to glue the suction tube and the impeller housing in to the body as a first step. Since I didn't have the body, I thought that I could start with that and work my way up to it. WRONG. The impeller housing needs to go in first. Now I have to mount the exit vane from the inside, which means Graupner's carefully designed jet vane that's supposed to sit flush will instead be the tops of screws. This is the OPPOSITE of the way it was designed. I also managed to burr up the driveshaft, meaning the simple act of trying to move the bearings around took nearly an hour of forcing and very sore hands.

Everything took a lot longer than it needed to, my hands hurt like hell, but at the end of it all, I have something that's ready to be sealed into the housing and glued into the duck.

Tomorrow morning I'm going to glue the suction tube and exit vane to the platform, drill holes for the servo arms, and reinforce the driveshaft tube, then come home at lunch to cut an appropriately shaped tail out of the duck, and install it there. Hopefully, by the time I get home at night, it'll be ready to install servos and water test.

I'm excited - for the first time since I actually started building, I have a roadmap to a product and I can see how it'll work.

it's time to build ducks and chew bubblegum

2010-09-07T11:51:00.003+10:00

And as our hero Duke so pathetically complained, I am all out of gum.

Here we have our duck this year. Although really I should say 'one of' our ducks. Last year was a fairly simple bolt-on affair, but this year we realised we were going to need to do major surgery on the body in order to get this duck down the line. Four ducks were purchased, with the intention of possibly destroying them in the prototyping process. A lot more cutting was going to be necessary.

I guess the first place to start is the plan - I've been holding off on this for a while for two reasons.

1. To stop people from copying it until it was too late, and,

2. To be intentionally obtuse.

This was the first step.

Any guesses? Glue these two halves together. But that looks like a water tube.. Oh screwit, here's the drivetrain when assembled.

What that is, ladies and gentlefolk, is a Graupner G2340 Jet Propulsion system. This jet-pump is usually installed in high-speed dragsters and racing boats, and is prized for quick acceleration and maneuverability. Sounds insane? It is, but that's just the start. The engine, a 9.6v Speed700 Race motor, is capable of putting out just over 400w. If the duck weighs 2kg when it's finished, that'll be a power to weight ratio of 200kw/tonne. Two hundred. That's more than my car, and that does 0 - 100km/h in 5 seconds.

So not only do we have an absolute buttload of power, we've got an efficient, low-resistance mechanism of putting it down without raising the centre of gravity.

Back from last year's design, we also have a wireless camera system which we'll hook up to a laptop for remote piloting capabilities.

Unfortunately, it hasn't all gone to plan at time of writing. One very important part that we'd ordered was missing - the main drive belt. I tried going to Hobby Engineering Supplies but the only belt they had that would fit was too big, and it turned out it didn't even fit. I had planned to make up a custom mounting bracket but the belt itself couldn't sustain the bend radius of the drive cog.

Sadface. Today I'm on the hunt for something else I can use to connect the drivetrain and the motor. I can't mount them end-to-end because it would be too long to fit in the duck, and because the high-revving motor is more efficient at high revs - this gearbox lets it spin freely up to its high RPM goal, providing more torque, better acceleration, and lower current draw.

For more photos, hit up the Flickr photostream. Stay tuned for the answer to the drivetrain problem.

The new arrivals

2010-07-29T00:08:00.002+10:00

The first of the parts for the duck turned up today. On the left we have our motor, and in the middle a steering servo. The pen is just there for scale, but what's that on the right? Does anyone know what that is? If you can figure it out, it's a strong clue towards this year's design. I'm almost sad that the duck is going to be this simple, but really it hasn't been that simple.

Last year we did it backwards - we looked at what we started with, and then tried to figure out what we could build with it. We had a pair of motors that we could balance force between, and that could be used for thrust and steering, so it was. We had a hammer and in turn, the problem looked decidedly nail-like. This year, the design process was followed properly. We identified the strong and weak points in our design from last year, analysed where we went wrong, and then found all the things we shouldn't do. With those fresh in mind, we set about all the other possible ways we could solve the problem.

But what did we come up with? And will it even work? All I know is, that's a sodding big electric motor. Stay tuned.

The illusion of control

2010-07-21T23:29:00.007+10:00

Regardless of the actual implementation, all design requires control. While initial project meetings floated (LOL) the idea of an Arduino-powered drone, it was decided that without the ability to secure a beacon at the finish line, we wouldn't be able to reliably point the duck in the right direction and make sure it stayed there. So, as per usual, a helicopter had to die. This time, it was my well-worn out 6-channel Twister CP Gold.

My Twister Gold had its final throes in our car park over a period of about a month of flying, crashing, flying, crashing, fixing, crashing, flying, crashing, fixing, repeat, as eventually a key part of the swash plate broke off, and one of the blades lost the ability to adjust its pitch. Needless to say, this rendered it useless, and it was more work than it was worth to fix.

After a bit of work, some disassembly, and a single tear, the bird had been reduced to its useful components, the rest disposed of. One thing that does improve things over last year, however, is the comms situation - the CP gold comes equipped with a six-channel transmitter and receiver. Currently the transmitter is configured for a helicopter but will work a lot better if I can reconfigure it to separate the channel mixers. Helicopters are complex in that moving one stick needs to trigger several different reactions from servos motor, so we need to make it behave more like a plane - one axis == one servo motor. If we can do that, we'll have a drive channel, three independent servo channels, one possibly-useless 3D switch, and a cool feature - the final channel of this transmitter seems to be a landing gear switch.

That will prove useful.

The driving force

2010-07-20T14:45:00.001+10:00

After a few design meetings and some basic agreements on what we wanted to do, we've decided a few things:

1. Buying an RC boat and gluing the duck to it is cheating. If we were going to go that route, we may as well have just hired a Jet-ski and sat the duck on it. We need to use the duck as the centre of the design and build out from there.

2. Dual-drive is not the way to go. Last year's attempt showed us a few things about dual-drive propulsion systems in water and the main ones were that it was only a balanced system when both motors were at full power. Many problems came from the fact that if one motor had more power than the other, there was no hull buoyancy to resist it, causing rolling. If you intend to have multiple motors, they need to be INSIDE the extents of the hull. It made for a servo-less steering system, but the complications of then having frame-mounted motors meant having wings, further destabilising an already unstable system.

3. Keep everything internal. Everything that sits outside the duck is moving the median distance from centre of gravity, making it easier to tip. In addition to that, everything that sits outside the duck needs its own waterproofing, making things fiddly and difficult. This also created a higher CoG, destabilising even more. Can you tell stability was a problem?

4. Drive angle. Last year we angled props into the water from a frame, which caused an opposing force angling out of the water. This caused the duck to rotate around its CoG, angling the nose down and lifting the engines out of the water. We countered this with skids, but that's not an effective solution as it just caused us to waste energy opposing force that could have been used for acceleration.

So, with all those things in mind, we've placed an order for the drivetrain and the motor. Once those things have arrived, we'll start figuring out how to control them, what we need to do in order to get them in the duck, and how it's going to move once it's all in there. We're going to be a little light on the details until we have something to show you, so it'll be more impressive when we do, but the goal here is to stay one development cycle ahead of the competition - by the time we show what we're working on, it will be too late to replicate it.

We looked at our solution briefly last year, but never investigated it to the depth that would have revealed that it was an excellent idea.

Stay tuned..

Great Brisbane Duck Race - Project Blackbird

2010-07-16T12:03:00.001+10:00

Well, it's that time of year again - the Great Brisbane Duck Race has rolled around and as usual, we've entered a team. We're still in the planning stages for the moment but we've got some good ideas and a direction to move in, so more details to come when we have them.

For the mean time, some reflections on last year. We learnt a lot, and more importantly, saw what the competition was like - other teams saw perfectly fit to buy a high-powered RC boat off the shelf and tape their duck to it. Contrasted against that we designed and built our duck from scratch, we felt quite ripped off, but we learned from our mistakes. While I'm still not prepared to buy an off the shelf solution, it became apparent that in a field where people are sticking their ducks on boats and not just propelling them, so too must we augment our duck's water characteristics in order to overcome this disadvantage.

We're still building it from the ground up, but we'll be taking it more seriously this year. These aren't your dad's ducks, these are energy ducks. Turbo ducks.

Stay tuned..

Even good engineers make stupid mistakes sometimes

2010-02-17T16:44:00.002+10:00

Every now and then, we lose our wits for a second and don't thoroughly think through the repercussions of something we're about to do. Then, disaster. It's embarassing, and although most of these blog posts focus on my (eventual) successes, this time it's time to talk about failure.

I've owned four RC Helicopters now, and I have this theory that's held true for all of them; it takes about 80% of the time to learn to fly a new helicopter as it does to destroy it. Of the four, three of them became near-unusably difficult to fly before I finally got the hang of them - so much so that I'm almost considering buying two of whatever I get next; one to learn on, one to fly for real.

This has never been more true than my one current bird, a Twister CP Gold. This helicopter appealed to me as it's targetted at people who can fly a Coaxial helicopter (two counter-rotating blades) and want to move up to the real thing, and it was cheap for a big one - $350.

When I got it, it didn't fly. I couldn't understand why, they're test-flown in the factory to make sure they work. I was building Bolt at the time and didn't have time to figure it out, but over successive attempts, I just ended up damaging the main drive shaft by bending it slightly and scraping the blades trying to get it to take off. It just spun endlessly, and to top things off, one of my crashes sucked grass up into the tail rotor's actuator, causing it to not return to the centre when the stick was released.

When I moved into my new place, I set about trying to figure out why it never flew. I took the tail rotor apart and cleaned and re-assembled it, getting all the gunk out of there, and re-tensioned all the appropriate screws. I fired it up on my desk, slowly, just to get a picture for what was happening. It jumped suddenly so I killed the throttle and the blades slowly rotated to a stop. That's when I noticed it - something was wrong with this picture. What didn't fit? Something about how things moved was wrong .. but how? It was test-flown in the factory and worked, I hadn't change anything .. and then it hit me.

I HAD changed something. I also bought a pair of coated white wooden blades, because the fancy carbon-fibre blades that it came from in the factory were going to get destroyed. When I fitted them, I lined them up under the existing blades, setting everything out where it was to be bolted on, but I was having trouble, because the bolts on the bottom of the blades weren't held in separately - they kept just falling out. So they needed to be on top. So then I turned the whole thing upside down...

Goddammit. Yep, the blades were on backwards because I'd flipped the bird when I was swapping them for easier access to the locking bolts. Putting the blades back the right way and they were actually generating lift properly. It'd still take off by over-juicing the motor in get enough lift to get off the ground, but the tail rotor wouldn't spin fast enough to counteract that rotation, so it just spun. I took it downstairs and walked/hopped it around the garage. It flew. It motherfucking flew. After six months of doing nothing, it flew. Amazing.

Since then, I thought I broke it again with similar symptoms, but it turned out to be just that I didn't understand some of the switches on my remote, and accidentally turned the rotor speed up way too high again. I've also blade-struck the concrete doing low-altitude flying enough to strip the lining off the previously white blades, and last night, snapped one of the skids off on a very hard landing.

It is currently at home, held on by glue and zip ties. I love fixing helicopters almost as much as I love flying them. Convenient, because you should never buy a helicopter for anyone who DOESN'T like fixing them.

How to sync your Gmail calendars to your iPhone while still syncing to Exchange.

2010-02-03T23:56:00.001+10:00

Problem: Google's preferred method of syncing an iPhone to Gmail is using Exchange ActiveSync. You have to sync your iPhone to your work's Exchange server. You can only configure one Exchange account per iPhone.

Now, I've been doing this for a while via various methods over the course of my iPhone ownership but upon purchasing a 3GS that didn't require hacktivation, I realised that my old method of using a hacked app to sync in the background wasn't going to work.

It's incredibly simple though, and the answer is here, buried in Google's documentation that doesn't really show up when searching. Using CalDav, you can add the calendar.

http://www.google.com/support/mobile/bin/answer.py?answer=151674

"CalDAV: CalDAV Calendar Sync

Sync your Google Calendar with your iPhone or iPod Touch and always have your schedule at your fingertips. Note: You can also choose to use Google Sync to sync your Calendar and Contacts. It is possible to run CalDAV sync and ActiveSync simultaneously.

Requirements

Syncing via CalDAV is only available for devices with version 3.0 firmware or above. You can check your current version by going to Settings > General > About > Version on your iPhone or iPod Touch device.
To upgrade, follow the instructions at http://www.apple.com/iphone/softwareupdate/

Getting Started

To sync with your device, follow these steps:

Open the Settings application on your device's home screen.
Open Mail, Contacts, Calendars.
Select Add Account...
Select Other.
Select Add CalDAV Account.
Enter your account information:
- In the Server field, enter [ google.com ]
- In the Username field, enter your full Google Account email address.
- In the Password field, enter your Google Account password.
- In the Description field, enter the name you'd like to appear on the account.
Select Next at the top of your screen.

After you've completed setup, open the Calendar application and syncing will automatically begin.
By default only your primary calendar will be synced to your device. You can sync additional calendars by visiting one of the following pages from any browser (via desktop or mobile device):

For non-Google Apps users: https://www.google.com/calendar/iphoneselect
For Google Apps users: https://www.google.com/calendar/hosted/YOUR DOMAIN NAME/iphoneselect

Select the calendars you'd like to sync, and click Save. The selected calendars will display on your device at the time of the next sync."

So goddamn easy! If only it were well-known.

Duck synopsis

2009-09-29T20:36:00.005+10:00

note: I've just been informed this isn't entirely correct - will update when I get more info.

So, as most of you know, a wedding in the Bunya Mountains prevented me from actually attending the duck race - my good friend Sara was left at the helm. I've finally talked to everyone involved and have gotten the full story on the duck and a reasonably clear explanation as to what happened. So here goes.

On saturday morning the last of the new changes were tested. Some minor modification was done with tape to clear the area around the propshaft entry, but the duck was driving fast and handling wonderfully. The new propellers were working well and everything was peachy. That's about where the good luck ends.

From what I can gather from various sources, one competitor's entry was a bodyboard with two ducks trailing behind it with fishing wire and various other decorations. Our duck was in prime position prior to the starting line being raised, upon which time said competitor's trailing ducks have floated off and gotten tangled around approximately six other ducks - about 1/5th of the field. The fishing wire became tangled around our duck as well as some tinsel that got caught around the propellers. A man in a kayak was sent to retrieve the ducks, however instead of bringing them back to shore, he decided to pick them up and THROW them over the fence onto the riverbank, several metres vertically and horizontally. Rather than untangling the ducks or cutting the line, he has also thrown our duck, complete with others entangled, in the same manner. This has had a three-way effect of bending the entire frame around the duck, loosening the connections internally, and dragging it through the water non-vertically has caused the water-resistant (not waterproof) sealing to be under water for some time, causing the duck to take on water. The impact has loosened the connections internally and the battery which was formerly taped down inside the duck away from the bottom of the hull has come loose.

At this point the duck is still technically working, so it's put back in the water, and given the opportunity to run again. However, the damage to the frame has meant that the motors are no longer even close to aligned with the propshafts. The severely mismatched angles of propulsion caused the propshafts to come loose from the motors and remove themselves from play, first one, then the other. This appears to have been the final straw that broke the internal electronics' back, as the joint between the power switch and the Lithium Polymer battery pack has broken off and caused the battery to earth itself into the water floating loose in the bottom with the 9v battery for the camera. This is a Bad Thing. Long story short, the Li-Po battery exploded and the inside of the duck is now filled with highly toxic battery materials. With no power we were officially dead in the water and had to be rescued. Again.

By throwing.

Autopsy photos here.

So that’s it - our duck race was a bust, but I still had an absolute blast building it. I had planned to reassemble everything and make repairs for a test-drive, but unfortunately the Li-Po battery explosion killed everything inside the duck. By all accounts, the entire race was a bit of a non-event on the day, but I had my fun beforehand anyway. So thanks for reading, and keep an eye out for our entry next year.

Ready to race

2009-09-25T01:15:00.003+10:00

The testing on the weekend revealed a few things. We were a lot slower than I'd anticipated - the tiny propellers weren't presenting enough resistance and the motors were reaching their full RPM. So they needed more load - bigger props got installed. Unfortunately more rotational inertia meant a whole host of other issues. More vibration through the driveshafts only exascerbated the balance problems we were having. The driveshafts had to be reinforced half-way, had their entry and exit points glued down, the silicon joints glued down, and due to an unexpected overheating problem in the entry/exit points, high speed bearing lubricant added.

Only electronics were left after that. The switch I'd bought was unreliable at best and the LED seemed to alter the voltage enough to cause the receiver not to work, so it was replaced with a smaller toggle switch. I wired up a lead-acid battery pack in place of the 12V AC adapter for the video receiver, resoldered all the weak joints on the right motor and encased everything in heat-shrink tube. After sealing it all up with tape and the appropriate plastic (read: freezer bags), it was painting time. All extraneous metal except for the mounting plate for the flag were painted yellow, as well as the keel, which needed additional glue reinforcement due to the battering it received in transit. The skis were painted orange to vaguely resemble feet. Cause, y'know. Ducks are yellow, with orange feet.

I also used the remaining balsa-wood that we'd bought for the skis to make a proper dry-dock to actually sit the thing in, since it no longer sat comfortably on a desk. It got painted too - can anyone guess what colour?

Matte black. Of course.

Open water testing

2009-09-20T08:41:00.000+10:00

Cameras, and balance

2009-09-18T23:37:00.004+10:00

After the keel glue was dry, it was time to get the engines relocated. This should make the thrust vector better (more forwards, less down). It won't fix the problem, but it should mean that it needs less stabilisation and I can get away with less rolling (floating?) resistance. I had a lot of opinions and ideas being thrown around all day, although less than yesterday. I've moved the duck off the main table at work and onto my desk so I can get some work done, as everyone just wants to stop and talk about it. Nevertheless, a lot of good ideas and useful comments. Everyone's been really getting into it and it's good to get people thinking about the possibilities. Great to see people analysing and weighing things up. I didn't know anything about hydrodynamics when I started this either but I sure do now!

Anyway, so the motors had to move. I built a new bracket around the front of the duck and adjusted the position of the old rear bracket. I couldn't get a deep enough propeller position, or get them equal, so I ended up abandoning the original (which had been a guessed hack anyway) on account of too much counterbending looking and holding like crap. A new rear bracket and the removal of the old front bracket and we were on our way. In addition to that I added a coathanger to bridge the gap between the skis, to avoid other ducks getting caught and pushed along (wouldn't that be a shitty way to lose?) and also to give the pilot a reference point for where we were going.

After that it was camera work. The first test of the camera worked okay at first but highlighted a few issues. After deciding to install a power switch for the camera to avoid the need to get elbow-deep in the duck to turn the transmitter on and off, I cut some holes, wired up the switch and mounted it in the port hole. It turned out the 9v power for the receiver was a bad idea - terrible in fact. The receiver's sensitivity was way down and we only had a few metres range. After going home for a 12v power supply and hooking up the switch, all worked fine.

Nothing left for it but to secure everything, seal all the water-sensitive bits, remount the flag with a new piece of bracket, reverse the polarity of the motor connections because I am a retard who mounted them backwards, then clean up and take a photo, pictured above.

Plan for tomorrow:

1. Take the duck to southbank and tune the performance of the skis.
2. Fit the passive stabilisers and test.
3. Fit a power switch to the drive system.
4. Profit.

That's it, we're on the home stretch. The new drive system is much more stable and less prone to vibration than the old one and I theorise that tomorrow I'll be able to reach full throttle. Then it's just a matter of fine-tuning exactly how fast we can go at said 100%. Stay tuned for test videos tomorrow night.

Further modification

2009-09-17T23:41:00.007+10:00

After our successful water test the other day, it became obvious that we had issues with the duck tipping forward. The way forward from there was pretty clear - the force needed to be countered. So I cut some skis out of balsa wood and started sealing them. That was going to be difficult to dry, so I took some of Trina's bobby-pins and fashioned them into an impromptu point stand. While that was drying, I marked out a plastic lid that Sara provided for a keel. After leaving them to dry, I left it for the night.

The next day I picked up a USB video capture card and went back to work to install the camera properly. After confirming that it worked and that the receiver worked on 9v instead of the prescribed 12v, the NSW sales manager walked past for a chat. He suggested that with such rolly-polly handling characteristics, it should have passive stabilisers - wings that hung in the water and dampened any particular movement. That gave me some ideas.

That night I went home and with Heather's help, cut out the two halves of the keel and two wings for the stabilisers. The down side of using a Dremel for a grinder is that you get covered in bits of melty plastic. I'd had an idea that I could use the old original prop shafts for the proof-of-concept to build the stabilisers, and a little reworking later, it proved acceptable in theory.

We glued the two halves of the keel together and cut some styrofoam out for skis. The keel looked like it'd hold, so it was cutting time. First the passive stabiliser mounts, then the frame for the skis. Dremels are fun.

Stopping for a second to lol at the fact that my camera mistook the M-Man for a face, it was time to see where the whole thing sat now. Unfortunately the duck no longer fitted in the sink. So I set about making my own pond in a box. The box test revealed what I suspected - the duck was a lot better balanced than before and the water line was acceptable.

After that it was just time for some branding and we were done for the night. I came in the following night to finish the keel and glue it on.

Next step - relocate the engines further forward, and water test for a water line so we can adjust the skis, and then figure out where to fit the passive stabilisers. After confirming that all works and refitting the camera (along with a switch to turn it off), we should be done.

Beta testing - round 1

2009-09-15T15:33:00.005+10:00

Last night I brought the duck in to work with the intention of testing at lunch in the fountain pond in Anzac Square. I was expecting some stability issues to do with the thrust vector and tipping forward under power, and that was exactly what I got. As you can see in the video, the times I got too throttle-happy, the whole thing lurched forward and threatened to tip over. Fortunately, this is a problem we have four possible solutions for.

- Front stabilisers, similar to a snowmobile, to keep the nose from dipping
- Angling the drive at the tail with a geared or flexible driveshaft
- Installing a lengthy winged keel which will hold onto the water
- Moving the drive portion forward to move the point of body rotation forward more, to resist the roll forward

I'm going to go home tonight and try the stabilisers, and I have some flexi-silicone tubing that can do for the drive, but how to straighten it after the bend? I'd need another section of driveshaft .. not impossible, and I could clamp down with the bearings from the heli shafts but whether or not they're water resistant enough to withstand living underwater for any length of time without locking up later remains to be seen. Other things to try include adding a winged keel, which I plan to do tonight, and

Current plans:

Wednesday lunch - testing with keel
Thursday lunch - testing with stabilisers
Friday lunch - testing relocation of motors

Wish me luck.

racing duck enters Beta testing

2009-09-13T21:26:00.001+10:00

Sara and I spent the second half of the day working on the duck. We installed the propellers and prop-shafts that Rachel and I bought earlier in the day, and then set about cleaning up for a water test. The laundry sink seemed like the place. After getting it in the water, and performing a thrust test to make sure the motors worked, we found two things.

1. The duck was extremely unbalanced, biased to the rear.
2. The length of the extended prop-shafts meant that with said rear bias, it still touched the bottom.

Both bad things. So, it was off to the city for tools and equipment. A dremel, spare bracing, and floatation devices to build stabilisers with. We came back with balsa-wood, a massive dremel kit and more alu bracketing. After cutting off the excess metal in the bracket, great news - the duck floated normally. It was time to test that it could move.

And it could.

After that it was just a matter of cleaning up and optimising. The camera cabling was routed through a port hole in the duck’s head, the control box and the battery were internally mounted through a port hole in the back. After that we sealed everything up with freezer bags and it was done. No time for a live water test as it was dark by the time we were done, so will test during the week.

As usual, complete build photos here.

(The workspace was a bit chaotic afterwards)

building a racing duck

2009-09-08T14:23:00.002+10:00

So, we’re entering a duck in the Great Brisbane Duck Race. And by ‘we’ I mean mostly ‘me’. I had two E-Sky Lama V4 helicopters I could use for parts as a proof of concept, so here’s the visual record I’ve been promising everyone for the Type Zero. Still missing - two propellers, waterproofing, fixtures that aren’t mostly made of cable ties, but I have one more spare duck to test with before the race. Which, by the way is Saturday the 26th of September. Click on the photo for the build photos. Alternately, there are some videos of the first servo/motor test and an installation test after the build.

So, we’re entering a duck in the Great Brisbane Duck Race. And by ‘we’ I mean mostly ‘me’. I had two E-Sky Lama V4 helicopters I could use for parts as a proof of concept, so here’s the visual record I’ve been promising everyone for the Type Zero.

First things first, survey the supplies and figure out what needed to be done. After getting the bodies off both the helicopters, I was left with the meat in the sandwich - two functioning chassis' with batteries, control boxes, antennas, motors and blades. The only parts I cared about were the control parts, so I set to work stripping them out. Two servo motors, two drive motors, and a control box per bird. One set would be all I needed in theory, but it was good to have the second there just in case. Lucky, because the second control box had epilepsy and would randomly completely spaz out. After that I had to try to figure out how the thing was going to move.

We'd had a bunch of different options on the table for getting the power into the water including 90 degree gearboxes and universal joints but about this time it became obvious that with the heli parts stripped away, the drive-shaft for the top rotor was actually going to be the weapon of choice. After stripping it down and getting the shaft out, which included a little cutting and some filing to smooth the burs and allow removal, we had them - one from each bird. The linkage wasn't great as previously they were gear driven, but it worked enough for testing.

Next, off to Mitre-10 to look for mounting hardware. I got some wood screws and found these nifty aluminium brackets. I'd seen them before and wondered what I might be able to use them for - evidently, I'd found my answer. As it was just the alpha test (and I lacked proper measuring equipment) I screwed them on and secured the motors and control box with zip ties. Only problem was, the power cables weren't long enough to reach the controller. And the only wire I had was cat5 - not enough to carry this kind of current. So, I found an old PC power supply that for some reason I hadn't thrown out, and cannibalised it for its cables. A little soldering, and we were done.

The following night, I added a wireless camera to the head - our duck could see.

Still missing - two propellers, waterproofing, fixtures that aren’t mostly made of cable ties, but I have one more spare duck to test with before the race. Which, by the way is Saturday the 26th of September.

Click here for the full build photos.

Alternately, there are some videos of the first servo/motor test and an installation test after the build.

guitar hero

2009-04-05T22:59:00.000+10:00

a few weeks back my guitar hero controller started flipping out something hardcore. i don't know why. whenever it had batteries in it it would constantly trigger and then minimise the xbox menu. it's as if it were pressing the xbox button on the controller over and over really fast. when i took the face off to see if there was anything obviously wrong, the sticker that said 'warranty void if removed' covering one of the screws had been ripped in half. oh well, no warranty, may as well dive in.

so i took it apart..

i found the circuit board that the ring of light button attached to and removed the buttons, and powered on the controller. same behaviour, even with nothing attached. the button itself wasn't malfunctioning or out of shape, the switch on the board was really just fucked. it's that round thing in the middle of this photo:

so i followed its path back along the circuit board until it came to a ribbon cable. there was no way i was going to be able to easily install a new switch off the pins where it terminated and i didn't have any solder handy so i traced it back to the middle of the wires and cut them to disconnect them from the busted switch on the board, stripped them back, and found a power button from an old computer i junked a while back. i cut its pin mounts off and stripped the wire back, and wrapped them up together, and wouldn't you know it, it worked. a little electrical tape to hold them together and we were ready to start installing properly.

i got so involved with the next part i forgot to take photos but long story short, stanley knives are ineffective for cutting through thick plastic unless you push down hard and rock back and forth on them. slicing is completely ineffective. using a combination of this technique and using a soldering iron like a light sabre i managed to cut holes in the face and the body of the guitar to mount my new power switch. eventually it all lined up and we were ready to reassemble and go, and presto!

i decided to leave the old RoL button on there because, well it looks like it's supposed to be there and it looked weird without it. all in all, pretty easy and extremely rewarding ie i can play again now. still, can't pass dragonforce on hard anymore..

edit: i can now.

How to set your Cisco 7940 and 7941 IP phone up to work with Asterisk

2009-04-05T22:57:00.002+10:00

7940:

These phones still use the old style configuration files so first things first, you'll need to trick it into updating itself to some SIP firmware.

Things you will need:

- the SIP firmware itself, several files, in this example: POS03-08-2-00.loads
POS03-08-2-00.bin
POS03-08-2-00.sb2
PO003-08-2-00.sbn
- OS79XX.TXT, which contains a single line with the sbn file name in it. in this case: PO003-08-2-00
- a TFTP server and a DHCP server. any will do as long as it can supply option 66 or 150 to deliver the IP of a boot server. if you intend to do this under linux, here is a working example of my dhcpd.conf:

option boot-server code 66 = string;
ddns-updates on;
ddns-update-style ad-hoc;
option domain-name-servers 192.168.0.70;
option subnet-mask 255.255.255.0;
option domain-name "hgh.local";
option routers 192.168.0.1;
option boot-server "192.168.0.70";
# Local LAN
subnet 192.168.0.0 netmask 255.255.255.0 {
range 192.168.0.15 192.168.0.30;
}
# extension 100 - 7940
host ext100 {
ddns-updates on;
hardware ethernet 00:15:2B:22:31:56;
fixed-address 192.168.0.10;
}
# extension 101 - 7941
host ext101 {
ddns-updates on;
hardware ethernet 00:16:C7:64:23:14;
fixed-address 192.168.0.11;
}

Obviously substitute for your own IP ranges, in bold I have given the address of the TFTP server that is also my Asterisk server. Please substitute 192.168.0.70 throughout this document for whatever your phone server is going to be. I've put in manual assignments for my phones there, the MAC addresses can be found on a sticker on the back of the phone itself. Other than that, check the dhcp leases for information about what IPs have been given to what phones.

Alternately, if you want to do this under windows, download tftpd32 from http://tftpd32.jounin.net/ as that works too.

Step zero, return the phone to factory defaults. I don't want to be fighting with whatever shit you've got on there already. There is a good walkthrough this procedure on the Cisco web site.

First things first, trick the phone into downloading its firmware files. Plug the phone in, power it on and let it start looking for a callmanager. If your DHCP option66 has worked properly you should see it requesting files from your TFTP server. You can watch this happening live in the Log tab of tftpd32 or in the log file from your tftp server. To watch it happen, SSH to your linux machine and type something like tail -f /var/log/tftpd.log. It will look for a bunch of them but when eventually it will look for a default setting for any phones, since it can't find its own. This file is XMLDefault.cnf.xml so create it and here is what it should contain:

The firmware version I'm using is listed there for 7940 under loadinformation. the processNodeName tag further up the file tells it where to try to load the new firmware from, in this case my tftp server. Stuff should happen on the screen to indicate it's gone looking for the firmware. If it hasn't, check back to the TFTP server logs and see which files the phone is looking for. This will be your biggest clue as to what the phone is doing since the phone will probably have its configuration locked down. There are some status messages you can get to in the Settings menu but they generally aren't useful. Anyway.

Look at the phone and the tftp logs to make sure it's actually downloading the firmware. Once it's got the SIP firmware on it, it will install it on its own, and then it will reboot itself. Whatever you are doing DO NOT TURN OFF THE PHONE OR REMOVE THE POWER while the firmware is being updated. This goes for ANY device. You will create a paperweight.

After it's rebooted it will start looking for its configuration file. Step 1 is to set the default configuration for all phones. For 7940s that take the old style ".cnf" configuration instead of the newer XML configs, this is called SIPDefault.cnf.

My SIPDefault.cnf looks like this:

# Image Version
image_version: "P0S3-08-2-00"

# Proxy Server
proxy1_address: "192.168.0.70"

# Proxy Server Port (default - 5060)
proxy1_port: "5060"

# Emergency Proxy info
proxy_emergency: "192.168.0.70"
proxy_emergency_port: "5060"

# Backup Proxy info
proxy_backup: "192.168.0.70"
proxy_backup_port: "5060"

# Outbound Proxy info
outbound_proxy: ""
outbound_proxy_port: "5060"

# NAT/Firewall Traversal
nat_enable: "0"
nat_address: ""
voip_control_port: "5061"
start_media_port: "16384"
end_media_port: "32766"
nat_received_processing: "0"

# Proxy Registration (0-disable (default), 1-enable)
proxy_register: "1"

# Phone Registration Expiration [1-3932100 sec] (Default - 3600)
timer_register_expires: "3600"

# Codec for media stream (g711ulaw (default), g711alaw, g729)
preferred_codec: "none"

# TOS bits in media stream [0-5] (Default - 5)
tos_media: "5"

# Enable VAD (0-disable (default), 1-enable)
enable_vad: "0"
# Allow for the bridge on a 3way call to join remaining parties upon hangup
cnf_join_enable: "1" ; 0-Disabled, 1-Enabled (default)

# Allow Transfer to be completed while target phone is still ringing
semi_attended_transfer: "0" ; 0-Disabled, 1-Enabled (default)

# Telnet Level (enable or disable the ability to telnet into this phone
telnet_level: "2" ; 0-Disabled (default), 1-Enabled, 2-Privileged

# Inband DTMF Settings (0-disable, 1-enable (default))
dtmf_inband: "1"

# Out of band DTMF Settings (none-disable, avt-avt enable (default), avt_always - always avt )
dtmf_outofband: "avt"

# DTMF dB Level Settings (1-6dB down, 2-3db down, 3-nominal (default), 4-3db up, 5-6dB up)
dtmf_db_level: "3"

# SIP Timers
timer_t1: "500" ; Default 500 msec
timer_t2: "4000" ; Default 4 sec
sip_retx: "10" ; Default 11
sip_invite_retx: "6" ; Default 7
timer_invite_expires: "180" ; Default 180 sec

# Setting for Message speeddial to UOne box
messages_uri: "*97"

# TFTP Phone Specific Configuration File Directory
tftp_cfg_dir: "./"

# Time Server
sntp_mode: "unicast"
sntp_server: "192.168.0.70"
time_zone: "EAST"
dst_offset: "0"
dst_start_month: "Mar"
dst_start_day: ""
dst_start_day_of_week: "Sun"
dst_start_week_of_month: "2"
dst_start_time: "02"
dst_stop_month: "Nov"
dst_stop_day: ""
dst_stop_day_of_week: "Sunday"
dst_stop_week_of_month: "1"
dst_stop_time: "2"
dst_auto_adjust: "1"

# Do Not Disturb Control (0-off, 1-on, 2-off with no user control, 3-on with no user control)
dnd_control: "0" ; Default 0 (Do Not Disturb feature is off)

# Caller ID Blocking (0-disabled, 1-enabled, 2-disabled no user control, 3-enabled no user control)
callerid_blocking: "0" ; Default 0 (Disable sending all calls as anonymous)

# Anonymous Call Blocking (0-disbaled, 1-enabled, 2-disabled no user control, 3-enabled no user control)
anonymous_call_block: "0" ; Default 0 (Disable blocking of anonymous calls)

# Call Waiting (0-disabled, 1-enabled, 2-disabled with no user control, 3-enabled with no user control)
call_waiting: "1" ; Default 1 (Call Waiting enabled)

# DTMF AVT Payload (Dynamic payload range for AVT tones - 96-127)
dtmf_avt_payload: "101" ; Default 100

# XML file that specifies the dialplan desired
dial_template: "dialplan"

# Network Media Type (auto, full100, full10, half100, half10)
network_media_type: "auto"

#Autocompletion During Dial (0-off, 1-on [default])
autocomplete: "1"

#Time Format (0-12hr, 1-24hr [default])
time_format_24hr: "0"

# URL for external Phone Services
services_url: "http://192.168.0.70/xmlservices/index.php"

# URL for external Directory location
directory_url: "http://192.168.0.70/xmlservices/PhoneDirectory.php"

# URL for branding logo
logo_url: "http://192.168.0.70/cisco/bmp/l4wlogo-new.bmp"

# Remote Party ID
remote_party_id: 1 ; 0-Disabled (default), 1-Enabled

I've bolded the parts you care about - specifically, your firmware version, the proxy (SIP server) address, an instruction to, yes, actually register with the proxy, to tell it it's not working with nat (unless you don't have your own asterisk server and are registering directly with an internet SIP server, then this should = 1 and the IP below should be your WAN IP), enabling telnetting to the phone (default password, cisco), there's some stuff about the timezone here, if you're in Australia on the east coast you'll want to leave it at Eastern Australia Standard Time EAST. There's settings for daylight savings and also at the end, the option to upload a logo to the phone to display.

Now when the phone should start looking for this file, and then for its own configuration file. This is pretty simple since you did all the main configuration that would be the same for all the phones at your site. It really just contains the registration information. The name of this file is SIP followed by your MAC address. Example config from my phone:

; phone-specific configuration file sample
phone_label: "Chris Pollock "
line1_name : 100
line1_authname : 100
line1_password : 100

Now this is very important. The line name gets sent in your SIP registration so for the love of god, do not do what I did and set it to something silly and arbitrary like the name of the server. If you're using Asterisk and FreePBX (or trixbox) this will probably be your extension number, so leave the authname and name the same in this config. You will be angry at yourself when you realise. Trust me. Please set a more secure password.

Voila - your phone should request these files from the TFTP server and download them, then set itself up with its config. You should've configured Asterisk by now to handle this extension, that is beyond the scope of this document. Please see one of the other hundred FAQs about how to configure Asterisk, probably at voip-info.org which is SO great. Alternately, if you don't want to use these other files, the phone itself can be configured from the handset. Unlock the config from the menu, the default password is cisco, then go in and edit the SIP lines. But it's easier this way. You can also telnet to the phone after it's been defaulted, this is pretty useful for debugging registration issues. This is how I found that the line1_name meant something so go in there and dig around.

NOW. ON TO THE 7941.

Factory reset the phone. This will erase the firmware that's on it so be prepared for a rude shock. When this phone boots up the first thing it will do is go looking for new firmware. The 7941 firmware is quite different to the 7940 so it doesn't do anything useful like request a whole bunch of files from the TFTP server, it'll just look for one file indefinitely - term41.default.loads. If you were smarter than me you may have tried to put the 7941 information into the XMLDefaults.cnf.xml - this may have worked, I don't know, but my 7941 never asked for that file. term41.default.loads is part of the 7941 firmware, which has a lot of files in it. Extract your firmware into the tftp server's root directory where all these other files are. term41.default.loads knows which firmware it came from and will instruct the phone on how to download all the other stuff it needs, the application launcher, all of it. Once the phone finds term41.default.loads it will just eat the rest of it automatically so go get a coffee and come back.

When it eventually finishes doing all that stuff and reboots, it'll restart and start looking for other files. It should have also enabled its web interface, and its SSH interface. These may be useful if you fuck up the next step. The files it'll look for will be CTLSEPmacaddress.cnf.xml and SEPmacaddress.cnf.xml - ignore CTLSEP, we're going to use the other one. It's quite big so bear with me. Most of this stuff can probably be put in XMLDefaults.cnf.xml but I'm not sure. If you're not doing this for a big office deployment (and lets face it, you're probably not, otherwise you'd be behaving like a professional and using either another brand of phone, or using a CallManager) then it really doesn't matter.

Here's an example of my SEP0016C7642314.cnf.xml file:

<device>

<deviceProtocol>SIP</deviceProtocol>

<sshUserId>admin</sshUserId>
<sshPassword>cisco</sshPassword>
<devicePool>
<dateTimeSetting>
<dateTemplate>D/M/Ya</dateTemplate>
<timeZone>E. Australia Standard Time</timeZone>
<ntps>
<ntp>
<name>192.168.0.70</name>
<ntpMode>Unicast</ntpMode>
</ntp>
</ntps>
</dateTimeSetting>

<callManagerGroup>
<members>
<member priority="0">
<callManager>
<ports>
<ethernetPhonePort>2000</ethernetPhonePort>
<sipPort>5060</sipPort>
<securedSipPort>5061</securedSipPort>
</ports>
<processNodeName>192.168.0.70</processNodeName>
</callManager>
</member>
</members>
</callManagerGroup>
</devicePool>

<commonProfile>
<phonePassword></phonePassword>
true
<callLogBlfEnabled>2</callLogBlfEnabled>
</commonProfile>

<loadInformation>SIP41.8-4-3S.loads</loadInformation>

<vendorConfig>
<disableSpeaker>false</disableSpeaker>
<disableSpeakerAndHeadset>false</disableSpeakerAndHeadset>
<pcPort>0</pcPort>
<settingsAccess>1</settingsAccess>
<garp>0</garp>
<voiceVlanAccess>0</voiceVlanAccess>
<videoCapability>0</videoCapability>
<autoSelectLineEnable>0</autoSelectLineEnable>

<webAccess>1</webAccess>
<spanToPCPort>1</spanToPCPort>
<loggingDisplay>1</loggingDisplay>
<loadServer></loadServer>
</vendorConfig>

<networkLocale>Australia</networkLocale>

<networkLocaleInfo>
<name>Australia</name>
</networkLocaleInfo>

<deviceSecurityMode>1</deviceSecurityMode>

<authenticationURL>http://192.168.0.70/cisco/services/authentication.php</authenticationURL>
<directoryURL>http://192.168.0.70/xmlservices/PhoneDirectory.php</directoryURL>
<idleURL>http://192.168.0.70/xmlservices/index.php</idleURL>
<informationURL></informationURL>

<messagesURL></messagesURL>
<proxyServerURL></proxyServerURL>
<servicesURL>http://192.168.0.70/xmlservices/index.php</servicesURL>
<dscpForSCCPPhoneConfig>96</dscpForSCCPPhoneConfig>
<dscpForSCCPPhoneServices>0</dscpForSCCPPhoneServices>
<dscpForCm2Dvce>96</dscpForCm2Dvce>

<transportLayerProtocol>4</transportLayerProtocol>

<capfAuthMode>0</capfAuthMode>
<capfList>
<capf>
<phonePort>3804</phonePort>
</capf>
</capfList>

<certHash></certHash>
<encrConfig>false</encrConfig>

<sipProfile>
<sipProxies>
<backupProxy>
<backupProxyPort>
<emergencyProxy></emergencyProxy>
<emergencyProxyPort></emergencyProxyPort>
<outboundProxy></outboundProxy>
<outboundProxyPort></outboundProxyPort>
<registerWithProxy>true</registerWithProxy>
</sipProxies>

<sipCallFeatures>
<cnfJoinEnabled>true</cnfJoinEnabled>
<callForwardURI>x--serviceuri-cfwdall</callForwardURI>
<callPickupURI>x-cisco-serviceuri-pickup</callPickupURI>
<callPickupListURI>x-cisco-serviceuri-opickup</callPickupListURI>
<callPickupGroupURI>x-cisco-serviceuri-gpickup</callPickupGroupURI>
<meetMeServiceURI>x-cisco-serviceuri-meetme</meetMeServiceURI>
<abbreviatedDialURI>x-cisco-serviceuri-abbrdial</abbreviatedDialURI>
<rfc2543Hold>false</rfc2543Hold>
<callHoldRingback>2</callHoldRingback>
<localCfwdEnable>true</localCfwdEnable>
<semiAttendedTransfer>true</semiAttendedTransfer>
<anonymousCallBlock>2</anonymousCallBlock>
<callerIdBlocking>2</callerIdBlocking>
<dndControl>0</dndControl>
<remoteCcEnable>true</remoteCcEnable>
</sipCallFeatures>

<sipStack>
<sipInviteRetx>6</sipInviteRetx>
<sipRetx>10</sipRetx>
<timerInviteExpires>180</timerInviteExpires>
<timerRegisterExpires>3600</timerRegisterExpires>
<timerRegisterDelta>5</timerRegisterDelta>
<timerKeepAliveExpires>120</timerKeepAliveExpires>
<timerSubscribeExpires>120</timerSubscribeExpires>
<timerSubscribeDelta>5</timerSubscribeDelta>
<timerT1>500</timerT1>
<timerT2>4000</timerT2>
<maxRedirects>70</maxRedirects>
<remotePartyID>false</remotePartyID>
<userInfo>None</userInfo>
</sipStack>

<autoAnswerTimer>1
<autoAnswerAltBehavior>false
<autoAnswerOverride>true
<transferOnhookEnabled>false</transferOnhookEnabled>
<enableVad>false</enableVad>
<preferredCodec>none</preferredCodec>
<dtmfAvtPayload>101</dtmfAvtPayload>
<dtmfDbLevel>3</dtmfDbLevel>
<dtmfOutofBand>avt</dtmfOutofBand>
false
false
<kpml>3</kpml>

<natEnabled>false</natEnabled>
<natAddress></natAddress>

<stutterMsgWaiting>0</stutterMsgWaiting>

<callStats>false</callStats>
<silentPeriodBetweenCallWaitingBursts>10</silentPeriodBetweenCallWaitingBursts>
<disableLocalSpeedDialConfig>false</disableLocalSpeedDialConfig>

<startMediaPort>16384</startMediaPort>
<stopMediaPort>32766</stopMediaPort>

<voipControlPort>5060</voipControlPort>
<dscpForAudio>184</dscpForAudio>
<ringSettingBusyStationPolicy>0</ringSettingBusyStationPolicy>
<dialTemplate>dialplan.xml</dialTemplate>

<phoneLabel>Chris Pollock</phoneLabel>
<sipLines>
<line button="1">
<featureID>9</featureID>
<featureLabel>101</featureLabel>
<name>101</name>
<displayName>101</displayName>
<contact>101</contact>

<proxy>192.168.0.70</proxy> <port>5060</port>
<autoAnswer>
<autoAnswerEnabled>2</autoAnswerEnabled>
</autoAnswer>
<callWaiting>3</callWaiting>

<authName>101</authName>
<authPassword>101</authPassword>

<sharedLine>false</sharedLine>
<messageWaitingLampPolicy>1</messageWaitingLampPolicy>
<messagesNumber>*97</messagesNumber>
<ringSettingIdle>4</ringSettingIdle>
<ringSettingActive>5</ringSettingActive>

<forwardCallInfoDisplay>
<callerName>true</callerName>
<callerNumber>false</callerNumber>
<redirectedNumber>false</redirectedNumber>
<dialedNumber>true</dialedNumber>
</forwardCallInfoDisplay>
</line>
</sipLines>
</sipProfile>
</device>

As per usual, the bolded bits are the ones you care about. Put your info in there. Keep in mind that the latest version of Trixbox has a Cisco deployment tool in it. This may have worked for the 7940, I don't know as I didn't find it until after doing it manually, but it DOES NOT WORK for the 7941 with version 8 of the SIP firmware! DOES NOT. The config files it generates are designed for version 7 which, amongst other things, was had a few different config items that version 8 doesn't understand, and also used 0 instead of false to disable features. ALWAYS USE FALSE with Version 8 of SIP code on a 7941! Yes I'm aware that the auth values get repeated - don't ask, I don't know.

Basically what happens if you use the Trixbox version is that it will spit at your config file. It will just say that your config's invalid and remain Unprovisioned. This is impossible to debug from the web interface of the phone, but very possible from an SSH session. If you've done it properly, the phone should read its config, reboot, register, and be set up. Unlike a 7940, the 7941 cannot be configured from the handset. The SIP configuration is all done from these config files, whether you like it or not.

So. SSHing to the phone. When you attempt to SSH to the phone it will ask the TFTP server for a file called authorized_keys. You don't have this file so nothing you type in will work, even if the password is right. So you'll need to generate one. Download Putty, PuttyGen and pageant from the putty download page. Open up puttygen and hit Generate. save the private key somewhere useful on your computer, then save the public key too. Copy the contents of the public key file and dump it into authorized_keys on the TFTP server. Now open pageant and load your private key into it. This will send your key to the phone for validation.

SSH to the phone's IP address. It'll request authorized_keys and then ask you for a username. The username is Default.

Then it'll ask you for the real username and password. the defaults are:
default/user - normal SSH shell, lets you into the phone's file system, not really useful)
log/log - shows normal phone status messages
debug/debug - useful for seeing whats going wrong

If your phone is rejecting your config and saying it's invalid, type debug then ? and hit enter. It'll show you a list of all the commands you can use but there's some useful ones in there about debugging XML, turn all the xml debugging that you can on. It'll go through piece by piece what the phone's doing and if it's complaining about your files or the registration is failing this will be where it lets you know. In my case, this is where I discovered the difference between the version 7 and 8 SIP firmware.

The other important thing is that if your SSH-ing doesn't work, you CAN get to the errors through the console dumps on the web interface, it's just very tricky to find. Every message in the world comes up on the console screens and it generates a new file every minute almost, so you'll be opening files and scanning them for 'xml' as fast as you can to find what line of the config it's choking on.

If you got all this right, this should happen:

So that's it! You're done. You were probably done several paragraphs ago but I thought I should include that troubleshooting bit in there. Maybe this will be useful to someone at some point.

FYI all the files used in this tutorial EXCEPT THE CISCO FIRMWARE will be available from http://loss4words.com/sip shortly - do not contact me asking for Cisco firmware. Get your own support contract, they are well worth it even if they don't support SIP in a non-CCM environment.

I'd love to be able to say this is the process I followed, but this is a subset of things that were successful. Figuring all this out was interspersed with lots of failure which explains why this took six days.

project matte black 2: this time, it's blacksonal

2009-04-05T22:54:00.000+10:00

project matte black - part 2

before - a 2007 apollo volare.

step 1 - remove wheels.

.. then prime the hell out of them.

and paint!

i'll need this before i go any further.

time to remove the gear assembly and fail at removing the cranks. sanding is
difficult and not very much fun so i didn't do much of it, or a very good job.

primed!

painted! this was the second of three coats.

wait, what?

AWWWRIGHT surprise time attack!! finished.

i kinda got a little dirty. the before and after photos next to each other are really not that amazing but now it has a lot more character. it definitely doesn't look like it's worth anything haha just the way i like it. a real road warrior. the downside is that my gears which were already in bad need of a service are in even worse need now due to that i don't know what the hell i'm doing, and i took my entire bike apart without documenting all the moves and re-arranges i did of all the unremovable parts, so when i went to reattach everything, nothing made
sense. again, i got burned by not following my instincts on whether or not to snip a cable and re-attach it, and in the process have left myself without 1st and 2nd cogs on the rear, and the front is a tad touchier now. so i pretty much need to go get that service i've been putting off for ages. preeeeeetty sure i'm gonna get yelled at by the dude at the bike shop but FUCK HIM HE'S NOT MY DAD

project matte black

2009-04-05T22:46:00.000+10:00

project matte black

so, i've had my old bed and linen for a long time. in the spirit of fresh starts i decided i should get new linen. something that could be mine. and everyone knows i've been obsessed with the idea of painting things matte black for a while so when i decided to paint the bed as well it was an easy choice.

i managed to avoid the lighting aisle of ikea but not the FIRE SECTION.

before. i need to clean under my bed more often.

the headstock had to come off to get the bed out the door and the top had to be removed so i could get the beams out to resilver them.

beams before.

beams after.

headstock freshly and unevenly painted. a second coat was required to even out the dense spots.

then i ran out of paint and had to go back to mitre10 for more. unfortunately they were out of the dulux matte that i'd been using and i had to buy generic. DO NOT WANT. always buy brand name, this stuff was crap.

after the initial pass on everything. a final touchup pass was done after this.

much more even this time. ready to be put back together.

success! looking much better now.

double success! new linen looks good.

over all including hardware store trips it took about two and a half hours. could've been worse and looks WAY better.

hello

2009-04-05T22:42:00.001+10:00

Hi.

So if you've found this you probably already know that my name's Chris and that amongst other things most of the crap on loss4words.com is my fault. I like to break things. Or rather, I like to take things apart, find out how they work, figure out why they don't work and try to fix them, then throw what's left in the bin when I realise it's not fixable. At least not by me. But every now and then, these little DIY experiments actually end up working out for me, and that's what I've started this blog for - recording this stuff.

I like riding, computers, guitars, musical instruments, recording, video games, martial arts, networking and hockey. Most of what comes up here will probably be related to that stuff.

Enjoy!