After over a week of waiting, I finally got the wind to try my panoramic KAP rig with the Canon T2i. In short, I’m torn…
On the face of it, it worked ok. Here’s a two shot sample:
But digging deeper, it leaves a good bit to be desired. The same day I made this, I also set the camera up to make movies and flew it while the rig panned around. It answered a couple of questions I had from going through the images. Here’s what I really learned:
The magnetic encoder on the pan axis really does work. But the kit lens on the T2i is soft at its widest setting if 18mm, so I habitually run it at 24mm. That’s just a little too narrow to work with the shot spacing I built into the pan gear (12 magnets = 12 shots per rotation). I either need a wider lens, or more shots per rev.
More shots per rev means a slower overall rotation, so the wider lens really is the right way to go. I just can’t afford the glass right now. It’s cheaper for me to pick up a new pan gear from Robot Zone, drill out a tighter pattern of holes, and make a finer pitched rig. It would be even cheaper if I just re-programmed the controller to stop twice per magnet – once on magnetic maximum, and once at magnetic minimum. Voila, double the shot density for no actual increase in price. Can’t beat it. So that’s next.
The other thing I learned is that most of the variation in orientation, shot-to-shot, comes from random movement of the rig in the air and not from variations in the pan axis. There’s a lot of roll and pitch, but there’s also a fair bit of wind-induced yaw. Something on the KAP rig is acting like a sail. In some orientations the wind tries to blow it a little further around, and in others it tries to blow it back. After staring at the video for a while I figured it out. It’s my lens! The lens on my camera is poking way out in space, just itching to be blown around by the wind. ARGH!
There’s a fix for the wind-induced yaw as well: replace my Picavet suspension with a very tightly constrained pendulum. Since I’ve been itching to build a damped pendulum suspension, this just adds fuel to that project. If I’m careful with the design, I should be able to take out the shakes in pitch and yaw, and damp both as well. Time will tell if that’s the right way to go. Meanwhile, back to the drawing board. Literally.
Santa was nice to me, and got me an RC airplane for Christmas: a Bixler 2. The guys at Flite Test happily said it takes about an hour to put together, and that the plane will be ready to fly before your battery charges. It took me closer to four hours, and all three of my batteries were charged before I’d even put the wings on. From this I conclude that I’m either far more meticulous than most model airplane builders, or that I’m just plain slow. I’m pretty sure it’s the latter.
But I did finally get it built. I initially set it up four-channel: elevators; ailerons; rudder; throttle. The Bixler 2 can also be set up for flaps, but four channels was already more than I could handle. In the afternoon, I took it out.
Flight 1: The Bixler is a hand-launched plane. You face into the wind, throttle up, and toss it into the air at a slight angle. I did all these things, and suddenly I was flying! Or rather the plane was. But it was obviously nose heavy. I had to keep applying pressure to the stick to keep it in the air. When I throttled down, it had a horrid glide slope, close to 4:1. I brought it around the field a couple of times, then took it in for an only slightly cartwheel landing. I’d done it! I’d flown, and not crashed! Total time in the air: ~180 seconds.
Flight 2: After adjusting the battery to bring the center of gravity aft, I launched again. In retrospect I can see I launched with too little throttle, and added back-stick without thinking. I stalled. It rolled left and lawn-darted into the ground. In the process I ripped off one of the wings. Total time in the air: ~4 seconds. End of day 1.
When I got home I learned the wonderful art of EPO foam repair. The plane is pretty easy to fix after a crash, but I vowed I’d do whatever I could not to crash again. After the wing was glued, I went over the fuselage, checking for damage. Control surfaces all survived. Servos all survived, too. Motor, ESC, everything fine except for the wing spar. In addition to the damage to the wing’s foam, I’d cracked the carbon fiber spar that runs through both wings. Luckily it’s a 6mm tube, similar to many kite spars. I had a spare in my kite spar collection.
By afternoon of the next day, the glue had set and the wing was whole. I put everything back together to check it out. This time I put the airplane on a balancing stand and double-checked the COG. The new location of the battery was perfect. I loaded everything into my car and drove down to the Kohala Coast. It’s not the best place to fly since the ground is mostly dead grass and large rocks. But it’s got some of the cleanest low-speed air on the island, and gobs of space.
Flight 3: After checking the plane one last time, I ran the throttle up to 75%, held it at a slight up-angle, and threw it into the wind. The launch was perfect. “Get the plane three mistakes high,” was my first thought. My second was, “What’s that gawdawful noise coming out of my airplane?!” When I saw the propeller fall off the plane, I knew full-well what that noise was. The prop had come loose! Thank goodness I got the COG right this time. The Bixler 2 is a motor glider, meaning it has a motor. But more important it means it’s a glider. This time my glide slope was closer to 10:1. I got it into flat, level flight, and did my best to bring it down gently. Total time in the air: ~15 seconds.
Unfortunately this meant it flew out of sight, over a gulch, and nose-first into a boulder. But by golly my wings survived! After some rock scrambling I was able to recover my plane. With the exception of a seriously torn up nose and an utter lack of propeller, it came away without a scratch. When I got home I broke out the cement (again) and repaired the foam where it had split. But the prop was lost forever. So then I ordered two new prop adapters and eight new propellers. And six new wing spars. I’m just trying to put off having to order a whole new plane!
The Lesson: If there is one, I guess the lesson is that flying a model airplane is a heckuvalot harder than it looks. I have a lot of respect for people who can fly these planes and make it look effortless, with hundreds of hours of flight time to their name. But I can only guess how they got there: by crashing just like I did, and eventually learning how not to.
Ever since installing it, I’ve had some issues with my video downlink. For anyone doing FPV on an RC aircraft, this will all be old hat. But for me doing KAP, some of it came as a surprise. What kind of issues? You guessed it – crummy signal.
Lots of things can contribute to bad video: RF interference, noise on the power supply, antenna feed line losses, a poorly tuned antenna, the list goes on and on. Over the past several months I’ve checked most of these, but didn’t find any real answers. Yesterday I rebuilt the entire ground-side system, from power to receiver to monitor. As far as I can tell it’s now solid as a rock. Which leaves the air side.
My video transmitter is currently powered off of the same battery that drives my RC receiver and servos. (If you’ve been doing FPV, this will already come across as a bad sign.) I’d seen some warnings against this, but all of them tied back to electrical interference with an electronic speed controller in the “aircraft”. KAP rigs typically don’t have an ESC since there’s no prop to turn. No ESC, no interference, right? Well… Possibly. But there are other ways to kill a video signal than interference.
The problem I ran into didn’t have to do with noise so much as running into a limit of how much current the battery could source. The battery pack on my rig was powering a radio receiver, two full-sized servos (including a high torque metal gear servo), a GentFocus cable (which draws milliamps, at most), a 12V boost-buck power supply for the video system, and the video transmitter itself (a 200mW transmitter… no small beast.) Add it all up, and when everything was up and running it was drawing my batteries down pretty hard. After trying the same setup using a bench power supply, I came to the conclusion that the single battery pack just couldn’t handle the draw.
So I’m giving up on the single battery idea. I’m going to rework the transmitter to use its own battery – a three-cell LiPoly – and to be almost completely independent of the RC electronics. The only connection will be an RC power switch for the video hardware. (It’s nice to be able to power down your video remotely when you’re not actively using it.) If the bad signal persists, I’ll yank that connection, too. I’m planning to package the battery, transmitter, and RC switch into a nice, neat enclosure that I can install when I want video, and remove when I don’t. Easy peasy.
Meanwhile I’ve gutted the video from my KAP rig. It’s amazing how simple and clean KAP gear can be when you’re not constantly pushing it to do something more. Maybe having the video packaged as an option rather than as an integral part of the rig is the right way to do it. I’m curious to see how much I’ll use video once it’s no longer a built-in feature of the rig.
Yesterday there was too much wind. This evening when I went out to test the changes to my pano rig, there was no wind. My son and I drove as far south as Anaehoomalu Bay trying to find even a breath of wind to put a kite up with. But no dice.
Soooo… Yep, you guessed it. More long duration exposure photography! This time I photographed sunset from the south end of Anaehoomalu Bay. There are some nice rocks on that end of the beach. If you walk out far enough, there are some nice vantage points from which to work. I was fortunate enough to find a set of three rocks so my tripod could stay (mostly) dry. Unfortunately there wasn’t a rock for me to stand on. So I stood in the surf and tried not to step on any sea urchins.
Despite a soldering session earlier today, for some reason my remote shutter release is still not working in the field. (I swear it worked fine when I tested it at home!) So I was stuck doing 30-second exposures again. About an hour after I first stepped into the water, the sun set. I packed my gear and headed home.
This is a stack of six 30-second exposures, all made back-to-back. Total open shutter time is three minutes:
I tried to get out and test my upgraded panoramic KAP rig. I really did. With less than an hour to go before sunset I drove down the mountain to the ocean, hoping for good wind. I didn’t get it.
What I did get was some of the nastiest offshore wind I’d ever seen. Even miles from the ocean, I could see it was black. Not really Homer’s wine dark sea. More like a wind-whipped daemon. “This can’t be good,” I thought.
As I drove past the Mauna Kea Resort, I checked the palm trees. These are some of my favorite trees to check for gauging the wind at Hapuna Beach. I had never seen them so bent over and flattened. Hapuna was a no-go. As I came over the rise, I could see the pattern extended all the way down the coast. And with less than half an hour of light to work with, I realized the game was up.
So I did the only thing I could: I zoomed down to Waialea Bay, left my KAP gear in the car, and grabbed my tripod instead. Finally! A chance to play with long duration exposures!
I’ve had this 10-stop ND filter for a couple of weeks now, but except for a few sessions at Hapuna Beach, I hadn’t really had a chance to play with it. Oh, and those sessions at Hapuna? Not so good. Hapuna Beach State Park is a people park. But it’s not people I’m after for long duration photography. It’s scenery. If any beach could provide that, it would be Waialea Bay.
That’s when I remembered I’d broken my remote shutter release a few weeks ago. “Broke” is really too strong a word. I… wounded it. The cable had come apart where it enters the case, so it wasn’t tripping the shutter reliably. (Yes, this is yet another small project I’m going to take on during my vacation.) “No matter!” I thought. “I’ll stick with thirty second exposures and stack them!”
Stacking is a trick that amateur astronomers have been using for aeons to get long exposures without incurring too much dark noise. Even better, by stacking multiple exposures you can beat down photon noise as well. So it’s actually a win. The only catch is that the read noise adds with each exposure, so there’s a point of diminishing returns. I planned to stack four 30 second exposures to get two minutes of total open shutter time.
In short, it worked. I got set up just as the sun began to set, and I made 30 second exposures until the light was gone. When I got home I started to play. And this is what came out:
Not the KAP test I was hoping for, but somehow that’s ok.
I’m on vacation! W00t!! For the record, this is my first real (meaning more than one day) vacation since Rydra was hospitalized last year, about 17 months ago. And for the record, this was way way too long. I’m a little crazy right now. I’m sure my co-workers would agree.
Yesterday, my first day of vacation, I pulled apart my KAP bag and took stock of where I am. I also dumped a couple hundred grams of sand from my bag, took out bits of gear I’d accumulated but didn’t necessarily use, and overall made my bag a good bit lighter. But the real purpose of the exercise was to catalog all the minor annoyances that have been bugging me over the past few years, and address them. Aside from the KAP bag inventory, here’s some of the stuff I did:
The shock cord in my BBKK rig legs has been fraying on a zip tie I installed backward. So thing #1 was to replace the shock cord. While I had the legs off I also sanded over some of the rough spots on them to make them a little more smooth. Thing #2 (moving clockwise) was to cut off the offending zip tie and replace it with one that’s facing the proper direction. No more rough edges biting into the leg shock cord!
Thing #3 was completely serendipitous, but I’m glad I did it. Some while back on the KAP Forum, Dave Mitchell posted to a thread on line work about replacing the typical knot at the end of your kite line with a splice. Dave uses hollow-cored Dacron line similar to the stuff that Brooks Leffler sells. So of course I had to give it a try!
The main reason for using a splice, aside from the neater aesthetics it offers, is that it’s stronger than a knot. Every line, rope, and string has a breaking strength. Every knot will weaken the line to some degree, usually expressed as a percentage of line strength. Even a good knot, like a bowline, will appreciably weaken the line at that point. In the case of the bowline it’s something around 80% line strength. Other knots present far lower numbers. Of course there are knots with exceptionally high percentages. But in this game the splice always wins.
Dave posted a nice set of pictures showing how to splice hollow-core braided Dacron line. I used his procedure to splice mine, and love the results. Yesterday afternoon I took my gear to Hapuna Beach and flew my Nighthawk in full-blown offshore tradewinds. When I brought the kite down the pull on the line was more than I could wind, so Rydra and I walked it down. The splice hadn’t shifted or stretched at all. I’m impressed!
The last of the little things I did, to complete the circle, was to add a sticker to the bottom of my BBKK rig showing what function each channel on the radio handles. I had one of these in the past, but somehow lost it. It’s good to have it back, especially since I’ve been trading out between the Canon A650IS and the Canon T2i. Each takes a different shutter cable, so it helps to know where the thing needs to plug in.
But that’s the small list. The annoyances. Now for the big list.
I’ve had a couple of KAP projects hanging around for far too long. One big one is to gut my video downlink and re-vamp the power system to get rid of some persistent noise I’ve been dealing with. That’s going to happen this weekend. I’ve also been planning to re-visit my panoramic KAP rig to add a rotary encoder to the pan axis. That’s what I did today.
The first step was to remove the pan gear and drill out twelve holes to take twelve neodymium magnets. This went relatively smoothly using my Foredom as a live spindle on my Taig lathe. (Sorry, no pictures of the setup. My shop is still a mess.)
The magnets were epoxied into the holes using West Systems epoxy. Any epoxy would have done. But I’ve been using the West Systems epoxy on my boat, so it’s what I’ve got on-hand. I also love their pump system for measuring. I haven’t had a single batch fail to kick.
Years ago I picked up a bunch of Honeywell SS495A1 ratiometric Hall-effect sensors. These are analog devices that can be used to sense a magnetic field. They’re pretty flexible about input voltage, but I’m driving mine using +5V/0V with the output varying between 0.2V for -670 gauss to 4.8V for +670 gauss. No magnetic field is 2.5V.
All of this is being sensed by the Pololu RoboticsMicro Maestro servo controller. I picked one of these up a couple of years ago, after a fellow KAPer posted to the KAP forums about how easy it was to press them into service as autoKAP controllers. It’s been the controller for my panoramic KAP rig ever since I built it.
The Micro Maestro can use its six I/O pins to drive servos, for analog or digital input, or for digital output. I set up Pin0 on mine to read the Hall-effect sensor. The magnets I’m using are only 1/8″ diameter x 1/16″ thick, so they’re not all that strong. From playing around with the sensor and a magnet, I found a 3mm sense distance to be just about right. (My apologies for mixing units like this… I drive everyone nuts, no matter which system they prefer to use.)
All that was left was to make a bracket for the sensor, hot-glue everything into place, and put it all together.
The control scheme is pretty simple: Rotate until there’s a rising edge on the Hall-effect, and decelerate the pan servo to a stop. Once it has stopped moving, trip the camera’s shutter for half a second, then accelerate into the next move. No real smarts at all. The very essence of autoKAP. The advantage this system has over a strictly time-based setup, however, is that the position changes are quite accurate, there’s no chance of a missed shot, and the distance moved is completely independent of motor speed, battery level, etc. Here’s what it looks like in motion:
Unfortunately it’s raining right now. Cats and dogs. Welcome to winter weather in the tropics. So the first flight of this rig will have to wait. Meanwhile I’ve got more KAP upgrades to do. Video transmitter, here I come!
I had the perfect plan. I really did. I left home an hour before work so I could spend some time doing photography in early morning light. And I failed. It wasn’t the light. It wasn’t the subject or the gear. It was me. Plain and simple.
Let me back-track.
A few weeks ago I decided I wanted to try my hand at long duration photography. There are a couple of photographers I follow on Flickr who use the technique, and their long exposure work is truly beautiful. If I had to pick a word to describe it, it would be “ethereal”. Another word I might pick is “fun”. So a little over a week ago I gave it a try. I didn’t really have the right equipment, so I did what I could with the light, the shade, and a really tiny aperture on my lens.
A two-second exposure was the longest I could get. Certainly longer than the typical 1/1000 sec exposures I’m used to while doing KAP, but nowhere near the exposures I saw on Flickr. With the daylight getting brighter second by second, I finally had to give up. Later that day I ordered a ten stop ND filter, which arrived a few days ago. Initial testing showed it to be slightly darker than 10 stops, so I did some exposure testing and got a more precise number. In the case of this particular filter, it requires an exposure 1120 times as long as the un-filtered exposure. Perfect!
This morning was to be The Morning. The creek where I did my un-filtered test is right behind the building where I work, so grabbed my camera gear, drove to work, and parked. And that’s when it happened. The parking lot was empty. The shop door was sitting there, closed and locked. And as I took my gear out of the car it hit me: If I started early, I could get all the machining I’d planned to do that day done by noon.
The creek was right there. I could hear the water burbling over the rocks. But the allure of an early start without meetings, without phone calls, without interruptions of any sort won out. I stashed my gear in my office and went out to the shop instead. My camera never even came out of the bag.
Today is Friday. Tomorrow is a weekend. If I spend the weekend working, or doing all the gotta-do things that never seem to come off my list, I’ll know where my head is: the wrong place. I need a break.
More progress on the cameras I’ve been building at work! I spent the past several weeks testing and building the cameras’ getters. (Yes, a “getter” is a noun. “Gettering” is the verb. Even after having done this for years that still feels funky when I say or type it.) Anyway, the getters have finally joined the ranks of the parts that have gone from CAD to reality. It’s show and tell time!
A getter is a type of sorption pump. It typically includes some source of cold (a cryogenic gas, a closed-cycle cryocooler, or magic – your pick), a gettering material (activated charcoal in this case), and the bits and pieces to bring the cold to the gettering material. By cooling the gettering material down to cryogenic temperatures, it becomes a cold trap for residual gases in a vacuum system. Get the thing cold enough, and as molecules of residual gases come in contact with the getter, they tend to stick to it. Voila! Pump!
These cameras were originally designed to take a commercially available getter. But as the project progressed, we realized we needed more performance out of them than the commercial units offered. So we designed and built our own.
The first design criterion was that it pump faster than the commercial unit. The commercial getter we originally intended to use in these cameras is basically a cup with a bunch of activated charcoal glued in it. The easiest way to double the pump speed is to double the amount of surface area that’s doing the pumping. Answer? Use two cups and glue charcoal to the other cup, too! But space is limited, so we stuck the two cups back to back with one facing up and one facing down.
The second design criterion was that we didn’t want carbon dust getting all over the inside of the camera. For starters it’s dirty, and we’re building these things in a clean room for a reason. But carbon dust is also conductive. This is not what you want to have rain down all over a circuit board! Answer? Add covers!
So here’s the design we came up with:
Two getters are shown. The one in the foreground is right side up. The one in the background is upside down, showing off the counter-bored screw holes that allow the getter to be bolted to the cold head. Inside are the cavities that take the activated charcoal. On top of each cavity is a cover plate that’s meant to hold filter material in place to keep the carbon dust from leaking out.
Having a design is nice, but it doesn’t do you any good if it doesn’t actually work. The last two weeks have been spent testing different charcoal loads to see what works best. Monday we pumped down our latest setup, and so far it looks good. Our recipe at the moment is 2-3mm of Stycast 2850FT epoxy, poured into the bottom of each cavity with as much activated coconut charcoal dumped in as will fit. Once the epoxy has cured, the excess charcoal is poured out and the cavity is sealed with a sub-micron fiberglass filter to catch any residual carbon dust. This is then followed by a 60 mesh .0065″ diameter wire stainless steel screen to protect the filter (which is quite fragile), followed by the cover plate as shown in the above rendering. So far the lab tests look good. So do the getters:
As exciting as it is to see each part of these cameras in finished form, it’s just as exciting to see the various bits and pieces on my workbench slowly disappearing as they’re incorporated into final form. The cleaner my bench is, the closer I am to done. I can’t wait.
But I also can’t wait to see this instrument in action. The cameras are only a tiny tiny part of what this thing will be. It’s exciting.