The View Up Here

Random scribblings about kites, photography, machining, and anything else

  • Flickr Gallery

Archive for September, 2012

Engineering – From CAD to Camera

Posted by Tom Benedict on 26/09/2012

It’s a good week for projects coming together. First I had the successful flight of my new panoramic KAP rig. Today I closed our prototype camera for the first time and put it on the vacuum pump.

The last time I wrote about this project the picture I posted was a CAD rendering of the camera design as it stood that day.

Sitelle Camera September, 2012

This time I can actually show off the camera!

Sitelle Camera Pumping

Some parts are missing in this photo – namely the cryo cooler isolators and hose support, and the pre-amplifier box. Those parts are done, mind you. They’re just not installed because they’re not part of this test. Except for some of the internals, the cameras are just about complete. (And yes, those internals are what I’m working on now. More pictures and stories to come!)

This photo also represents a new idea we’re trying with this instrument: a prototype. Normally everything we build is a prototype because we only ever build one of them. This instrument calls for two cameras. In the old way of doing things we’d build precisely that: two cameras. This time we decided to build three. It came down to timing: building three was faster.

Yeah… that requires some explanation.

In almost every way these cameras are a new approach to a problem. It’s a detector we’ve never worked with before. We’re using a new pre-amplifier design. None of our existing cameras fit in the space we had available, so the form-factor is a new design. It’s the first camera we’ve designed around the Polycold PCC cryocooler rather than retrofitting at a later date. And it’s our first instrument to use this kind of vibration damping on the cryocooler. Because of some space constraints even the cold strap and cold foot are new. Any time you have even one new thing in an instrument, it pays to test test test. But testing takes time. Finding out your initial design didn’t work takes even longer.

Early on we realized having a prototype camera we could hack on would let us test each idea as it was integrated while still allowing development on the two production cameras to proceed. Lessons learned could be rolled into the final design as we went. It also let us do time-intensive tasks – like anodizing all the parts for the production cameras – while development and testing work was still being done.

As a result the anodizing is happening at the same time the electronics development is being done. All of that is being done in parallel with testing of the cryocooler and cold strap. And that is being done in parallel with the machining of the camera internals. No one task has to wait for any other. As a result, the cameras are being developed in much less time than it would have taken if we’d only built two. See? Building three was faster!

To add icing to the cake, the camera pumped down fine the first time. It has no leaks (a first for me on a first-pump!) a nice fast pump rate, and at just under 10kg fully built, it’s easy to handle. If the rest of the project goes as smoothly as this, I might actually get to take some time off to fly a kite at some point. I can only hope.

– Tom

Posted in Engineering | Leave a Comment »

Made Good #1 – Panorama Rig

Posted by Tom Benedict on 24/09/2012

I can actually say that in the past tense now: I made good. I finished that panorama rig I was building. I even had a chance to take it out and put it through its paces. In short, I’m stoked!

The Pano Rig

As you can see, it’s set up to take a variety of cameras. On the left is my Canon T2i, currently my KAP and ground workhorse. In the middle is my Canon A650IS, my KAP workhorse for many years. And on the right is a new combination featuring my cell phone, a Samsung Galaxy S2.

The real trick with this rig is the controller. It’s a Pololu Robotics Micro Maestro servo controller. It costs just under 20 USD and will drive up to six servos through a pre-canned program, or you can use some of the I/O pins as inputs and have the program respond in true robotic fashion. This rig is set up to use two of three servos, depending on which camera is installed in the rig.

In the first mode of operation the pan axis is plugged into Channel 0. This channel is set to rotate the rig at a fixed speed of 5RPM. It’s for use in conjunction with cameras that feature an intervalometer. Set the intervalometer to 1 second per photo, set the rig to spinning, and you have a very capable burst autoKAP setup. This is how the Samsung Galaxy S2 was set up.

In the second mode of operation the pan axis is plugged into Channel 1 and a shutter cable is plugged into Channel 3. The pan axis is set to make twelve smooth moves per full rotation of the pan axis, tripping the shutter after each move. Because the moves are smoothed (ramped acceleration and deceleration for each move), there’s practically no settling time at the end of a move. It simply stops. It’s a little slower than burst autoKAP, but there’s less camera motion in this mode. This is how the Canon T2i and A650 were set up.

To test this I ran all three cameras up into the air and gave them a whirl.

The Galaxy S2 performed, but right now I’m using a free intervalometer program that limits the camera’s resolution to 640×480. Considering it’s an 8MP camera, this is pretty cheesy. But now that I’ve seen it work, I’m more inclined to shell out the money for a real application.

The A650 IS performed like a champ:

Hapuna Prince Mid-Afternoon

This was done about two and a half hours before sunset using six photos. Nice shadows, but no real challenge on light. The A650 has optics that are quite good, even by today’s standards. At 12MP it has plenty of pixels to work with. And while smaller than most high-end compact cameras, the detector in the A650 is still larger than many on the market today. This gives it reasonably good noise characteristics for a compact.

The T2i performed equally well:

Hapuna Prince Late in the Day

This was done over an hour after the A650 IS panorama using five photos. It was done well into the golden hour, when the shadows start to get long and the light takes on a warmer hue. The original for this came out at just over 13k pixels wide. I scaled it down several times before uploading it to Flickr. The original is quite sharp.

But what I really wanted to see was how late in the day I could still make a panorama. I’ve had a goal of making a true post-sunset panorama of a city skyline. I think (I THINK!) I may finally have a setup that will let me pull this off.

Hapuna Prince at Sunset

This was also a five photo panorama. Unfortunately I had my ISO clamped at 100, so it’s darker than I’d like. I bumped this up by a stop in Photoshop – enough to bring  out some detail, but not enough to make it not look dark. It’s also not as sharp as I’d like, but letting the ISO float a little more should help with that as well. Obviously more experimentation is required.

Regardless, I think this rig is a hands-down win for me. It’ll take every camera I own, including my cell phone. It’ll make excellent aerial panoramas. And it’s small enough to fit in my kite bag along with my A650 so I can leave my full KAP bag at home on those days where the full RC-KAP video rig is just too much.

Now on to the damped pendulum suspension!

– Tom

Posted in Kite Aerial Photography | 2 Comments »

Making Good

Posted by Tom Benedict on 17/09/2012

I’m still under a lot of pressure from this project at work, and an unfortunate amount of my weekend was devoted to that instead of weekend-ish type things. But I still set some time aside to go through my spare KAP bits, think about the rigs I want to make, and tinker.

Among other things, I made a start on the panorama rig. I found my Pololu Micro Maestro servo controller and loaded it with enough code to rotate the pan axis and trigger the shutter on the camera. So far so good. I had a spare Brooxes Utility Bracket, which I plan to use as the frame of the rig, and found a Futaba servo I’d already modified for continuous rotation. Unfortunately that’s as far as I could get. I don’t want to suspend a DSLR from a servo, so I really need a geared pan axis for this rig. I also didn’t have a spare Picavet to suspend it from, so I couldn’t even set up a dummy test.

Rather than call it quits, I built a panorama rig around my A650 instead. It’s light enough to suspend from a servo, and the rig uses all the same parts I’d need for the T2i pano rig. It’s still lacking a Picavet cross, but that’s all. Other than the Picavet, it’s ready to run. I ordered the geared axis and a Picavet cross from Brooks, and set the rig aside until the parts come in.

Afterward I toyed around with the idea of gutting my T2i rig, but I don’t have enough recent experience under my belt to make the rebuild worthwhile. The whole point of the rebuild is to address any existing problems with my gear. I’ve forgotten what those problems are! So rather than risk making my rig unusable, I took it out for a session at the beach.

In the past, 99% of my KAP sessions at Hapuna Beach have been from the north end, near the Hapuna Prince Beach Resort. But ever since Rydra’s surgery, we’ve tended toward the south end of the beach. The path to the north end of the beach includes a fairly hairy downhill section that Rydra found hard to navigate with only one eye. She used hiking sticks for a while, but finally said enough is enough. Besides, once we found the rocks at the south end of the beach were perfect for cliff diving, the kids were sold on the idea!

The wind has been spotty of late, but this time I lucked out. The light was only so-so, but the kite had plenty of pull to lift the heavy DSLR rig. I had some problems with the video link (hey! problems! I can address those!), so some of the time I was flying blind. But it was a good session.

Looking the Other Way

Early on I let the camera fly high, and took a picture looking toward the north, back where we used to go. If you click on the photo, it will take you to where it is hosted on Flickr. (This is true of every photo I post in my blog.) On the Flickr site there’s a note pointing out the location where most of my other Hapuna KAP photos have been made.

Hapuna Beach from the South

Since I’ve got most of a panorama rig sitting on the bench, I couldn’t help but make one manually using my existing rig. KAP panoramas are actually quite straightforward to make, provided the photos are made close enough together in time for the camera not to have much of a chance to move around in the sky. Since the wind was quite steady, this was easy to do even while aiming manually. But one of the ideas with a panorama rig is to minimize the time between shutter clicks. The code I put into the Micro Maestro times out to about one photo every three seconds. Still not as fast as the burst-KAP I did with my A650 HoVer rig, which could cycle the shutter every 1.1 seconds, but not bad. And if my plan works out the way I hope it does, the photos should be sharper with the new rig. (Sorry, that’s a topic for another post.)

But neither of these pictures really captures the fun of the south end of the beach: the cliffs. Hapuna Beach itself is a gently curving sandy beach with a shallow bottom and decent waves in the winter. The cliffs, on the other hand, are hard basalt, with sheer drops and crashing waves. If you pick your location carefully, there are spots where the water is quite deep. Deep enough to make for some good cliff diving.

Making the Break

My son found a good spot, but hadn’t scouted the water beforehand. I pointed out some rocks, and the fact that neither one of us knew how deep the water was. By then we were both too tired to scout, climb back up, and jump back in, so neither one of us jumped. Truth be told I think it’s a good spot. Next time I’ll bring snorkel gear and check out the landing zone before climbing up. Already I can see a whole set of cool pictures I could make from the air of people jumping off those cliffs.

I didn’t get as much weekend as I wanted. But now I’ve got a better idea about what I want to fix when I rebuild my BBKK rig, and I’ve got the parts on order to build out the panorama rig. Even better, I’ve got plans for my next outing. Any KAP session where you walk away excited about your next session is a good one in my book.

Yeah. I’m back.

– Tom

Posted in Engineering, Kite Aerial Photography | 2 Comments »

Glimmers of KAP

Posted by Tom Benedict on 12/09/2012

I’ve been out of kite aerial photography for the better part of a year. In that time I’ve had a couple of wonderful KAP sessions – namely when the World Wide KAP Banner came through, and when I got to fly with Ramon Pallarés at the windmills near his home in The Netherlands. But as far as flying for the sake of flying goes, the last year has been a dry well. Likewise, my tendency to constantly tweak, re-design, and re-build my KAP gear has also gone by the wayside. I don’t know when I’ll move past this and get back into KAP, and I’m not sure if I will ever get back into KAP to the degree I have in the past. But recently I’ve had the itch to make new gear. I’m taking this as a good sign.

It started off with some recent realizations. At the risk of boring the reader, here’s some history:

A couple of years ago I was flying a Canon PowerShot A650 IS – a very capable, if hefty, compact camera. It served me well, but I wanted something better. To some degree this was driven by the gear lust almost every photographer suffers from – KAPers are just as subject to this as ground photographers. But my decision to upgrade had another reason behind it: Getty Images.

Even before Getty began their program with Flickr, I’d been looking into their contributor requirements. Getty’s library of images is broken down into different collections. Each has its own requirements for its images and the equipment used to create them. I read through all the camera requirements, and figured out that there was a sweet spot in their royalty per pixel table: the 18MP DSLR. It offered the most buck for the bang, so to speak, in terms of image royalties and collections a contributor could participate in. For some collections, if you weren’t using a camera at least that large, you couldn’t play the game at all.

Then the Getty Flickr Collection came along. This was a brand new collection, using photographs their curators found on the Flickr site. No camera requirements, aside from a 5MP minimum image size. I wasn’t in the first pool of KAPers who participated in the Getty-Flickr program, but I wasn’t far behind. In that first year I earned good royalties from the photos I’d made with my A650. But Getty uses the same royalty schedule for all of its collections. That meant the 18MP break in the royalty pricing was still there, even for Getty-Flickr.

But Getty wasn’t the only driver behind my decision to upgrade. For myself, I wanted a camera with a bigger chip, better noise characteristics, and the ability to shoot real 14-bit RAW images. At the time I also thought it would be nice to get a camera I was equally happy to use on the ground. One camera for all reasons and seasons, so to speak. I’ve used Canon gear for years, so my first inclination was to get a Canon DSLR. Around this same time, though, Sony released the NEX-3 and NEX-5. These were mirrorless crop-sensor cameras that took interchangeable lenses, just like a DSLR. My choice came down to two cameras: The Canon EOS T2i and the Sony NEX-5.

At almost half the weight of the T2i, the NEX-5 was the clear winner for KAP. It offered an honest to goodness crop sensor – far larger than the chip in the A650. It had very good low light performance with much less noise than the A650. And like the T2i it also offered 14-bit RAW images. But it lacked two key features I wanted at the time: At 14.2MP, it didn’t fall inside the 18MP sweet spot Getty was willing to pay extra for. And unless I bought a bag full of Sony NEX lenses to augment my existing bag full of Canon EOS lenses, I didn’t think I would get as much out of it on the ground as I would the T2i. The NEX-5 was an attractive camera, but in the end I went for the larger, heavier, Canon.

I love my T2i. No doubt about it. The images it produces are superb, and it’s a joy to use both in the air and on the ground. But therein lay half of the problem: it became just as much fun to do photography on the ground as it was to do it in the air. Already KAP started to take a back seat. Throw in the complexity of the video down-link hardware I was trying to build and the constant hacking on my ground-side radio unit, and KAP took more and more of a secondary role in my photography. Eventually I spent so much time messing with my gear that I stopped doing KAP altogether. In time I stopped working on the gear as well. My KAP gear stayed in its bag, idle.

During this time  my wife was diagnosed with a brain tumor. A year ago this month, she was operated on. The tumor came out intact, and as of her last MRI she’s still 100% tumor-free. But in case anyone has any doubts about about how long it takes people to recover from a craniotomy, it’s not out-patient surgery. Not even close. It was weeks before she could walk more than a few feet, months before she could drive. The operation left her blind in one eye, so we had to spend time re-thinking the things we did and the ways we lived. We replaced the mirrors in her car with wider ones. I got her hiking sticks so she could keep her balance on steep trails. Bit by bit we got back to the business of living life.

I went back to work after the first month, but I kept my phone on me, ready to bolt for home in case she needed me. It was even longer before I felt I could stop hovering at home every weekend. Even now, a year later, I haven’t returned to any of the evening activities I used to do, like going to photo club meetings or trying for the ever-elusive sunset KAP session. I basically turned into a care-giving hermit. Though to be fair I turned into a hermit who spends a lot more time with his family now.

Recently, though, I’ve started taking my gear out. Mostly it’s when I go to the beach with my family, but about a week ago I lined up another session at the Mala`ai Garden in Waimea. It felt strange even talking about doing KAP. To my surprise I felt guilty.

Guilty? For even thinking about doing KAP? How weird is that?

It gave me pause. And it made me think about where I’d wound up. First and foremost, I felt guilty for doing something strictly for me. My kids aren’t interested. My wife isn’t interested. Except for two line sport kites, I’m the only one in the family who flies kites, much less hangs a camera from one. But I also felt guilty for photographing something I couldn’t then market to Getty. I was doing KAP for the sake of doing KAP. And somehow that felt wrong.

Where I’d wound up was somewhere very very dysfunctional. I talked about this with my wife, and she assured me that she was well past the point where she needed me hovering every second of the day. I can’t disappear for entire weekends, but taking off for an afternoon of KAP isn’t a problem. And feeling guilty about photographing something just because I want to? That’s something I’m more than happy to take on and conquer. Photography should be fun.

All of this made me look at my gear with fresh eyes, too. What I saw didn’t make me 100% happy. It was time for a change.

Every piece of engineering has its shortcomings. Every piece of equipment can be improved. This was the first time in a long time I was interested in improving my KAP gear. Even more than the go-ahead from my wife, this let me know that maybe I was coming out of the hermit cave. Maybe it’s real.

In hindsight I don’t think my one size fits all approach was necessarily the right way to go. It makes for feature-heavy gear that compromises weight and complexity in favor of flexibility. But there are other ways to achieve flexibility than to throw in every bell and whistle that comes to mind. Rather than have one set of gear that must serve all needs, I think a small set of semi-specialized gear might be the better way to go. It’s still too early to say what I’ll eventually do, but right now I’m leaning toward three changes:

The first is to gut my DSLR rig and rebuild it with simplicity in mind. I’m keeping the pan and tilt axes, and I’m re-vamping the power supply system for the video downlink. I pulled one of the joysticks from my ground-side radio, and plan to install the video receiver in its place. The radio will look exactly like a two-stick RC radio, except the right stick will be replaced with a monitor, and there will be a second antenna jack for the video antenna. Small, simple, and reliable, with no loose wires to get caught on.

The second is to build an autoKAP panorama rig for the T2i. This would be a pan-only rig similar to Brooks Leffler’s BEAK, but built for a DSLR. I have a Pololu Robotics Micro Maestro built out as a rig controller, and I have a spare Picavet with PeKaBe blocks sitting around from an earlier project. I still need a couple of parts from Brooks to finish this off, but the idea is to have a lightweight rig I can bolt the T2i into and send aloft with no further interaction required from the ground. I’ve done enough panorama photography with my A650 HoVer rig to have a pretty good idea what I want from this thing. I just need to do it.

The third rig is the most challenging, but also the most enjoyable of the bunch: I’m going to make an autoKAP rig for my phone.

In a word, the T2i rig is heavy. It weighs in just over 1kg, or more than two pounds. The wind has to be just right, or it won’t even fly. I want something lightweight. Even more, though, I want something that means bringing less gear with me instead of more.

No matter where I go, I carry my phone with me. It’s light, it’s got a capable camera on board, and turning it into an intervalometer camera is a simple matter of downloading the appropriate app. Bolt it into a super lightweight pan/tilt rig, and it’s an 8MP KAP workhorse. Add some smarts, and it may be even better than that.

The Pololu Micro Maestro has six ports on it, any of which can act as a servo output port, a general purpose digital I/O port, or an analog input port. Headphone audio jacks typically output 1V peak-to-peak. A generous audio signal should be detectable through the Micro Maestro’s ADC input. Every camera app I’ve tried on my phone makes a loud shutter “click!” sound when a photo is made. My plan is to make a cable to connect my phone’s headphone jack to the Micro Maestro. I should be able to use the sound of the shutter click to trigger the next rig move. The Micro Maestro is flexible enough to let it drive a pan-only rig, a pan-tilt rig, or even a HoVer rig. It’s just a matter of sensing the shutter sound from the phone and getting the rig positioned before the next image is made.

The end result will be a very small rig – a little larger than my phone – that I can toss in my kite bag. When the wind is light, I can strap my phone in and send it aloft while I relax on the ground. And for travel KAP, it would be a tough combination to beat. No radios, no DSLR bag. Just a chunk of hardware smaller than a paperback book.

I’m still swamped at work. I’m still trying to work through the feelings of guilt when I think of going out by myself to fly a kite. And at the moment I have very little money to throw at new KAP gear. So I don’t know when I’ll get most of this done. But the changes to my existing DSLR rig are already in the works, and I’ve started drawing up the bits for the phone rig. Time will tell if I’m really getting back into KAP. But for now it sure feels good.

– Tom


Posted in Kite Aerial Photography, Photography | 2 Comments »

Ready for a Weekend!

Posted by Tom Benedict on 08/09/2012

For a short work week, it’s been a long week at work. But it’s been a good one! Most of the external parts for the cameras are done. If fortune continues to shine on us, it’s looking like we might send out all the external parts for anodizing some time in the next few weeks. We still haven’t settled on a color for these things, but it’s looking like the original orange I used for making the CAD renderings may win out in the end. The only requirement is that the cameras be brightly colored. One of them will be located almost knee-height when it’s on the telescope, in an area where forklifts are in use. If I could make it fluorescent orange with tiger stripes, I would! But I’d settle for fire engine red or hot pink, too.

This is, I hope, the last rendering I post of this camera design. The next picture I post should be the real deal:

Sitelle Camera September, 2012

I’ve still got some internal parts in the queue, though, so the work is far from done. I’m working on the vacuum feed-throughs next week, and hope to make a start on the cold finger/foot/strap assemblies as well. That is a story all its own:

Any time you build a cryogenically cooled camera, you need some way to isolate the cold bits so they don’t let heat from the room-temperature camera body leak in. Along with this, you need some way to attach your cold source to the cold bits so any heat they generate or collect through ambient radiation can be taken away. The first is accomplished by using some thermally insulating material to support the cold bits. In these cameras I’m using G-10 fiberglass tabs to separate the warm parts and the cold parts. The second is accomplished by using some thermally conductive material to draw the heat off. In these cameras that’s done through an unfortunately complicated assembly of copper and aluminum.

Usually these assemblies are comprised of three parts: the cold head (the cold source), the cold strap (the bit that connects the cold head to the rest of the assembly), and the cold foot (the bit that actually connects the cold to the back of the detector chip). In this case our cold head is a Polycold PCC cold head. Our cold strap is a piece of copper 0.20mm thick x 12mm wide x 90mm long. Our cold foot is a block of copper about 35mm square and about 15mm thick.

The trick is that you have to be able to remove the cold head to access the rest of the camera. So somewhere in there you need some sort of a removable joint. In one camera I’ve worked on, this was done using a spring and two pressure plates. One of the plates is connected to the cold head. The other is connected to the cold strap. Push the two together, and the force of the spring keeps things in good solid contact. Voila! The cold head / cold strap / cold foot circuit is complete!

The original design for this camera also used a spring, but concerns about mechanical vibration coupling between the cold head (which does vibrate) to the detector chip through the spring ruled that out. I had to come up with something else. Over the course of the last three months I tried all sorts of ideas, none of which panned out. The best approach I had used a single bolt to bolt the copper cold strap directly to the cold head. Only problem? There was no way to reach the bolt with a wrench! No matter what I did, there was no way to get a tool in there. Time and again, I found myself back at the drawing board, starting over from scratch.

Then one of my co-workers came up with a brilliant idea: Why not use magnets? Stick a magnet on the cold head and a magnet on the cold strap, and let the magnets do the work of holding the two in contact. The idea is similar to how the clasp on a lot of handbags work these days. We tried it on the bench in the lab, and it worked great. So yesterday I mocked up a set of hardware to try this in a real cryostat. Today I installed everything and started pumping it down.

Installing the cryo head was dead nuts simple. As soon as the two magnets got close enough, “Clack!” they came into contact. The magnets are self-centering, so some of the off-center contact problems we had on the spring designs are also solved. Even before getting things cold, I could already tell the idea was working. In order to drive some of the water off the cryostat walls, I stuck a heater on it while it was pumping down. Both the cold head and the cold foot registered the heat being applied. Under vacuum, the only path that would let both parts see the same heat is the cold strap. Only way for that to work? For that joint to be solid. Yaaaay!

I plan to start cooling Saturday morning and see how it cools through the weekend. By Monday we should have good numbers for how it’s likely to behave. If it all works out well, I’ll start cutting parts for the real cameras on Tuesday.

But enough work! It’s time for a weekend!

And speaking of weekends, I finally ordered the tools and parts to replace the main halyard on our boat. My plan now is to finish drawing up the new main halyard winch in CAD, put the old winch back into the mast, and get the boat ready for when the halyard bits all show up. We should be sailing by next weekend. Meanwhile I can start machining the new winch, which is being designed not to let the cable jam ever again. When it’s ready, swapping them out will be about an hour’s work: the perfect project for a weekday evening with the promise of a weekend sail at the end of it.

Time to play.

– Tom

Posted in Astronomy, Engineering, Sailing | Leave a Comment »