What is LENSE? To begin, it stands for Low Expense Near Space Experiment. By near space, I mean 100,000 feet, and by low expense I mean less than $1000. LENSE is my stint in High Altitude Ballooning, the hobby of using weather balloons to tow boxes of electronics to the upper atmosphere for the purpose of photography or scientific study. I got interested in this field when I got my first camera in December 2010. After taking tons of photos and becoming somewhat proficient with my camera (and my dad's Nikon D3100) I wanted to experiment with different fields of photography. I started with panoramas of mountains, then downsized to capturing Lincoln's nose on the back of the US penny. As with my cycle of big and small projects, I decided to once again try to take big pictures. I looked into taking pictures from the tops of buildings, then remote control airplanes, then kites, and eventually I stumbled on using balloons to capture the holy grail of aerial imaging: the Earth itself! I was hooked and researched every available source to get a sense of scale for this project that would soon consume my free time for the following 8 months.
Many well documented sites show use of Canon Powershot point-and-shoot cameras. These cameras are inexpensive to find on Ebay and Craigslist, and are simple, light, take pictures plenty good enough for us, and allow you to use CHDK. Canon Hack Development Kit is a firmware upgrade for Canon Powershot cameras that is booted from your SD card. It allows you to access hidden camera functions and unlock features unavailable on normal camera menus. The kicker is that you can write uBasic scripts to operate the camera once it is under the control of the CHDK software. Scripts like an ultra-intervalometer that takes a picture on a set interval, which is exactly what we need to capture space. This system is free, easy to use, and does not damage the camera in any way. No modifying is required and the camera is completely independent from all other systems. Delete the CHDK software from the SD card and the camera functions normally. Quick tests with a Canon Powershot S2IS were very successful and impressive. I moved on to my sister’s smaller, higher quality Powershot SX110IS, which will most likely be the camera in flight. My dad said, “Maybe it’s time for your sister to get a new camera.”
That’s the still photos, but what about video? Majority of the high altitude balloon projects don’t consider video because of expense, but a project on SparkFun Electronics details the use of GoPro HD Hero cameras. They are helmet-mounted cameras aimed at the snow-sports and motorsports industries. They shoot in 1080p, have extremely wide angle lenses, and are incredibly compact. Downside? Cost. One camera is around $300 which is completely out of my price range. Searching on Ebay revealed I could get a “spy camera” that shoots in reasonable quality for $25, but I was skeptical. After putting this off for awhile, I decided to look into the Flip and other pocket video cameras that have allowed the public access to high quality video in a compact package. I chose the JVC Picsio for the cheap package that still shoots 1080p video. I could find it on Amazon for $55, a great deal, however the reviews say that quality is terrible. Many suggested the Kodak Zi6 as an alternative. The Kodak had much better reviews and wasn’t much more at $65 used. My mind was made, or so I thought. The Zi6 shoots 720p video, but Google Instant showed me the Zi8. The only main difference is HD quality. The Zi8 shoots 1080p, has equally good, if not better reviews, and can be found for $75 used on Amazon. Only downside is that 1 hour of HD video takes up 8 gigs of card space because of the 15 megabit data rate. No problem, I’ll order a 32 gig SD card. (side note: when nerds electronics engineers see modern technology able to make it possible for them to hold half the storage of a typical laptop and they get to hold it, they become immensely happy and the normal person can't figure out why. Maybe that's why we're treated differently. Then again, it might just be me. But that's 32 gigabytes, the size of a quarter! Amazing!)
Total cost? $105, 345 grams.
The flight string is the payload box, the parachute, and the balloon all in line and attached with mil-spec nylon cord. FAA regulations state that the payload must not exceed 6 pounds unless they are notified. 6 pounds is about 2750 grams, which should be easy to stay under. For balloons, party balloons aren’t an option. Really the only supplier in the US for weather balloons is Kaymont. They have several sizes ranging from 200 grams to 3000 grams in weight. The K1200 can lift 4 pounds easily, and is the most widely used by balloonists. It requires a standard cylinder of helium, 110 cubic feet, and fills to 8 feet on the ground but expands to 31 feet before bursting at an altitude of 33.2 kilometers, which translates to just over 100k feet. You can’t order online, but other balloonists report pricing around $70 plus shipping when you call in to place your order. Parachutes used can be homemade from garbage bags or custom, but popular a popular and proven supplier is RocketChutes. I assume my assembled payload will be 2-3 pounds, and the recommendation on their site is for a 48” parachute, at a cost of $25.
Traditional choices for payload containers are Styrofoam coolers to keep in the heat from equipment while remaining lightweight. It is important to consider that the container must not be airtight, or the drop in air pressure outside will cause the container to explode. Coolers seemed too big for the few things that will be inside, so I decided to go another common route and make your own out of polystyrene insulation board. Available in 2 inch thick sheets 4 feet by 8 feet at Home Depot, availability should be no issue. And they only cost $10. Assembled with hot glue and ports cut for cameras, I should have a lightweight, stable box for my precious equipment. Rigging is nothing special for the flight string. From top to bottom, the balloon is attached to the crest of the parachute, and the leader lines of the parachute pull on the payload. Two-mil nylon cord should do it, and be lightweight at 1.2 grams per foot. $0.15 a foot at REI. I’m thinking about coating every side of the container with a space blanket (aluminized mylar, highly reflective to heat radiation), except for the top to allow GPS reception. This should keep in the heat and protect the electronics. As a bonus, it will also reflect radar from aircraft.
Total cost, $265 including quotes for helium and regulator rental.
Radio communication was my primary concern. There are plenty of web pages about HAM radio groups that use a device called a TinyTrak to interpret GPS data and send it through HAM radios for the chase team to pick up on, but this requires a license and transceivers are expensive. 900mhz is the only other radio option I came across that would allow full-time communication with the GPS as it rose into the atmosphere. Models such as the Digi Xtend have a range of an impressive 40 miles line of sight and are very compact as well as easy to interface to. Cost was the limiting factor; just 1 transceiver would run me $200, the pair over 50% of my budget. Plus, they suck large amounts of precious battery power while transmitting. This led to my final option (for a while) of relying on a cellphone-based setup like the MIT group. Many phones have integrated GPS, and free apps allow you to read GPS data and send it via text message to another phone at intervals of minutes. This would be a great option and significantly lower the cost, but I’m pretty sure I would lose contact, especially in the vast farmlands of Kansas. Then my only option would be to guess where the payload is and wait helplessly while it came back down into coverage range, hoping it would continue sending text messages. That option is not reliable, and I don’t want to lose this expensive project! I dropped this subject for a few days, but I had an Aha moment.
Walkie-talkies to kids, two-way-radios to professionals. To my surprise, Midland makes a 36 mile version for only $60 a pair. To send information across, I could build a modem to send data bit by bit via audio modulation, and a module in the chase car to decode the data. Voilà! And simple geometry reveals that I could be 30 miles away from the payload and still have reception during any part of its journey. Since communication is two way, I could conceivably even request data from sensors onboard or check the battery level. Total cost, $65, 400 grams projected after lightening.
Besides just cameras, other instruments will take measurements aboard the payload. Most important, a serial GPS for providing key location data. Originally, I planned to use a Microsoft Pharos 500 for the low cost. Further investigation revealed that consumer GPS units have one or both of two restrictions. The US government requires all GPS devices to stop functioning at a speed of 515 meters per second and/or above 18 kilometers (60,000 feet). GPS devices are not required to have both, although many do. These regulations are designed to prevent the use of commercial units in intercontinental ballistic missiles. LENSE probably isn’t going to break any speed records, but it will definitely surpass the 60k altitude limit. The Pharos 500 is based on the SiRF Star III chipset, which is limited in both ways. This makes it ineligible for use because I would have no position information for half the journey. Instead of commercial products you can plug into your computer or use to go hiking, I looked for modules designed for embedded electronics.
SparkFun has a great selection in many price ranges, and the cheapest I could find was the 12 Channel Lassen IQ GPS Reciever by Trimble. It’s teeny tiny, highly sensitive, low power, uses a dual-channel serial interface, and compliant with 3.3-volt electronics. The downside is that requires an external active antenna with a specialized Hirose HFL connector, which sells for $20, making the total package about $72 with shipping. Not bad! In the course of my search, I found that any GPS units based on the VENUS, MT3329, MT3318, or SiRF Star III chipsets also have an altitude limitation.
I also plan to include an array of sensors to take periodic measurements. The humidity/temp sensor is 3.3 volt compliant, has a digital output, and is extremely low power. It has a range of -40 to 80 degrees Celsius, and accuracy to +/- 0.5 degrees. The humidity feature also measures from 0-100% RH. They run $10 a piece from the SparkFun website. I will use one to measure internal temperature, one to measure external.
The BMP085 is an excellent value. It operates at 3.3 volts, and can measure from 300 to 1100 hPa with accuracy down to 0.03 hPa. It also communicates on the I2C interface. Cost: $20.
The LSM303 DLH carrier includes a digital 3 axis magnetometer and a digital 3 axis accelerometer on one board. Voltage regulators on the board allow it to operate at any voltage from 2.6 to 5.5 volts. The accelerometer can be set for sensitivity ranges at +/- 2, 4, or 8 Gs. The magnetometer also has a range of selectable sensitivities from +/- 1.3 to 8.1 gauss. Any microcontroller can access the data through a level-shifted I2C interface. A similar module is available on SparkFun, but Pololu offers one for a cheaper price. Cost: $30, headers included.
Since I will be collecting large amounts of data, an off-chip data logger is a must. SparkFun sells the OpenLog for a measly $25. It runs at 3.3 volts, and records serial data at 9600 bps to a FAT16 or FAT32 micro-SD card in a slot on the backside. You can also send commands like ‘Ctrl+z’, ‘new’, ‘.md’, and ‘?’. Ctrl+z 3 times will put the module into command mode, in which you can make a new file, new directory, or use ‘?’ to bring up a list of commands. Even recording at 57600 bps, no characters are dropped and the OpenLog only uses 6ma!
The microcontroller I used to synchronize the GPS to Radio connection and reading sensor data is the Parallax Propeller. I happen to have it from another project, and it is extremely powerful. Any microcontroller with enough inputs and 1 output for the OpenLog would work, and an Arduino would probably be ideal. With the Propeller however, you have access to the Propeller Object Exchange. The POE is a section of the Parallax website where people can upload programs, or “objects”, that can be downloaded for others to use or modify. I recently came across an object that allows the propeller to take in GPS data via RS232 and convert it to .kmz files that GoogleEarth uses. Thus, I can see my balloon’s entire journey and share it with others. The board I have is the Propeller Prototyping Board, but after testing I will have a compact PCB made professionally.
For batteries, there’s pretty much two choices. Either rechargeable lithium packs or disposable AA lithium cells. Energizer Ultimate Lithium batteries each boast a capacity of 3000 mah at 1.5 volts and can handle temperatures down to -40 Celsius. The video camera needs 3.7 volts from its included rechargeable lithium, and the electronics need 3.3 volts. I plan to have 6 batteries in the payload supplying 6000 mah at 4.5 volts, regulated to each circuit. One “gotcha” I’ll probably encounter later is that the radio uses significant power when transmitting, which may temporarily cause the voltage to dip, resetting the Propeller. Decoupling capacitors maybe?
I ordered the GPS unit from SparkFun, $73 in total. It arrived today, in the familiar red box. It took a little searching to find an accurate pinout, but soon I had connected the GPS and antenna to a logic level converter, since the GPS operates at 3.3v and USB is 5v. I was then able to read the serial data transmitted in NMEA sequences using RealTerm. It turns out that NMEA sequences are output on pin 5, ground is pin 2, and 3.3v power is pin 7. The NMEA data also alternates between VTA and GGA sequences.
The unit was working great until I tried it in a box with batteries and a serial connector so I could test the unit outside. For some reason, the Lassen IQ refused to output data to the logic level converter. I took it out of the box and connected it just the way it was the first time, only to find that it was drawing an unusual amount of current. Assuming it was a faulty unit, I emailed SparkFun to negotiate an exchange, and they were happy to have us send I back for testing. Things are at a standstill until I can get a replacement, but I’m eyeing the Midland 36 mile radio for communications testing.
On a side note, I want to have the position-transmit system composed of the GPS coupled to the two-way radio independent of the Propeller microcontroller. However, I also want to use timestamps and position/altitude data to log in the OpenLog to cite sensor data. Quick research reveals I’ll need a mux/demux IC to spread one output across several devices. SparkFun’s got one for $0.95.
Lots of time has passed, lots has happened. I have right now a Spot Satellite messenger as a backup tracker, have all the sensors and cameras, as well as the polystyrene insulation and the parachute. All I need to buy is the 32 GB Class 10 SD card, a final SparkFun order, and to have the PCB I designed mailed back from the fabrication house. The balloon is on its way, the batteries will have to wait until just before we leave for the trip. Changes I made include switching from a Propeller to the PSoC C8Y29466-24PXI and ordering a 3.3v LCD for testing. I also found a 9-dof sensor board from InvenSense to replace the accelerometer and magnetometer. In addition to 3 axis models of both sensors I wanted, it includes a temperature sensor and a 3 axis gyroscope, for just $70! As a bonus, it operates over I2C and processes the data internally, removing much of number-crunching. It took a while to get here though. Also, there is enough RAM on the PSoC that a data logger is unnecessary.
The parachute is high quality, as expected, but doesn’t have a connection point at its apex, so I’ll have to sew my own. The polystyrene sheet is 1.5” thick, I may do a double wall for extra insulation and paint it black to add some solar-heating, with a space blanket in between layers. The PCB is designed to route power to the video camera as well, since tests reveal it can only record 1 hour on its built-in battery. My budget is just were I expected; over-budget at a projected $820. But well worth it for the increased reliability! I have to increase the battery count from 6 to 12, giving me 6 volts at 9000 mah, hopefully enough to run everything for about 8 hours.
Back to the PCB, everything is on the 3.3v power rail, except the video camera’s power, which is good, and eliminates any voltage conflicts. Poking around the SparkFun forums has led me to decide on DTMF modulation for sending data across the Midland Radios, which I have now. The #highaltitude IRC webchat has been very useful for answering my questions in real-time, as have others posting their adventures. Paul Verhage wrote a 15-chapter guide on a Parallax post, which has been VERY insightful. He also has articles he wrote for Nuts and Volts magazine for free download with his guide.
The biggest factor, heating, likely won’t be a problem. Nate Seidle, Sparkfun founder and CEO, attempted this and found his electronics had enough thermal capacity to avoid using a heater. Sadly, his payload was lost. The camera generates a large amount of heat while recording, sucking 1 amp while recording. 1000 milliamps going through a PCB trace is likely to keep things very toasty for the first 3 hours or so. If necessary, I found some crazy sodium-acetate super-saturated chemical hand warmers that don’t require oxygen, are reusable, and give off 130 degrees of heat for 30 minutes. Now I may have to worry about cooking my electronics!
I’m a bit overweight with the introduction of several unpredicted components, but lightening is easy, and experienced balloonists report that the KCI 1200 gram balloon can easily lift 12 pounds to 85,000 feet, but this requires an over-fill to 300 cubic feet of helium, about 3 tanks, which costs lots of money. Speaking of helium, I have begun assembling the fill valve and tubing with 1” PVC and air hose.
As of today, the box is cut, the PCB has been ordered, and all the parts are here. I leave for Boy Scouts summer camp in 3 days, for a week until the Sunday after next. This will only give me 10 days before we leave for Washington D.C., 17 including working on it during the vacation. The enclosure is quite small on the inside, only 8x7x5 internal dimensions. It could have been smaller, but the video camera is 4.5” tall. Disappointingly, the field of view on the camera is only about 35-40 degrees! I want to order an external lens that is wide-angle without distorting the image, but my dad complains it is unnecessary. I might even shrink the payload further, There’s more space than I thought there would be. The plan is to use only two packs of 4 batteries, granting 6000 milliamp-hours at 6 volts. This drop in battery count will save money, weight, and still have plenty of capacity. The internal battery of the camcorder will work for the first 45 minutes, then the PCB will switch to external power, providing 1 amp at 5V, for the remaining two hours of the mission. This will allot 4000 milliamp-hours for the electronics. Several balloon calculators state that the mission will only last 3.5 hours maximum. It also said that if I end up launching in Seattle then it won’t land in the ocean, but on top of a mountain.
With respect to the payload box, I plan to mount the PCB on the ceiling along with the batteries to give the GPS antennas the best possible reception. The Spot, I have found through much trialing, is very particular about having the fewest number of obstructions to get a satellite lock. Also, the pressure has been shown by others to suck all the air out of the device, pushing down all the buttons. This is bad, because holding down the “power” button turns the device off! Also, it wouldn’t be good to accidentally call in an SOS! Mounting the electronics on the roof empties a good amount of space in the payload, so I might make it even smaller. Today I hope to sew a loop to the parachute, cut holes for the cameras, and refine the payload box. Final construction will be hot glued together, coated and reinforced with reflective duct tape, then painted black on all sides. The top must remain without duct tape reinforcing to allow radio penetration. Speaking of penetration, if I end up using the two way radios for telemetry (which I hope I do) then I plan to cut holes for the antenna out the side. The filling adapter is finished, it just needs a hose and tank fitting which I will get at a welding supply store on-site in Kansas. Right now I’m just waiting for the balloon and PCB to arrive, in the mean time I can finish up the software with my dad and then hopefully begin on the DTMF communication through the two way radios.
The day before we leave has arrived! Tomorrow we leave for the airport at 5 AM. The balloon swooped in from UPS at the last moment, dropped off at 7 PM. The PCB finally arrived and I assembled, tested, and have begun working on the software. Learning Eagle was an adventure, but it paid off and the board looks great. As expected, I had to make many “green wire” fixes. I forgot to put in pull-up resistors, and there were some power wiring issues for the video camera’s auxiliary power, as well as a trace that didn’t get fabricated! The huge 3.3v LCD doesn’t work for whatever reason, but it was simple to adapt the 5v one. We got the GPS running, and hope to get every sensor online by launch in about a week. It’s a good thing I decided to design in a header for unused pins, since we are going to order an OpenLog after all, and an indicator LED. The electronics are mounted in the box, albeit very tightly, and it looks a bit hastily done, but it’s very robust. Cutting the standoffs and project box holes was a nightmare, I HATE working with plastic and small threaded parts.
I finished the payload container and made it about ½ size to conserve weight. The video camera was easy to mount, held in place with blocks of foam and a small hole for the very narrow field of view. I finally got to visit the Gasworks Park Kite Shop to inquire about rigging and sewing. The lady there set us straight about sewing the loop with a piece of flattened paracord, and donated 10 feet of Spectra 30 lb. line for the parachute-balloon rigging. Though it is strong, it is small enough to be used as dental floss! I also cut holes for the temperature/humidity sensor to be exposed to the outside environment. Lastly, after gluing the sides and bottom, I aluminum-tape wrapped the seams, and will do the same for the top just before launch. I did find a hand warmer to put inside the payload, the “MegaWarmer”. It’s basically a conventional hand warmer, but flat and really big, providing 12 hours of 135 degree heat. Combined with predicted 90s weather, I may just cook my electronics!
With respect to the SPOT issue, I don’t think it will be a problem since a group online uses them for primary tracking means without issues or modifications necessary. I don’t think I’ll have time to incorporate radio system, sadly. Since the SPOT only provides updates every 10 minutes, and it cuts out at 18,000 feet, I will only get 4 hits about location to go on, not even with altitude. My dad is looking into putting my horrible, if I do say so myself, cellphone inside and using a T-mobile service to triangulate it as a backup plan. However, worst-case, a farmer will be tilling his fields and see the bright red parachute on his land, then call the enclosed number and the payload is recovered.
I visited Frys only yesterday, and picked up a 32gb Patriot LX series Class 10 SDHC card for $39.99 after rebate, and 12 Energizer Ultimate Lithium batteries. They were cheaper than I thought, only $8 for a 4-pack. Sadly, the Canon Powershot’s screen stopped working after I tactically shut off the backlight by replacing the LEDs with a common diode to replicate the voltage drop, which worked successfully for quite a while. Then one day, the LCD quit, but the camera still worked. I had memorized the CHDK button activation, so there’s no problem there. I can’t change settings, but I can just place the camera in full-auto mode. I also discovered that an internal boost circuit took the 3v input of two AA batteries and bumped it up to 12v to drive the LCD, which explained why there is such poor battery life with the backlight activated. The lithium batteries combined with no backlight should provide plenty of time for the 4-gig card to fill up with 1300 pictures. Cameras done!
I started concocting a simple fill valve, but can’t finish it until I get to Kansas and see the tanks I will be working with. It is made of a 1” PCV tee fitting that has a ¾” threaded fitting on the cross-opening, perfect for a threaded barb hose fitting. I then plugged the bottom of the tee and will screw a hook in for the counterweight. The open end will go to a short piece of 1” PVC with grooves to fill the balloon. I also found some latex gloves to use when handling the balloon during filling.
I ended up shipping all extra parts, tools, the oscilloscope, payload box, and balloon/flightstring to Kansas, only keeping the video camera and electronics and programmer with me to debug in DC. I hope it all arrives intact, as the balloon bag split open, causing need for a hasty double-bagging. It added another 38 dollars to my tab to ship the 38 lb. box to Kansas in only 5 days.
All I can say is that I hope I didn’t forget anything!
No, the project actually ended out spectacularly! On the final flight lots of things went wrong. First, the external humidity/temperature sensor quit working, so that had to be thrown out. Second, the video camera wasn't responding at all to the external charging system I had in place, so it could only record for an hour. I had to hastily make a dedicated battery pack out of C-cells that brought the final weight to 3.5 lbs, much heavier than I initially budgeted. The radios weren't an option since we didn't have time to implement them, but the SPOT is a fantastic device, thankfully. We received live updates every ten minutes on my dad's iPad via a Verizon 3G data connection, even in the rural farmland of Kansas.
The results: a launch height of 88,671 feet! The pictures were incredible as you see in the gallery, and the payload drifted 80 miles from launch to landing thanks to the jet-stream. Oddly, it only lasted 2 hours and 12 minutes, almost an hour shorter than we planned. When we arrived at the landing site we saw something unusual, the entire balloon had dragged down the payload with an extra 3lbs of load on the parachute instead of shattering like it does in theory! Nonetheless, everything landed in perfect shape, still functioning and recording. At home I took the logged GPS data and plotted it in Google Earth thanks to an online program i found. What great results! I am definitely planning to do this again, with many revisions, in the summer of 2012.
To top it off, while in Washington D.C. I auditioned to speak at TedxRedmond 2011, a youth event that lets kids talk TED-style about a theme. Upon getting notice I would get to talk for 3-4 minutes about my project, I put together a presentation and speech that you can watch here. On to the next project!