AZHAL-1 Payload Design

Yes, we are going up again.  Our 4th balloon mission, AZHAL-1, will be for the Global Space Balloon Challenge (GSBC) (http://www.balloonchallenge.org/). There are currently over 100 launches planned around the globe for this long-weekend of ballooning, and we are one of the two teams competing from Arizona. Well, competing is maybe not the correct term, as we don’t really expect to win any of the challenges (highest altitude, best picture…), but it is a great excuse to go fly a balloon again, and to build a new payload with more goodies inside.

We have a new team for this mission: Arizona High-Altitude ‘Looners (AZHAL), six kids ranging from age 10 to 16, and four dads, three (soon to be four) of whom are hams. So yeah, we plan a lot of amateur radio stuff on this flight.  On our last mission (LE-3) we forged a partnership with Byonics (http://byonics.com/), who loaned us the tracking and RDF transmitters. That partnership continues on AZHAL-1, with Byonics providing two APRS tracking transmitters, two beacon transmitters for Radio-Direction-Finding (RDF), and a new crossband repeater, for amateur radio operators to communicate via the balloon while it is in flight. Byon, N6BG, will not only provide RDF capability on the chase team, but he has some new APRS tracking tricks up his sleeve. Should be a fun flight!

Fundamental to tracking our balloon, and to success of recovery, is APRS, the Automatic Position Reporting System (aka Automatic Packet Reporting System). The APRS system uses transmitters on 144.390 MHz to send GPS (Global Positioning System) information to amateur radio ground stations that are located on mountain tops, downtown buildings, and even on houses of hardcore hams. The position (latitude/longitude/altitude) of a transmitter can be followed in real-time on internet sites such as http://aprs.fi/ (and others). Any amateur radio operator can run an APRS transmitter on a car, jeep, bicycle, airplane, or even a balloon.

On our last flight we built a little styrofoam box to hold a backup APRS transmitter, and we plan to fly it again. This backup payload box will be suspended below the parachute, and the main payload will hang a few feet below the backup box. This little fellow will be transmitting with callsign AF7EZ-12, and the main aprs tx will be on AF7EZ-11 — you plug either of those callsigns into http://aprs.fi/ when tracking our balloon (be sure to set the tail and track length to something like one-day, and not 1-hour).

The backup payload is an independent system with a battery pack, an efficient buck regulator to provide a clean 5VDC with minimal waste of battery power, a GPS receiver, a Byonics MF-15 loaded with APRS firmware, and a spring-steel wire dipole antenna.

This is only a 15 mW transmitter, but we got reliable APRS data from 92,000 feet on our last mission. After losing our second mission’s payload, we consider a backup tracker invaluable.

We made one change from the previous-flight configuration: a little pic controller has been added for the sole purpose of providing a timing signal on one pin. The 28-pin package is way-overkill when an 8-pin would do, but this one was sitting around on my shelf anyway. The timing signal, in conjunction with the MF-15′s primary/secondary configuration, allows us to send APRS data on two frequencies: 144.390 MHz for getting into the public APRS system so everybody can watch us on the internet, and a separate and quiet 2m channel so we can get more-frequent updates on our radios and laptops in the chase cars.

The battery pack holds six AAA Energizer Ultimate Lithium batteries, which have great capacity, are lightweight, and specified to operate down to -40 degrees F (-40 C). These batteries are awesome, though a bit pricey. The AAA size has a capacity at room temperature of about 1200 mAh, when depleted down to 0.8V per cell. We will only use it down to about 1.2V per cell (6×1.2=7.2 which is about where the buck will lose regulation), but since the lithium voltage knee is quite sharp at end of life, the remaining capacity between 1.2V and 0.8V per cell is a pretty small fraction of total capacity. If we derate the capacity to about 900 mAh to adjust for some loss at colder temps and a bit higher terminal voltage, this backup tracker is still expected to run for about 20 hours.

While we use AAA batteries in our backup tracker, we use AA Ultimate batts in our main payload and Canon camera. The AA size has a room-temp capacity of about 3000 mAh! We derate these down to about 2200 mAh for our balloon use. Energizer also makes an “Advanced Lithium” family, which is not as fully specified as the Ultimate batts, but they are rated down to -40 F/C as well, and look to have similar capacity. The main difference is lower peak current (and higher battery impedance), and a shelf life of (only) ten years (the ultimate is good for 20 years). The 9V battery package is only available in the Advanced family. I think the Advanced batts would work well for balloon applications (and they are cheaper) but we are flying Ultimate batts this time.

Note that not all GPS receivers are suitable for balloons, as some stop reporting altitude above 60,000 feet. GPS devices are designed so they cannot be used in ballistic missles (look up COCOM), and should disable tracking when they are moving faster than 1,000 knots (1,200 mph) AND at an altitude higher than 60,000 feet. Some GPS units disable on “either” limit and hence are useless for balloons, so check your device specs.

Let’s take a look at some of the radio stuff in our main payload. When we started back on our first mission, we designed a “jack” antenna using 1/4-inch wood dowels. One pair of dowels was wrapped with an adhesive-backed copper foil tape to create a 2m dipole antenna on one axis.

We later designed it for two antennas, and now have evolved it to use all three axes for three separate 2m dipoles. In past missions, we strapped our main payload box inside the jack frame. This time we will let the jack dangle below the main payload, suspended by the three coax cables, which my wife braided for me. The idea of the jack was originally to suspend the dipole on an angle (so vertical/horizontal polarization was kinda split for better ground reception at low and high altitude), and so the antenna did not drop into the dirt on landing, so we could receive the signal and RDF on it. But the best thing about the jack is that is just looks cool — it is our trademark ballooning identity now. The styrofoam balls protect our eyes from the obvious pokiness of this little guy, and make it look sorta retro-spaceage.

The antenna hub is just a block of wood, with four dowels permanently glued in place, and two dowels (A and B) designed for a press-fit, and removable for easier transport. The permanent antennas are soldered directly to the coax cables, which run through holes in the block before exiting through an eye-hook. The removable antenna dowels connect via lugs and screws. You will also see how we use hot-melt glue to hold things in place. We use it everywhere to provide strain-relief, retain wires, glue circuit boards to foamcore…

We don’t crimp SMA plugs to cables, we just buy pre-made cables twice as long as we need and cut them in half. Take a careful look at the connectors though. On our first flight we could not RDF on the landed balloon with our tape-measure yagi — we later found that the center pin of the SMA plug was very short and not making contact with the radio antenna jack at all. Since most everything is made in cheap factories these days, we need to be extra diligent to monitor quality of everything, even a simple connector.

Why three 2m antennas (in addition to the 2m wire dipole on the backup tracker)? Well, the jack antenna supports three more transmitters: the main APRS tracking transmitter, a beacon transmitter for RDF use, and the output channel of our crossband voice repeater.

The RF module of our payload is a little foamcore slab that slides into a compartment in the main styrofoam box. In the pic above at the top, you see the Byonics MT-400 APRS transmitter. This is a 400 mW unit, and we will drive it both on the public 144.390 MHz channel, and on a different freq for chase-car logging, as we are doing with the backup tracker. In the center is a Byonics MF-15 beacon transmitter sending tones and a morse-code ID for RDF use. This is the same 15 mW unit as used in our backup tracker payload, but loaded with firmware for beacon (instead of aprs) use.

The 70cm-to-2m crossband repeater unit is glued to the underside of the foamcore slab. The receive input is on 434.075 MHz, with a 250.3 Hz access PL tone, and the 300 mW transmit output is on 144.800 MHz.

On the front styrofoam cover is (besides the bright-orange bit of poster board) the 70cm dipole receive antenna for the repeater, and holes where the coax cables for the other antennas poke through for connection to the various SMA jacks.

A balloon flight would be a bummer without some pictures from above. On our last mission we flew an old Canon A560 camera loaded with CHDK firmware to snap pics every 5 seconds or so. It is a heavy camera, and only 7Mpix, but was very reliable, so we are flying it again. We now have a newer/lighter A1400 which recently got a port of chdk; we may try that on a future flight for higher-res pics, but we will stick with the old 560 for this flight.

The pictures we got last time were kinda blue, so we are adding a UV and haze filter (Tiffen UVHAZE-1) on the outside of the payload to hopefully give us some clearer shots. The camera is glued into a foamcore framework to protect the buttons and let us slide it into the payload box and have it stay put.

We have also found a new video/still camera, a Mobius ActionCam, which is an evolved version of the good old 808 keyfob cameras we used to strap to model rockets. The Mobius takes 1080p video and saves it to an SD card, and also feeds ntsc std-def video out the usb port. We have hacked it to connect to the pushbutton switches with some open-drain lines from the controller.

OK, well some video on SD card will be cool to look at later, but how about some live video on the ground as well? We toyed with the idea of using an Amateur TV (ATV) transmitter, as there are some ATV repeaters on mountains around here, but the only common freq was 433 MHz which would interfere with our voice repeater, and the available transmitters were ancient and large designs. It would have been interesting as the repeaters here can stream live video to the internet, but we were not ready to figure that all out for this flight. There is a possibility of some ATV using the 2.4GHz band, which might be worth trying out some time.

But we found a compromise. The folks who are into remote-control (R/C) model airplane and quadcopter stuff fly FPV (First-Person-View) video systems on their aircraft, so they can fly by watching a video screen (or even video goggles). We picked up a 5.8GHz video transmitter/receiver setup to try out. The video transmitter is 200 mW, and we are going to drive a circularly-polarized cloverleaf antenna with it. We have two receivers, for both launch and chase teams to use, and little portable DVD players that will serve as a video monitor.

The Mobius camera pokes out the rear of the payload box, and the video transmitter (VTX) antenna sticks out the side (the cloverleaf will have a null horizontally, and a toroid-like pattern around the coax axis).

I made a cloverleaf tx antenna for testing, based on a design by one of the FPV gurus, Alex Greve (aka IBCrazy) — see his stuff at Video Aerial Systems (http://videoaerialsystems.com/). The thinnest welding wire I had on hand was 0.045″, which is pretty thick for these dimensions, but it got the job done for our proof-of-concept test antenna. While this antenna let us prove at least a mile of ground transmission (when paired with a helical, see below), we subsequently bought one of the IBCrazy cloverleaf antennas so we are flying one that has been properly assembled and tuned.

If it had four segments it would be a skew-planar antenna, but the three-segment design is a cloverleaf. The segments are leaning in the direction for right-hand circular polarization (RHCP), so our receive antenna needs to be RHCP as well — a LHCP rx ant would have a 26 dB loss, like a vertical-to-horizontal linear polarized setup would have. However, a circularly-polarized tx antenna (RH or LH) can be received by a linear rx antenna (vert or horiz) with only about 3 dB loss, so that is another option. There are some linear-polarized dishes available, but we are going to make a RHCP helical receive antenna, also a design provided by IBCrazy.

The helical is interesting in that you can adjust the gain/beamwidth by the number of turns. I picked nine-ish turns, just because it seemed like a good starting point. In testing, it did not seem to be overly-directional (needing tricky aiming) so I think it will be a decent handheld rx antenna.

A bit of epoxy and a few minutes in the Bessey K-Body clamp gives us a solid mount to the ground reflector.

The little flag of copper tape is a “wavetrap” for impedance matching. Crazy stuff indeed Mister IBCrazy.

So we don’t really know what kind of range to expect with our 5.8GHz-band video transmitter (we will be running it at 5905 MHz), but I ran some numbers and figure maybe 10,000 feet or so, which could be some fun stuff at launch and landing. Both the camera and transmitter get hot, so we will be cycling them on for some video and off for a cool-down period.

The pic above is the Auxillary Power module. Underneath, there is a battery pack containing 8x AA Ultimate Lithium batteries providing power to the RDF Beacon, the Crossband Repeater, and the Video Transmitter. Even with moderate use of the repeater, we should get about 30 hours of operation, so we can still RDF if we need to recover the payload the next day. Even if the 8xAA battery pack dies, the RDF transmitter has an additional 9V Advanced Lithium bat that kicks in (shottky-diode-OR; good for perhaps 500 mAh), which will give us an extra 12 hours of RDF time. Yeah, we don’t plan to ever lose another payload.

We ran Arduino Mega boards as the controller for previous missions, but never really felt comfortable with the reliability of the system (we found shorted pins on one of them), and the pcb is a simple design that is not the best from an electromagnetic standpoint (radiation/susceptibility). So this time it is one of my own boards, a 4-layer design using a PIC controller. It is handy to have a surface mount placement machine to quickly make up a special like this.

It has a PIC18F46K22, a TMP101 temp sensor, Honeywell 0-15 psi pressure sensor, an SD card holder on the back, and so on. I love the little JST 2mm connectors. Sparkfun has them in 2-, 3-, 4-, and 5-pin versions. They are 2mm pin spacing, but can be spread to fit 0.1″ (2.54mm) holes. I got some other ones from Seeed Studio (they are not compatible with the Sparkfun connectors) which are also quite nice, and have retention clips. Seeed (yes, that is three Es) calls them “Grove” connectors. The 2- and 3-pin ones we are using are JSTs from Sparkfun, and the 4-pin ones we are using are partly Sparkfun JSTs, and partly the Grove connectors with the locking clips from Seeed.

Sitting on a bit of 1/32″ hobby plywood (light and strong), and mounted on light nylon standoffs, is the Ultimate GPS and a compass module from Adafruit.

We are also going to fly a Geiger Counter, to see whether we can detect elevated radiation levels at altitude. We got a 1950s-vintage Soviet Geiger-Muller tube, a popular SBM-20 (we have played with these before). The SBM-20 has pretty decent sensitivity to beta and gamma radiation. It does not detect alpha particles (helium nucleus), but those can be stopped by a sheet of paper so they are not usually of concern. Beta particles (electrons) can penetrate into your skin layer, but can be shielded by very heavy clothing, or a bit of wood, and maybe by our payload box to some extent. Gamma particles are highly-energetic and can penetrate several inches into human tissue; it would take thick layers of lead or concrete to shield them. At altitude, we are looking for Cosmic Rays, which are particles that originate outside the Earth’s atmosphere, and should be gamma-class. We will be watching pulses generated by detection events, and logging Counts-per-Minute (CPM) vs. altitude.

We mounted fuse clips to the foamcore backer with nylon hardware, to let the tube snap in place. We also found a nice little high-voltage inverter that takes 5V in and lets us adjust the output to the 400VDC needed by our tube. All exposed high-voltage conductors need to be sealed to prevent arcing when the pressure drops at altitude, as the insulating properties of air diminish with lowered air pressure (to a certain point, and then increase again). Yeah, it does seem counter-intuitive that a near-vacuum would be a worse insulator that air, since arcing requires ionization of the atoms and molecules in the gas, but there you have it. Anyway, I calculated that up to about 1000VDC needs about a half-inch spacing worst-case for exposed conductors, so all high-voltage conductors on our geiger counter are sealed with hot-melt glue, except the hot end of the tube where it snaps into the fuse clip — here we wrapped it with electrical tape so the tube is still easily removable.

I recall seeing on some other sites that each detection event resulted in a short burst of multiple pulses — I thought this made no sense since a particle entering the tube chamber, and causing an avalanche of current to occur, should result in one fat pulse of current. However, I saw the multiple-pulse thing too! Upon closer inspection it turns out the power supply was seriously drooping during the current pulse — adding a 1000pF 1000V film cap to the HV supply output cleaned it all up nicely. Each detection event results in a single pulse of about 400 us or so.

We made a little High-Voltage Test Probe to measure and set the 400VDC supply. Connecting a 10-meg scope or meter to the inverter will bring it to its knees and kill the hv output, but this little gizmo is a negligible 100-megohm load, and about a 100:1 divider for easy measurement (400V in is about 4 V out). A string of resistors a bit of heat-shrink, and, as always, some hot-melt glue. Easy-peasy.

We can see some blips of background radiation, but we want to test our geiger gizmo with some real radioactive sources. A bit of searching and we found a couple of hot test sources:

Old camping-lantern mantles from Coleman used to have Thorium in the fabric to let them glow brighter. These days, Coleman no longer uses Thorium, but offshore suppliers don’t really care what they sell, so Thorium mantles can easily be found today.

Does your wife like Fiesta dishes? How old are they? If you have red or orange Fiestaware dishes from pre-WWII, they are glazed with Uranium added for color. Very pretty. Yum. Let’s eat. During WWII, Uranium was not available, as the government was busy with all of it, but after WWII, Fiesta stuff was again made rather hot, with depleted-Uranium this time. After the 1970s or so they finally removed the radioactive glaze from their products. The reasonably-new Fiesta stuff in our cupboard seems harmless enough.

There are other sources around as well, like Radium-painted glow-in-the-dark watch dials from years ago. Anyway, getting either the mantles or the Fiesta fragments near the tube made my circuit flash like crazy!

How do we calculate CPM (Counts-Per-Minute)? Well, doing a full 60-second count every minute is the obvious way to count pulses, but then the system responds slower to changing conditions (eg: a small burst of pulses would get averaged down in a 60-second window). Going to a small moving-window, like 5 seconds, can make the counter more-responsive to short bursts, but at the expense of resolution, since you now have to multiply that number by twelve (5×12=60; the cpm readings will be 0, 12, 24, 36…). So as a compromise, we are using a 15-second moving-window, which is multiplied by four to get CPM (Counts-Per-Minute). This means our counts will be multiples of four (0, 4, 8, 12, 16…), and the system seems to respond reasonably-well to changing conditions, like moving test sources closer and farther. There are likely better cpm algorithms, but this seems to work pretty well. You can dig into the theory of iir and fir digital filters to learn more about digital-signal-processing in general, which is a great subject for another day, but this simple little moving window approach works just fine for our geiger-counter needs.

So what kind of numbers are we getting? In our test lab, background radiation levels averaged about 29 CPM over time (median was 28), ranging from 4 to 64 CPM. Our test sources were contained in two plastic bags and tested by laying them over the tube. Our Thorium source averaged 2204 CPM (median was also 2204), ranging from 2040 to 2392 CPM. Our Uranium source averaged 5802 CPM (median was 5876), ranging from 5088 to 6216 CPM.

The CTRL module has the main battery pack (6xAA lithium batts), the geiger module, and the controller board.

We also added a particulate sensor, a PPD42NJ from Shinyei Technology (we got it at Seeed). This is an interesting little board that has a sensing chamber (containing a current air sample) with an led (I am guessing infrared, but I am not certain) which shines down through the test chamber allowing a phototransistor aiming down into the chamber to detect light scattered from particles in the air sample.

Apparently, the manufacturer is focused on sensing nasty stuff from cigarettes, and their specified test source is a “Japanese 10 mg Mils-Seven or Mevius cigarette, in a 20-to-30 cu-meter room, stirred by an electric fan, with the sensor 40-to-80cm above floor…” Well, I don’t have access to such high-end laboratory facilities, or any cigarettes other than butts in the gutter, so I guess my low-tech version would be to let my fluffy cat chase a bird in the backyard with overgrown and flowering grass for 33.5 minutes, then bring the cat into the test room, elevate the cat to 17.3 cm, and tickle his nose with an ostrich feather until he sneezes. Do not forget to keep stirring the room air with an an electric fan. Sheesh. Regardless of the design intentions, I actually do think this sensor can give us some useful information on particulate concentration.

The assembly is intended to be mounted vertically, and the sensing chamber has a small power resistor which promotes a convective air flow to refresh the chamber with, well, a “current, air current.” We did not need a power-robbing resistor depleting our batteries, and presumed that we would get periodic refreshing of the test chamber air due to wind direction/magnitude changes, so we cut the resistor out to save power.

This unit has two analog-comparator outputs, fed by the same phototransistor input, but using different reference voltages, for detecting different sized particles. The “P1″ output triggers a pulse for chamber particles over about 1 micron (1E-6 meter = 0.039 mil) diameter. The “P2″ output triggers a pulse for chamber particles over about 2.5 micron (2.5E-6 meter = 0.098 mil) diameter. One problem with this sensor is that the supplied cable only provides the P1 output signal, and the P1 signal triggers for both large and small particles. Here in the US, and perhaps elsewhere, we have concern over the health effects of “PM10 particulates” (over 10 micron, typically dust), and “PM2.5 particulates” (2.5 to 10 micron, typically combustion products). These are both nasty-sized particles which can embed deep into your lungs and cause a variety of health problems.

We soldered wires to the board to get access to both outputs, and also wrote a sensing algorithm that deleted the P2 counts from the P1 counts, so we had better data:
– P1 counts are now only 1-to-2.5-micron (approx. PM2.5)
– P2 counts are >2.5-micron (approx. PM10)
The data is read as active-pulse-percentage over a short time window, and then converted to a calibrated concentration level using a table supplied by the manufacturer. The concentration is in units of particles-per-0.01 cu-ft, which seems to be an odd unit, but can be scaled to any other measurement system. As with all of these balloon experiments, the absolute numbers are less important than any magnitude-vs-altitude trend that we may find, indicating a need for further study.

To make the styrofoam payload container, we start with sheets of styro, and cut them to size on the table saw. It is an incredibly-messy process, and you need a compressed-air hose to get the little statically-charged bits off of you, but it gives very clean cuts.

As recommended by a craft site, we glue it together with “Glidden Gripper,” found in the paint aisle of Home Depot. One quart is the smallest size I could find, which should last me for at least a hundred years of balloon payloads (provided it is tightly-sealed, of course). It needs to dry overnight, so we tape it in place. The joints are very strong.

The payload box will stand vertically for flight, but here is a sideways shot. At the top of the payload is the ATTIC, a compartment that we are only using to fly special trinkets this flight. We have other plans for a future use of this compartment. Below that is the CTRL (controller/geiger) module. Next down is a compartment painted black inside with two air holes on each side. It will get rather cold in here, so we call it the BRR compartment — this is where we will place a particulate sensor (still in process). Below that is the CAM bay, with the cameras, video transmitter, and aux power supply. At the bottom of the payload is the RF module. The antenna cables will connect here and be tied to the bottom of the payload, letting the jack antenna hang down below.

We will add some blue tape stripes before we go — the orange and blue really stand out when you are looking for your payload in the brown and green desert. We realized just how bad it was to have a pure-white box, when looking for our lost LE-2 payload. There is an incredible amount of white trash out there to confuse your search.

On the right side you can see the little UV (ultaviolet) sensor on the top of the box, cabled down to the little console board that has external temperature and humidity sensors. You can see the holes into the BRR compartment, and the UV/haze filter we scabbed onto the front of the Canon camera port. The Mobius camera looks out the rear and has no filter (we will see how the video looks later, and may decide we need a filter here too).

The UV sensor board we got from seeed had an absurd amount of gain that just drove the output to the rail, so I tweaked it in several spots to place the output about half-supply in mid-day Arizona full sun. This should give us a decent range. This voltage output sensor just runs into an ADC input on the PIC.

We just hacked 1/8W resistors onto it.

The cloverleaf made by IBCrazy is much nicer than our test unit.

Here is the Controller Module installed in the payload.

The kids gave our payload a personal touch!