JOURNAL  (II  very long)    go to journal (I)   go to journal (III)

• We could follow the launch and ascension of Fieldgen meteo balloon (please consult journal page 1) this morning over their video-stream. Some notes and screen-shots:
  • excellent PR-action
  • interesting ham activity
  • remarkable educational initiative. Well done guys!
  • the balloon and the payload rotated enormously during the ascension. This surely was the main reason for the bad picture quality. (Question: could there be added some inertial stabilization to ours?)
  • once passed 10000m, the hygrometer returned -4% (negative relative humidity !) The sensor perhaps was dependent either on the temperature or the pressure (200hPa), or both
  • above 8000m the temperature inside of the payload started to be negative (-2°C-decreasing at -48°C outside temperature)
  • at 12000m they measured -45°C oustide and -9°C inside
  • at 23000 the outside temperature was -38°C and the inside value was -4°C
  • when the payload descended at 12000m the inside temperature was -18°C

• Calibration of the new 10k NTCs connected to Gasperi's RCX input multiplexer. Probably due to the package and the metallic head, the LEGO temperature sensor has a slower response. The graphs show that both sensors evolve in parallel. The difference in temperature can be estimated to 10°C.

• Francis's document about the PT100 calibration
• LUXPAK now needed a lot of work to make everyrthing pass into the box and respect all the requirements. Unfortunately the final design even without any silica-gel now weights 1600g. Eric's last mail tells us:
  • "We now have our complete payload set at 10480 grams:
    
    1) David Levy (USA) 1440 grams
    2) Eugene Tsai (Taiwan)  1630 grams
    3) Brian Davis I (USA) 1580 grams
    4) Brian Davis II (USA) 1130 grams
    5) LEGO Mindstorms team (Denmark) 800 grams
    6) Tufts Univ (USA) 800 grams
    7) Claude Baumann (Lux)  1600 grams
    8) David Martinez (Sweden)  1500 grams
    
    As we cannot reasonably carry anymore, this will be the final list. I am
    begging you all to attempt to reduce weight as much as possible. The lighter
    the mission, the higher it goes! Every gram helps!!!
    
    I am in the process of updating the HALE website now with information on all
    the payloads and more information about building payloads.
    
    In case you didn't see the email, David Levy has setup a Google groups page
    at: http://groups.google.com/group/hale-teams/
    
    This is going to be an exciting experiment!!!!"
    
    Regards,
    Eric Wang
• So, we will try to find those grams that we can save.
• Here a long series of pictures of the payload assembly as decided by Francis and the Convict-team:

1. How the sensor wires are fixed to SNOOP_LOG (from the left to the right: pressure sensor, ozone sensor, PT100, light-sensor)

2. air inlet and outlet fixed to the new ozone sensor box that replaces the transparent one. Note that on the bottom of the box (not well visible) another NTC temperature sensor is glued.

3. outlet glued to the polystyrene box

4. outlet stud outside

5. green plate helps fixing the accus pack

6. so does black plate (studs are cut)

7. two yellow beams are glued to the top of the ozone-sensor box. They serve as support for the accu pack. (Inverted disposition)

8. the tube for the balloon string needs to pass through the middle of the payload box

9. By contrast to the group's decision, Laurent suggests to put the accus on top in order to allow easy access.

10. an NTC temperature sensor is glued to Mootz_board

11. Heat-resistor wires, air-tube and silicon isolated NTC sensor wires

12. the black box beneath the pressure-sensor contains the Gasperi RCX input multiplexer

13. cutting the box material for the "remove before launch" device

14. easy to cut and to press out

15. idem

16. the button and the support pieces are glued to the box

17. a beam is used to cut the hole for the LED support

18. inner view

19. the transparent brick is glued to the box

20. how the main switch is fixed

21. the plate bridge to the main structure will later be glued to the button (perhaps we will use a smaller one)

22. the PT100 cannot be fixed next to the light-sensor (bottom) because the wires are not long enough

23. air-outlet stud

24. inlet-stud (ATTENTION : not to squeeze the flexible silicon tube that makes the bridge

25. View from the side; the silicon-protected NTC sensor has not been fixed yet.

26. Another NTC sensor that measures the temperature in the box

27. the balloon string will pass through this tube

28. the complete payload (the brown paper contains the silica-gel; but for weight reasons, we cannot carry this pack on board.)

• Francis's document on air heating modelling and experiments    ==>   and the update
• Now the work consists in:
  • learn about accu charging
  • rewriting the RCX firmware for LUXPAK with the new requirements:
    • one heater powered from two RCX ports
    • heater control :
      • keep temperature in the tube at 100°C (perhaps less, as Francis suggests)
      • if the temperature in the ozone-sensor box is above 10°C (exact cut-off value has to be determined), reduce the temperature in the tube
    • control the sensor input port multiplexer:
      • changing the "active" powered sensor state shortly to "passive" unpowered state reduces the voltage from 9 to 5V and also sets high impedance (10k). This gives an impulse to the 4017 counter/multiplexer and anothe 4066 port is activated. The normal short sensor reading of 0.1ms is not sufficient to trigger the counter.
      • The first 4066 port to be recognized is the one that only sends back the voltage through the 1k resistor (raw values below 338). Now every pulse switches the NTC sensors that are connected to the multiplexer board. The raw values can never be below 430 as may be described in the equations and the equivalent schematics for one "Gasperi" port:

           

    • if port2 values are transmitted to SNOOP_LOG, the two most significant bits are used to determine the active 4066 port (thus we only log temperatures on port 2 every 40 seconds). We also log the 338 reference voltage in order to detect bad data, if ever.
    • If during flight the gasperi mux should present some mulfunction, we reduce the tube temperature to 50°C not to risk any LUXPAK overheating.
    • add self-tests of the new sensors and also the battery level, in order to detect low battery (it is a good thing that the SNOOP_LOG heater is working right from start, because low accu power will be seen better, if some current is drained.
    • only do the SNOOP_LOG data pointer reset on pressing the VIEW/PRGM button at the same time.
    • add a conversion function for those out of range values of the RCX ROM function concerning the temperature sensor.
    • other points?
• Here a picture of the air inlet where we added an insect filter:

• Yesterday's equations model the real circuit a bit too much. They presuppose that the diode resistance is zero, which is not the case. Also at low voltages, the voltage-drop on the diodes doesn't equal 0.6V, because of the non-linear response. Therefore we proceeded to another experiment. We built up the circuit as shown above with a potentiometer in series with a 2k2 resistor. We measured the potentiometer resistance and noted the corresponding RCX raw value. Then we used the indicated approximation formula to establish a calibration curve (Steinhart & Hart):

• The function graph has been drawn in Excel and a linear approximation with 3 intervals has been computed in fix-point equations to facilitate the RCX's task:

• If the NTC is used directly on an RCX-port, e.a. not with the Gasperi mux, then the approximation function :

• First test with the new RCX firmware:

  •  For this hot-test, the cut-off values are:

    • ozone-sensor temperature: if T_oz>30°C ==> reduce the control temperature for the tube gradually down to 10°C, else gradually increase to 100°C
    • box-temperarture: if Tbox>60°C ==> set the tube control temperature to 10°C (risk of overheating the whole box contents)
    • at any reading malfunction, set a safe tube temperature of 40°C
  • During the longest part of the test, the air at ambient temperature was pressed with a LEGO pump (Hempel's design) through the tube. (We will estimate the air-quantity/second tomorrow)
  • At the end of the test, we used a N₂-capsule with a cream-bottle. The air was quite cold and pressed through the tube.

• The results are very interesting, although there still was a bug in the firmware. That's why the regulation took so long to decrease the power, even, when 30°C was measured at the ozone sensor.

  • The ozone-sensor IS sensitive to N₂, as can be seen on the graph. At the end, the increase is drastic (... observe the scale plse.). We had no moisture in the sensor box after the experiment.

• PT100 run at ambient temperature. A few times, the sensor was taken in one hand. At the end the cold N2 capsule was placed next to the sensor:

• A lamp was pointed at different distances towards the light sensor. The negative values obviously are wrong. They are the result of the calibration curve and should be pruned.

• The RCX-voltage is extremely stable, although during more than 30 minutes the RCX had to produce total currents of 1200mA during most of the time.

• The tube temperature grows exponentially at the beginninng, then there are two moments (visible oscillations), where the regulation reduced the motor power. The increase at the end was due to the stopping of the air-pump. With continuous air-transport, it will hardly be possible to reach 100°C. Thus the heat stabilzation at 100°C can not be activated.

• Obviously the PWM-values are NOT expressed in %, but represent the instant output power level. Twice the regulation was triggered through external events.

• The multiplexed values are not yet representable in a time graph. The box-sensor was heated twice over 60°C (the peaks are visible, not the real extrema, because the events were so short.) The ozone-sensor box temperature first is reduced, as the air-pump starts, then steadily, but weakly grows. Note that the cold N₂ is heated as well. There is no noticeable drop!!! The RCX is seriously heating under the extreme currents. The firmware should also consider this temperature in order to protect the RCX and Mootz_board from overheating. Note that neither the PT100, nor the light-sensor are drifting visibly.

• SNOOP_BURN PIC16F88 firmware:  .asm  .hex
• Yesterday evening we proceeeded to the "cold test":
  • The RCX firmware version 22 (.asm   .srec)  was adapted with a better temperature regulation:
    • now the tube heater will be powered at 255 during the first 12 minutes of the flight in order to pre-heat everything and overcome the thermal "inertia" of the big resistors. The accus are supposed to be fresh and there should be no risk what ever of overheating during this short phase.
    • if the RCX voltage level drops below 8.3V, then the heater is switched off independently of any other function in order to protect the datalogging process.
    • after 2 hours the heater task is switched off.
    • if the ozone-sensor temperature is below 20°C then the setpoint temperature (also called here "goal temperature") is increased (maximum 80°C), else the setpoint temperature is decreased (minimum 10°C)
    • if the box temperature is below 10°C then the setpoint is fixed to 80°C. If this temperature is more than 50°C, there is a risk of overheating and the setpoint is fixed to 10°C
    • if the RCX temperature is below 10°C then the setpoint is fixed to 80°C. If this temperature is more than 45°C, there is a serious risk for the RCX to overheat and the setpoint temperature is reduced to 10°C
    • if the GASPERI multiplexer does not work correctly, then the set-point is fixed to safe 30°C
    • the firmware is programmed in order to have a hierarchy of importance:
      • lowest importance: ozone sensor -temperature
      • box temperature requirements override the ozone sensor one
      • RCX temperature requirements override the previous ones
      • first 12 minutes and overall time (2 hours) have equal priority
      • accu level has highest priority
    • now the firmware sends the setpoint value to SNOOP_LOG instead of RCX portB state, which always is identical now to portA.
    • we won't be able to use the VIEW/PRGM button combination to reset the EEPROM data-pointer, because we want to maintain the normal VIEW-button function.
• The test was made under worst conditions:
  • weak accus
  • humidity 98%
  • temperature -18°C in a big freezer
  • duration 40 minutes
  • the LEGO compressor pumped the air that passed through a box where we put a silica-gel pack to reduce the humidity.
    • first the quantity of air per second was measured and calculated:
      • a rotation sensor was coupled to the pump motor (12738 pulses in 3 minutes)
      • the pump motor rotates at a speed of 12738/3/16=265.375 rot/min
      • the pistons were measured: inner cylinder: 6mm diameter, height 16.5mm ==> volume=0.4665cm³
      • one motor rotation produces two piston actions ==> air quantity/min= 257.6cm³/min = 4cm³/sec

• After 3 minutes the accus (that already had served during several hours without being recharged) lost their voltage, which first is compensated strongly by the RCX stabilization (voltage grows to almost 10V). Then the voltage can no longer be maintained and it drops below the cut-off value of 8.3V, where the heater is switched off. During the next few minutes the heater switches on and off because the battery voltage drops as the heater is on and recovers when it's off. After 12 minutes the heater setpoint is reduced to 10+C, which reduces the RCX output power to 0. The accus now recover and LUXPAK can continue working. After 39 minutes, the accus voltage drops below definitely switching off the heater, because the accus are unable to recover.

• The RCX output PWM reacts accordingly and oscillates at first weak accus detection. The attempt to relaunch the heater at minute 39 fails, because of the low battery state. (QUESTIONS: does the oscillation affect the sensor readings? Could we use a smooth decrease of the power instead of radically switching off the heater? In that case, we would have to add a second cut-off point, where the heater defintely is switched off.)

• The temperature setpoint is 10°C during most of the time, because the ozone-sensor temperature is above 20°C and neither the box interior nor the RCX are in low temperature state. However, after 38 minutes, the setpoint starts oscillating, because the ozone-sensor temperature oscillates around the 20°C cut-off value. (QUESTIONs: is this oscillation desirable? Especially the decrease of the setpoint seems odd. Perhaps, we should simply allow two different setpoints (either 80°C or 10°C)? )

• The ozone-sensor temperature values reach the cut-off value after 38 minutes. The discontinuity at minute 15 is due to our curiosity: we opened the freezer in order to have a glance. The hot air streamed into the freezer and reached the ozone-sensor. The second opening however did not affect this temperature at all. Note thaht the sampling rate of the GASPERI multiplexed sensors is divided by factor 4!

• During the second freezer opening we moved the box a bit. Because the cover did not completely close the box -the air-tube pass between the box and the cover-, the box-sensor reacts drastically. The increase at the end was produced, when we took LUXPAK out of the freezer.

• The RCX temperature increases during the heater activity, then decreases slowly.

• The tube temperature readings are more continuous. It increases as the heater is switched on and decreases aftwerwords.

• The PT100 values on the next graph still are raw and don't represent °C !!! The battery voltage drop affects the sensor readings which must be corrected. We must discuss the calibration of this sensor.

• Observations and conclusions:

  • Except the oscillation discussion above, the RCX firmware is most satisfying.

  • The temperature regulation will probably be sufficient to do the job under frost temperature

  • Moisture didn't cause any problem: the ozone-sensor box wasn't wet at all. However on the inner side of the light sensor, the metallic support was a condensation point. So was the transparent LEGO piece that is used to protect the LEDs!

  • We recommend that once the balloon string has been pulled through the special tube, the holes should be filled with silicon in order to prevent additional cooling

  • The accus seem to have enough capacity for this mission. The protection in the case of low battery allows the system without the heater to work about 30 minutes beyond the point of power-drop. In previous experiments we could notice that any ULTIMATE ROBOLAB firmware continues running, even if the battery level drops below 7V.

• SNOOP_LOG and SNOOP_BURN are in their definite form both from the electronics and the microcontroller programs. We put everything into one file that people can download now.
• A 710-4 LEGO wood-storage building set (1963) from the flea market delivered a few white 1x1 bricks with printed letters that we glued together to form a real historical and well visible "REMOVE BEFORE LAUNCH" panel. (Eric, could you manage that we get it back after the event?)

• We measured the different currents that are drained (all measurements at 12V power supply voltage):
  • 150mA at start (with O3-sensor batteries empty)
  • after 3 minutes this current is reduced to 120mA
  • the RCX firmware download needs 5 minutes, after which the 800mW heater is switched on and the current grows to 240mA
  • If the tube heater is switched on, the current is increased to exactly 1A
• Francis determined that the pressure sensor is linear and the calibration function can be set up easily:
  
  0 hPa = 0.5V
  1200 hPa = 2 V
  
  thus hPa = aV+ b
         0 = a*0.5 + b
         1200 = a*2 + b
  ------------------------------------
  ==> hPa = 800*Volt - 400
  
• The intriguing 2008_05_29 graph of the RCX regulated voltage asked for deeper analysis. So, we powered LUXPAK with a well stabilized power supply and got the following plot that in fact is the result of the ULTIMATE ROBOLAB battery level limitation (<=9999mV). The issue is that with values greater than 9999 the battery level is burned in the last valid value smaller than 9999. This obviously is a bug. We will fix it and replace this function with a new one that does not limit.

• The only sensor that is dependent on the applied voltage is the PT100. Since Mootz_board tries to produce a constant current of 1mA through the PT100, it is somehow obvious that the ouput will be proportionally dependent on the voltage. The calibration function will have the form:

• The new RCX firmware (version23) now respects all the described requirements except that, if the battery voltage drops below 8.3V, there is no smooth stop of the heater, because the battery voltage would continue dropping and we would loose the RCX. Therefore, at low battery, the heater is immediately switched off, which unfortunetely leads to the undesired oscillation that we already enountered. But, it seems that the only alternative is to switch the heater off definetely, as soon as the low voltage is detected. This remains on the TODO-list.
• With the new RCX battery firmware module we overcome the issues descibed above. The folllowing graph has been realized while gradually changing the power supply. During the first 12 minutes and between seconds 1300..2200, the heater was on. (From sec 720 to the end of the experiment the power supply voltage was kept constant.)

• Applying the following equation, we get adjusted PT100 values:

• It also is visible that the PTvalues are dependent on the Mootz_board temperatures that also are measured:

• From Jean Mootz and Francis Massen's' calibration study we compare two of the calibration functions. The first one is accurate with negative temperatures, while the second is applyable with positive values and yields the correct temperature at start, while the first returns the lowest temperature that was measured. We therefore propose a third calibration function that will spread the values all over the operation range.

• A few serious issues appeared yesterday (during the meeting with Francis and the evening team session):
  • Hardware:
    • The main LUXPAK device still is not totally stabilized within the polystyrene box ==> the students already solved this : photos will follow
    • Francis is right objecting that the small stud in front of the HMOS sensor (ozone nose) should be clipped off in order to liberate the nose. A plastic nug must be added then to secure the O3-sensor.
    • Brian Davis, one of the HALE participants has suggested to somehow use the heated air from the ozone-tube to heat the box interior. Good suggestion! We therefore will not close the O3-sensor box in an air-proof way.
    • We already received the NiMH replacement batteries from CONRAD electronics.(Thanks for the incredible speed!). However the NiMH batteries present some disadvantages:
      • Weight is still an issue, because we need to find those grams that Eric Wang needs for other participants.
      • We have no official data about self-discharge. Francis' experiment shows a loss of 0.32V loss over 5 days, which does not encourage our optimism, because of the characteristic discharge graph. The only solution is that the accus should be recharged just before launch.
      • We have no indication about the conformity to air transportation legagy neither internationally nor specially for the US for these batteries.
    • As ENERGIZER is one of the HALE sponsors, we had a glance at there products and found a Li-battery that could be an interesting alternative:
      • Product-name : L91
      • Voltage : 1.5V
      • Capacity : more than 3000mAh
      • Wide operation range : -40 .. 60°C
      • Weight : 14.5g (we would have a gain of 164g for LUXPAK !!!!)
      • Inner resistance : 90 to 150 milliohm (dependent on temperature)
      • Long shelf life : 15 years !!! (no need to charge; LUXPAK even could be sealed.)
      • CONRAD electronics : article number : 626830-62
      • costs : ~40€ for 8 cells (we only would need 8 cells to get 12V) We have a gentle donator, so we can buy 2 sets of those cells, one for test and a seocnd one for the flight
      • ==> the different characteristics under high current load, needs a general protection: we will add thermo-fuses (cut-off at 90°C) to the heater circuits. If the batteries run the risk of overheating, these fuses will dramatically cut the current. Both fuses will not be placed on the same side of the battery pack and thus will not react at the same moment. We will not necessarily loose both heaters in this extreme emergency case.
      • ==> we also will move the box-NTC closer to the batteries to sense the battery heating.
  • RCX firmware:
    • The selftest-function at program start may result in a situation, where the Nevada operators won't be able to start LUXPAK correctly and run the datalogging. The trouble is that the error-handling just issues the error message on the display, the LED blinking and the acoustic signal, but does not add any fail-safe action. ==> we must add a fail-safe mode.
    • If we are doing the job with the ENERGIZER batteries instead of the NiMH, then we will have to add another temperature control, because those batteries risk to overheat under high endurance constant 1A current. The idea is:
      • since we now get the correct information about the battery voltage (minus the loss at the internal RCX rectifier bridge), and we know that most of the current is drained through resistive loads, the we can deduce the current from the voltage. If the inner resistance of the battery drops through the temperature increase, then the available voltage rises and with the current. If we add a regulation that will try to maintain the current stable by reducing the PWM-value, then the battery heating will not be dangerously important.
    • During the high endurance test we found that the NTC conversion functions as described above present points of discontinuity tha we absolutely must avoid.
    • The hysteresis of 1°C is too wide for the O3-temperature regulation, resulting in a too slow regulation frequency
    • The threshold of the O3-temperature regulation must be reduced around 5°C. (We only should avoid frost. We don't really need higher temperatures.)
• High endurance test :
  • Conditions:
    • O3-sensor precharged during 1 hour
    • -19°C freezing chamber
    • 1h30
    • PL91 ENERGIZER batteries
    • no air pump this time, just the simple stream that will pass through the O3 sensor system tube
  • Results:
    • The battery voltage is reduced at start, because the heater is immediately switched on during 12 minutes. We we consider both 3W heaters as parallel resistances, we have a total heavy heater resistance R=10Ohms and the heater current at start is 9.9/10=0.99A. The increase of the voltage after 12 minutes gives us the delta voltage of 11.6-9.9=1.7V and thus an internal battery resistance of 1.7/0.99=1.7ohm, which is much more than indicated on the datasheet. However already during the first 12minutes pre-heating phase and surely during the second long-time heater-on phase the inner resistances drops due to the increase of the battery temperature. At instant 1h25, where the heater is shortly switched off, we have only : 11.6-11.05=0.55V difference. The heater current grows to 11.05/10=1.105A and the inner resistance is 0.55/1.105=500milliohm. At the end of the experiment the battery heat was about 50°C.

• The device temperature graphs (please ignore the discontinuity, which is due to the error of the NTC conversion function that will be changed soon.) One can see that after one operation our the heater regulation is triggered, because the ozone-sensor temperature droped below 19°C. After 30 minutes the tube temperature reached the cut-off value of 80°C. The temperature drops again and the regulation starts to maintain the 80°C in the tube.

• The ozone concentration values and the light-sensor suffer a bit from the voltage changes as can be seen on the next graph. We therefore will add a compensation that will be applied to the raw values.

• The pressure sensor values -still raw- don't corrolate directly with any of the graphs except the temperature. As the box was well closed this time, the drop of the pressure can be seen as a function of air-temperature (although there is a certain delay between the box temperature and the pressure, probably due to the pressure sensor tube.) Note that the pressure sensor is temperature compensated and does not react on changes of the supply voltage.

• Francis suggested that we could add a voltage stabilization for the ozone-sensor only. So we powered the ozone sensor (and later also the Mootz_board) with a regulated voltage of 10.40V. But this does not change the influence of the heater activation to the ozone sensor values. This is a bizarre behaviour that we cannot easily explain. So, we can only apply a correction to the measured raw-values after the flight. Comparably to the PT100, we can recalibrate the O3-sensor and the back scattered light sensor as well. It must be underlined that the calibration curve shows that the O3-sensor must be extremely sensitive to any change at low ozone concentrations.

• We added the thermo fuses (cut-off at 91°C+/-3°C). These sensitive devices will be directly glued to the battery pack on different places.

• We adapted the NTC calibration functions to:

• We found out that the voltage stabilization absorbs rapid fluctuations. However, the influence of the heater on the 3 concerned sensors still needs a correction on the raw data.
• ==> Due to the smaller supply-voltage on the Mootz_board, the third PT100 calibration function now is changed to:
  • x₃ = 0.275 * x₂ - 231 (the x₁ and x₂ function are inaltered.)
  • Note that this function is much closer to the one suggested by Jean Mootz and Francis Massen in their calibration document.
• Another calibration test confirms these functions:

• For this experiment one of the well-calibrated NTC sensors (the formerly called "box temperature sensor", now the "battery temperature sensor") was placed in a water-proof small bag together with the PT100. Then both sensors were plunged in ice-water. At the end of the test, the sensors were taken off and could heat up to ambient temperature. (Note that we will make a further test in the freezing chamber on Wednesday to see, how the PT100 reacts then.

• Since we know that the ozone-sensor is most sensitive for variations at weak ozone-concentrations, we probably must recalibrate this sensor.

• The light sensor also must be recalibrated. So does the pressure sensor, as can be seen on the following graphs. Note that the test has been executed with the new RCX firmware that has a few important changes:

  • The preheating phase only has 10 minutes anymore. The PWM only works at 50% duty cycle during that phase. The reason for this is to protect the batteries from overheating, since the system has to run before with the 800mW heater on during 1hour (before flight).

  • The battery-NTC sensor now delivers important information about the battery temperature. If it exceeds a certain threshold, the heater output power is reduced.

  • The firmware shuts down non-vital functions at low battery to protect SNOOP_LOG

• During the test the heater was deliberately switched off by unplugging the wires from the RCX from t=18 to 22minute and t=30 to 33. As can be seen on the graphs, the influence on the curves is less strong.

• At the end of the test the RCX output driver circuitry regularly reduced the current. This became visible on the Ampere-meter that was added to the battery cicuit

  • The ozone-concentration should be 0.03ppm, but is indicated as 0.3ppm on the graph

  • Important observation: according to the inverse of Francis' calibration function, 0.03ppm is equivalent to a measured voltage of 0.031V (=16 [raw]). Remember that the raw values extend over the range 0..1023 proportionally to the voltages 0..2V. Now 0.3ppm correspond to a measured voltage of 0.074V (=38 [raw]), which means that in the lower voltages small errors have an impact of factor 10. This is not the case at higher voltages. Since the deviation must be considered as an offset of a few mV, we probably can adjust Francis' calibration function by subtracting the raw offset 22 from the original data.

• The ESR400 light-sensor is operated at very weak voltage (irradiance of 10W/m^2 corresponds to a measured voltage of 46mW according to the calibration function. However, this is out of the linear range of this sensor (and also of the LM324 operation amplifier), so that we can ignore those small values and errors. A throughful study of errors will be done on the final values.

• The pressure-sensor should return ambient air-pressure of yesterday evening that passed from real 995 to 996 hPa during the experiment. ==> This sensor absolutely needs to be recalibrated.

• The new RCX firmware now uses the improved calibration functions for the NTC sensors. The discontinuities have disappeared in the graphs.

• Francis worked out the equipment description paper.
• The new RCX firmware can be found here: ==> LUXPAK_RCX_firmware

• Our project has found a generous private sponsor. Thanks a lot!!! The financial support is most welcome.
• CONRAD electronics offers more detailed datasheet about the ENERGIZER L91 batteries. And we learn from these papers that the L91 batteries have a special built-in PTC thermistor that increases the battery internal resistor as the temperature reaches 85°-95°C. This represents a very useful protection against overheating, since the current will be limited at high temperature. However, we will maintain the additional thermo-fuse protections and the temperature regulation, because reduced battery inner resistance will cause higher currents through the RCX... and this will happen already at much lower temperatures. The result will be current limitation in the RCX output drivers, but also undesired heating of the RCX interior. If the batteries reach very high temperature, besides the PTC, the permanent disabling of the heaters -one by one- will help protecting both the batteries and the RCX.

• We met today for fine calibration of the pressure sensor and the ESR light sensor. (Photos and new calibration functions will follow from this activity.) Here the recalibration of the ESR values in function of the LUXPAK raw values that are delivered by SNOOP_LOG's 10bit ADC:

• ... and we found two bad bugs in the data extraction software that, besides voltage stabilization issues, were the major problems of bad results:

  • SNOOP_LOG does not generate a reference voltage for the ADC of exactly 2V, but rather 2.078V. This was not respected in the conversion functions and thus produced bad data.

  • The transcription of Francis' ozone calibration function had a simple calculation error that now has been fixed : we asserted recently in this journal that the sensor was most sensitive to errors at low voltages. This was a simple, but nasty mis-interpretation of the calibration function. In reality, it is exactly the inverse. Sorry, Francis ! ==> Now the ozone-sensor values have the correct aspect:

• With this we reviewed all our ozone data that we got so far and we can conclude that the sensor needs pre-heating and -charging only during 10 minutes. Thus we will reduce the recommended LUXPAK STAND_BY BEFORE FLIGHT TIME to 30 minutes that should not be exceeded in order to save electric power.

• LUXPAK will be closed by Velcro hook-and-loop fasteners that will be glued to the box. This is an excellent mechanical way to fix things together that has proven its validity under extreme conditions. That's why NASA is a big Velcro user.

• We now started the user guide paperwork, a second one about the pedagogical aspects and finally on the control stuff that completes Francis' work on the equipment.

• This evening will be the moment of definite assembly of LUXPAK

• Paul-Nicolas has already worked on the logo picture.

• The adjusted pressure sensor calibration function now is : p [hPa] = 1.712 * RAW_VALUE - 419.44
• Here the photos of the final sensor calibrations:
  • with the pump we produce vaccum:

• LUXPAK's chirurgical operation:

• pyranometer vs. ESR sensor calibration:

• help is welcome:

• the definite assembly:

the RCX power plug:

the RS232 cable:

preparing to fix the buttons and LEDs:

switch, LEDs and RCX touch-sensor from the rear:

before placing beneath the balloon tube:

the balloon tube is visible:

the outlet of the ozone-sensor:

the Teflon-balloon-tube is glued to the ozone-sensor case:

the resistors disappear in the isolation:

idem:

the inlet to the heated tubing:

the inlet to the ozone-sensor case:

the RS232 cable in the box:

general aspect of the interior with Erik and Laurent's LEGO construction:

with the batteries:

the communication side:

how to install the batteries:

see the thermo fuse on the bottom of the pack:

... and the thermistor glued to the outer wall::

the batteries must be fixed with a long cable binder:

how this is done:

next step:

one more:

cut the end:

turn it around and place it in the socket:

now everthing is well in place:

another view on the whole thing:

Velcro is fastening the top of the box:

with additional isolation pads:

the LEDs light:

• The batteries need to be fixed with two cable binders. Otherwise there is a risk that they loosen through vibrations. This would corrupt the whole mission. The first one absolutely must be strongly tied. But attention to the seocnd one, not to damage the holder.

• Now LUXPAK has its LOGO, which was designed by Paul-Nicolas. (He borrowed Eric Wang's photo. (Eric, hope this is OK.). The Konvikt's and the LCD's logos are well visible and the sponsors' too. The web-site is indicated to help potential finders to return the payload to their owners. The container body and the cover have marks that must be respected. LUXPAK has an orientation that also must be observed.

• Finally, before fixing LUXPAK to the balloon string, the Velcro fasteners must be secured with adhesive tape (the wind could loosen the ends):

• Three documents are ready now: LUXPAK_user_guide.pdf, LUXPAK_control.pdf and LUXPAK_LOGO.pdf

• The LabVIEW8.1 viewer program for the LUXPAK data is also ready.
• The Center of Engineering Outreach (CEEO) from Tufts University has published an article about our contribution to HALE.

• The bill of material is available.
• LUXPAK's transport to Reno will probably be done via UPS.

• Francis worked hard on the albedo problem. Doesn't seem easy. But his approach seems most valuable. The new document on the sensor-equipment and the calibrations is now available. This completes the LUXPAK documentation.

• Laurent while preparing the very final test in the freezing compartment

• Eric already passed to his next project a sophisticated fully automated crane, which definetely has nothing to do with LUXPAK :)

• From the data we retain that the pressure sensor either needs a power-up time or has a certain dependency on voltage or container temperature variations that we will analyze on the HALE event data later. The air-temperature gradient is flatter, because most of the PT100 sensor is placed in the isolation material. Only the very top has contact to the air.

• LUXPAK is being shipped to its destination Reno NV via UPS. The package should arrive by July 7th.
• We should agree with Francis :-)  :
  • *** let's hope and pray the Lord:
    1. to boost battery-life
    2. to create a local short lived stratospheric warming
    3. to remove that sand-corn blocking the ozone inlet/outlet channel
    4. to make the landing site smooth and spongy
  • Perhaps we should add:
    • to overcome Murphy's law

• HALE launch date has been fixed to July 29th at dawn likely at location: N 40° 20’  W 119° 20.5'
• LUXPAK survived the "horizontal trip" from Luxembourg to Reno as can be seen on one of Eric's photos that we borrowed from him:
  • Hi guys, don't forget to remove both paper protections, that we placed over the ozone sensor inlet and the outlet stud, before launching the balloon... and of course the most important panel "REMOVE BEFORE LAUNCH" !!!!