This page relates the parallel advances of the Power of the Mud project made respectively by Paul Granjon in Cardiff and Michka Melo in Brussels between their intensive work session which took place in Cardiff in May 2017 and September 2017.

  • Paul prototyped and made a series of clay pots to host our MFCs.
  • Paul designed the final version of the robots, and built the first one.
  • Paul started up the MFCs, and performed power tests.
  • Paul wrote our contribution to the V&A Digital Design weekend publication.
  • Michka performed tests on the Power Harnessing System (PHS) circuit.
    • The best circuit we got so far is the following one :
      • the + pin of the power supply is connected to the + pin of the supercapacitor.
      • the + pin of the supercapacitor is connected to the + input pin of the 1381E.
      • the - pin of the supercapacitor is conneced to the ground of the power supply.
      • the - input pin of the 1381E is connected to the ground of the power supply.
      • the output pin of the 1381E is connected to the gate of a KSD5041QTA transistor.
      • the collector of the transistor is connected to the
      • the emitter of the transistor is connected to the 3V pin of the BBC microbit.
      • the ground pin of the BBC microbit is connected to the ground of the power supply.
    • With two 2.8 V 10F supercaps in parallel, we obtain a 56 seconds run of the blinking BBC microbit.
  • Paul made some clay prototyping and cooking in Claire’s oven in the summer, testing a container design for our V&A DIY microbial fuel cells.
  • He will try to make a series in Cardiff University’s ceramic workshop over the summer.
proto-pots-Claire


  • Paul wrote a first version of our contribution to the V&A Digital Design Weekend documentation.
mudDrawing


  • After some logistical tribulations to get a few missing components, Michka tried (and failed) to replicate the power harnessing system circuit designed during our last working session.
  • Paul worked on:
    • Making the new clay pots series for the V&A MFCs.
piles_vides


main mud module


  • The clay pots had to cook over 36 hours, the temperature gradually rising to 1200 °C. This high temperature allows for waterproofing of the pots without the use of glaze or additional treatments.
  • Testing stainless-steel-scrubbers electrodes - which at least partially worked.
  • Over the long term, they seem to work as well as the test battery with graphite felt electrodes.
  • Designing & prototyping the robotic system design
new_robot_design


  • The current structure is made of two BBC microbit modules which communicate, placed in a clay/metal structure.
  • Paul sent Michka an article on GMO Shewanella optimized to digest xylose, and produce more electricity thanks to this new source of food.
  • Paul made a quick prototype of the horizontally-moving part of the robot, which runs at 1.9 V.
mudBotCar


Tests with resistor-based voltage divider

  • FoAM's power supply does not go below 2.8 V of voltage.
  • In order to control the voltage to a lower threshold, I built a resistor-based voltage divider.
  • First version of the resistor-based voltage divider was made of a 100 ohms and a 10 ohms resistors in series. It worked but, smelled funny. After few runs, the 100 ohms had turned a nasty shade of black. It seems that the amount of current flowing in the resistor was too important because the resistor value was too low.
  • The second version of the resistor-based voltage divider was made of a 2.2 kohms and 1 kohms resistors in series. No weird smells, it worked.
  • However, I kept having funny results. I tested many of the components, and finally concluded with a test without the voltage divider that it might be perturbing the correct working of the circuit.
  • I will therefore not present here the hectic results of these first tests.

Tests directly powered by the power supply

Test #1 - 1381E only
  • When powering the 1381E voltage detector directly with the power supply (power supply + pin on 1381E pin 2, power supply - pin on 1381E pin 3, we obtain a 2.8V voltage on its output pin (pin 1), which is the same value as the one measured between the + and - pins of the power supply.
Test #2 - 1381E powering BBC microbit directly
  • When connecting the BBC microbit 3V pin to the output pin of the 1381E (the GND pin of the BBC microbit being connected to the - pin of the power supply), the BBC microbit does not light up. The measured voltage on the output pin of the 1381E is 0.64 V.
  • It seems that the 1381E does not supply enough current to its output pin to power the BBC microbit.
Test #3 - 1381E controlling 2N3904, which powers BBC microbit
  • When connecting the gate of the 2N3904 to the output pin of the 1381E, then 2N3904's collector to the + of the power supply, and the 2N3904's emitter to the + of the BBC microbit, it does light up.
  • The voltage measured at the + end of the 1381E is 2.80V.
  • The voltage measured at the output of the 1381E is 2.64V.
  • The voltage measured at the emitter of the 2N3904 is 1.82V, which is sufficient to bleakly but steadily light up the LEDs of the BBC microbit.
  • We can measure an overall -0.98 V voltage drop between the power supply + pin and the BBC microbit + pin, through the 1381E (2.80-2.64=0.16 V) and the 2N3904 (2.64-1.82=0.82 V).
Test #4 - Power supply feeding 1F supercap, powering the circuit
  • When connecting a 1F 2.7 V supercap in parallel of the power supply, we obtain the same values as in test #3.
  • When turning off the power supply, once the supercap charged, we have rougly 1 sec autonomy of the BBC microbit. The BBC microbit goes off when :
    • the voltage on its input pin goes roughly below 1.76 V.
    • the voltage on the output pin of the 1381E goes rougly below 2.6 V.
    • the voltage on the + pin of the power supply goes below 2.76 V.
Test #5 - Power supply feeding 3F supercap, powering the circuit
  • When connecting a 3F 2.7 V supercap in parallel of the power supply, we obtain the same values as in test #3.
  • When turning off the power supply, once the supercap charged, we have rougly 3 sec autonomy of the BBC microbit. The BBC microbit goes off when :
    • the voltage on its input pin goes roughly below 1.72 V.
    • the voltage on the output pin of the 1381E goes rougly below 2.52 V.
    • the voltage on the + pin of the power supply goes below 2.67 V.
Test #6 - Power supply feeding 10F supercap, powering the circuit
  • When connecting a 10F 2.7 V supercap in parallel of the power supply, we obtain the same values as in test #3.
  • When turning off the power supply, once the supercap charged, we have rougly 8 sec autonomy of the BBC microbit. The BBC microbit goes off when :
    • the voltage on its input pin goes roughly below 1.71 V.
    • the voltage on the output pin of the 1381E goes rougly below 2.52 V.
    • the voltage on the + pin of the power supply goes below 2.67 V.
Test #7 - Adding an additional 2N3904 transistor controlled by the BBC micro itself
  • We add up a second NPN 2N3904 transistor, which base is connected to the 0 output line of the BBC microbit. The BBC microbit outputs a 1023 (max value) analog signal to its 0 output line to activate the gate of this transistor. The collector of this transistor is connected to the + of the power supply, and the emitter to the + of the BBC microbit.
    • The goal of this second transistor is to allow the BBC microbit to feed itself directly from the power supply (the supercap) once the voltage of the power supply (the supercap) drops below the one activating the output pin of the 1381E. Indeed, if the power supply (supercap) voltage drops below the lower activation limit of the 1381E, its output signal will drop to 0, which will shut the gate of the first transistor, thereby cutting the power line to the BBC microbit, even though the power supply voltage is still (theoretically) around 2.2 V. The BBC microbit being able to run as low as 1.71 V, as our prevous results show, it would be a pity not to use the burst of energy delivered by the supercap between 2.2 and 1.7 V.
  • (A.) When connecting this two-transistors version of the circuit to the power supply, we obtain the following values :
    • 2.80 V at the + pin of the power supply, as in test #3.
    • 2.63 V at the output pin of the power supply, very much as in test #3.
    • (1.) 1.82 V at the emitter of the first 2N3904 transistor, as in test #3, and therefore the same at the emitter of the second 2N3904 transistor.
    • (2.) 1.78 V at the 0 output line of the BBC microbit, and therefore the same at the gate of the second 2N3904 transistor.
  • (B.) We know that current flows through an NPN transistor when :
    • (1.) It has a relative positive voltage on its gate.
    • (2.) Its gate voltage is at least 0.6 V superior to its emitter voltage.
    • (3.) The collector voltage is superior to the emitter voltage.
  • We can therefore deduce from the values (A.1) and (A.2) that the second transistor cannot open, because the voltage difference between the gate (1.78 V) and the emitter (1.82 V) of the transistor is below 0.6 (criteria B.2).
  • WAs the voltage provided by the power supply (supercap) drops, the voltage at the gate of the first 2N3904 transistor drops, and so does the voltage supplied to the BBC microbit, which in turn affects the voltage supplied to the gate of the second 2N3904 transistor, which will never opens.
  • This circuit typology therefore cannot work.
Test #8 - Higher power supply voltage
  • We clearly see from tests #3 and following that one of main issues reducing the autonomy of the BBC microbit is the voltage drop between the power supply and the BBC microbit (about 1 V). When charging the supercap to its nominal voltage of 2.7 V, this brings us to 1.7 V at the 3V pin of the BBC microbit, at which the BBC microbit barely lights up.
  • We tried to overcharge the supercap at 4.0 V to measure how much the voltage drop across the circuit would be, but the supercap overheated quite a bit, so we stopped the experiment to prevent explosion.
Test #9 - 1381E characterization
  • Our power supply not going below 2.8 V, it was tricky to check the max and min activation threshold of our 1381E.
  • Using a slowly discharging 10F supercap allowed us to observe that :
    • The output of the 1381E would jump to the value of rougly 2.35 V when the input + pin of the component would reach 2.55 V.
    • The output would sink down to 0 at around 2.35 V on the input + pin.
    • The voltage drop between the + input pin voltage and the output voltage seems to be around -0.15 to -0.2 V.

Discussion & conclusion

  • The current typology of the one-transistor version of the circuit does not give the BBC microbit a long autonomy because there is a mismatch between the nominal charge voltage of the supercap, the minimal power supply voltage of the BBC microbit, and the voltage drop across the circuit. Possible options to overcome these pitfalls include :
    • Increasing the capacity of the supercap to a higher value.
      • We already observed that going from 1F to 10F increased the autonomy of the BBC microbit about 10 folds.
    • Changing the supercap to a higher nominal charge voltage to allow for longer bursts
      • but whould we adapt the 1381x too to have appropriate triggering thresholds ?
      • this means increasing the amount of microbial fuel cells to stack ahead of the supercap to charge it…
    • Changing the transistor to have a lower voltage drop across the circuit.
      • do NPN transistors with a lower voltage drop exist ?
    • Accept the shortness of the BBC microbit bursts…
      • Which might be even shorter once the BBC microbit powers a motor, or any higher load than a LED…
  • The two-transistors version of the circuit displayed here cannot work, as the output given by the BBC microbit does not seem to go above the power supply it gets.
    • To overcome this pitfall, an option could be to try again the two-transistor version similar to the one displayed in the 1381 solarbot circuit by Beam robotics.
    • Would the digital output mode of the BBC microbit go higher than the analog output mode - and higher than its own input ? We can try out.
  • We could also try out what happens when we change the 1381E to 1381J.
  • Paul filled up the 13 batteries a few days ago.
piles


piles_ensemble


  • This morning :
    • Two of them do not work.
    • One has a voltage of 0.13 V.
    • The others have a voltage between 0.55 and 0.76 V.
0.572VTopVIew


  • When combining all in series, Paul measures 5.7 V, which allows to run the BBC microbit displaying text during 1 to 2 minutes !

Video : powerOfTheMudShortEdit

Test #10 - Changing the transistors
  • I tested several transistors I had on a shelf after my unfortunate joule thief experiments.
  • I replaced the 2N3904 with each of them and watched how the circuit performed.
    • Regarding the SS850BBN transistor:
      • The voltage at the + pin of the power supply was 2.81 V.
      • The voltage at the 3V pin of the BBC microbit was 1.98 V.
      • The voltage at the output pin of the 1381E was 2.69 V.
      • The collector-emitter voltage drop was 0.71 V.
      • When running the BBC microbit on the supercap, the BBC microbit stopped to work when the voltage across the supercap reached 2.51 V, after about 16 seconds operation.
    • Regarding the BC63916 transistor:
      • The voltage at the + pin of the power supply was 2.81 V.
      • The voltage at the output pin of the 1381E was 2.71 V.
      • The voltage at the 3V pin of the BBC microbit was 1.98 V.
      • The collector-emitter voltage drop was 0.73 V.
    • Regarding the KSD5041QTA transistor:
      • The voltage at the + pin of the power supply was 2.81 V.
      • The voltage at the output pin of the 1381E was 2.75 V.
      • The voltage at the 3V pin of the BBC microbit was 2.09 V.
      • The collector-emitter voltage drop was 0.66 V.
      • When running the BBC microbit on the supercap, the BBC microbit stopped to work when the voltage across the supercap reached 2.41 V, after about 25 seconds operation.
  • Conclusion : the KSD5041QTA is the transistor which allows the longest run time for the BBC microbit.
Test #11 - Switching the 1381E for 1381J
  • The 1381J voltage detector does not trigger at 2.8 V, probably because the voltage is too low, we therefore get a 0 voltage on the 1381J output, and a 0 voltage on the BBC microbit 3V power supply pin.
  • Conclusion : using the 1381J does not improve the autonomy of the BBC microbit.
Test #12 - Changing the BBC microbit output mode from analog to digital
  • I modified the code of the BBC microbit to have set a '1' digital output on pin 1.
  • When powered through USB, we obtain 3,11 V on pin 1, the same as on (analog) pin 0.
  • When powered through our one-transistor circuit, we obtain 1,75 V on pin 1, the same as on (analog) pin 0, input power pin 3V being powered at 1,84 V.
    • We therefore cannot hope that the BBC microbit will open the gate of the second transistor controlling the line powering itself, as the gate voltage will always be a little bit lower than the emitter voltage in this version of a two-transistors circuit.
  • Conclusion : using the digital output of the BBC microbit does not improve the autonomy of the BBC microbit.
Test #13 - Changing the supercapacitors
  • For some with higher nominal charge voltage ?
  • For some with higher capacity ?
  • Elak needed one week delivery to provide supercaps, so experiment cancelled for now…
  • But, in the shower, an idea occured to me: to increase the operation voltage of the circuit or the overall capacity, I could just pile up supercapacitors !!!
Test #14 - Piling up supercapacitors in series
  • I put two 2.7 V 10F supercapacitors in series, and charged them up to the point where the power-feeding 3V pin of the BBC microbit would be fed with 3.02 V. This means that :
    • The voltage at the output pin of the 1381E would be 3.68 V.
    • The voltage at the + pin of the power supply would be 3.72 V.
    • The voltage between the two supercaps would be at 2.65 V, just below the its nominal charge voltage of 2.7 V.
    • The voltage across one of the supercaps would only be 3.72 - 2.65 = 1.07 V, which will mean that we would not use it at its full 10F capacity.
  • The BBC microbit discharged for a beautiful 45 sec before it went off.
  • The BBC microbit went off because of lack of power (below 1.70 V on its 3V power-feeding pin), but not because the 1381E shut down. We checked, and the 1381E output drops to 0 only several seconds after the BBC microbit went off.
  • Nota bene : this test has been performed with a KSD5041, our best transistor so far, and not with a 2N3904.
  • Another measurement of voltage across each of the supercaps gave a 0.48 + 3.20 repartition, the one closer to the - pin of the power supply being the most charged.
    • The charge repartition is therefore quite different than the one measured on the first test.
    • It therefore seems quite a random repartition…
  • Autonomy of blinking BBC microbit (+ max analog output on P0 and max digital output on P1) with 2N3904 is 37 sec, 46 s with KSD5041.
Test #15 - Piling up supercapacitors in parallel
  • I put two 2.7 V 10F supercapacitors in series, and charged them up to the usual 2.8V minimal voltage of the power supply.
  • We had a brilliant 60 seconds autonomy of the blinking BBC microbit.
  • As the supercaps might start to discharge already when there voltage reaches 2.55 V, which seems to be the light-up “rising” voltage of the 1381E according to our (lousy) characterization (Test #9), we measured the autonomy of the blinking BBC microbit from 2,55 V to shut-down : 30 seconds.
  • Autonomy of blinking BBC microbit (+ max analog output on P0 and max digital output on P1) with 2N3904 is 13 sec, but reaches 58 s with KSD5041.
Test #16 - Determine at which voltage the 1381E lights up during parallel two-supercaps charge
  • The BBC microbit seems to light up when the voltmeter shows 2.67 V on the + pin of the power supply.
  • Paul finalized the rail-based robot.
  • Paul finalized the ceramic container for the second robot, which ended its oven cooking over the afternoon.
  • Paul plugged a 10 F supercap on the 12 working clay MFCs.
    • It charged up to 1.8 V in 6 hours.
    • Afterwards, it only charged 1 mV every 3 minutes.
  • When plugging directly our first robot on the batteries, nothing happens.
  • When connected to the power supply, the robot consumes about 100 mA at 2 V.
  • Microbit alone displays text for about one minute in the same configuration.
  • The 10 F supercap plugged on the batteries reached 2.690 V 23 hours after being put to charge. It had an initial voltage before charge of 0.35 V, when the batteries without any load where giving 5.9 V.
  • This 10 F charge made the first V&A robot run for seven minutes.
  • When the robot stopped, the 10 F supercap was still giving 1.667 V.
  • Paul wants to try to replace the 10 F supercap for a 0.22 F supercap, which would allow 12 seconds operation for 40 times less charging time.
  • Paul is also preparing :
    • a manual charging system for the V&A installation for demonstration purposes
    • a simple switch which will allow us to control ourself the charge/discharge pattern, with terminals for voltmeter measurement
  • Test autonomy while running a motor on BBC microbit
  • Try again the two-transistors topology with parallel two-supercaps charge
  • Testing again the two-transistor Solarbot design