Hi again !

Friday was supposed to be a great day, we were supposed to finally receive the last components needed to solder our PCBs.
Unfortunately, UPS didn’t plane it that way, and due to an “exception” we will not receive these last components untill Monday …

So as I was a bit pissed of, I didn’t feel like coding and thus started soldering the LEDs for our balloons (you saw only one ribbon), I took me quite a while (about 10 hours) but had the merit to change my mind as it allowed me to listen to some good music 🙂

I was rather happy when I tested the ribbons and realized that I only had 3 bad solders on about 2000, and 7 out of the 9 worked perfectly on the first attempt. I have to admit it made me rather proud.

Here are some picture of the assembling of the ribbons :

IMG_20140411_203257 IMG_20140411_203303


IMG_20140411_224042 IMG_20140411_224156

Some kind of time lapse, of a ribbon assembly :

IMG_20140411_224343 IMG_20140411_224600 IMG_20140411_225024  IMG_20140411_225210 IMG_20140411_225407

And the final result :



While soldering I also had a hangout with drix who explained me how to properly power the TWI (and by the way making me realized that I should have started by looking at the schematics instead of trying all the wiring combinations that came to my mind ^^), I’ll be working on it this afternoon hopefully it will work this time ^^


LEDzep – Title of the post

Hi everyone

We won’t have our PCB until monday evening for sure, inflatable slide so we were all a bit disappointed this friday.

Anyway, these past days, I’ve been working aufblasbare wasserrutschen with Félix, mainly, and I assume my work has already been pretty much summed up in his posts.

In the mean time, I’ve been doing research about the inflatable motion feedback of the balloon. As forecasted, it’s not going o be a piece of cake, but we’ll make it work.

LEDzep – More nRF51 and Android!

Hi everyone,

Since last time, I continued to work on Android. I am now able to exchange any kind of characteristic using read, write or notify operations (we won’t need indicate) between the nRF and Android. The Android app is still very basic, it is able to connect to a selected BLE device and it contains some buttons and text fields to let the user exchange data with the nRF.

I then began working on the nRF chip. It took me some time to get started as the code is quite complicated. I was able to create the localization service and to add the compass and altimeter characteristics. I am now working on transmitting the data coming from the UART to Android through BLE.

We also received our PCB today! Thomas also did a great work on the balloon’s structure. Hopefully, we’ll have some remote controlled flying balloons by next week.

Stay tuned!


LEDzep – TWI, LEDs, nRF, refactoring and presentation

Well it’s been a while, but I haven’t been inactive !

This Week-End I mostly (and only) worked on the final presentation as I was busy at a hackathon Saturday (which served as a great day off with lau and former ROSE students).

Since then I read some other application notes on the nrf51822 (and managed to remember the full name of the chip doing so ^^). Thanks to that I’ll (hopefully) be orders of magnitude more efficient writing code for it.

Later we planned to work with Thomas on using the Digital Motion Processor of the mpu9150 (which we overlooked at first), which could do all the angles calculations and tap detection, and thus saving us a lot of time. To do so Alexis and Sam lent us two TWIs (a great board made by drix !) which already has a nRF51822, a mpu9150 and a serial port : all we needed to get to know the DMP. Unfortunately we haven’t managed yet to find a way to lower the voltage of our JLINK probes, as the nrf has to be flashed with a 1.8 voltage and our probes are configured to give 3.3V. We tried quite a long time (about a day) different tricks to find how to do so, eventually we will have to write a jlink script to reflash our probes (we hoped to find an option to do so in command line, but no success so far).

Today, I remembered that we had a PSSC to fill : having a balloon structure with LEDs, so I ‘improved’ Thomas great structure by putting LEDs on it. So I had to spend a while soldering, but the result is rather nice, as you can see 🙂

Capture d'écran de 2014-04-10 02:42:11 IMG_20140410_003917 IMG_20140410_003929  IMG_20140410_004220 IMG_20140410_004439


EDIT : as Alexis asked, here is a little video


LEDzep – WSIP (Work Still In Progress)

Hi there

This week-end, I’ve been quite busy preparing Monday’s presentation, which will finally take place this Friday. I wanted this presentation to be inflatable water slides insightful and good looking, and it took some time before I was satisfied.

Anyway, yesterday and today I worked on the TWI, ( for the bounce house curious one). The goal being to make it work and particularly to enable the DMP (Digital Motion Processor), the magical part of the MPU9150 which would do almost all the work …

But it doesn’t seem to work yet, so I spent aufblasbarer hindernisparcours some time cleaning the project repository.

The PCB will soon be on thei way and we’ll have to be ready to welcome them as they deserve it.

More news as soon as the TWI works.

LEDzep – MPU & Production costs


Lately I worked on two aspect of the project. First I tried to recover datas from an IMU device throught and I2c connection in order to filter then accelerations, angular accelerationand and compass angles of our zeppelin. Indeed, We want to use the acceleration datas in order to detect collision. Angular acceleration will help us projecting the compass angles. And we will be able then to accès to the absolute angle of our balloon. Thus said, I have not able to properly configure the right MPU registers. And I realized that I had forgotten a key element present in my model that I must also configure: our mpu-9150 is already aquiped with a Digital Motion processor hardware accelerator engine. It’s a good point, this means that we will not have to filtre 🙂
We decided today to resume a tutorial proposed on the InvenSense forum. It seems to configure well the DMP processor.

Secondly we also worked on Scénarios and on budget. We tried to find as accurately as possible the unit cost of our balloons. We based it on a 10,000 units order. We evaluate the cost of each componnents, of the plastic for the structure and of the delivery. The total amount is 34,93 euros and it is not including the cost of helium (if we rely on Air Liquid fares, we will need 3,6 euros per balloons of 0.17 m3).LEDZEP - coût de revient

We decided then to possibly offer two commercial proposals: First we will sell the balloon by 10 for little firms that want to acquire for cheaper something to make their own show. Secondly we wan’t to sell a real kinetic art show for bigger firms that want it for a bigger and punctual event.


LEDzep – WIP

I haven’t been very talkative in a bounce house with slide while, and it has been kind of a slow week.

I read a lot a stuffes about BLE and custom services, and also studied the code of Benjamin / Félix providing an UART for the nRF. In the aufblasbare spiele mean time, I made me up to date with the current state on our aestetically flawless android app, and inflatable obstacle course currently working on custom services and characteristics.

That’s all so far



LEDzep – Some work on the BLE

This weekend and the beginning of this week I worked with Benjamin on the nRF51 chip, we mostly worked on having a working UART while still being able to use the BLE (which wasn’t a piece of cake as we cannot do multithreading with Nordic’s software). So we had to use interrupts, we were later really pissed of as we realized we had a working code which didn’t throw any errors even though we didn’t observe any communication with minicom, until Benjamin realized that we forgot to put a pull down resistor on the CTS pin …

Today I did some research on the BLE when I discovered (rather lately to my taste), Nordic’s application notes which were much more clearer than the source code. I managed to understand how to do simple advertising so that I could deal with the altitude issue (with one altimeter advertising its altitude).

And as you have been very nice here is a picture of Benjamin’s and I working station (as you can see we are ready to kick some asses) :


I then debugged the LED strip code and it is now functional with SPI (damn you hardware bug).

LEDzep – Pictures !!!!

Be careful, this post will be full of pictures !

First of all, we send our PCB schematics to PCB Pool yesterday afternoon and we received the first pictures of it yesterday night 😀

ledzepBIE533ae6c20a6d0_02_seite2 BIE533ae6c20a6d0_02_seite1 Dzep2 Dzep1


It’s being made much faster than we even hoped, and that’s awesome !

I also worked a while ago on the barometer in order to get the altitude of the balloons and the results were rather encouraging. Following Sam’s advice I let it run for a night (between 4am and 11am) and here are the results.


As you can there is a little issue about the stability of these measures, hopefully after seeing this graph I did a new measurement and the altitude stayed the same, which means that this bias is due to the pressure and not a problem with the sensor. Using a second barometer kept on the floor will allow us to fix this issue, as both will have the same bias (as it’s due to the pressure and not the sensor), and the difference between the two altitude will give us the real altitude of the balloons. We will connect it to a computer and advertise its measure with a BLE dongle and each balloon will be able to easily determine its altitude.
Furthermore it would allow a fun way to control the altitude if the balloons by simply moving the reference barometer (the one on the floor), lifting this one would have all the balloons fly higher 😀

LEDzep – LED ribbon & battery level check


Today the PCB was finished. We sent it to the production line and we look forward to receive it. I worked Today on two functionnality of our system. First, I completed the code that now allows us to control the ribbon color LEDs. We use a SPI output in order to use the cascading port transmission control. We send by regular intervals, into a single line all RGB codes for the LEDs and a “reset” signal activate ribbon with the colors set. I worked then on ADC input, in order to recover the voltage delivered by the battery, to monitor it continuously and to alert if there is any discharge. Thus, as soon as it crossed a certain threshold we can land the Zeplin and prevent any underload problem with our Lipo battery. We are moving in order to be ready when the PCB will arrive. We want to have the most pieces of codes allowing us to control the devices.

Good evening,