I swear I could connect

Now that the Hall effect sensor is working and while the coils are still being tested with the marbles, we started working on our WiFi module. As explained in our project’s presentation, we want our devices to be able to communicate through MQTT with a backend and with other Touch.

The WiFi module :

To do our test we took an ESP32 Wroom-32 Dev kit C. The main purpose of this module will be to connect to the WiFi, receive through MQTT the images that the box will display and then transmit them to the STM32. In some cases, the STM32 will send to the ESP the state of the box and it will have to transfert those data through MQTT.… Read more

3D printing : behind the scenes

As mentionned in A solid foundation , we printed our first full-scale sculpture last friday, and it was a success.

We were relieved because there were multiple factors that were not 3D printing-friendly :

  • Our scultpure needs to be hollow, because we need to be able to put all the PCBs and LEDs in it
  • Not only should it be hollow, it needs to be translucent, and thus very thin
  • The thickness should also be relatively constant in order to have uniform lighting
  • The sculpture has several parts which are almost horizontal

Why were these problematic ? Well, let’s see how 3D printing works


The first step in 3D printing is to create a .stl… Read more

Let me hear your vox

For nearly two weeks now I’ve been working on a voxelizer, to convert a 3D model into an image that can be displayed by LitSpin. The goal of voxelization is simple: we need to display an image on a grid of leds, which means that the input image needs to be divided into voxels (3D pixels), each voxel representing a led.

Our grid looks like this:

The number of voxels corresponds to our desired resolution (20 circles, 128 angles and 32 leds from top to bottom).

The voxelization algorithm consists in tracing rays across the model to detect intersections with the triangles of the model.… Read more

We keep feeling it

We previously tested a Hall effect sensor that, was working, but which gave us a voltage from which we could get the marble’s side. The issue with this is that in the final box, with all the marbles and the coils, we could have had strong pertubations that could alter the data. To avoid this issue we decided to use a sensor that would send the data through a communication protocol that could resist. We chose the TLE493DW2B6A0HTSA1 Infineon Technologies. This sensor sends its data through the I2C protocol so it should resist to our perturbations.

Let’s connect it !Read more

Final LED tests

Since we started doing research on which LEDs to choose for our project, there three different LEDs / type of LEDs we considered:

  • The LE RTDUW S2WP , requiring 1.4 A
  • The ASMG-PT00-00001, requiring 200mA
  • Four similar SMD PLCC 2 (the package compatible with the flexible waveguides we were considering) LEDs: ASMB-MTB1-0A3A2, HSMA-A431-Z50M1, CLM2D-GCC-CC0F0783 and ASMB-BTE1-0B332, all requiring around 20mA

The LEDs for the PCB petals (see our latest architecture post) were to be either the LE RTDUW or the ASMG-PT00, and the LEDs for the top PCB were to be the ASMG-PT00 or one of the four SMD PLCC2 LEDs.… Read more

A solid foundation

Last week we met with Alain Croullebois, the go-to guy for mechanical questions at the school. We went to his workshop with the motor we received during the holidays : this Turnigy Multistar 4225-390Kv 16 Poles Multi-Rotor Outrunner. We discussed what the physical structure of the Phyllo could look like.

Here is a global diagram of what we came up with :

The mechanical structure of the whole Phyllo

The Base

For the base of our Phyllo, he suggested stacking two 30cmx30cm aluminium plates 8mm thick, with a separation of 6cm between the two plates maintained by 2cm wide pillars at each corner.… Read more

IRoning out the Phyllo detection

Back to school after the holidays, we ran tests to validate our idea for detecting other Phyllos with IR radiations. Quick reminder, we had ordered two powerful IR transmitters (LTE-R38381S-ZF-U and SFH 4441) and several receivers, some in AGC2 (TSOP2256 and TSOP4856) and others in AGC4 (TSOP4456). Signal management is a little different depending on the type of gain control (AGC): some receivers are more suitable for noise reduction and other lower detection times.


We will briefly go through the IR detection protocol once again. 

First of all, Phyllos give themselves a unique identifier by communicating via Wifi.… Read more

Flipping Test n°25158636

Back to our 2 marble-box

After enjoying nice holidays knowing we had a well working configuration and current value, it was time to get back on it and move further ahead.

Or we thought so. We tried to have it work again with the same values but it did not work. So we needed to some more tests.

Test test test

Alexis has told us to do some measures : we needed to measure the current rise time in the coil, the overvoltage and the voltage drop.

First of all we measured the overvoltage when brutally disconnecting the coil from the power source.… Read more

New Year, new skin

Happy New Year everyone, hope you enjoyed the holidays !!

Today we are going to talk about the latest trends of our hardware architecture.

New PCB disposition to drive the petals

We used to try to drive each LED with its own processor. However we could not manage to get every component to fit on the PCB petals. Moreover, the processors were underused as we barely needed half of their timer outputs. Therefore, further inside the Phyllo, we decided to add a layer with slightly bigger PCBs (let’s call them the petal controllers). They have a STM32F207VIT6 which can drive up to 10 LEDs.… Read more

FPGA architecture, episode 2

The architecture of the FPGA has evolved since the last post.

General architecture

First, since we don’t plan on controlling the rotation speed from the HPS anymore, this feature has been removed from the architecture.

Moreover, the SCLK and GCLK are now generated by each of the 20 LED band controllers since we have enough ouput pins on the FPGA. This results in a simpler design and better signal integrity on these two signals.

Each LED band controller controls a LED driver. More information on the way this driver works can be found on this post.

For the inner memory of the LED band controller we use a dual port, dual clock synchronous RAM, which is proposed in Intel’s recommended HDL coding style for memory inference.… Read more