This week, as well as the previous ones for most of us, was devoted to PCB design. During this work, several changes were made to our design. Here is a summary and some details. The overall structure remains the same as what Marc described in this post. Let’s go through the modifications for the 5 PCBs.
We have made many modifications to this PCB.
- First concerning the Wifi module : we wanted to use the AMW006 module because the OS seemed convenient and it is supposed to have less bugs than the ESP32. But we discovered that the compilation chain has to be used in windows, so we decided to use the ESP32 instead (the DevkitC so as to flash easily).
- IR detection : our IR tests were promising, but we realized that we will probably have a lot of troubles with all the PCBs between the receptors and the emitters. That’s why in addition to the powerful transmitters under the hull, we will place other side-looking IR LEDs under the PCB (enough to cover 360°), with the receivers, as in the diagram below:
- Furthermore, we change the power transmission voltage as explained in Marc’s post. The main board will receive 5V from the balls bearings, transmit 5V to the processor_PCBs and have a 3.3V switching regulator for the processor and IR LED.
We are still working on placing and routing this PCB, and we will inform you of the troubles we have soon.
There are just a few changes to these PCBs. They will have a more optimized shape so as not to overlap :
They will be able to drive 8 LEDs, have capacitors (4*100nF) to store more energy for the LEDs, a 3.3V LDO-regulator for the processor, and a lot of pads :
We are still working on this design :
- For the 5V power supply, we aren’t sure of the best way to connect PCBs together. Right now, we have 4 SMD pads, two for VCC5 and GND supply, the other two right under them for to send VCC5 and GND on to another PCB. Ideally we would be to be able to have four through-hole pads, and we manage to make it all fit we’ll do that. If we cannot fit four additional through-hole pads, we are thinking of using two big through-holes pads instead, and either use T-shape connection between wires ( as explained in this video), or solder two wires on the same through-hole pad.
- We are also preparing a back-up plan for the SPI connection between the processors-PCB because we don’t know if the connection summarized in the following drawing will work.
That’s why we will make this circuit, allowing us to choose between a direct connection (like above, with one master on the main_board and several slaves) or a SPI connection with one master and one slave between each processor-PCB.
If it turns out it is necessary to regenerate the signal with the processor, we’ll remove the resistors 0-ohm resistors between the two pads, leaving the outgoing signal to be generated by the processor. On the other hand, if the precaution proves unneeded, we’ll remove the 0-ohm resistors on the (now unnecessary) wires coming from the processor, so as to avoid connecting two output pins together.
- Finally, we would like to replace the 8 pads for SPI connections by 2 small connectors of 4 pads each. It would be much convenient, but we haven’t found anything yet.
As Vlaya explained in her post ‘Final LED tests’, our idea was to place 16 LEDs (ASMG-PT00-00001) on this PCB to illuminate 16 petals, and a large LED (LE RTDUW S2WP) in the center which lights up the remaining small petals, and to drive everything with two processors.
The placement result wasn’t encouraging and after discussion with Alexis, we finally decided to put only one processor on the top pcb to drive the most internal LEDs, and that the eight most external LEDs would be driven by the closest processor-PCBs. For this reason we updated the processor-PCBs to be able to drive 8 RGB LEDs – previously only 7 were needed. We also considered the idea of placing a 13th PCB-processor under the top_pcb to drive the external LEDs but it was easier to have one processor on the Top_PCB than 9 x 3 additional pads to drive the LEDs and 9 x 3 additional wires to control them, on top of the LEDs already driven by nearby processor-PCBs.
As explained earlier, we need a 3.3V-regulator to supply appropriate voltage to the CPU.
We also need to put as few components as possible on the edges of the upper face of the PCB (only the LEDs) because Vlaya has prepared a 3D-printed mask that will serve to guide the light from the LEDs to the petals as on this picture :
We have placed and routed with those modifications, and we decided that this PCB will be a 4-layers PCB. Alexis finished the review and correction of this PCB this week 🙂 There are still components on the upper face, but hopefully, they won’t hinder the mask.
No changes for this one, it still has a LED (a large one : LE RTDUW S2WP), transistors and resistors, a capacitor (100nF) and pads for 5V, PWM signals and GND arrival.
Only two changes on this PCB.
First, as explained by Marc in his post , it will have a 5V switching regulator to transmit 5V through the balls bearings.
We also realized that Alain Croulebois (the school mechanic) wanted to put a 8mm aluminium plate between our IrDA transceivers. We wanted to solve this problem by drilling a hole in the plate, but Alexis advise us to add a ESP32 also on the fixed PCB, to be able to communicate with the main board even in case something goes wrong with the IrDA.
We will very soon update you on a lot of other subjects and questions regarding IR, mechanics and power transmission, ESC, Wifi… Any opinion or help would be very much welcome 🙂