An AMD unit is contained inside a grey enclosure made of metal sheets which are electrically connected to the system’s ground.
Inside the AMD are two main PCBs containing 200 components, mainly Surface Mounted Devices (SMD), whose sizes vary from 0,5mm to several centimeters :
–The main PCB (MPCB), it controls all the other elements, except for the sensors. It contains components such as a CPU, USB and motor drivers, and other connectors to communicate instructions to the connected devices.
–The secondary PCB (SPCB) : It controls the sensors, then forwards the measurement data to the MPCB. It is made of components such as a CPU and multiplexers, whose role is to relay the informations between the CPU and the sensors.

The enclosure was designed by us on Front Plate Designer, a software used to design panels by taking into account the structure, the materials, the colour and the finishing. After designing them, we sent the files to get them manufactured.
The other pieces were made by my colleagues, with our CNC machine. The pieces were designed with RhinoCAM, on Rhinoceros 3D. The software sends the 3D model to the CNC, as well as some instructions like the zones to be milled and which actions to perform. The main actions are :
-Pocketting : which consists of removing matter at the center of a material by “digging”.
-Outlining : which consists of removing matter while following a line, like a pair of scissors would.

One of my main tasks was soldering the components on the PCBs. In order to place the components, you first have to apply the soldering paste on the pads with a stencil. Placing the components, and especially the chips, requires great precision to avoid merging the pads together or missoldering a pin.
Once the components are placed, you can start putting the PCB inside a reflow oven and letting it cook. You might have to be careful about where to place the PCBs in the oven, as the components can swell if the hot air hits directly on the components. After cooking them, you might need to fix some soldering, with a heat gun or a soldering iron.

Once the SMD soldering is done, you might have to solder through-hole components, such as connectors or USB ports. For this to be done, you need to melt some solder wire on the tip of the soldering iron. By melting the soldering wire on the iron, the drop of melted solder with coat the pin and conduct the heat better. Then, heat the pad with the iron so that the solder melts on the pad and the component’s pin. To have a good soldering, you need the joint to be cone shaped instead of round shaped, otherwise there would be no contact between the pin and the pad.
At the beginning, I needed 5h to solder one 150 components PCB, I can now do five in 5h.

Once the PCBs are done and working, we need to connect them together using cables. I was also tasked of making the cables and connecting everything. I worked on three main connector types : AMP, Molex and Micromatch. In order to crimp a cable, you must strip the wires, then put the crimp in the crimp tool, insert the wire that you need to crimp and press it with the tool and insert it in the connector.
Finally, I’ve also soldered wires together or to components. You first have to strip the wires, put some solder on them and heat the solder while bringing the wires together. You have to be quick because wires are heat conductors and can burn your fingers. After the cables are soldered, you can use I heat shrinking tube to isolate the wires.

The other main task of the project was fixing the problems. Fortunately, most problems were easy to solve with an oscilloscope, but some of them were bad enough to damage the system, and sometimes forced us to start over. Here are two examples :

One evening, as we were done assembling one unit, we turned it on. After a few seconds without response, a thin line of smoke escaped the CPU of the MPCB. To see where the problem was coming from, my colleagues suggested changing the PCB and trying again, but I decided to be more cautious and wait for the next day. Then I disassembled the entire unit and tried the peripherals one by one with a new MPCB. Doing so, I understood that the problem was not caused by any of us, but by the manufacturer of the LED buttons that were used for the control panel. Indeed, one button had the 12V pin (needed to power up the LED) and the output of said button connected. The output being directly connected to a 3,3V pin of the CPU, it damaged it. After this, I decided to test every button independently and noticed that 30% of the button had some kind of defect.

We also discovered a problem that was much harder to diagnose, and that I’ve not been used to. There is a port used to flash the device, and sometimes touching it with the flashing device causes the AMD to power off and on instantly. It is at the very least annoying, but can become dangerous if done repeatedly. Indeed, powering off and on a device before the capacitors have time to discharge can damage a circuit. We tried fixing the problem, but touching the port with the flashing device doesn’t often cause this reboot so it was difficult to recreate the problem. We tried changing the port connector, the MPCB, and the power supply, but the problem remained. By studying the AMD’s power supply, we realised that it was a floating ground transformer, as is often the case with laboratory equipment. That means the ground of the device is not connected to the Earth. And we also realised that the control panel being varnished, the port was not connected to the device’s ground through the enclosure, meaning that the electric potential difference from the flashing device (that is connected to the ground) had to go towards the MPCB instead of to the enclosure. We then removed the varnish near the port, and the AMD was working perfectly again.

The project was considered by the team as one of the hardest they’ve made. The reason being that some features (that I can’t disclose) caused a lot of problems, especially mechanical ones, as the precision needed for our project to work was too difficult to obtain with our current tools. Furthermore, we are in the middle of an electronic components shortage that would’ve made us wait several weeks to receive components if we hadn’t found a smarter way of doing things.
We were also severely understaffed, which means that sometimes I was alone in the workshop doing the troubleshooting.

It was difficult being by myself but also very educative. I never troubleshooted such a complex PCB before, and was surprised about how fast I got the hang of it. I also learned how to use several tools, such as reflow ovens, milling machines, and many more.
I learned a lot during this project, not only about electronics or mechanics, but also about myself. I had no problem staying up late and working hard with no breaks because it was something that I enjoyed doing, and it convinced me that I have made the right choice working in engineering, and even more so in a workshop.