I have mentioned before that I have had a hard time to produce the Freak filter due to the semiconductor shortage. The parts used by the Freak are scarce. Some of the parts are available but they have a very inflated cost. I have been monitoring my usual part suppliers to see if they become available. Little by little I managed to purchase some of the parts. After a few redesigns of the main board and manually assembling the module I have a new Freak module.
Board redesigns
I designed the original Freak board back in 2018. The intention was to have a versatile board that I could use in different designs. It had a bunch of analog inputs (for knobs and modulations) and stereo audio inputs and outputs. The Freak board did not use all the features.
This board was made using large SMD components (0805) and it was somewhat easy to assemble by hand (for an experienced builder). In the beginning of 2020, I started redesigning the board using smaller components and also removing the unused components. I also moved my design from EasyEDA to KiCAD. It was a bad timing because the pandemic made impossible to produce the Freaks with the redesigned board.
Since then I have redesigned the board 5 times in order to adapt to the available components. When one of the components was out of stock, for example, an op-amp in a certain package, I would find a replacement in a different package, redesign the board and make a prototype. By the time the prototype was ready, some other component would be out of stock, forcing me to remake the design. Here you can see a picture with the old board and some of the redesigns I made.
The only strategy that would allow me to produce boards was to look for any compatible parts, buy them immediately and redesign the board later in order to adapt to the available parts. But even if you pay for the parts it does not mean that you will receive them any time soon.
Mouser loses the parts
One day I saw that some parts I needed were available in Mouser so I immediately placed an order. Everything went fine, some days later I received the box and I opened. I was a bit shocked... the box contained a bunch of inductors and none of my parts.
It turned out that the shipping clerk at Mouser made a mistake and sent me the order destined for a different person. Most probably my order was in the hands of the other person. I quickly contacted Mouser thinking that, because they are a big American company they would quickly solve my problem. It took them two weeks to answer back. After that it was more than a month until they sorted out my problem. I had to provide a bunch of evidence that I was in possession of the incorrect components and they made me ship them back. Unfortunately, it seems like the person receiving my package did not returned it. So when finally they shipped again my parts, some of them were already out of stock and I did not receive them. They are going to ship the missing parts some time in 2023.
Those missing parts forced me to find an alternative version and redesign the board.
Even after this big screw up I still love you Mouser.
Manual assembling
In the good old days before the semiconductor shortage, I would create a list of parts alongside with the buying links and my assembly house would order all the parts and put together the boards. This became impossible because of the uncertain landscape which forced me to track and buy the parts from different suppliers and with different time frames. That made me think that I could assemble the boards by myself, just like I do when producing small batches of modules. That way I could mix and match boards using some of the alternative version of the boards if necessary. That way I would not have to pay the assembly house to build different variations of the board using different components. If you have some knowledge about electronics manufacturing you may know that the most cost effective way of producing boards is in large batches where the build machines have to be setup one time.
I have built a lot of boards myself so I felt confident that I could assemble them without mistakes. The main problem was the sheer volume of components that I needed to pick and place by hand.
My day consisted of waking up very early in the morning and applying soldering paste to a batch of 5 boards. Then I would manually place every single component of the boards while I listen to some music or an audio book. Some hours after I would have the boards ready to bake. In order to simplify this task, I added small markers (per component value) like, circles, squares and triangles next to each component in order to locate them quickly. For example, I know that every circle is a 100nF capacitor.
Additionally, I used an interactive BOM to highlight all the locations. I had a lot of references to place the components. But at this point, I know by heart the location of every component of the Freak boards.
For next time, I'm planning on building a pick and place machine that I can use to build modules on special occasions like this.
Once I had all the components placed, I had to solder the boards. For that, I built a SMD reflow oven. I used a small commercial oven and added a solid state relay. I wrote the firmware in a couple of days and ran tests in order to get the "perfect" soldering profile for the solder paste I was using. Here are the first two boards coming out of the oven and smelling better than a baguette.
After carefully assemble all these modules, I have the sensation that I should have been surgeon instead of an electrical engineer. These modules are really handcrafted.
Defective panels
My last issue when building these modules was that, after I regenerated the design files of the panel and sent them for manufacturing, something went wrong. The panel I received came missing the hole for the screen.
This mistake forced me to reorder a new panel and reuse some of the spare panels I had. Hopefully in the future I will have the tools to make the holes by myself so I don't have to trash these nice panels.