Comment by CheeseFromLidl | Hacker News Reader

Comment by CheeseFromLidl | Hacker News Reader

Parent Full thread

I’m going to go against the standard advice of this book or that course.

Pick up (broken) equipment and start disassembling it to figure out how they turn A into B [1]. Go down the rabbit hole of hunting down the service manual of the thing or one of its siblings. Look at how the pcbs follow the same pattern in competing design. Look at how all the yamaha, sony, medion, … amplifier/tuners are made in the same way and learn from it. Notice that one that is costlier and has those few quircks in its design. Notice how different variations of a theme achieve the same result, but died out because the tech doesn’t scale or simply proved to be suboptimal. Try to repair your broken equipment by understanding the path that the signals and power lines follow.

Rinse and repeat a few years and you’ll get a grasp on what the innards of an unknown electronic thing looks like without opening it. Then open it and be amazed that there was a different, cheaper, simpler way to turn A into B.

All along the way you’ll experience that most educational resources aren’t actually that good at explaining, or that they follow a different school of mathematical notation, or that they’re really good at explaining this detail but the rest is missing.

Design your own pcbs. Remember that - like software - hardware design is iterative. Remember that - unlike software - hardware iterations cost money.

Hope this helps.

[1] the ways to turn A into B are rather limited and it relies heavily on electromagnetism and conservation of energy.

I think you can take a bit more directed approach too.

You can build a micro-controller with a circuit that controls a stepper motor. Let the motor do something simple/fun. Connect the fun doohikie to give feedback to the microcontroller — e.g. using some kind of an encoder chip that converts motor's rotation amount to numbers that will tell the microcontroller how much to move, initialize the doohikie's start state.

But you can understand a lot more if you don't use the HBridge chip for the motor. Build the bridge circuit yourself. Build your own power supply for the microcontroller too(if you want to).

You can pick a path to go down on and focus on specific parts:

1. For the h bridge there is lots to learn. Designing the operational amp for D2A conversion as well as amplification/signal modeling — which you will need for the motor for current limiting in the h bridge per motor spec for the doohickie you want to power — That will teach you quite a bit about analog electronics and design. What kind of currents you need to protect and how e.g. using opto electronics. Limiting noise from power supply and parasitic noise so that your circuit does not misfire. You will likely need a set up of an oscilloscope, soldering irons and breadboards to prototype. Learn some basics from a book about design then go back to the circuit and build.

2. If you build your own PCB for this. It is a multi month project. You can learn a out CAD and chip layout. But I think you can do this in parts for example you can design the initial PCB only for the digital components and then connect it to a breadboard where you can prototype the H bridge you want.

3. If you choose to learn digital design and embedded system programming then maybe you can build the tougher analog parts for motor control using store-bought components and chips and focus mostly on the programming the microcontroller. That is a totally legitimate part too. You could use an old MCS-51 microcontroller and learn about data and program memory addressing and interrupt handling from scratch.

Reply on news.ycombinator.com