日本語
C****lie
Which is better, embedded or PCB design?
706
1
Apr 08.2019, 15:00:31
Embedded systems is a rapidly growing field. PCB design, on the other hand, is a relatively mature field.Which is better?
- Comments(1)
Upload a photo:You can only upload 1 files in total. Each file cannot exceed 2MB.Supports JPG, JPEG, GIF, PNG, BMP
L****cas
Apr 08.2019, 15:04:10
I would imagine your question is in terms of a career, so I will answer with this assumption. Of course, anyone will predict the answer will be “it depends.”
Embedded systems is a rapidly growing field. In many senses, it is still in its infancy, as we make more mobile and IoT devices. There are endless possibilities that you could focus on. Relatively similar to PCB, you could study the hardware behind FPGAs, ASICs and even newer hardware types (like the CGRA - course grained reconfigurable architecture).
Embedded systems has a lot of architecture, programming, communication protocols/proxies (multicore and multi-processing), etc., considerations since the designer is trying to minimize power consumption, area (size) and speed. Some researchers and students are implementing deep learning on wireless devices, so there are many fun and interesting projects.
PCB design, on the other hand, is a relatively mature field. You might do some CAD design (Eagle) and try to make as many layers as possible while considering the best locations of components. Why are the power supplies on this part of the board and how do the other components need to be mounted to prevent the board from snapping due to a mechanical moment (torque)? How much should components be derated to prevent failure without being over-engineered and costing more? These studies are pretty well known already, so there is arguably not as many new areas to explore (unless you start trying to 3D print circuits).
I studied PCB for undergraduate studies, but I became bored after a few projects. Embedded systems is different, I find myself investigating hundreds of research papers, tutorials, and models for a single project. It is fun to take some EEG application with a data set and compare its energy usage on multicore FPGAs, then some ARM9-ARM9-ARM15 setup in OpenMP, or exporting C++ algorithms like ApEn (FFT), Hurst, and PSD for a Butterworth filter to CUDA and running on GPUs with AWS. Then feeding some training set and doing machine learning to find clustering or patterns.
I will never be bored with embedded systems.
Embedded systems is a rapidly growing field. In many senses, it is still in its infancy, as we make more mobile and IoT devices. There are endless possibilities that you could focus on. Relatively similar to PCB, you could study the hardware behind FPGAs, ASICs and even newer hardware types (like the CGRA - course grained reconfigurable architecture).
Embedded systems has a lot of architecture, programming, communication protocols/proxies (multicore and multi-processing), etc., considerations since the designer is trying to minimize power consumption, area (size) and speed. Some researchers and students are implementing deep learning on wireless devices, so there are many fun and interesting projects.
PCB design, on the other hand, is a relatively mature field. You might do some CAD design (Eagle) and try to make as many layers as possible while considering the best locations of components. Why are the power supplies on this part of the board and how do the other components need to be mounted to prevent the board from snapping due to a mechanical moment (torque)? How much should components be derated to prevent failure without being over-engineered and costing more? These studies are pretty well known already, so there is arguably not as many new areas to explore (unless you start trying to 3D print circuits).
I studied PCB for undergraduate studies, but I became bored after a few projects. Embedded systems is different, I find myself investigating hundreds of research papers, tutorials, and models for a single project. It is fun to take some EEG application with a data set and compare its energy usage on multicore FPGAs, then some ARM9-ARM9-ARM15 setup in OpenMP, or exporting C++ algorithms like ApEn (FFT), Hurst, and PSD for a Butterworth filter to CUDA and running on GPUs with AWS. Then feeding some training set and doing machine learning to find clustering or patterns.
I will never be bored with embedded systems.
