In Preparation for the AVNET MiniZed (Xilinx Zynq XC7Z007S SoC) RoadTest. What Do You Need?

I see the tradeoffs of MiniZed versus a simple FPGA board to be similar to Raspberry Pi versus a simple microcontroller board like Arduino or ST Discovery.  In terms of raw power for the price, MiniZed and RasPi (especially including VideoCore IV) are clearly superior.  However, it requires a great deal of knowledge to use that power effectively.  Most people use RasPi as a tiny mainframe, running a mainframe operating system like GNU/Linux.  This means that accessing I/O devices is difficult: just giving yourself permission to access memory-mapped devices is a challenge for the new user.  I can only imagine the amount of code required to initialize RasPi, and I've never tried to write and download my own VideoCore IV software.  So you have a great deal of computing power, but actually using it is a steep learning curve.

In contrast, with a microcontroller like an STM32 or Cypress PSoC 4 it's really easy.  The GPIO is fully documented and you just read and write memory locations documented in the data sheet and tech ref manual.  You can blink an LED with zero lines of code if you come in through the well-documented "single wire" debug (SWD) port.  For the most part, it's simple and clean and like playing with a PDP-11 minicomputer.

Similarly, it's really easy to understand a simple FPGA like a Lattice iCE40.  It's a toy compared to a Zynq, but for learning FPGA principles it's a very educational toy.  With an Xilinx FPGA newer than a Spartan-3A, you face a steep learning curve where even simple things are complex and you have multiple tech ref manuals to cover clocking, block memory, I/O pins, etc.  Zynq, Spartan-6, and newer FPGAs are very powerful and have attractive prices for their capabilities.  But they can be pretty daunting for a new user.  Like RasPi, you can ignore the underlying complexity and work at a high level.  But then you're not really learning what's going on.

JMO/YMMV

I see the tradeoffs of MiniZed versus a simple FPGA board to be similar to Raspberry Pi versus a simple microcontroller board like Arduino or ST Discovery.  In terms of raw power for the price, MiniZed and RasPi (especially including VideoCore IV) are clearly superior.  However, it requires a great deal of knowledge to use that power effectively.  Most people use RasPi as a tiny mainframe, running a mainframe operating system like GNU/Linux.  This means that accessing I/O devices is difficult: just giving yourself permission to access memory-mapped devices is a challenge for the new user.  I can only imagine the amount of code required to initialize RasPi, and I've never tried to write and download my own VideoCore IV software.  So you have a great deal of computing power, but actually using it is a steep learning curve.

In contrast, with a microcontroller like an STM32 or Cypress PSoC 4 it's really easy.  The GPIO is fully documented and you just read and write memory locations documented in the data sheet and tech ref manual.  You can blink an LED with zero lines of code if you come in through the well-documented "single wire" debug (SWD) port.  For the most part, it's simple and clean and like playing with a PDP-11 minicomputer.

Similarly, it's really easy to understand a simple FPGA like a Lattice iCE40.  It's a toy compared to a Zynq, but for learning FPGA principles it's a very educational toy.  With an Xilinx FPGA newer than a Spartan-3A, you face a steep learning curve where even simple things are complex and you have multiple tech ref manuals to cover clocking, block memory, I/O pins, etc.  Zynq, Spartan-6, and newer FPGAs are very powerful and have attractive prices for their capabilities.  But they can be pretty daunting for a new user.  Like RasPi, you can ignore the underlying complexity and work at a high level.  But then you're not really learning what's going on.

JMO/YMMV

John,

I don't know the STM32 but I can say about the PSoC4 that it's not so easy. I am thinking to a maker, someone used to make projects on Arduino boards or similar and I imagine that approaching a project on a PSoC model by Cypress can found it extremely difficult. This is the reason that despite the rumors when Cypress launched the PSoC ecosystem supposed to be the "Arduino killer" this never occurred. I think for this reason.

Take in account that I am not an expert of FPGA but I am almost sure that if a system offers wider options I tend to prefer it as expecting more flexibility.

It is true that in most of the cases (and projects) also the Raspberry Pi is used with just its basically features; but occurs also with the Arduino - just to mention to most famous micro controller. It is a sort of false positive; reality is that you can find a lot of preset projects easy to adapt with few changes, something totally different than proficiently using a micro controller IMHO. For example robotics: you can make a relatively complex robot based on Arduino but not because things are easy; the reason is that you find all the libraries you need for download, the designs for 3D printing the parts if needed, the ready-to-use Arduino shields and more. I bet that the number of Arduino makers or "inventors" able to make their own C library or create some new are under the 10-20% of the makers you can find.

The same happens with Raspberry Pi. With a positive difference: this embedded linux board has more flexibility and more options to evolve and has a good support to be used in the educational world.

Enrico

Enrico, maybe this is because many makers know software programming well but aren't strong on electronics principles.

For the PSoC, the real power comes when you understand comparators, opamps and the likes.

For FPGA and also to get the maximum out of the PSoC, digital electronics is a required skill.

I think that these skills are less common with makers than programming skills.

(I also think that with little programming knowledge, google and a debugger, most people can make something that works on a cpu without external help - not so with an fpga)

Perfectly agree. The limit (if we can say it is a limit) of FPGA is that also "something that do something" is not obvious with FPGA

Enrico