Do we need more open source SBC designs?

Do we need more open source SBC designs?

The release of the Raspberry Pi 5 brought to the forefront a quiet issue with many SBC designs, the fight between closed and open source designs. Since the industry has been all over the place as far as open sourcing their designs.

SBCs range the gamut of designs and capabilities, and whether they’re open source varies a lot as well. Let’s look at the Raspberry Pi, which has is an odd animal. Although in general the the Raspberry Pi is considered to be “Open Source”, in reality this depends. You can find limited schematics and designs for the various RPI boards. BUt those schematics are limited, typically to the I/O sections you will interface with. The SoC, Memory and various other parts are not released.

The Raspberry Pi 5 itself doesn’t have any schematics released. You can find a 3D STEP file (for developing enclosures) and I/O information to interface to it, build HATs, and extension boards.

The end result is that people can’t create custom Raspberry Pi boards themselves. That’s a choice the RPI Foundation decided to make, and force users to use their boards. It’s also likely that Broadcom influenced this, as they don’t want small entities to need support and complain when their designs fail. They’re simply not setup to help small customers.

Now, the hardware design itself isn’t the only thing that’s not disclosed. For many years, many proprietary aspects for the Raspberry Pi has been limited.

Could this be a factor to the Raspberry Pi’s popularity? Giving people just enough open source while preventing them from making custom boards if they want to? Possibly. RPI has many industrial customers that want the boards, but many of these customers wouldn’t necessarily want to build their own hardware. It’s difficult, has a high cost and takes quite a while. The RPI 5 has been in design in one way or another since 2016. Clearly, unless you have some very clear requirements, you wouldn’t want to build your own hardware. It’s unlikely you’ll reach enough volume for it to even make sense.

You’ll notice that the Raspberry Pi’s goal is to ultimately sell more boards. Giving away all the secret sauce is not necessarily in their best interest. The one benefit of this is that it enables the foundation to develop more and more boards (and pay millions to design custom chips like we mentioned in our article on the RP1)

On the other hand, the Beagle Board and Beagle Bone foundation designs are fully open source. You can easily find the schematics and source design files (in OrCAD format) in their gitlab repo, even for the latest boards. Some of this is due to the fact that BeagleBones use TI processors that are relatively open source and well documented. BeagleBones themselves serve as a sort of reference design for using TI chips, as opposed to leveraging TI’s development boards (which are usually quite expansive and quite expensive).
Texas Instruments is different than Broadcom and sells catalog parts to the mass market, which means support and designs are much more open because TI is incentivized to sell chips, not boards.

What’s interesting is that the RPI has far outsold the Beagle Board and Beagle Bones, for a variety of reasons. We can’t point to open source as being a large factor in this, but the fact the RPI foundation makes money from selling boards does mean that it can provide significant support.

On the other hand Beagle depends significantly on resources from Texas Instruments.