I'm currently evaluating an embedded linux system for a specific use-case with my company.
What's necessary is a lot of specific interfaces, like MIPI-DSI/DisplayPort/HDMI, USB HS (+PHY) and the standard RAM and other peripherals. I've found the STM32MP257FAK3, or more specifically the Discovery Board STM32MP257F-DK from ST to be sufficient for my task and would like to keep working either with it directly or with something very similar - that's where the issue starts.
In ST's license agreement, the commercial use of a DISCO board is prohibited, so buying them directly from ST and using them as components in their designated machines is not an option.
Out of curiosity, I've used the publicly available BOM and manufacturing files for the DISCO board for a lower-volume quotation, just to see where a custom and perhaps reduced board would get me - let's just say a very popular pcb manufacturer would like 3.5k for 5, yes, 5 boards. The components make up for most of the cost, but buried vias and thin traces also do play a part. So just 'orienting' on the public files, slightly modifying them to my needs and having it manufactured as custom design is not economically feasible.
A complete and cost-prioritized redesign is out of the question due to the (comparatively) high workload associated with it. Obviously, industry-common aspects like longevity and steady part supply (not just the MPU but RAM [which Mouser tells me is NRND anyways] and others, too) are also an important aspect.
I have two questions:
1) If the DISCO board is not to be used in commercial products (which, just to clarify, is respected), why is it even CE certified and all the other stuff? I assume it should be used as a reference design, but using it only as a "reference" is, as said, economic mayhem (at least <100 pcs). Why make all this effort and not offer a commercial SBC-option?
2) Where to go now? I'd like to keep working with ST's embedded Linux. I do not have the time nor resources to build a main board for this MPU from scratch. Are there any industry-proven ways to move from DISCO board to finished product as time-efficient as possible?
