Good day,
I would suggest you review and analyze a number of Industry standard controllers and the Industry itself before you go too much for forward with your hardware design. For controls whether it be for commercial or residential applications, there are a number of important considerations along with various Governmental approvals that may be required. Building hardware is very straightforward, but these other items are considerably more challenging to address.
With regards to the hardware’s core processors… The RPI CM4 is a great platform, but has a number of issues that may be a concern to you. The first is that the CM4’s design uses connectors that were never designed for how they are used on the CM4/base board type of application. In fact, the manufacturer (Hirose) specifically says this. Dual fixed placement connectors and respective mating connectors are not designed to accommodate the vertical and horizontal stresses when the CM4 is mated to a main board. This issue is amplified do to the very fine pitch of the connector which have very precise tolerance which makes it challenging for assembly. In a dual connector arrangement, Hirose provides a separation (x) error of 0mm… and a vertical error of+/- 0.01mm. Complicating things further is that the connector design has an “overhang” which makes it extremely difficult to touch up or correct if there are solder reflow issues. On my side I tested 10 assemblies with high $ automated assembly equipment only 8 required no rework and even these I noted mechanical issues that were a concern for me (e.g. potential CM4 dislodging and/or problems with long term connector contact).
Another key factor to the CM4 has been significant supply issues even before Covid. Given the importance and criticality of HVAC controls, having an adequate supply and availability is important. Although supply issues have affected most companies, industry standard controllers even from different manufacturers could be used if needed with some level of work (e.g. programming, commissioning, etc). If you have too specialized controller and/or have a need to have a extensive setup or development platform, then migration to another controller could be quite challenging.
Another key consideration that is often overlooked is I/O protection circuitry and/or automatic recovery from a variety of electrical transients and events (Brown outs, power bumps, etc). One of the key requirements (in my and my customer’s view) is a high reliability and longevity. In many applications controllers are expected to operate 24/7 for 10-15 years. In fact many buildings have controllers that are still working well past 15 years with many still with manufacturer support (although limited). There are very good reasons why one does not see arduino or other hobbyist level hardware architectures in building controls. Depending upon the region and customer’s insurance, installing controllers with the required approvals could void a customer’s insurance should things go awry.
Lastly, you mention that one of the reasons for your efforts is for a Net Zero objective. I would ask how energy efficient it is to use a CM4 or beagle bone for doing straightforward controls? Even the ESP32 is quite thirsty on the Wifi side of things and so all of these processors consume a fair amount of energy. Remember many HVAC Controllers are installed in enclosures without ventilation, fans, etc and in confined spaces. The CM4 can require a heat sink or active or passive cooling for it to work reliably and for long term needs. It seems a bit ironic to use a uber powerful processor that is way overkill for a number of applications, and thirsty as well, with ultimate goal to achieve net zero. I can certainly understand the desire and allure to use very powerful controllers, but these come at a price and so one must balance the requirements and needs accordingly.
Anyway, the above it not to deter you, but to highlight a number of things that may be of interest.
Cheers,
Sam