I'm working on a project with a remarkable amount of overlap with this!
I don't know why there's the green box. I haven't found any note explaining it. But if you've copied the schematic exactly, it'll work as you expect.
Some other comments:
- why the dual 3W resistors on the sense optocoupler inputs? As configured, you're using a drive current of 0.575mA, so you're only dissipating 132mW across two resistors (so 66mW per resistor).
- the optocoupler datasheet rates it at 1mA, but the characteristic curves and typical Vf values are rated at 1mA up to 20mA+. Given that you're using these as switches, you're better off letting the output transistors saturate by using a bit higher. (When using BJTs as switches, one normally "overdrives" them to 2-5 times the saturation current.) I'd perhaps do 1.5mA. At 1.5mA you're still only dissipating around 350mW, easily handled by a few small resistors in series (e.g. three 51K resistors in series) or one 1/2W resistor. You can make the resistor happy by using fat traces to and from it, which will help it dissipate heat.
- the USB side of the CP2102N needs
very careful layout. If, like me, you don't know exactly what you're doing (which you say you don't in the review document), then copy the exact layout used in the dev board. A colleague of mine tried laying out a copy of the same dev board, but didn't know that USB traces need care, and his boards simply did not work. An exact copy worked fine. USB traces need to be length-matched, and ideally need to be controlled-impedance traces. So copying that layout is the lowest-risk option.
- if you're using electromagnetic relays, then you need freewheeling diodes across them. Dave just did a video on this exact topic. (I haven't watched it yet). In my project, I opted for solid-state relays (SSRs) to reduce switching noise (since most SSRs have built-in zero-crossing circuitry, so the relay always switches at the point where the AC waveform is at 0V). SSRs are also optocouplers in and of themselves, so they eliminate the need for a driver since you only need to drive a small LED instead of a coil.
No obvious reason why those nets should be in a green box, but they're all things the CP2102 requires:
- nRST pulled high, taking it out of reset
- nSUSPEND pulled low, presumably taking it out of suspend
- VBUS connected to the middle of a voltage divider, so that the chip can measure USB voltage greater than the 3.3 V VDD
SUSPEND is an output, not an input.
I think I don't need SUSPEND, so I will not add that resistor. Will copy the rest.
I assume the “NC” next to it means “no connection”, i.e. not populated. As best I can tell, the suspend signal ("Active") isn't used anywhere in the dev board design.