after many hours of work, the layout was complete. then the enclosure was drawn and does not fit. the layout(which was provisional) is now undergoing major rework

the mezzanine structure has been scrapped as it would impose too many cutouts, the two levels of PCB will be retained but their stacking will be directly above with wire to board connectors allowing the position offset

this has also triggered a rethink on the internal contact connectors as the existing vertical molex KK are fairly tall and the new layout allows more vertical freedom, the current connectors under review are molex microfit 3.0

front panel connectors are not under review as they still meet all requirements. i did think about instead fitting some D-sub instead but the lower cost connectors dont have the same current rating and this is a non-negotiable feature

layout work is going well. there have been several changes to the initial plan:

• the connectors on the PCB(other than power) are going to be exposed through the chassis wall
• there is a large cutout section to allow stacked PCB to fit against the chassis wall

the USB interface will now need a PCB mount common mode choke as the original plan to wrap a toroid over the interior chassis cable is no longer feasible

it might be worth adding better ESD protection now

right now, there are only 32 unrouted nets so changes are probably not going to be major...

XP power ITR0324S05 is a reasonable looking power module
will need to look into how it performs as it is an unregulated supply. this may be fine, the datasheet says 10% regulation from 0-100% load...

had a look around a supplier for suitable USB hub IC, there is the microchip USB2514B/M2 which has four high speed ports. its configuration is based on an embedded i2c controller which interfaces directly to a 256x8 EEPROM (24AA02/24LC02B style)

looks like the ADuM3165/3166 is a reasonable isolator to choose too

power is the main question now. this will need 5 isolated zones. the four outputs need to provide 5V@500mA+ simultaneously while the hub is full belt. this is at minimum ((5v*0.5A)*4)+((5v*69mA)*4)=11.38W
thats about 474mA at 24Vin when 100% efficient! more likely will be 11.38W/0.8 = 14.225W (80% efficiency) which is 592.71mA at 24V(a significant amount of power)
power usage aside, the topology will be interesting. it would be possible to use a total of four flybacks outputting 5v (four downstream facing zones. upstream port will be self powered and also power hub controller) but using a single transformer with four postregulated outputs may be a more robustr solution. that said, it wouldnt be impossible to use some 24Vin5V.5Aout modules as they will be a more reliable/easy to work with solution

there are only five missing footprints for the PCB! and only one components that need BoM information! the ordering onfo is almost over. this is probably going to be the most strongly defined schematic ive ever done outside of work though so its worth it

looking forward to the layout as thats actually fun