Thanks for this tear down Amir, I wanted to try and answer some of the points and why I did what I did.
Firstly it is always worth noting that there is always more that one way to skin a cat, some engineering decisions lead to necessary compromises and taking other decisions would always lead to another set of compromises. When designing something that is a flexible as can be without essentially being a ground up build each time it is a given that there will be some compromise so with that in mind I'll answer some of the point above
When I approached this design a few years back I took a good look round at the current offerings with the same modules, mostly they were similar to DIY designs but were cheap and then some much better designs from the bigger manufacturers but they were expensive. The thing that I really like about these modules is that Hypex provide an on board communications buss that can communicate temperature, power supply state/error, amp state/error, clipping and DC detection. They also rather handily have various aux voltages that can be used to power other stuff. I really wanted to be able to use all of that neat stuff and that meant having some sort of MCU reading those data and deciding what to then do with it. I decided to build a PCB with the MCU on board and also use it to do the routing etc. This makes it much quicker to assemble that wiring every XLR with hook up cable, and every power cable back to the IEC connector. Sure it is much more expensive and components that fit to PCBs are also more expensive but it saves time and makes things much more repeatable. Ultimately it gave me what I wanted, all the functionality I wanted to preserve, a semblance of neatness and still keep the price low. The final iteration after much messing about writing software I am pretty pleased with. Pressing the standby switch on the front powers the MCU (from the aux power on board 1) which then polls all the modules to ask them to switch out of standby one by one, when one module reports back that it is happy and stable, the MCU moves to the next one, preventing large in rush. If there is an error, the board is powered down and an error state is displayed in red on the front button ring. Once all the boards are on, they are constantly monitored for power errors, DC errors, temperature and clipping. All errors are again displayed on the front ring with a different flashing sequence. All errors are recorded in case of a return to base so diagnosis is easier. Temperature is monitored and if it goes above a fixed temp, the fans will start to turn slowly at first and increasing in speed with temperature until the temperature is bought under control. That said, it is a very rare set of circumstances that would bring the fans on. As you will all know these modules are very efficient (just over 80%) and they are very well heat sinked meaning little heat is produced. That said a couple of kW can still produce a significant amount of heat when the amps are pushed hard and I wanted a belt and braces approach. My main customer type has been for studios and for custom install home cinema where multiple amps are stacked in racks in equipment cupboards or rooms. I wanted to make sure that those customers had peace of mind that their amps would not over-heat. Nobody else offered this so hopefully it is a good USP. Finally We I could also use the MCU to look after 12V triggering. The LED round the button is also dimmable with by pressing and holding it in while it cycles through different brightnesses, letting go will store the brightness until next time it is changed.
I chose to use ribbon cable as the modules have a ribbon/IDC connection themselves so at some point there has to be a ribbon cable involved. I used the best quality copper ribbon I could find at 28 AWG (biggest that will fit) and IDC connectors with gold plated pins for what it is worth. While not quite as good as twisted pairs the signal lines all have GND wires between them which is how sensitive PC cabling works. In tests this was good.
The PCB itself is split into 3 different GDN pour areas so I could control the way XLR connectors were connected to the chassis. The XLR shield/shells need connecting to the chassis which in turn needs to be connectected to the safety earth and the 0V needs to be connected to safety as well but critically at a different point so different PCB areas are essential. It is much easier to repeatedly get this right with a PCB that with cables. The power also needs distributing as well so this is also in it's own PCB section kept well away from any signals, effectively just a pair of wires form one side of the case to the other without having to run more cables around the case. It was pointed out above that the sticky cable tie bases aren't great, and it is true, they are not ideal so having as few cables as possible running around is the best way to do it. In busy multi channel build like this you will always end up with cables where you don't like them. Lastly I wanted to pay particular attention to the earth bonding, many of these cheaper boxes just rely on the connection of the panels together and then don't really bother getting any certification done in an effort to keep prices down. I wanted to be able to CE mark these so I needed to ensure this was bullet proof, Each panels gets it's own earth cable, all are cinched to the chassis with star washers. Amir you mentioned about the one on the back panel, if you took the screw out you would see the anodising removed with a reamer and the star washer used results for all panels were less than 0.0126 Ohms.
I'm sure more stuff will come to mind that might be interesting for me to post and I will try and keep an eye on the thread but there are about a million audio forums
Stefan
