Not sure I understand what you are trying to do, but yes contrary to x86 boards where you just need to shortly trigger the power switch connection, the clusterboard seems to require a permanent connection of the two pins in the powerswitch connector. That is what I was referring to above with the piece of copper wire wedged between the pins.
I find this very strange to be honest, but that is the only way my Clusterboard v.2.2 works.
It has been a while since I posted about trying to get the clusterboard up and running. I spent most of that time waiting for parts. So I would rate where I am at overall build at about a 35% currently. So just going to hit the highlights:
ATX Power Supply (EVGA 750W) - This power supply does not need the capacitor included with the clusterboard to power on; however, all of the cables are black coming off the power supply, so you cant just "jump the green wire' to trick it into coming on. The clusterboard connector is a 2-PIN JST PH connector. I was able to order the correct cable from Amazon with some to spare. So I sacrificed a couple to try various supposedly non-momentary switches, all of which failed to work. So currently, stripped back the wires and just twisted them together to get the board powering up. Not going to be my final solution, but I need to spend some time talking to someone that doesnt sell momentary switches as 'latching switches'. Currently ordered some slider style and rocker style light up switches, which are taking forever to arrive.
6 Nodes of Fun - Sopine64 Networking Mess - First and foremost, THANK YOU to everyone whom has posted in the forum. I spent the better part of two hours going, why does this work on the dev board and not on the clsuterboard. Simple answer is that Clusterboard needs the 'networking fix' while the dev board DOES NOT WORK with the 'networking fix'. Luckily I have a BUNCH of working Debian builds all over my house, so editing the file was as easy as installing device-tree-compiler and then following the edit for the ethernet, forward and reverse. EVEN after the networking fix, imaging the "fixed image', and applying that image to ever SD Card, I still ended up with two boards that just WOULD not boot. So I used a process of Boot Sopine module on the Devboard, unedit the network fix, reboot, does it work?, re-edit the network fix, does it boot off net, move Sopine Module to the clusterboard. Boot Clusterboard with additional module, does it work yet?, repeat.
Which puts me where I am today: Clusterboard powers up, 6 modules (all of the ones I currently have) boot up, and are network accessible.
Currently working on:
Ansible Scripting for everything (already running jsut getting hosts files and vars setup for automation)
Rancher K3S installation pending.
01-01-2021, 10:10 AM
(This post was last modified: 01-01-2021, 10:12 AM by poVoq.)
Cool. Do you have a working eMMC with yours?
I noticed that sometimes on a cold reboot modules fail to start, but rather rarely. Not sure if that was your issue.
I have my setup pretty much at the same level but with 7 modules. But also installed some nice aluminum coolers on the A64 chips and added a big fan. Definitely helps keeping the temperatures down at 1.1ghz.
Anyone know how/if it is possible to push the clock a bit higher even? I think the A64 is used in a slightly bigger than 1.2ghz in some setups, no? But on Armbian it doesn't do higher than 1.1ghz with the standard settings.