I’ve tried many different power adapters to power my Axon. From standard PD, QC, VOOC, SuperVOOC, WARP etc (yes, I know there’s difference between charging & powering… and their protocols, but)
All of them work normally (as a common user, I don’t see any problem, including benchmark results); except for the fact that they all fallback to using 5V ONLY; which is not recommended.
Apart from the PCIe peripherals (or 3.5" HDD), I don’t see anything using 12V anyway.
So I am not sure what’s the downside of continuing to use 5V only; however I was told (on discord) there are downsides to that.
So, with that preamble, I have two questions:
Which power delivery mechanism does the board support anyway? Knowing this will help find the adapter that’ll work according to the recommendations (instead of shooting in the dark and crossing fingers, hoping it works this time).
Knowing the PD controller IC will also help diagnose it further, to see what’s going on.
If 12V is not provided, and the board is continued to be run on 5V (while no downstream powered peripherals are attached) what would be the actual harm to the usage or the board? Knowing this will help understand the impact of continuing to use the fallback 5V operation.
As an aside:
Having the negotiated VIN as the passthrough for +12V rail isn’t ideal. There should’ve been a loadswitch or UVLO controlling the +12V rail, monitoring if the VIN is actually 12V or not. Even better, it should have allowed any of the negotiated PD voltages from +12-20V, and then run it through a buck converter (with UVLO/SS/PG etc.) to step it down to 12V to power that rail.
I’m not on the Dev team for these boards but I have tinkered around with the Vaaman board a bit (sadly my Axon hasn’t been delivered yet).
Don’t think that they would have changed much in the Input Power section of both of these boards keeping procurement and development effort in mind.
Since this is only PD 2.0 enabled, it can only request the 7 supported PDOs (out of which 5 are relevant for us) 5, 9, 12, 15, 20V. The resistor setting seems to be 011 (will validate this later) which is the 12V setting. No Programmable Power Supply (PPS) functionality like the USB PD 3.2 spec (APDO).
My suggestion would be to stick to USB-PD certified power plugs that support the 12V PDO. You can find this information printed on the power plug itself.
Attaching an example of the PDO information on the RPi 5 Power Plug and the IC on the Vaaman Board (Excuse the bad quality image)
The PCB tracks would have been sized keeping 12V in mind. Supplying 5V would result in higher currents on these tracks leading to worse thermal performance.
The DC-DC converters would have been optimised to operate at maximum efficiency for a 12V Input. From my limited time messing around with the Vaaman board, it runs really really hot and I don’t think this would have changed with Axon considering it’s a powerful chip. So stacking inefficient power conversion on this would be a bad idea.
The Vicharak team needs to work on thermal management of these boards quickly. Either start including a fan or better heatsinks. The current ones don’t really do much.
The LDT6015TS mentions VSENSE and ISENSE, but there seems to be a discrepancy between the Pinout Diagram and the Pinout Table. I don’t see any shunt resistor either on the Vaaman board. Highly doubt they are doing OCP, OVP using this IC. I can’t comment on that functionality being there on the DC-DC converters. They seem to have used non-descript Bucks everywhere. So I’ll let the hardware development team chime in on that.
Ideally yes, you are justified in saying that these boards need to have hardware cutoffs
Extensively, there is none OVP and OCP methods are there which are being used in standard consumer computing devices, i think i will definitely try to use in other hardwares. but the bucks and other devices are protected by diodes and stuffs which is standard practice in SBCs.
Running 5V won’t do a damage, some of cores of RK3588 won’t work eventually whole system runs on 5V only. and tracks are sufficient enough to draw current. though this is an open question for us as well, and will figure something out in next hardwares.
