Voltage issues

I’m trying to power the VF2 with a NVMe and I am using the Arch Image.

Powering with 5V 2.4A gives me NVMe probe timeout but 4 devices shown with lspci. Probably due to weak and unstable voltage.

[    2.272208] nvme nvme0: pci function 0001:01:00.0
[    2.272273] nvme 0001:01:00.0: enabling device (0000 -> 0002)
[   62.433598] nvme nvme0: I/O 13 QID 0 timeout, disable controller
[   62.439038] nvme nvme0: Removing after probe failure status: -4
0000:00:00.0 PCI bridge: PLDA XpressRich-AXI Ref Design (rev 02)
0000:01:00.0 USB controller: VIA Technologies, Inc. VL805/806 xHCI USB 3.0 Controller (rev 01)
0001:00:00.0 PCI bridge: PLDA XpressRich-AXI Ref Design (rev 02)
0001:01:00.0 Non-Volatile memory controller: Toshiba Corporation XG4 NVMe SSD Controller (rev 01)

Powering with 15V 2.6A

00:00.0 PCI bridge: PLDA XpressRich-AXI Ref Design (rev 02)
01:00.0 USB controller: VIA Technologies, Inc. VL805/806 xHCI USB 3.0 Controller (rev 01)

and no output on dmesg | grep nvm.

I can reproduce this indefinitely.

What is this strange behavior loosing the NVMe while using a 15V/2A+ with PD (Nintendo Switch PowerSupply) thats actually in terms of voltage listed as supported in the documentation? Or doesnt it switch to 15V and its undersupplied in 5V 1.25A Mode?

Sadly the AXP15060 does not deliver any voltage readings within the OS.

2 Likes

Strange. With 15V @PD it shall make 5V/3V3 rails dead hard. However, PCIe 3V3 is separate and fed from system’s 5V rail by a SCT2250FPA buck which is rated 5A max.

My only guess can be that your SSD (Toshiba X64) somehow browns out rail for a while because it has quite unusual maximum amperage rating (3A). For example, I have Apacer AS2280P4 installed with peak power draw rating of 4W (/3V3=1.21A) and I also have quite old “sata” SN520 256GB SSD from WD which did not work with plain M.2 socket of VF2 laying around which has maximum rating 1.5A.

Try another SSD, with better power rating.

2 Likes

Its really strange. I’ve 3 of those toshiba nvme. I swapped the powersupply to a 5.2V/2A (nvme shows up but timeout), 5V/3A (9 V/2.2 A)PD (nvme does not show up).

The nb679 buck converter doesn’t regulate with 5V input so you’re dependent on the external power supply and cable with a small drop through the regulator.
NB679.pdf (1.0 MB)
The nb679 is designed for 5.5V → 28V input voltage.

Would like StarFive to offer a more suitable power supply, USB-C compliant.
This problem pops up too often.
It needs to be dealt with when purchasing the board.

I’ve to find my 5.5mm/2.5mm Barrel Jack to USB-C Adapter. Then I can try 12V/2A.

But then why are 5V on list?
VisionFive 2 Datasheet (rvspace.org)

it will ‘work’ minimally on 5V,
but add an nvme and it gets a bit iffy,
I have previously asked for updates to official documentation,
pulling more power through the nb679 increases the drop,
it’s really slight but has effect through the power supply circuit.

1 Like

So I’m currently powering it with stable 12V (2A). Same behavior. As soon as I go above 5V, the NVMe and USB Port/Hub on 00:00.0 disappears.

[user@ArchVF2 ~]$ lspci
00:00.0 PCI bridge: PLDA XpressRich-AXI Ref Design (rev 02)
01:00.0 USB controller: VIA Technologies, Inc. VL805/806 xHCI USB 3.0 Controller (rev 01)
[user@ArchVF2 ~]$ dmesg | grep pci
[    1.596583] pcie_plda 2b000000.pcie: Failed to get power-gpio, but maybe it's always on.
[    1.596764] pcie_plda 2b000000.pcie: host bridge /soc/pcie@2B000000 ranges:
[    1.596812] pcie_plda 2b000000.pcie:      MEM 0x0030000000..0x0037ffffff -> 0x0030000000
[    1.596842] pcie_plda 2b000000.pcie:      MEM 0x0900000000..0x093fffffff -> 0x0900000000
[    1.935545] pcie_plda 2b000000.pcie: Port link up.
[    1.935892] pcie_plda 2b000000.pcie: PCI host bridge to bus 0000:00
[    1.935909] pci_bus 0000:00: root bus resource [bus 00-ff]
[    1.935928] pci_bus 0000:00: root bus resource [mem 0x30000000-0x37ffffff]
[    1.935944] pci_bus 0000:00: root bus resource [mem 0x900000000-0x93fffffff pref]
[    1.935989] pci 0000:00:00.0: [1556:1111] type 01 class 0x060400
[    1.936016] pci 0000:00:00.0: reg 0x10: [mem 0x00000000-0xffffffff 64bit pref]
[    1.936080] pci 0000:00:00.0: supports D1 D2
[    1.936092] pci 0000:00:00.0: PME# supported from D0 D1 D2 D3hot D3cold
[    1.939905] pci 0000:00:00.0: bridge configuration invalid ([bus 00-00]), reconfiguring
[    1.940058] pci 0000:01:00.0: [1106:3483] type 00 class 0x0c0330
[    1.940093] pci 0000:01:00.0: reg 0x10: [mem 0x00000000-0x00000fff 64bit]
[    1.940192] pci 0000:01:00.0: PME# supported from D0 D3cold
[    1.943791] pci_bus 0000:01: busn_res: [bus 01-ff] end is updated to 01
[    1.943833] pci 0000:00:00.0: BAR 0: no space for [mem size 0x100000000 64bit pref]
[    1.943850] pci 0000:00:00.0: BAR 0: failed to assign [mem size 0x100000000 64bit pref]
[    1.943867] pci 0000:00:00.0: BAR 8: assigned [mem 0x30000000-0x300fffff]
[    1.943888] pci 0000:01:00.0: BAR 0: assigned [mem 0x30000000-0x30000fff 64bit]
[    1.943914] pci 0000:00:00.0: PCI bridge to [bus 01]
[    1.943929] pci 0000:00:00.0:   bridge window [mem 0x30000000-0x300fffff]
[    1.944015] pci 0000:00:00.0: enabling device (0000 -> 0002)
[    1.944036] pci 0000:01:00.0: enabling device (0000 -> 0002)
[    1.944828] pcie_plda 2b000000.pcie: msi#0 address_hi 0x0 address_lo 0x190
[    1.950778] pcie_plda 2c000000.pcie: Failed to get power-gpio, but maybe it's always on.
[    1.950954] pcie_plda 2c000000.pcie: host bridge /soc/pcie@2C000000 ranges:
[    1.951001] pcie_plda 2c000000.pcie:      MEM 0x0038000000..0x003fffffff -> 0x0038000000
[    1.951030] pcie_plda 2c000000.pcie:      MEM 0x0980000000..0x09bfffffff -> 0x0980000000
[    2.357683] pcie_plda 2c000000.pcie: Port link down, exit.

I’m having constant power problems only when supplied 5VDC from header (of course you have to count voltage drop over wire there too), so far my experience is not much good there.
If my 5V rail goes down to 4.8V or less, my Apacer SSD immediately goes down with that nvme nvme0: I/O 13 QID 0 timeout, disable controller nasty dmesg line. So, for backup supply, I have to raise voltage over a 20cm backup cable to 5.1V (at least).

On the other hand, feeding the board from relatively cheap Samsung EP-T1510 10W USB-C (capable of 9V supply but VF2 never uses this level voltage for some reason and switches to 5V instead) supply never raised me any trouble. Never. I’ve got only these infrequent SSD timeouts, but everyone gets them, for now (just search this forum). Anyway, I also fed it from a Fairphone A2009-03 USB-PD 3.0 12V 30W (capable up to 20V, but VF2 again never uses this level voltage too, switching to 12V) supply, and had exactly same experience. But I never tried any other SSD, because I have two of them and one is used there (another later went to upgrade my laptop).

Your best option could be grabbing multimeter and checking voltages. Another option could be to have another board or SSD. Or staying for now at 5V input. This dmesg does not show anything informative, unfortunately.