Topton Jasper Lake Quad i225V Mini PC Report

[klosz007]

Hi,

I have Topton's TVI7309X N5105-based MiniPC and I have been dealing with same page fault kernel panics in OPNsense that ware mentioned many pages back in this thread with no apparent resolution.

In my case OPNsense runs as a VM on ESXi 8.0u2. No other ESX VM running on this MiniPC (Windows 10, various Linux distros) was experiencing any crashes nor ESXi itself had issues, just OPNsense VM was rebooting once a day due to aformentioned "Fatal trap 12: page fault while in kernel mode" error. Moreover, this same VM ran perfectly on ESXi 8 when on my "primary" home server utilizing i7-10700K CPU. It all started when I moved this VM 'as is' to this N5105 PC.
This must be a corner case between N5105, virtualization and FreeBSD.

What, I believe, helped in my case was one of the two changes, not sure which one, maybe both helped together:
- bumping up VM allocated RAM from 4GB to 6GB (OPNsense was a bit short on memory with 4GB, especially because 1.5GB was consumed by ARC - my setup uses ZFS inside)
- reserving entire allocated memory and checking ESX VM option "Reserve all guest memory (All locked)"

OPNsense VM has been up now for a few days with no panic. Previously panics were as frequent as once a day.
Maybe it will help you.

 

[klosz007]

I'm setting up Proxmox and when I try to pass in my SATA controller via IOMMU (VT-d on intel cards) my BIOS has apparently everything related to it enabled: VT-d is enabled and Memory Remap (not sure if related but this is also enabled: IGD VTD, IPU VTD, IOP VTD). The Chipset/SystemAgentConfig(SA) states that VT-d is "unsupported", which if what I read around here is true it could be a bug from my BIOS and it should still work? but my Proxmox does not recognize IOMMU yet (my grub: "quiet intel_IOMMU=on" and added the modules changes for vfio, I don't think this is the issue here?).

I have different chinese MiniPC on N5105 (TVI7309X), with ESXi 8i on it.
VT-d/IOMMU certainly works here (albeit I'm not using this feature). Still it would not be possible to redirect SATA controller to VM here, apparently because it does not have own, dedicated IOMMU group. SATA controller shares IOMMU group with other devices hence all devices in a group cannot be redirected then. I can redirect (if needed be) four I226 controllers, NVMe SSD and integrated GPU only - devices with their own IOMMu groups.

You may try to disable PCIe ACS via Proxmox kernel option and check if SATA controller redirection is possible then.

But disabling ACS is not recommeneded approach since it may affect stability of the entire system. I tried this on my another system to individually redirect 4 PCIe devices connected behind non-ACS PCIe switch and result was ESX purple screens/kernel panics caused by VM to which these 4 PCIe devices were redirected to.

 

[VLMT]

I'm setting up Proxmox and when I try to pass in my SATA controller via IOMMU (VT-d on intel cards) my BIOS has apparently everything related to it enabled: VT-d is enabled and Memory Remap (not sure if related but this is also enabled: IGD VTD, IPU VTD, IOP VTD). The Chipset/SystemAgentConfig(SA) states that VT-d is "unsupported", which if what I read around here is true it could be a bug from my BIOS and it should still work? but my Proxmox does not recognize IOMMU yet (my grub: "quiet intel_IOMMU=on" and added the modules changes for vfio, I don't think this is the issue here?).

You can do passthrough on logic level, not hardware level (device will be still presented in proxmox but entire disk can be managed by VM), this was my solution. Google it like "proxmox Passthrough Physical Disk to Virtual Machine"

 

[rosu]

Hi all,
I've seen a few posts in this thread about plugging in a PWM fan into the CPU fan header on these boards (specifically the GF-1338NP-12 v1.4). I believe the header is a JST 1.25mm and I have the connectors available to build an adapter to convert to 4-pin PWM fan header - was just wondering if anyone has successfully done this for this specific model motherboard and what the pinouts are on this JST header?

 

[VLMT]

Hi all,
I've seen a few posts in this thread about plugging in a PWM fan into the CPU fan header on these boards (specifically the GF-1338NP-12 v1.4). I believe the header is a JST 1.25mm and I have the connectors available to build an adapter to convert to 4-pin PWM fan header - was just wondering if anyone has successfully done this for this specific model motherboard and what the pinouts are on this JST header?

Hi, have same model of motherboard
I wrote about it here https://forums.servethehome.com/ind...e-quad-i225v-mini-pc-report.36699/post-401780

 

[TvdH]

Currently my CWWK N305 (V2) with the NVMEx4 expansionboard is causing my ram-memory to overheat quite severely: 137 degrees Celcius. This happens when running memtest and at these temps memory issues seem to ocure. Not sure yet that the temperature is the cause of the errors, or that it is that I'm running a 48GB module.

There are mounting holes for a fan, but due to the expansion board there is no room inside to actually attach a fan. The room inside is just a few millimeters.

Did someone else experience temperature issues witih the nvme expansion board installed?

Possible solution's I'm considering:
- Place fan on the underside -> will have to find a solution because the fan will be thicker than the cases' feet.
- Place fan on the inside -> might have too use some standoffs to lower the mouting plate, and will cause the bottom to be open.
- Use a fan on the top of the case/heatsink -> No attachment point and only cools the heatsing itself which is not directly attached to ram. Might not help enough.
- Try different ram -> that might be causing less heat? Or is it just a 48GB issue?
- Replace the unit -> use a les power efficient solution in a regular case instead that has enough pci-e lanes for NVMe

 

[Mike9474593]

Currently my CWWK N305 (V2) with the NVMEx4 expansionboard is causing my ram-memory to overheat quite severely: 137 degrees Celcius. This happens when running memtest and at these temps memory issues seem to ocure. Not sure yet that the temperature is the cause of the errors, or that it is that I'm running a 48GB module.

There are mounting holes for a fan, but due to the expansion board there is no room inside to actually attach a fan. The room inside is just a few millimeters.

Did someone else experience temperature issues witih the nvme expansion board installed?

Possible solution's I'm considering:
- Place fan on the underside -> will have to find a solution because the fan will be thicker than the cases' feet.
- Place fan on the inside -> might have too use some standoffs to lower the mouting plate, and will cause the bottom to be open.
- Use a fan on the top of the case/heatsink -> No attachment point and only cools the heatsing itself which is not directly attached to ram. Might not help enough.
- Try different ram -> that might be causing less heat? Or is it just a 48GB issue?
- Replace the unit -> use a les power efficient solution in a regular case instead that has enough pci-e lanes for NVMe

The N305v2 seems to have issues with 48G RAM, see the Nxxx thread here.

 

[luxurywklw]

Hello everyone.

I need help.
I have this motherboard, but i don't know this connector,
The distance between the two contact pins is 1.25 mm

 

[Mike9474593]

Hello everyone.

I need help.
I have this motherboard, but i don't know this connector,
The distance between the two contact pins is 1.25 mm

Should be JST 4pin 1.25mm Do a quick search here in this thread, I think connectors have been linked.

 

[Stovar]

Hi did anyone by chance figure a way of stopping that annoying bios start up beep when there is no option in the bios menu to switch it off?

Its for an Cwwk V5 N5105 model, not tried any bios updates but I believe many N100/N305 versions have the option in bios to switch it off.

 

[TvdH]

Hi did anyone by chance figure a way of stopping that annoying bios start up beep when there is no option in the bios menu to switch it off?

Its for an Cwwk V5 N5105 model, not tried any bios updates but I believe many N100/N305 versions have the option in bios to switch it off.

I have both a CWWK N5105 V5 and a CWWK N305 V2. In the BIOS of the N5105 I there should be an pootioin somewhere to disable the startup beep, as I did disable mine. Did not disable it on the N305 yet.

 

[Stovar]

I have both a CWWK N5105 V5 and a CWWK N305 V2. In the BIOS of the N5105 I there should be an pootioin somewhere to disable the startup beep, as I did disable mine. Did not disable it on the N305 yet.

I checked through every page and sub option but could not see it, do you by chance recall where it is on the cwwk V5 5105?

Not unless you did a bios update by chance....

I believe on the N100/305 ones many had beep option under the Boot menu

 

[TvdH]

I checked through every page and sub option but could not see it, do you by chance recall where it is on the cwwk V5 5105?

Not unless you did a bios update by chance....

I believe on the N100/305 ones many had beep option under the Boot menu

I did look at the N305 it was under the chipset menu, if I remember correctly. Will check on the N5105, will report back later.

 

[tortugaconqueso]

Any way you could measure or even guess at the thickness of that copper shim? I used a .005" thick shim on mine. My shim was placed under the large copper heatsink with paste between case / shim / heatsink / CPU which brought idle temps down 10°C from 40°C to 30° for Win11 when testing. With pfSense bare metal, I get 40°C idle temps. But that is using two different monitoring methods. Also, was there thermal paste on both sides of the copper shim - or not?

BobS

Do you remember what size were the shims you used? I have the 5 NIC box running OPNsense right now, and it's idling at no less than 48°C with two 120mm fans blowing at it (one above, one below with the lower panel off) in ~30°C ambient temps, which I think can be improved a bit with shimming and repasting.

 

[Fd0]

Update after post : Fixed the sending of corrupted characters problem which I initially reported by lowering the baudrate from 115200 to 57600. so this has been updated in this post to reflect this.

I have been trying for ages to get the serial console to work via the RJ45 ethernet connection labelled as "COM" on the back of my Topton unit.
After trying several RS232 cables which all did not work for me I finally figured out a setup which works for me which I will share below.
It seems like the Topton uses 12V levels to send and requires (a minimum of) 5V levels to receive.

This setup supports a 57600 baudrate VT100+ terminal via serial console using clients on a PC/laptop like for example Screen or Minicom for:
- The BIOS
- The GRUB menu (using Linux)
- Kernel logging (using Linux)
- Login prompt (using Linux)
- via USB Serial and optionally also via Bluetooth or even both at the same time !

Items used:
- CP2102 based USB RS232/485 "6 in 1" dongle, available on Aliexpress
- HC-05 Bluetooth UART dongle, available on Aliexpress
- RJ45 Ethernet cable with only GND,RX and TX connected, inserted in to the "COM" port on the back of your unit.

BIOS preparation:
- See the two BIOS screenshots on how I configured Serial Redirection
- Of interest is that I had to modify the "SIO configuration" page to set IRQ 3 with IO=2e8H address. Setting it to anything else also worked but especially in the grub screen I observed corrupted keystrokes being send to the Topton. I suspect likely due to some inadequate serial port processing by the Topton unit when using a different IRQ combination. Changing the SIO configuration also required using a different tty for the kernel + login. In my case it ended up being /dev/ttyS3. I used "dmesg | grep tty" to find out which one was active.

CP2102 USB RS232/485 "6 in 1" dongle preparation:
- After many hours of trying to setup the dongle trying all sorts of DIP switch combinations I finally figured out the right combination for robust TX sending. It required to connect the "B" pin (!) of the CP2102 dongle to the "232 TX : pin 6" as shown in my RJ45 ethernet connector cable picture. For the other DIP switch settings see the picture. The settings are different than what you would expect from the small manual of the CP2012 dongle specification (see attachment as well)

So disclaimer: be aware connecting the setup like I did is at your own risk. I did not properly check matching any voltage levels. You might be shortcircuiting or overvolting any pins or create any other type of problem if you do not do your own research first.

- The labels in RJ45 ethernet connector picture should be interpreted as:
- "232 RX: 3" -> pin 3 of the ethernet connector should be connected to the 232 RX of the CP2102 dongle
- "232 TX: 6" -> pin 6 of the ethernet connector should be connected to the "B" label of the CP2102 dongle
- I connected the CP2102 USB dongle to a free USB port on my lapto and used screen /dev/tty<USB name> 57600 to view the serial output

HC-05 FC114 Bluetooth UART dongle preparation:
This option allows for serial console via bluetooth when you connect this HC-05 dongle to the CP2102 USB dongle, see attached pictures.
The serial console also remains available via the USB at the same time by the way.

I first connected the HC-05 dongle to a PC using a FTDI cable (3.3v) at 38400 baud using screen in order to program it.
I kept the black button pressed when connecting it to a USB port in order for it to go to the programming mode (led starts to blink slowly)
Likely the CP2102 dongle can be used as well as an alternative method instead of the FTDI cable but I didnt try this.

So using > screen /dev/tty<BT name> 38400
and while keeping the small black button pressed all the time I executed the following commands in the terminal window in order to program it for 57600 data transmission:
AT+ORGL (reset to defaults if already used in the past)
AT+NAME=BTJCOM
AT+PSWD=0000
AT+ROLE=0
AT+UART=57600,1,0
This set it up as a BT slave with a password of "0000" required to connect to it.
It starts in serial tranmission data mode after a disconnect/reconnect to a USB port.

Perhaps it is even possible to use BT dongle standalone while completely skipping the CP2102 USB dongle but I didn't try this.
This would btw also require to source 3.3v from somewhere.

When using this BT option it worked for me to connect the CP2102 USB dongle :
- to a USB port on the topton
- as well as plugged into a separate 5v USB charger connected to a wall outlet.
I did not need to change any DIP switches on the CP2102 USB in order to switch between these two configurations.

For GRUB kernel + serial console I used the following config for /etc/defaults/grub:
GRUB_CMDLINE_LINUX="console=tty0 console=ttyS3,57600n8"
GRUB_TERMINAL="console serial"
GRUB_SERIAL_COMMAND="serial --unit=0 --speed=57600 --word=8 --parity=no --stop==1" (<- not sure if this line is really needed)

I applied this GRUB config with:
update-grub
proxmox-boot-tools refresh (when using EFI secure boot with Proxmox like me)

Finally a (wireless) serial console !

 

[thepsyborg]

It seems like the Topton uses 12V levels to send and requires 5V levels to receive.

That is so, so, so weird.

Anyway. Great post. Impressive sleuthing.

It starts in serial tranmission data mode after a disconnect/reconnect to a USB port.

Does it start up properly on reboot or do you have to go through the press-the-small-black-button-program-unplug-replug dance again?

 

[Fd0]

That is so, so, so weird.

I probably should have rephrased it as: requires 5V minimum for receiving from a client (afaik as I know rs232 doesn’t mandate a mininum of 12v being used)

Does it start up properly on reboot or do you have to go through the press-the-small-black-button-program-unplug-replug dance again?

It starts up properly on reboot.

 

[Stovar]

Has anyone by chance found this issue, I have a wireless pci-e card (wallytech DR7915) works fine on desktop PCs, but zero detection on 3 different mini routers from cwwk/topton?

This is with latest biose's installed and latest wallytech dr7915 firmware & driver in openwrt also.

 

[AnthonyUK]

Has anyone by chance found this issue, I have a wireless pci-e card (wallytech DR7915) works fine on desktop PCs, but zero detection on 3 different mini routers from cwwk/topton?

This is with latest biose's installed and latest wallytech dr7915 firmware & driver in openwrt also.

Not detected by the BIOS or by the OS?

 

[Stovar]

Not detected by the BIOS or by the OS?

OS, tried windows 10,11 (usually would say missing pci device) and openwrt latest x86 build with latest 7915 driver and firmware.

lspci utilities does not see it either.

Is there a way to enable it or check in bios?, there was only wake on lan BT/wifi so I tried on with that. Also tried factory defaults on bios and bios updates on the cwwk boards with no luck.