[N100] - Firebat T8 - Inner stuff and some tests

Just wanted to chime in, mine has the same BIOS version CasualNerd reported (2.22.1288, 5.26 core, T8P 0.01 x64 project), I bought it mid-April. I made a BIOS dump back then I could upload somewhere if needed, but it seems to be the same.

The SSD in mine died after ~3 months of usage and didn’t even see much use of it. It began showing remapped sectors on the 2nd of July, from what the logs of Hard Disk Sentinel show. And it seems it’s hardwired to report 40ºC.

The seller offered no warranty replacement, or SSD exchange either, but I think I may still return it. Would it be too much to ask for a year of warranty on things off of AliExpress?

I am just debating whether to return it and look for something else, or replace the drive at my expense. Honestly undecided, it was cheap, the same unit right now would cost me 50% more at least, but if everything else is of the same quality I don’t know if it would be worth it.

Just want to confirm what Shonk said. I replaced WiFi card with Intel BE200 and it works fine.

Did anyone try replacing thermal paste and noticed any difference? I see on reddit a few people did it, and supposedly it has some positive benefits. From IreuN photo in this thread, seems like the factory did a poor job applying it.

I considered replacing the paste but didnt bother in the end its only a low power cpu

Slight update on the Wifi cards i have tested

Tested working Wifi Cards By Me
Intel® Wireless-AC 9260 (PCIe)
Intel® Wi-Fi 6 AX200 (PCIe)
Intel® Wi-Fi 6 AX201 (CNVio2)
Intel® Wi-Fi 6E AX210 (PCIe)
Intel® Wi-Fi 7 BE200 (PCIe)

Should Work Wifi Cards
Intel® Wi-Fi 6E AX211 (CNVio2)
Intel® Wi-Fi 6E AX411 (CNVio2)

I pulled the ssd on this when i got it
and its in a usb to m.2 sata enclosure now

i have seen a few posts about them dying
so decided to find out about the chipset and nand

Mine is 72 Layer SK Hynix TLC with and a Dramless SMI2259XT
with a very low bad block count from the factory

Maybe some people got QLC drives?

Model: N900-512
Fw : W0825A0
Size : 488386 MB [512.1 GB]
From smart : [SMI2259XT] [W0825A0 00] [HYNV4]
Controller : SM2259XA bufferless
FlashID: 0xad,0x5e,0x28,0x22,0x10,0x90,0x0,0x0 - Hynix 3dv4-72L TLC 8k 512Gb/CE 512Gb/die
Channel: 2
CE : 4
TotDie : 8
Plane : 4
Die/Ce : 1
Ch map : 0x03
CE map : 0x33
Inter. : 4
First Fblock : 2
Total Fblock : 1272
Total Hblock : 7632
Fblock Per Ce : 1272
Fblock Per Die: 1272
Original Spare Block Count : 85
Vendor Marked Bad Block : 0
Bad Block From Pretest : 8

I might do it regardless, if I get some free time. I also noticed in various reviews people doing something to ssd to make it cooler. I’m not sure are they using thermal pads.

SSD in my PC is almost the same as this one:

This was mine, I chose to keep the little computer, but I replaced the disk with another one I bought, so I no longer have it:

Model: N900-512
Fw   : V1207A0
Size : 488386 MB [512.1 GB]
From smart : [SMI2259XT] [V1207A0 00] [TB1ZM]
Controller : SM2259XA bufferless
FlashID: 0x98,0x3e,0xa6,0x93,0x7e,0xd1,0x0,0x0 - Toshiba 15nm MLC 16k 512Gb/CE 128Gb/die 2Plane/die
Channel: 2
CE     : 4
TotDie : 32
Plane  : 2
Die/Ce : 4
Ch map : 0x03
CE map : 0x33
Inter. : 16
First Fblock  :    2
Total Fblock  : 2126
Total Hblock  : 7632
Fblock Per Ce : 2126
Fblock Per Die: 2126
Original Spare Block Count :    96
Vendor Marked Bad Block    :     0
Bad Block From Pretest     :    79

I don’t think it’s related to any of the existing discussion, but I want to share my findings:

I had mine (16GB model) with Proxmox and 2 VMs. VM #1 (debian) was set to 10GB memory and VM #2 (haos) was set to 4GB memory, both without ballooning. What’s weird was that I kept having a weird crash on the VM #1 whenever there are some load. No traces, and just the VM crashes, not the Proxmox nor the VM #2.

After months of bandage fix (cron to restart the VM) and troubleshooting, I’ve tried lowering the VM #1’s memory limit to 8GB, and it’s been stable since, even under load. I thought the 14GB/16GB total allocation was well within it’s limits, so this was a surprise to me. Perhaps the soldered memory on my device is faulty and/or physically at lower capacity? Not sure how the host and VM #2 never crashed because of it. I have no clue, but hope this is helpful to some.

https://www.firebat.com.cn/index/activity?pageType=product&commodityId=67

My 1TB QLC SSD runs extremely hot so I cut and strapped on a heatsink to it.


The BE200 card works alright.

Now that I own a CH341a, I can certainly back it up. The AliExpress seller told me that BIOS builds correspond to batches. Although mine isn’t Firebat branded and I also have the same ACPI errors on Linux. Besides, I was able to update the Intel ME using a build from station-drivers.com.

Since as you are also seeing the errors I’m not sure it will be of much help sadly, but thank you anyhow! Might have to check out the site you linked though, thanks for that too.
I did find this workaround on reddit to prevent the ACPI error messages (at least on pfsense):

editing the \boot\loader.conf and adding the following line seems to have suppressed the alerts:

debug.acpi.disabled="thermal"

And in this reply, someone explains how to set that in pfsense during boot:
(this isn’t limited to only the installer)

In the boot menu of the installer, there is an option to “escape to loader prompt”.

You can then use the “set” command to set the variables accordingly, e.g.

set debug.acpi.disabled="thermal"

Once you have set all variables you need to set, you can use the

boot

command to launch the live system and start the install

However, this method won’t survive reboots. I had to edit the loader as per the first link/example once I booted into pfsense, so that I wouldn’t have to run the command every boot. I’m unsure if it survives updates.

Also, for others curious/worried about the supplied drive’s reliability (I received same 512GB N900 drive as OP on 2x units ordered a month apart):
I have 2x currently pending sectors @ 941 power-on hours and 48 lifetime power-on resets.


I can also confirm that the supplied N900 SATA drive reads a constant 40c (even with a fan pointed at it), as mentioned by Carlisle above.
(although my S.M.A.R.T data seems to show that it read 23c at some point in its life?)
smartctl temp data

I removed that drive from the Mini-PC the day I got it and installed a Toshiba NVMe drive i picked up cheap, so I’m not particularly worried if it fails on me as it’s only being used in a windows VM, so I’ll update you all when that time comes.

I too noticed the (NVMe) drive was getting toasty, no surprise seeing as there isn’t really any airflow in the top of the Firebat T8 Plus. Soooo, I chopped a hole in the top, stuck a Noctua 5v USB fan (NF-A4X10-5V-PWM) in there, and drilled several 3.5mm holes in the case along the top of the USB and HDMI ports. Bonus, now I don’t have to strip the Mini-PC down every time I want to remove the drive lol.

Here’s some pics for those that are interested. Not the most elegant solution, I definitely could’ve taken my time a bit more lol but it’s out of sight and does it’s job so I’m not fussed.

I keep mine with the top lid off and it’s alright. I do use it with a Kioxia BG4 sometimes (ChromeOS Flex) and that runs very cool. Too bad the port is limited to Gen3x2.

1 Like

I dont really need the 1TB Sabrent Rocket in this so ordered a WD SN740 256GB for £10 on ebay
anyway within a few mins of having it on the SSD temp was 66°C

i trimmed a slim nvme heatsink down with tin snips it dropped the temps by about 15-20°C


1 Like

Have a similar style one using a different branding but same SSD layout. It does some odd thermal throttle if attempting to do QuickSync at 2K capture, had been using it on an experimental teaching platform(golf but also works for hockey goal skill challenge).
Doesn’t seem to matter if the SSD is the stock QLC or a TLC SSD, cutting a thermal pad+mini heatsink barely gets temps to 60 C idle(65 C average/70 C full load).

Can I change the SSD N900-512 with any 2230 / 2242 nvme?
Any thermal pad or cooler recomended?

2242 is the size but the slot supports both SATA and NVNe drives.
@unsuspicious

i wouldnt have bothered replying to him he cant even read two posts up