Hi everyone,
Please don't roast me in the replies as I'm quite new to networking & OPNsense. I wanted to tinker with VLANs on my home network, as well as segregate my IoT devices for privacy reasons, so in my opinion, my setup isn't really all that complex, nor would I consider myself a heavy user.
I was running OPNsense on an EliteDesk previously for several weeks and it was incredibly stable. However, I wanted to upgrade to 2.5 GbE speeds as my internet plan is 3 Gbps, so I purchased A Topton device from AliExpress (https://www.toptonpc.com/product/intel-n150-7505-6305-6x-i226-v-2-5g-solid-firewall-router-fanless-mini-pc-ddr4-nvme-1com-type-c-pfsense-opnsense-mini-computer/) -- I specifically got the 6305 model, and I loaded the same config to the new device (I obviously set the new interfaces, but otherwise the VLANs, firewall rules, etc. stayed the same). I do not have IDS or IPS enabled.
I have seen a lot of chatter on the forums that i226-V NICs have quite a bit of instability, but almost entirely related to dropped packets, random disconnects, etc. However, my issue is that my device freezes entirely and is unresponsive to SSH, ping, WebGUI (even when directly connected to one of the devices available ports). I have tried a number of different settings to try to get this sorted and I'm at my wit's end. The longest I've been able to keep this box online was 3.5 days, but it usually freezes up once per day, and there is nothing specific I am able to tie it to.
Here are my questions for the community:
- Has anyone seen this exact pattern? (EEE fix helps, ASPM tunable presumably makes worse or does nothing)
- Is BIOS-level ASPM/C-state disabling necessary in addition to OS tunables? Or should OS tunables be sufficient?
- Any reports of Topton N150 + i226-V + bare metal OPNsense running stable? What settings were used?
- Is NIC firmware V2.32 worth the flash, or should I focus on BIOS tweaks first?
- Has anyone modified/updated the BIOS on a Topton N150? Board: HSX-TGLNP V126
Below is all the detail I can possibly think of that will hopefully help someone diagnose the root cause. Otherwise, I'm going back to the EliteDesk setup, because my network going down at least once a day is not something I want to be dealing with.
Hardware SetupTopton X6E (marketed as N150):
CPU: Intel Celeron 6305 @ 1.80GHz (Tiger Lake UP3, 2 cores)
NICs: 6x Intel i226-V 2.5GbE
Current EEPROM: V2.14-0 eTrack 0x80000290
Community max available: V2.32 (via BillyCurtis GitHub)
RAM: DDR4 SO-DIMM (16GB, ex-EliteDesk hardware)
Storage: WDC PC SN720 256GB NVMe (ex-EliteDesk)
Form Factor: Fanless + external case fan running 24/7
BIOS: AMI `HSX-TGLNP-12-V126` dated 2024-01-29
Network:
3Gbps fiber ISP (Telus, Canadian) (10G port in bridged mode connected to the OPNsense WAN port. 1G ports of Telus modem not connected to anything)
OPNsense runs bare metal (NOT virtualized)
Router position: between fiber modem and Proxmox host
CPU utilization: ~5-10% under normal load
SymptomsSystem freezes completely:
Cannot SSH or ping box
Watchdog does not trigger
WebGUI unresponsive
NIC LEDs still blinking (power + activity present)
Power cycle always fixes it (temporarily — 1-3 days uptime before next freeze)
No kernel panic messages in logs before freeze, in fact there is nothing useful in the logs as to why this is happening
Nothing major running when freezes occur (e.g., qBittorrent seeding only '1 torrent total])
Timeline:
2026-04-29: Daily crashes discovered → disabled EEE → uptime improved to 3.5-4 days
2026-05-02: Crash after ~3.5 days → disabled ASPM tunable → uptime dropped to ~15 hours
2026-05-03: Still crashing; applied additional interrupt delay tunables (no improvement)
Tunables Applied (System → Settings → Tunables)
All listed below are confirmed working via `sysctl` commands:
Tunable Value Purpose Verified
`dev.igc.0.eee_control` `0` Disable Energy Efficient Ethernet on NIC 0 `sysctl dev.igc.0.eee_control`
`dev.igc.1.eee_control` `0` " NIC 1 ✓ all 6 NICs
`dev.igc.2.eee_control` `0` " NIC 2 ✓
`dev.igc.3.eee_control` `0` " NIC 3 ✓
`dev.igc.4.eee_control` `0` " NIC 4 ✓
`dev.igc.5.eee_control` `0` " NIC 5 ✓
`hw.pci.enable_aspm` `0` Disable ASPM (link power saving) `pciconf -lvc | grep ASPM`
`hw.igc.rx_abs_int_delay` `0` Disable RX interrupt delay (prevent NIC hang on timer) Added 2026-05-03
`hw.igc.tx_abs_int_delay` `0` Disable TX interrupt delay Added 2026-05-03
Interface settings (System → Settings → Networking):
Hardware CRC offload: ✓ Disabled
Hardware TSO: ✓ Disabled
Hardware LRO: ✓ Disabled
Hardware state:
C-states: Currently set to C1
Speed Shift: Not configured
What's Been Ruled OutNVMe: FreeBSD's nvme driver does not implement APST; drive stays in PS0. `sysctl dev.nvme.0` shows zero failures/retries/recovery events.
RAM: Worked without issue in prior EliteDesk hardware. Not investigated further.
Temperature: External case fan runs 24/7; system never exceeds 29.5°C even under load.
HypothesisThe system is freezing at the OS/hardware level, not crashing. No kernel panic logs suggests either:
A power state incompatibility (C-states + i226-V on FreeBSD) that the tunable alone can't fix
Interrupt handling issue with the NICs during idle/light load
BIOS-level power management conflicting with OPNsense settings
Key observation: Disabling EEE improved uptime 24x (daily → 3.5 days), but ASPM tunable presumably made it worse--or did nothing (15 hours), suggesting hardware-level power management in BIOS may be overriding the OS-level tunable.
Next Steps Being InvestigatedBIOS power management settings:
Disable C-states in BIOS (not just OS-level)
Disable ASPM in BIOS hardware config
Verify Power Limit 1 (PL1) and Power Limit 2 (PL2) settings
NIC firmware upgrade (if BIOS changes don't help):
Billy Curtis community firmware: V2.14 → V2.32
Verified MD5 hashes before flashing
Will update via Linux + nvmupdate64e tool
BIOS firmware update (last resort):
Current BIOS dated 2024-01-29; no public update found
Considering contacting AliExpress seller
Any help would be greatly appreciated. Thank you in advance.
Have you checked the console?
Quote from: toonable on May 04, 2026, 05:36:03 PMI wanted to tinker with VLANs on my home network
1. freezes how? does the console itself freeze?
2. VLANs? this usually indicates tagged traffic. are you using .1q, or do you just have a subnet per iface?
Consider some pro-active changes:
mkdir /var/crash
chmod 700 /var/crash
rc.conf
dumpdev="AUTO"
kernel config
options DDB
When it freezes, go dig out why.
Then we can perhaps provide some fix recommendations.
If the System freezes up completely, I would try using the CPU microcode plugin.
Quote from: toonable on May 04, 2026, 05:36:03 PMI loaded the same config to the new device.
I would never do that when moving to different hardware, but I think that's not the cause of these crashes anyway so let's ignore that for now... :)
QuoteI have seen a lot of chatter on the forums that i226-V NICs have quite a bit of instability, but almost entirely related to dropped packets, random disconnects, etc.
However, my issue is that my device freezes entirely and is unresponsive to SSH, ping, WebGUI (even when directly connected to one of the devices available ports).
I think those two can still be related in this case...
QuoteIs BIOS-level ASPM/C-state disabling necessary in addition to OS tunables?
Or should OS tunables be sufficient?
I would always start with the BIOS/UEFI Settings if I would suspect that changing any of those is needed.
QuoteIs NIC firmware V2.32 worth the flash, or should I focus on BIOS tweaks first?
I would always check with the manufacturer first in this case and flash a newer BIOS/UEFI if they have one or think it would help!
QuoteOtherwise, I'm going back to the EliteDesk setup, because my network going down at least once a day is not something I want to be dealing with.
I think your TopTon would crash even when just running IDLE while not being connected to your network in this case so you could test it while using a different Router if you have one :)
This is just weird :
QuoteTopton X6E (marketed as N150):
CPU: Intel Celeron 6305 @ 1.80GHz (Tiger Lake UP3, 2 cores)
It's either the one or the other :
- https://www.intel.com/content/www/us/en/products/sku/208646/intel-celeron-6305-processor-4m-cache-1-80-ghz-with-ipu/specifications.html
- https://www.intel.com/content/www/us/en/products/sku/241636/intel-processor-n150-6m-cache-up-to-3-60-ghz/specifications.html
It can't be both !!
QuoteRAM: DDR4 SO-DIMM (16GB, ex-EliteDesk hardware)
Storage: WDC PC SN720 256GB NVMe (ex-EliteDesk)
These Mini PC thingies can be very picky when it comes to the RAM and SSD used !!
The fact that these parts worked in the EliteDesk Mini PC does not guarantee that they will work in the TopTon without any issues...
QuoteNVMe: FreeBSD's nvme driver does not implement APST; drive stays in PS0. `sysctl dev.nvme.0` shows zero failures/retries/recovery events.
Please use commands like :
- nvmecontrol
- smartctl
To check the actual S.M.A.R.T. details ;)
QuoteRAM: Worked without issue in prior EliteDesk hardware. Not investigated further.
NOFI but running
memtest should have been the first thing that you had done after experiencing these issues !!
Many Linux/*BSD distros have one included in their Live ISO Boot Image for USB Sticks ;)
QuoteNext Steps Being Investigated :
- Considering contacting AliExpress seller
- BIOS firmware update (last resort):
Current BIOS dated 2024-01-29; no public update found
- NIC firmware upgrade (if BIOS changes don't help):
Billy Curtis community firmware: V2.14 → V2.32
Verified MD5 hashes before flashing
Will update via Linux + nvmupdate64e tool
I would do those things in that order and not the other way around ;)
QuoteIs NIC firmware V2.32 worth the flash, or should I focus on BIOS tweaks first?
Quote from: nero355 on May 05, 2026, 08:46:06 PMI would always check with the manufacturer first in this case and flash a newer BIOS/UEFI if they have one or think it would help!
Just for some reader clarity, the nvm for i226 is not any bios/efi thing.
There are probably no cases where updating nvm causes issue for a bios/uefi. There are however possible pitfalls (low risk) where an older driver (klm and the like) does not work 100% with new nvm, but this is usually something in old driver that cannot take advantage of some new feature in new nvm. The general rule has been, use latest nvm with latest driver. An unfortunate of most distro's is use of the GENERIC build, making it difficult to just unload bad klm and load in new better klm. It's only my opinion, but I think it's better to compile kernel very stripped, then do the load in of needed klm's.
I don't think I have ever seen any kernel code version that was dependent on the (as example) igc hardware nvm.
If the kernel supports igc, then the nvm version should (should) make no difference.
What we expect from updated nvm is, security/bug fixes and in some cases added features.
I believe most of the functionality changes (problems) are seen at the driver level (klm or otherwise). The kernel tree gets some new/bad code, it compiles down, goes to distro, and the bug list begins, then some patch is released, then final fix in next kernel tree. Certainly, Intel devices that use the igc driver code, is code probably supplied by Intel. I am not aware of say Oracle having a dev team who is writing their own "igc" driver for i226. Custom dev drivers certainly do exist, but that's for specialty/boutique stuff likely running with custom compiled kernel.
Getting to nvm v2.32 to fix a problem, was likely just Intel goofing up some code along the way, then they finally fixed it.
Quote from: BrandyWine on May 07, 2026, 01:09:34 AMJust for some reader clarity, the nvm for i226 is not any bios/efi thing.
There are probably no cases where updating nvm causes issue for a bios/uefi.
I think you have misunderstood/misread my post because that's not what I meant at all :)
Quote from: nero355 on May 07, 2026, 04:17:09 PMQuote from: BrandyWine on May 07, 2026, 01:09:34 AMJust for some reader clarity, the nvm for i226 is not any bios/efi thing.
There are probably no cases where updating nvm causes issue for a bios/uefi.
I think you have misunderstood/misread my post because that's not what I meant at all :)
Perhaps misunderstood.
There's only a few vendors that appear to have any notes or downloads for i226 nvm. Contacting the others does appear to be fruitless. If it were fruitful then the other big i226 nvm thread would have been a very small thread.
We can disable ASPM per device, including specific ASPM states. I would start there if ASPM is suspected. However, ASPM issue for a nic device would not freeze/panic an OS (low probability), the console should still be active if the nic dies while you were in via ssh, etc.
I have seen such freezes/lockups in the past. It was caused by Memory faulty blocks.
I would advise to run memtest in a prolonged period.
Regards,
S.
if it fails memtest86, maybe try again with a big fan directed onto the top of the Topton. Then try regular deployed use with the big fan blowing on it. Do you see the same results, or different?
I have been interested in mini-pc devices like this, such as Protectli Vault models, but the lack of active cooling always makes me hesitate. I wonder, can it really be stable and provide high uptimes? I have no experience with these tiny PC models myself, so I hesitate and search for testimonials. More than once I was just about to click BUY!!! on similar items while browsing Amazon, and just in the corner of my eye I then see "frequently returned item." Yes, well perhaps I can guess why that might be, I think, and it all comes to a screeching halt.
On at least some models, Protectli includes tightly fitted slabs of aluminum heat transfer blocks between critical components and the topside heat sink. Are blocks like these present inside the Topton models? I have no idea. If they aren't present they could be fabricated by a sufficiently OCD afflicted owner. (TBH I would set about this at once without asking myself why really, or consulting with a licensed therapist) But if I were advising someone else, I would say, can lowering the operating temperature by increasing airflow across the heat sink improve system stability at all? If it does then maybe we're onto the cause of the instability, but if it doesn't significantly change things then it's going to be down to something else.
Topton is just unreliable cheap garbage. There is a good reason why Deciso, Thomas Krenn and Protecli cost that much. Ive seen people buy Toptons like crazy, and then mount a fan on them because their aluminum case can not dissipate heat properly because they use the worst quality materials. It complete defeats the purpose because this is advertised as passively cooled unit. Worst of all, their quality is so inconsistent, that some units can work reliably for years, and some die for no reason after few months. And thats the exact reason why you should avoid them. Especially when people that have these units start to recommend them and claim that they have it for X amount of years, and they are working just fine. If you are on a tight budget, go for Quotom. But dont be a cheap ass and go below that. It will cost you more in a long run.
Is the device overheating on cpu core? Most cpu's will try and protect themselves from overheating.
The cheap N150 device I have for OPNsense is sinked at the case, but has an integrated ext fan to keep the aluminum cool. It is not doing a lot of work, but so far it's been stable.
see https://forum.opnsense.org/index.php?topic=48166.0
Quote from: BrandyWine on May 11, 2026, 06:17:01 PMIs the device overheating on cpu core?
Its normal for CPU to run hot, but its the heat sinks job to take that heat away from CPU and then dissipate it. These units are not doing that because they have crappy metal case with terrible thermal conductivity. The CPU gets hot, starts to throttle, it heats everything inside the unit, and then that heat from the inside is radiating on the aluminum case through your storage, memory modules and PCB. This is why on passive Toptons people are forced to install additional fans, not to keep the unit cool, but to keep CPU from thermal throttling and premature damage. It is that bad. The unit with same specs from either Protectli or Thomas Krenn costs 3 times more. Ask yourself why.
Quote from: BrandyWine on May 11, 2026, 06:17:01 PMMost cpu's will try and protect themselves from overheating.
CPU will almost never die on these units because it has mechanisms to protect itself. It does that with thermal throttling or simply shutting down. Its the other components in that tiny case that will suffer and die eventually. Your storage, your memory modules and worst of all, motherboard.
Quote from: BrandyWine on May 11, 2026, 06:17:01 PMThe cheap N150 device I have for OPNsense is sinked at the case, but has an integrated ext fan to keep the aluminum cool. It is not doing a lot of work, but so far it's been stable.
see https://forum.opnsense.org/index.php?topic=48166.0
Thats not a good solution either. Opnsense appliances should always be passive machines with no moving parts. If you have a tiny fan inside, that thing is piling up the dust as we speak. If you dont clean it regularly, it can die in a few years. Even if its not spinning all the time. I dont want to think about this. I want to setup my firewall and forget that its there.
Quote from: Nullman on May 11, 2026, 06:36:34 PM[...]Opnsense appliances should always be passive machines with no moving parts.[...]
Opinions are like... (There's a "one size fits all" joke in there somewhere.)
There's nothing wrong with a good passive system. It just costs money. (Note the "good".) (And, going off-topic a bit, it's particularly hard to passively cool a 100+W CPU in an 86+F (30+C) environment.)
Quote[...]I dont want to think about this. I want to setup my firewall and forget that its there.
Understandable. A tradeoff I wasn't willing to make. My firewall is not my loudest device, but even if it was, I'd still make the same choice.
Speaking of thermal testing, mprime (https://www.mersenne.org/) is generally more appropriate than memtest, but I don't know of a bootable package that contains a newer version (the UBCD's is a bit old). It's easy enough to fire up a live Linux or FreeBSD image and execute it - it just takes a bit more effort than a bootable package.
Quote from: pfry on May 11, 2026, 08:05:21 PMThere's nothing wrong with a good passive system. It just costs money. (Note the "good".)
If money is issue, go for Qotom. If you cheap out, eventually, its going to cost you more.
Quote from: pfry on May 11, 2026, 08:05:21 PM(And, going off-topic a bit, it's particularly hard to passively cool a 100+W CPU in an 86+F (30+C) environment.)
100W is overkill for bare metal opnsense machine. If you are running virtualizied with bunch of other stuff, thats a different story. We talk bare metal machines.
Quote from: pfry on May 11, 2026, 08:05:21 PMUnderstandable. A tradeoff I wasn't willing to make. My firewall is not my loudest device, but even if it was, I'd still make the same choice.
We are talking normal home use. Not the rack filled with enterprise grade arista quad power redundant switches and 1kW rack mounted pc monsters running hypervisors. Please stay on track.
Quote from: pfry on May 11, 2026, 08:05:21 PMSpeaking of thermal testing, mprime (https://www.mersenne.org/) is generally more appropriate than memtest, but I don't know of a bootable package that contains a newer version (the UBCD's is a bit old). It's easy enough to fire up a live Linux or FreeBSD image and execute it - it just takes a bit more effort than a bootable package.
Going way of topic dude.
Quote from: Irredio_Revancho on May 11, 2026, 01:13:45 AMif it fails memtest86, maybe try again with a big fan directed onto the top of the Topton.
Then try regular deployed use with the big fan blowing on it.
I have been interested in mini-pc devices like this, such as Protectli Vault models, but the lack of active cooling always makes me hesitate.
I wonder, can it really be stable and provide high uptimes?
On at least some models, Protectli includes tightly fitted slabs of aluminum heat transfer blocks between critical components and the topside heat sink. Are blocks like these present inside the Topton models? I have no idea.
But if I were advising someone else, I would say, can lowering the operating temperature by increasing airflow across the heat sink improve system stability at all?
Quote from: Nullman on May 11, 2026, 07:23:15 AMTopton is just unreliable cheap garbage.
Ive seen people buy Toptons like crazy, and then mount a fan on them because their aluminum case can not dissipate heat properly because they use the worst quality materials.
It complete defeats the purpose because this is advertised as passively cooled unit.
Worst of all, their quality is so inconsistent, that some units can work reliably for years, and some die for no reason after few months.
And thats the exact reason why you should avoid them.
I think you guys need to know that there are some "Rules" so to speak about these units :
- There were some quality issues in the past with some of the cheaper Mini PCs.
I am not sure if they were all TopTon models or some others too like Qotom/Protectli/etc. but these should all be fixes by now AFAIK.
Main issues were misaligned heatsinks and/or thermal compound fuckups!
- There were multiple versions of them : A/B/C
In general you want to have a model with Y-shaped heatsink elements on top of the case.
The ones that were I-shaped or simply flat and/or T-shaped were less efficient and sometimes even simply bad !!
- Another issue that some people had when upgrading the amount of RAM or simply ordering them without RAM and wanting to add their own was the compatibility with various RAM modules.
If you want exact details then go and look it up because I can't remember the exact details anymore...
- Something I experienced myself with the used unit that I bought from someone who used the unit for about 6 months : The NVMe SSD made by SKhynix turned out to be unreliable.
Slapped in a Samsung 980 (Pro?! Not sure anymore...) that I had laying around and for almost a year now ZERO ISSUES :)
QuoteEspecially when people that have these units start to recommend them and claim that they have it for X amount of years, and they are working just fine. If you are on a tight budget, go for Quotom. But dont be a cheap ass and go below that. It will cost you more in a long run.
If you happen to be Dutch or feel like using some translating service in your browser go ahead and look around here : https://gathering.tweakers.net/forum/list_messages/1622491/last
You will find experiences with all sorts of units and not everything is as horrible as you seem have experienced it for whatever reason...
People running pfSense/OPNsense/OpenWRT/Full DIY Linux Router/etc. on Protectli/Qotom/TopTon/other brands that I can't remember without any issues at all !!!
Quote from: Nullman on May 11, 2026, 06:36:34 PMIts normal for CPU to run hot, but its the heat sinks job to take that heat away from CPU and then dissipate it.
These units are not doing that because they have crappy metal case with terrible thermal conductivity.
Mostly misaligned heatsinks and thermal paste mistakes from what I have seen and the design of the heatsink fins matter too!
QuoteThis is why on passive Toptons people are forced to install additional fans, not to keep the unit cool, but to keep CPU from thermal throttling and premature damage. It is that bad.
Most units have a spot for mounting a fan, but I have seen people only use them when the Mini PC would be placed in a hot environment or badly cooled place.
In normal cases I haven't seen many people use one to be honest!
QuoteQuote from: BrandyWine on May 11, 2026, 06:17:01 PMThe cheap N150 device I have for OPNsense is sinked at the case, but has an integrated ext fan to keep the aluminum cool. It is not doing a lot of work, but so far it's been stable.
see https://forum.opnsense.org/index.php?topic=48166.0
Thats not a good solution either.
Opnsense appliances should always be passive machines with no moving parts.
If you have a tiny fan inside, that thing is piling up the dust as we speak. If you dont clean it regularly, it can die in a few years.
Even if its not spinning all the time. I dont want to think about this. I want to setup my firewall and forget that its there.
I FULLY AGREE !!! :)
Quote from: pfry on May 11, 2026, 08:05:21 PMOpinions are like... (There's a "one size fits all" joke in there somewhere.)
The thing is that some people have soo strong opinions about certain things that they make things sound worse than they are and that's something to watch out for IMHO :)
Quote from: Nullman on May 11, 2026, 06:36:34 PMQuote from: BrandyWine on May 11, 2026, 06:17:01 PMIs the device overheating on cpu core?
Its normal for CPU to run hot, but its the heat sinks job to take that heat away from CPU and then dissipate it. These units are not doing that because they have crappy metal case with terrible thermal conductivity. The CPU gets hot, starts to throttle, it heats everything inside the unit, and then that heat from the inside is radiating on the aluminum case through your storage, memory modules and PCB. This is why on passive Toptons people are forced to install additional fans, not to keep the unit cool, but to keep CPU from thermal throttling and premature damage. It is that bad. The unit with same specs from either Protectli or Thomas Krenn costs 3 times more. Ask yourself why.
Quote from: BrandyWine on May 11, 2026, 06:17:01 PMMost cpu's will try and protect themselves from overheating.
CPU will almost never die on these units because it has mechanisms to protect itself. It does that with thermal throttling or simply shutting down. Its the other components in that tiny case that will suffer and die eventually. Your storage, your memory modules and worst of all, motherboard.
Quote from: BrandyWine on May 11, 2026, 06:17:01 PMThe cheap N150 device I have for OPNsense is sinked at the case, but has an integrated ext fan to keep the aluminum cool. It is not doing a lot of work, but so far it's been stable.
see https://forum.opnsense.org/index.php?topic=48166.0
Thats not a good solution either. Opnsense appliances should always be passive machines with no moving parts. If you have a tiny fan inside, that thing is piling up the dust as we speak. If you dont clean it regularly, it can die in a few years. Even if its not spinning all the time. I dont want to think about this. I want to setup my firewall and forget that its there.
I was asking how or if the unit from post #1 is overheating?
But just for clarity I will repeat what I wrote, "...
but has an integrated ext fan ...". Just a fan on the "sealed" case. There is no tiny fan inside.
The same device I linked also had a no-fan model. They advertised the unit with fan as "
it runs cooler with fan", etc.
If the ext fan dies or faults, it's a low cost fan and a 3min swap-out w/o ever shutting down the device.
As for dust where there is a fan pulling or pushing air through the device, all kinda depends on if it's filtered and flow rate. My old hefty xeon/asrock PC runs a lot every day, it has two ~5" push-in fans in the front bezel, two 5" pull-out fans in the top of case, fan on the cpu radiator, along with a fan in the PSU, and the nvidia gpu card has mini fans on it (wow, so many fans). They all run at various speeds depending on temp sensors, but usually low rpm's most of the time, it moves a lot of air but does it very quietly. The top of my case is 80% perforated, it's under a desk, not best for spill risk. The inside of the case does get slight dust, filters get dirty, but no dust buildup anywhere. I usually clean the inbound filters every 6mo. I think because of the airflow rate the dust that comes in just gets sucked out.
Quote from: Nullman on May 11, 2026, 06:36:34 PMI want to setup my firewall and forget that its there.
Set It and Forget It ?
That's like speaking for every end-user who plugs anything in and wants it to work 100% w/o ever having an issue. Never gonna happen.
Plus, with OPNsense you'll be updating frequently, no real way around that if you wish to maintain security posture.
Quote from: BrandyWine on May 12, 2026, 05:15:41 AMSet It and Forget It ?
Yes. I dont want to open my device just so i can remove dust carpets. I dont want to lubricate or replace failing/dead fans. I dont want to get my hands dirty with thermal paste.
Quote from: BrandyWine on May 12, 2026, 05:15:41 AMThat's like speaking for every end-user who plugs anything in and wants it to work 100% w/o ever having an issue. Never gonna happen.
Its already happening for me, and im happy.
Quote from: BrandyWine on May 12, 2026, 05:15:41 AMPlus, with OPNsense you'll be updating frequently, no real way around that if you wish to maintain security posture.
I have no problem with that. However, my set it and forget it statement was referring to hardware, not the software.
Quote from: Nullman on May 12, 2026, 09:58:36 PMI dont want to get my hands dirty with thermal paste.
'Special for you my friend' : https://duckduckgo.com/?q=zalman+stg1+paste&ia=images&iax=images
No need to get messy with that good old paste! :P
Quote from: Nullman on May 12, 2026, 09:58:36 PMYes. I dont want to open my device just so i can remove dust carpets. I dont want to lubricate or replace failing/dead fans. I dont want to get my hands dirty with thermal paste.
Is there any fanless hardware that has no holes to let in air & to let out hot air? All my fanless asic devices have holes in them.
Better get the dust out. Blast it through every hole with one of those little battery-powered blowers ;-)
Quote from: passeri on May 13, 2026, 03:08:57 AMBetter get the dust out. Blast it through every hole with one of those little battery-powered blowers ;-)
But I have run them for many years 24/7 and never used compressed air, the units never died. The firmware stopped getting updates, so I moved on to another device.
Plus, reader beware, never blow cold air onto the hot device, must power down and let it get to room temp, and even then the compressed canned air being cold can temp shock components.
Quote from: nero355 on May 12, 2026, 11:54:59 PMQuote from: Nullman on May 12, 2026, 09:58:36 PMI dont want to get my hands dirty with thermal paste.
'Special for you my friend' : https://duckduckgo.com/?q=zalman+stg1+paste&ia=images&iax=images
No need to get messy with that good old paste! :P
Thank you my friend. But as you probably dont know, high end mini pcs are using thermal pads instead of thermal paste. Thats a permanent solution.
Quote from: BrandyWine on May 13, 2026, 01:20:56 AMIs there any fanless hardware that has no holes to let in air & to let out hot air?
Yes.
Quote from: BrandyWine on May 13, 2026, 01:20:56 AMIs there any fanless hardware that has no holes to let in air & to let out hot air?
APU? (outdated now) Protectli?
Look ma, no holes! :-)
@brandywine.
You missed the ascii emoji.
Quote from: Nullman on May 13, 2026, 08:18:32 AMThank you my friend.
Any time! :)
QuoteBut as you probably dont know, high end mini pcs are using thermal pads instead of thermal paste. Thats a permanent solution.
In that case something like this : https://www.igorslab.de/en/thermal-grizzly-phase-sheet-ptm-test-how-good-is-the-honeywell-alternative-really-in-reality/ :)
Quote from: nero355 on May 13, 2026, 05:49:45 PMIn that case something like this : https://www.igorslab.de/en/thermal-grizzly-phase-sheet-ptm-test-how-good-is-the-honeywell-alternative-really-in-reality/ :)
Its seem like you dont know the difference between thermal sheet and thermal pad.
Quote from: Nullman on May 13, 2026, 06:01:30 PMQuote from: nero355 on May 13, 2026, 05:49:45 PMIn that case something like this : https://www.igorslab.de/en/thermal-grizzly-phase-sheet-ptm-test-how-good-is-the-honeywell-alternative-really-in-reality/ :)
Its seem like you dont know the difference between thermal sheet and thermal pad.
Ahh damnit... used the wrong link...
The Honeywell brand has the right one!
/JokeFail... My Bad! :P
Quote from: Patrick M. Hausen on May 13, 2026, 09:19:47 AMLook ma, no holes! :-)
I would want some breathing ability. But ok, they exist.
Verify uCode is loading in.
Verify the PSU is capable in both volts & amps under rated load.
If its a complete freeze, its 100% hardware.
My suggestion put windows on it, use it, let it idle overnight etc, run stress tools, benchmarks etc, try to make it fall over.
I do this will all my new NUC's.
My current N100, I brought with NVME and RAM preinstalled, and it wasnt stable (primary instability was data becoming corrupted on storage silently), however it passed every test I threw at it, so ended up just replacing both NVME and RAM and it has been stable since. The bios's on these units tend to be of low quality, so I wouldnt use any C-state above 1.
If temps in windows dont look great, then do something about it, bear in mind the thing runs hotter in FreeBSD than Windows. I have a couple of tiny fans placed on top of my unit blowing down on it, an easy 10-15C gain, I also have a attenuator attached so the RPM is slow enough that they effectively silent. NVME without this gets extremely hot it basically idles at its limit 70C, and ends up heating everything else inside. (on windows it has an idle state that makes it run cooler, but FreeBSD it idles in full power state)
Ideally use SATA for storage if its possible, in my unit I couldnt as there is no m.SATA and the normal SATA power/data connectors are next to each other making them not possible to use (design flaw on the PCB).
Quote from: chrcoluk on May 24, 2026, 11:40:36 AMIdeally use SATA for storage if its possible, in my unit I couldnt as there is no m.SATA and the normal SATA power/data connectors are next to each other making them not possible to use (design flaw on the PCB).
The reason they do that is because that same PCB is used on a completely different system that has no space constraints and classic SATA drive(s) fit perfectly. Qotom does this. They have mini PCs with PCBs that have 2 or more SATA connectors that you cant really use. I mean, you can, if you dont mind your drives dangling on the cables outside the case. And then, they use same PCB on a system that has huge metal case with drive bays where you can even mount full size SATA hard drives. They cut the costs this way.
Quote from: Nullman on May 24, 2026, 11:56:46 AMThe reason they do that is because that same PCB is used on a completely different system that has no space constraints and classic SATA drive(s) fit perfectly. Qotom does this. They have mini PCs with PCBs that have 2 or more SATA connectors that you cant really use. I mean, you can, if you dont mind your drives dangling on the cables outside the case. And then, they use same PCB on a system that has huge metal case with drive bays where you can even mount full size SATA hard drives. They cut the costs this way.
There is space in the case for the drive, the issue is that the power connector and data connector are so close together they cant be both in at the same time.
TopTon have done a board revision on later revisions, I posted a picture of my board on I think is snbforums and my PCB layout is different to later models where the connectors have been moved apart. Basically if I forced both cables in it likely would have ripped one of the connectors of the PCB or at least bent it partially off.
Topton is just garbage. This confirms it even more.
Quote from: chrcoluk on May 24, 2026, 07:30:45 PMthe issue is that the power connector and data connector are so close together they cant be both in at the same time.
And it's not one of those connector combinations that need a special cable when you want to actually use them ?
I think my TopTon has the same one that one of my older AsRock motherboards use :)