Yeah, that's fine. If it sits that high at idle, then you may need to do something, but for full load that's actually pretty good. For comparison, my big workstation (16-core, water cooled) will hit 85-90 C when I load it up with a 2-million-line compile ("make -j 32"), but drops back to 35-45 C (very dependent on ambient temp in the office) when it's done... My N5105 with bare metal OpenWrt sits at 40-42 most of the time, can spike up to 60 when I torture it (but routing tasks are pretty light torture), so yours look normal.
Topton Jasper Lake Quad i225V Mini PC Report
- Thread starter ReturnedSword
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
I had an issue where if I assigned all the RAM to the single VM that I planned to run originally. It would crash when running qm start.
Baremetal running 23.05 on a Topton N5105 system (16G RAMB/128GB NVME)
pfSense is using 16.1GB+ when in reality it should be 4.7GB or so.
Looks like my snapshots went very rogue...I deleted everything except ONE, but I did get an "Error 5" when deleting SOME if them. Didn't think much of it as they were gone from GUI Boot Environments page.
Now I'm seeing this
I suspect the snapshots left some referenced space I don't know how to recover
Excuse the dumb formatting, I'm blaming the stupid 7year old iPad.
pfSense is using 16.1GB+ when in reality it should be 4.7GB or so.
Looks like my snapshots went very rogue...I deleted everything except ONE, but I did get an "Error 5" when deleting SOME if them. Didn't think much of it as they were gone from GUI Boot Environments page.
Now I'm seeing this
zfs list -r -o space
NAME AVAIL USED USEDSNAP USEDDS USEDREFRESERV USEDCHILD
pfSense 98.2G 16.1G 0B 96K 0B 16.1G
pfSense/ROOT 98.2G 4.66G 0B 96K 0B 4.66G
pfSense/ROOT/June25_everything_working_zfs_cleaned 98.2G 4.66G 2.21G 1.35G 0B 1.09G
pfSense/ROOT/June25_everything_working_zfs_cleaned/cf 98.2G 61.2M 57.0M 4.15M 0B 0B
pfSense/ROOT/June25_everything_working_zfs_cleaned/var_cache_pkg 98.2G 1006M 1006M 144K 0B 0B
pfSense/ROOT/June25_everything_working_zfs_cleaned/var_db_pkg 98.2G 50.3M 43.5M 6.79M 0B 0B
pfSense/cf 98.2G 192K 0B 96K 0B 96K
pfSense/cf/conf 98.2G 96K 0B 96K 0B 0B
pfSense/home 98.2G 96K 0B 96K 0B 0B
pfSense/reservation 110G 96K 0B 96K 0B 0B
pfSense/tmp 98.2G 96K 0B 96K 0B 0B
pfSense/var 98.2G 5.25M 0B 3.21M 0B 2.04M
pfSense/var/cache 98.2G 104K 0B 104K 0B 0B
pfSense/var/db 98.2G 936K 0B 936K 0B 0B
pfSense/var/empty 98.2G 96K 0B 96K 0B 0B
pfSense/var/log 98.2G 852K 0B 852K 0B 0B
pfSense/var/tmp 98.2G 104K 0B 104K 0B 0B
[admin@pfSense.home.arpa]/:
I suspect the snapshots left some referenced space I don't know how to recover
zfs list -t snapshot
NAME USED AVAIL REFER MOUNTPOINT
pfSense/ROOT/June25_everything_working_zfs_cleaned@2022-08-21-19:58:38-0 64.1M - 1.00G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2022-09-04-13:41:30-0 47.8M - 1.04G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2022-09-05-16:41:47-0 35.4M - 1.01G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2023-01-05-16:02:47-0 53.5M - 1.02G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2023-02-16-18:05:42-0 33.6M - 1.26G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2023-02-21-10:19:53-0 33.4M - 1.26G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2023-02-28-16:21:03-0 33.9M - 1.26G -
pfSense/ROOT/June25_everything_working_zfs_cleaned@2023-05-27-13:47:01-0 57.6M - 1.35G -
pfSense/ROOT/June25_everything_working_zfs_cleaned/cf@2022-08-17-19:44:46-0 236K - 268K -
-----snip------
Excuse the dumb formatting, I'm blaming the stupid 7year old iPad.
Last edited:
Curious if others are seeing ~3% IO Wait on their Topton units under PVE? I’ve tried different nvme SSDs with no difference. I’m using the primary M.2 slot though tried the other via their ‘H’ bracket to the Wi-Fi slot, same result. I’m primarily using this for a Sophos XG firewall/router as a guest vm along with a Home Assistant guest vm. Everything is acceptable just curious what others are seeing. Thanks in advance
with my Topton box (Pentium Silver N6005), with 6 containers and 6 VM's running, I'm not seeing that sort of numbers. IO delay is 0,07% on average
Quick question, as I'm a bit confused as to the V1 .. V5 designations for these boards.
I have a Bluetech board but with 4x i226, GF-1338NP-12-V13. Is this still considered V2?
I also see there might be a V14 BIOS for this unit, anyone have a download link please?
Thanks!
I have a Bluetech board but with 4x i226, GF-1338NP-12-V13. Is this still considered V2?
I also see there might be a V14 BIOS for this unit, anyone have a download link please?
Thanks!
Thank you! I’d hoped my experience wasn’t normal, but I’m at a loss on why. Could you share what brand/model ssd you have? I too am at the latest kernel and recently upgraded to PVE v8 with no impact there. I doubt it’s anything in BIOS as I made some changes to defaults like enabling ASPM on the pcie lanes for these slots.
My own box (N5105) with a 256 GB SM961 (Samsung 960 Pro OEM) has hovered around 3% since forever. So far i only have a couple VMs on it, but the CPU usage is almost as high as the IOwait. Also, it doesn't increase with any CPU load, so i'm not too concerned about it.
A couple days ago, after the TopTon had my mini pc's motherboard for a month, I raised a ticket with aliexpress as I wasn't getting anywhere. In the end, I managed to get a full refund (discussion here). So, now I have an heatsink to reuse for something else, and the power supply itself.
As I had done some prior testing on the PSU in this post, today I've decided to actually take it apart and have a look at it, most importantly to see if it's a genuine LiteON and what components are inside.
Overall I've been quite surprised. The power supply looks to be an actual LiteON, as both the pcb and transformer are clearly marked. The overall build quality is very good. I've since closed it back again, I'll keep it as a spare or use it for other DIY stuff.
Those are the main components:
Input protection: two fuses, a general 6.3A which protects the MOV and a 3.5A for the psu itself
input filtering: common mode choke and a single polypropilene 0.22µF on the input side.
Bridge: KBP208G, 2A 800V.
Main capacitor: 130µF 400V Nichicon 105°C pt series.
Main switch: AOTF8N65, N-Channel MOS, 750V, 8A 1.15ohm RDS(on)
Transformer: Lite-ON SL130
Secondary rectification: SBR20A100CT
Secondary Caps: Taicon 16V 1000µF x2 + Nippon Chemicon 16V 220µF after the filtering inductor. All caps are 105°C rated.
Secondary protection: S8M02-600B across the output, as a crowbar. Possibly an OVP protection?
The whole back of the PCB the is conformal coated, the main IC is a SOT23-6 which I was not able to read anything off. There is a very good separation between primary and secondary.
Some pics:
As I had done some prior testing on the PSU in this post, today I've decided to actually take it apart and have a look at it, most importantly to see if it's a genuine LiteON and what components are inside.
Overall I've been quite surprised. The power supply looks to be an actual LiteON, as both the pcb and transformer are clearly marked. The overall build quality is very good. I've since closed it back again, I'll keep it as a spare or use it for other DIY stuff.
Those are the main components:
Input protection: two fuses, a general 6.3A which protects the MOV and a 3.5A for the psu itself
input filtering: common mode choke and a single polypropilene 0.22µF on the input side.
Bridge: KBP208G, 2A 800V.
Main capacitor: 130µF 400V Nichicon 105°C pt series.
Main switch: AOTF8N65, N-Channel MOS, 750V, 8A 1.15ohm RDS(on)
Transformer: Lite-ON SL130
Secondary rectification: SBR20A100CT
Secondary Caps: Taicon 16V 1000µF x2 + Nippon Chemicon 16V 220µF after the filtering inductor. All caps are 105°C rated.
Secondary protection: S8M02-600B across the output, as a crowbar. Possibly an OVP protection?
The whole back of the PCB the is conformal coated, the main IC is a SOT23-6 which I was not able to read anything off. There is a very good separation between primary and secondary.
Some pics:
Last edited:
Just chucking in my own experiences. I have a V5 N5105 i226-V, running Proxmox, OPNSense and a selection of containers doing various activities. Nothing with tremendously high workloads and it's been extremely stable over the last two months or so.
I'd noticed that CPU temperatures were quite...robust:
- Idle: 52C
- Transferring 2.5Gpbs through the firewall: 89C
Stripped down and observed that it had copious paste on the (copper) CPU heatsink and a thermal pad on the (aluminium) NIC drivers heatsink. While there was no gap visible, you could minutely move the mobo when it was pressed, suggesting that there was an airgap. I stripped it down and filed around 0.5mm from each of the standoffs, cleaned everything down (the heatsinks were well finished and did not need to be sanded) and reassembled with a more sensible amount of MX-4. Tests were...satisfactory ;-)
- Idle: 28C
- Transferring 2.5Gpbs through the firewall: 42C (*and* more stable transfer speed)
I notice that there is a spare, unused standoff hole on the motherboard near the CPU which would be perfect for keeping a more consistent pressure on the heatsink. I will probably fit a standoff to this when I have a moment.
One interesting observation: the case is much, much cooler now which surprised me. I'd have thought that this was a function of the power being dissipated, but it's clear that improving the heat transfer efficiency has clearly improved this too. No, I don't understand why.
I'd noticed that CPU temperatures were quite...robust:
- Idle: 52C
- Transferring 2.5Gpbs through the firewall: 89C
Stripped down and observed that it had copious paste on the (copper) CPU heatsink and a thermal pad on the (aluminium) NIC drivers heatsink. While there was no gap visible, you could minutely move the mobo when it was pressed, suggesting that there was an airgap. I stripped it down and filed around 0.5mm from each of the standoffs, cleaned everything down (the heatsinks were well finished and did not need to be sanded) and reassembled with a more sensible amount of MX-4. Tests were...satisfactory ;-)
- Idle: 28C
- Transferring 2.5Gpbs through the firewall: 42C (*and* more stable transfer speed)
I notice that there is a spare, unused standoff hole on the motherboard near the CPU which would be perfect for keeping a more consistent pressure on the heatsink. I will probably fit a standoff to this when I have a moment.
One interesting observation: the case is much, much cooler now which surprised me. I'd have thought that this was a function of the power being dissipated, but it's clear that improving the heat transfer efficiency has clearly improved this too. No, I don't understand why.
Have a look at the thermal resistance model on heatsinks and dissipation of these devices.
Before filing your standoffs, the thick thermal paste layer was acting as a big "thermal resistance" creating a large temperature hike before the heat could reach the heatsink and be actually dissipated away.
Thermal paste is really efficient only if it's applied in a very thin layer, once it gets thick it can turn into an hindarance, especially with devices like those which are direct-die and have a lot of heat concentrated in a small area.
Yesterday my device died. I investigated why it idled at 54°C and I was expecting bad contact or bad thermal paste application. Apparently it was three thick layers of thermal paste: one on the aluminium cover, one on the copper block and one on a thin copper shim.
After cleaning and applying thin layers of good thermal paste, I reassembled it and attached it to all connectors. Normally it beeps for a successful POST but it didn't. I turned it off to investigate, apparently the CPU didn't contact the copper heatsink or shim, about half a millimeter between it.
Apparently the VRM cooling (aluminium strip above the heatsink was also slathered with a thick application of thermal paste and a thermal pad. Cleaned it, applied a thin layer of paste and noticed the CPU still doesn't touch the copper heatsink. So I removed the thermal pad, applied a thin layer of paste and it doesn't even touch the chips. Reapplied another thermal pad than. Checked to make sure the CPU was making contact and it seems to do so. I'm not suprised these tolerances result in so much temp and performance variation.
But it doesn't power on. It supplies power as USB powered devices receiver power, but the power button doesn't light up. So I started configuring another router as this was getting me nowhere. This was last night.
Let's check today, while I'm typing that last paragraph and before I go look for a new permanent solution, lets just give it another go.
Attach the PSU and on it goes. What ?!
So I check temps and the idle is still bad, even a little worse. It's probably still not making good contact. I put a little more thermal paste on it and the idle temps are floating around 50°C so a small improvement. Maybe I need to tweak the power settings some more ? But at this point I'm just happy I don't have to find another router (yet).
After cleaning and applying thin layers of good thermal paste, I reassembled it and attached it to all connectors. Normally it beeps for a successful POST but it didn't. I turned it off to investigate, apparently the CPU didn't contact the copper heatsink or shim, about half a millimeter between it.
Apparently the VRM cooling (aluminium strip above the heatsink was also slathered with a thick application of thermal paste and a thermal pad. Cleaned it, applied a thin layer of paste and noticed the CPU still doesn't touch the copper heatsink. So I removed the thermal pad, applied a thin layer of paste and it doesn't even touch the chips. Reapplied another thermal pad than. Checked to make sure the CPU was making contact and it seems to do so. I'm not suprised these tolerances result in so much temp and performance variation.
But it doesn't power on. It supplies power as USB powered devices receiver power, but the power button doesn't light up. So I started configuring another router as this was getting me nowhere. This was last night.
Let's check today, while I'm typing that last paragraph and before I go look for a new permanent solution, lets just give it another go.
Attach the PSU and on it goes. What ?!
So I check temps and the idle is still bad, even a little worse. It's probably still not making good contact. I put a little more thermal paste on it and the idle temps are floating around 50°C so a small improvement. Maybe I need to tweak the power settings some more ? But at this point I'm just happy I don't have to find another router (yet).
@MarcoP82 Have you tried some thin copper pads? My v3 unit too had a gap between the copper block and the CPU. Jammed a thin copper pad between them, works better than bridging the gap with thermal paste.
I bought these, 0.8mm and 1.0mm
I bought these, 0.8mm and 1.0mm
10x laptop CPU GPU copper pads shim copper slide thermal conductor pad 20x20mm 0.3-1.2mm | eBay
Find many great new & used options and get the best deals for 10x laptop CPU GPU copper pads shim copper slide thermal conductor pad 20x20mm 0.3-1.2mm at the best online prices at eBay! Free shipping for many products!
www.ebay.com
Hi all. It seems everyone is adding their questions to this super-long thread, so I'll do the same: I bought this one: topton 12th gen i3 N305 8 Core mini-pc with 80x80 mm fan holes - and have now tried to buy 2 wrong fans, both have been returned now - the reason is the incredibly lame/incompetent chinese "support" (they wrote I needed 4010 fan, which is completely wrong and also they didn't reply until very late if I could use 12V PWM or 5V, so after reading in the forums I bought 4010 12V PWM fan - and had to return it, because it was wrong information they gave me - next I bought 8010 12V PWM fan, because I didn't receive any replies from the chinese aliexpress support and I also had to return that!)... Now I don't want to make more mistakes: The chinese aliexpress support wrote to me that I need a 8010 5V fan (several times, they've ignored my question about whether it was possible to use 12V PWM and they still ignore this question and just write "8010 5V" - no answer, I assume they're trying to tell that I must buy a 8010 5V fan)... I don't want to return more fans, I want to get it right this time. Does anyone here know if I should/can use:
1) https://www.amazon.de/-/en/Pack-Dou...=industrial&sprefix=801,industrial,645&sr=1-2
2) Pocaton 5V USB DC Cooler Fan 80mm x 80mm x 10mm Computer Case Fan PC Latop CPU Fan Heatsink : Amazon.de: Computer & Accessories
???
Both are 8010 5V fans - but I prefer without USB-cable, so I have an extra USB-connection. I read somebody also had to use a 3- or 4-pin adapter, because the internal 4-pin fan connectors on the motherboard are very small and non-standard (I think)... Anyone can help, so I get it right this time and don't have to waste more time on returning wrong fans?
1) https://www.amazon.de/-/en/Pack-Dou...=industrial&sprefix=801,industrial,645&sr=1-2
2) Pocaton 5V USB DC Cooler Fan 80mm x 80mm x 10mm Computer Case Fan PC Latop CPU Fan Heatsink : Amazon.de: Computer & Accessories
???
Both are 8010 5V fans - but I prefer without USB-cable, so I have an extra USB-connection. I read somebody also had to use a 3- or 4-pin adapter, because the internal 4-pin fan connectors on the motherboard are very small and non-standard (I think)... Anyone can help, so I get it right this time and don't have to waste more time on returning wrong fans?
I'm currently installing a fan on the same mini PC you have (i3-N305 with 4x LANs). The fan is 12V, but the connector is not a standard CPU connector. While researching for other 4-PIN standards I found out there is also a micro 4-pin (PH2.0mm) connector, for example, CRJ Micro 4-Pin PWM adapter.
PS: I bough this Fan 80x80x10mm SC8010KH12H-P from aliexpress
I will let you know if such a similar adaptor works, it might take 2-3 weeks to get the order
@newsboost have you seen this thread?
Sorry for my delayed answer (I'm currently travelling). I'm just fed up with that almost non-existing aliexpress "support", that either don't want or cannot provide detailed information about anything, despite numerous attempts from my side... I asked the seller from aliexpress many times and they kept refusing to answer if I could use a 12 V fan - instead they wrote: "Use 5V fan" and ignored that I asked if 12V PWM would also work or not (because I've also seen that recommendation in the forums here, that's why I'm so confused and I cannot even see what each of the 4 pins do, e.g. GND, 5V, 12V, PWM or whatever)... I would very much appreciate to hear if that 80x80x10 mm fan works, yes, please write back once you receive it! I myself bought a 120mm USB-fan to put outside, which I think is good - but I made the mistake to buy my minipc with chinese RAM and SSD-drive and my minipc should be on 24/7 and therefore I would also like to install an 80x80x10 mm fan, to keep everything healthy and at low temperatures... Looking forward to hear if you can make it work with 12V, when the "sellers support" wrote to me I should use a 5V 8010-fan...? So also now, I'm confused...I'm currently installing a fan on the same mini PC you have (i3-N305 with 4x LANs). The fan is 12V, but the connector is not a standard CPU connector. While researching for other 4-PIN standards I found out there is also a micro 4-pin (PH2.0mm) connector, for example, CRJ Micro 4-Pin PWM adapter.
PS: I bough this Fan 80x80x10mm SC8010KH12H-P from aliexpress
I will let you know if such a similar adaptor works, it might take 2-3 weeks to get the order
I don't remember if I saw exactly this thread, because I've tried to search a lot but it's still just so very confusing: I think there are so many different versions now of these minipc's that even if something works for someone, it might not work for a guy who buys a similar pc a year after, because then there's a new version out... And they don't come with *any* useful technical specs, which I think is extremely disappointing - so maybe I should just buy this: PH2.0 Mini 4Pin to PWM Fan GPU - then again, some people (from your link) talks about JST connector and one writes "It looks to be a JST with 1.25mm pitch" and the other writes "for me looks like JST ZH 1.5mm" - aaargh! So currently I really don't know for sure what to buy and there's long delivery time from Asia to Europe so I'm hoping someone could tell for sure what it is we need, but I fear that they keep changing the hardware so what worked >1-2 years ago won't work with the newer minipc's they ship... But it does sound like we bought our minipc's at around the same time, so I'm very eager to hear your experience, once you receive the fan and if you can install the 8010-fan 12V PWM without problems, thanks a lot for your feedback/input!
I have connected on both my units an N6005 and an N305 a 5v fan and made a connection with 1.25mm JST.
This is what I have purchased to connect:
This is what I have purchased to connect: