I wouldn't worry too much about the included brick. Mine was from a different seller, but a no-name 4A brick with a fake "Delta" sticker. I ordered a genuine Delta 3A brick (ADP-36PH A) to see if it was any better, but the original PSU was actually more efficient and seemed to be a better circuit design. At least it had a more complex output filter.. So I'm just using the original unit now, and wish I hadn't spent $15 on another PSU.Also can someone please recommend a compatible replacement AC to DC power brick adapter? The one provided they looks a bit sketchy.
By the way, what's the make and model of the 64GB of memory did you use?I now have 3 of these interesting little boxes:
1. N6005+4x i225 CW-N6000 (Topton) - This one has been my primary firewall device since May.
2. N5105+6x i225 CW-N11 (Topton) - This one has been a backup/toy.
3. J6413+6x i226 CW-J6-L6 (CWWK) - Just received this model this week.
The N6005 unit has, thus far, been the lowest power draw of them all sitting around 7.8w while mostly idle (just background IoT traffic).
The N6005 and N5105 both only "officially" support 16GB. As long as you don't have a monitor plugged in, they'll run with 64GB. Try to do anything with graphics and they'll fall over with >16GB installed.
The J6413 officially only supports 32GB according to Intel's product description page, but I've had absolutely zero issue with running this system with 64GB. The unit came with a BIOS dated 9/XX/2022 (I forget the exact day) and it appears to have had issues exposing the PTT/TPM to the OS. Downloaded a new BIOS from Changwang's website dated 11/11/2022 and the TPM is now exposed (and I think the new BIOS added the Restore after AC/Power Loss option - at least I couldn't seem to find it on the old BIOS). So far I've been able to install Windows 10, Windows 11, Fedora 37, pfsense, and Proxmox for various testing purposes.
The N6005 and N5105 both came from Topton and I had to repaste those units because both would immediately spike to 93C under prime95. CWWK seems to have done a much better job and the J6413 did not require repasting. The prime95 temps looked good although I don't remember what the numbers were off the top of my head. Using 1 stick or 2 sticks of memory didn't cause power usage to creep upwards, and adding a second stick of memory also didn't make an appreciable difference in memory bandwidth, at least in terms of the numbers memtest86 reported. Haven't had a chance to run something like Sandra yet to get more accurate numbers.
The J6413, in about a half a day of Proxmox testing, has not experienced the issue that I have with the N6005 and N5105 models. On the N-models, with the combination of pfsense on Proxmox with C-states enabled in BIOS and PowerD enabled in pfsense, I can cause those hosts to lock up within a few minutes. Proxmox + C-states disabled + PowerD disabled in pfsense has been stable for me for months at a time on the N-models. Bare metal pfsense on the N-models with C-states enabled and PowerD enabled (hiadaptive) also performs flawlessly.
I really like that the J6413 doesn't act sketchy with 64GB of RAM (memtest86 verified). The thing I don't like about it is that it takes mSATA rather than m.2 as the market for mSATA is pretty barren. CWWK shipped a SATA+power cable combo with the unit but the power cable has the wrong plug type. The unit provides a mini 2-pin for SATA power but the cable they shipped wants a 4-pin.
There's a very small on/off toggle on the J6413 that's described as "incoming call self-start switch" so I guess if you opt to put a wireless and sim card in it, you can remotely power it up.
All three units let you disable Intel ME from BIOS.
Can't help any with running a N6005 inside an enclosure like you're showing but I purchased the NUSN RJ03L with the N6005 mini. My open air testing results may provide some insight on what to expect. See my post:I'm considering buying either a 4x i226 N5105 or N6005 (barebone) unit, but curious if anyone got some experience wrt. thermals when placing the device inside a closet/wall box for networking devices? Most of the reviews I've read are based on placing the devices on a desktop, where there at least is some air circulation.
All the ethernet cables in my house are wired to a box similar to this ( , but only larger - the one I got is about 40cm width, 80cm height, and 7cm depth) that is embedded into the (dry)wall, thus, not too much air circulation in there. Today I got a Ubiquiti EdgeRouter X-SFP in there, together with a 8p L2 managed switch, a RPi4b and the fiber ONT. My plan/hope is to replace the ERX-SFP and the RPi4b with a one of these mini PCs running Proxmox with OPNSense/PFSens VM and one Ubuntu (or similar) VM for the Unifi controller+PiHole. Don't plan to do much else than what I do today (a couple of VLANs for main network/IoT/guest network/office network, etc). I do plan to setup a VPN so I can connect to it when on the out on business trips, etc, but no high throughput otherwise.
Also, is there any advantages buying from one "store/seller" to another except for that price may vary a bit? E.g. I see that CWWK Store explicitly mentions V5 (pcb?) in some of their advertisements, whereas I can't see it being mentioned for e.g. the Topton ones on Ali.
I was first thinking of this (Topton , 4x i226, €210 for N6005, €157 for N5105), a bit due to form factor, about 1cm shorter in height compared to e.g. this (CWWK, 4x i226, €242 for N6005, €184 for N5105, this also explicitly mentions V5 for pcb), but then again, the extra 1cm will probably be handy wrt. thermals, but I haven't heard people mention this CWWK store that much compared to Topton when reading reviews over the last year.
If anyone have experience with placing the devices inside a closed environment with less air circulation I'd be happy to hear some details about that. Also, any suggestion on device/store/seller (e.g. if there's anything I really should avoid, etc) is also greatly appreciated.
This module is the one that I am using a pair of in the J6413.By the way, what's the make and model of the 64GB of memory did you use?
This is a very relevant setting for powersavings, from my measurements I was able to determine that without option 3 the system never sleeps deeper than C3. With the ASPM set to auto i reaches easily C8.Figured I would post an update of the bios config I'm using with the N5100 box. Running OPNsense bare metal, with a pretty basic config, the idle power is 5.4W from the wall.
Chipset Tab :
3. PCH-IO > PCIe Config > Root Ports 1,4,5,6,7,8 > ASPM = Auto
4. PCH-IO > Sata > Disabled (if you are not using a sata drive)
I think the main advantage is the lower TDP. According to Intel Ark 1.95W for the i225V vs 1.3W for the i226V. Unclear if the i225 datasheet is for v1, 2 or 3, don't know if there are any differences between these in terms of thermals. Intel i219 has 0.5 W. So there is a little bit less heat which is always good with a passively cooled device.What is the consensus on these i226 nics vs the i225? I know for pfsense it's a little tricky but are these nics better in any way?
From a pfsense/opnsense perspective, i225 and i226 are functionally identical. Power difference seems to be negligible (1.95W TDP vs 1.3W theoretical) and actual behavior is the same. I think i225 exposed some unexpected behaviors in a lot of cheap switches like tp-link and d-link. I have some 1g switches that both i225 and i226 won't negotiate more than 100mbit on. HP and Aruba 1g switches, no problem at all. Netgear switches have been hit or miss. Aquantia/Marvell multi-gig NICs don't have issues with those same switches so its hard to point a finger at the switches.What is the consensus on these i226 nics vs the i225? I know for pfsense it's a little tricky but are these nics better in any way?
Glad it worked for youThis is a very relevant setting for powersavings, from my measurements I was able to determine that without option 3 the system never sleeps deeper than C3. With the ASPM set to auto i reaches easily C8.
Topton N5105 (FMI07 N5105 JK 4L-V02 CBOX )
C3 = 7.6 watts
C8 = 5.0 watts
Thanks for sharing this info.
I used a USB to Ethernet adapter to get 2.6 running then upgraded to 22.05. Worked fine. Did the same to test the Beta 2.7 version.yes it is, but then you need to go trough the chicken-egg phase. 22.05 doesn't have a native installer, its only 2.6.0 CE and after installing that you have no nics working.
Thanks for the link to your testing, i had missed that for some reason!Can't help any with running a N6005 inside an enclosure like you're showing but I purchased the NUSN RJ03L with the N6005 mini. My open air testing results may provide some insight on what to expect. See my post:
https://forums.servethehome.com/index.php?threads/hunsn-model-rj03l-w-n6005-v5-w-4-i226v-nics.38464/
For me, this box will only be running pfsense for now but I have tested (and stress tested it) with Win10 Pro and Win11 Pro and stressed it with Prime95. A fan on the top blowing down onto the aluminum fins get's great results and lowers the CPU temps by 10°+ in open air. You can get USB powered 60mm x 10mm and even 80mm fans that can be mounted onto the top:
( https://www.amazon.com/Ronyaoko-80x10mm-Brushless-Connect-Computer/dp/B08Z7JH38S/ref=sr_1_4?crid=2X1VF5RT673TZ&keywords=80mm+x+10mm+usb+powered+fan&qid=1671756490&s=electronics&sprefix=80mm+x+10mm+usb+powered+fan,electronics,71&sr=1-4 )
Then depending on your configuration, you could put a 60mm x 10mm 5/12V fan inside to help cool the NVMe.
I have experience with equipment being used in enclosures used in the ITS industry (Intelligent Transportation Systems) for both inside and outside use. You're gonna need airflow.... The temps in that enclosure will easily get to 120°F plus. As I mentioned in my post, the case got up to 125°F without any fans after several hours of stress testing. Add in your other equipment and you'll soon have a nice slow cooking oven. The vents on your box are great but you need air flowing to keep temps within operating range - or risk shortening the life of the equipment.
BobS