CWWK/Topton/... Nxxx quad NIC router
- Thread starter Becks0815
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
20% PWM should be 51, not 75 (~29%). I'll poke around in the BIOS some more. Or maybe I'll need to switch over to voltage control instead of PWM.
Oh, dangit. I found this spec sheet from Intel that says the fans must run at a minimum 30% RPM. Fan behavior below that is defined by the fan manufacturer. https://www.intel.com/content/dam/support/us/en/documents/intel-nuc/intel-4wire-pwm-fans-specs.pdf
So it looks like I'm out of luck with these fans which will run at their minimum RPM with no further ramp-down or stop (which is optional).
Thanks for pointing me in the direction of the PWM fan control spec.
Does anyone have any experience with or thoughts on either of the following two models? Going through this thread a bit I don't see either mentioned, but maybe I missed something. Intended use would be OPNsense on bare metal or possibly on top of Proxmox.
To me the standout feature over similar units is the console/serial port. Also the first one looks to have a x2 link to the m.2 NVMe slot instead of the typical x1. On the downside, I have reservations about the cooling ability of the case heatsink on each.
To me the standout feature over similar units is the console/serial port. Also the first one looks to have a x2 link to the m.2 NVMe slot instead of the typical x1. On the downside, I have reservations about the cooling ability of the case heatsink on each.
Unless those have active cooling, I would expect them to have throttling issues. That is probably why people aren't talking about them much, although I think there was a separate thread about them where someone complained about heat issues.
If you are running these boxes as a router/firewall, I imagine faster SSD speeds to be mostly pointless. What you will be writing or reading will be small files so what matters would be latency, not raw read/write speed.
I'm looking to buy that top one also, with 8gb ram and 128 m.2 I'm torn on which os to use. OPNsense, PFsense or IPfire. but I think it will be a good box. Does anyone have one or similar, that has heat issues. I see varied reports from the other models.
I found the thread I mentioned before. See below.
It seems pretty similar to the one below.
The gallium in liquid metal will migrate slightly into the copper, permanently staining it. However, this has no impact on thermal performance and does not compromise the integrity of the copper IHS and/or heatsink. It also doesn't continue indefinitely- the copper will absorb gallium to a certain depth and then stop.*
It does, however, mean that if you just apply once and go, you'll have (VERY VERY VERY) slightly less liquid metal after a few months, and that liquid metal will have (VERY VERY VERY) slightly less gallium and more (indium? tin? idk, ask Thermal Grizzly) than when new.
The upshot is that when using liquid metal on copper, for absolute maximum performance you should reapply it once or twice after 1-2 month intervals, but once the staining is complete, it'll be stable, and the final (usually second or third?) application of liquid metal should be good for years.
Not doing so is extremely unlikely to cause any actual damage; however, if you're bothering with liquid metal at all you might as well get the most out of it you can.
*My approximate understanding is that the copper and gallium form a stable complex or alloy of some kind that stops further gallium ion migration from the liquid metal into the pure copper once it hits a certain thickness (several micrometers? half a millimeter? idk), but I'm considerably more vague on the details of the metallurgy than the practical outcome.
I've got a CWWK N100 6 port router that I'm using for OPNSense. The BIOS version is CW-ADLN-6L Ver: 1.4. I've run iperf3 to check throughput and although it starts off ok (2.3Gbps), the throughput drop off significantly after 5-10 seconds. This is without any additional processing like Zenarmour enabled. From what I can see, the CPU frequency drops to 400-800mhz at this point. It doesn't appear to be temperature throttling as the core temps never get beyond 38°C. Someone has suggested that the PL1/PL2 settings may need tweaked. However I don't see these in the BIOS. Does anyone have a version of the BIOS for the 6 port router with PL1/PL2 settings enabled?
I found cwwk response to this is to return the product. Just purchased this unit and haven't even installed anything so this is concerning.
This theoretical information doesn't match my practical results, using a 360mm water cooling system and a 5800x3d CPU. After repasting it twice within 3 months, in all cases the liquid turned into solid metal film, baking the CPU and cooler into one block. It always took a while to separate each other. On top, this combo showed a very bad cooling performance after a short period.
I replaced everything with standard cooling paste from Grizzly, and ended in .5°C to 1°C higher temps (73°C vs. 72°C) under full load of the 5800x3d, drawing 100W.
If you take these results as starting point, there is zero chance you gain anything on a CPU drawing 20W peak, and then dropping to maybe 12W power consumption. The limiting factor always is the case and not the heat transfer from the CPU to the cooling block. You will have to attach a 120mm fan on top of the and constantly blow cool air over the cooling fins to cool it down as much as possible, and then you might end with a CPU core temperature of 60°C under full load, if you run the system long enough under full, far away from the critical one where it slows down to avoid damage.
On the other side you will end with a system which requires constant maintenance, doesn't run at a lower temperature and with the risk of creating shortcuts with the liquid metal spilling somewhere you don't want to have it. So imo: not worth at all. I have attached a cooler on top of the case and used an ESP8266 to trigger it whenever the case reaches 43°C, and it only reaches this temperature when I transfer high amounts of data over an extended period of time.
Last edited:
I have the first one and it's running pfSense on Proxmox. I was a little worried about the cooling capabilities of the case as well but I just put a USB powered fan on top of it like I did with a similar N5105 box and I haven't had any issues with heat.
The N100 is at 33 C and pulling 15w at idle.
I sent an email to CWWK but I’ve not heard anything back. Has anyone wall mounted these boxes? I’m looking to either 3D print a bracket or find one to buy that’s metal. Just looking to see what others have done. Have a single n100 and three n305 on the way. Thanks!
What email did you use to contact them?
I tried support@cwwk.com but got bounce back in gmail inbox.
Code:
"Delivery incomplete. Maximum number of MX connection attempts exceeded"Does anyone know how to auto startup the device after a powerloss? I just can not seem to find it in the BIOS.
Addtition:
After quite some time searching, there seems to be a little switch at the back which allows to unit to autoboot, set it to ON.
I also switched the jwpr jumper on the board, seems to proved power to the USB ports at all time.
Addtition:
After quite some time searching, there seems to be a little switch at the back which allows to unit to autoboot, set it to ON.
I also switched the jwpr jumper on the board, seems to proved power to the USB ports at all time.
Last edited:
Oh wow. Good find. Anyone else able to confirm that the switch is for autoboot after power losses?
Reporting my experience with a 2 port N100 CWWK model - CW-X86-P5.
Installed an Intel Optane P1600X 58GB NVME drive, Crucial 32GB 4800MHz CT32G48C40S5, and using stock power supply.
Memtest failed pretty quickly. Temperature was warm but not HOT. I installed proxmox anyway, fired up a VM and let a lot of slowly running packages install overnight. The unit was crashed in the morning. I tried adjusting several BIOS settings, focusing on reducing power consumption, with the same crashing results.
Next I ordered a Crucial 8GB 4800MHz CT8G48C40S5 and installed it. Memtest passed with flying colors, no issues in Proxomx.
I’m a little bit at a loss as the specific 32GB stick model was reported as working up thread. I may just have a bum stick or maybe my CWWK unit is finicky. Next troubleshooting step for me is a replacement 3rd party power supply, and maybe trying a 48GB ram stick for giggles.
Installed an Intel Optane P1600X 58GB NVME drive, Crucial 32GB 4800MHz CT32G48C40S5, and using stock power supply.
Memtest failed pretty quickly. Temperature was warm but not HOT. I installed proxmox anyway, fired up a VM and let a lot of slowly running packages install overnight. The unit was crashed in the morning. I tried adjusting several BIOS settings, focusing on reducing power consumption, with the same crashing results.
Next I ordered a Crucial 8GB 4800MHz CT8G48C40S5 and installed it. Memtest passed with flying colors, no issues in Proxomx.
I’m a little bit at a loss as the specific 32GB stick model was reported as working up thread. I may just have a bum stick or maybe my CWWK unit is finicky. Next troubleshooting step for me is a replacement 3rd party power supply, and maybe trying a 48GB ram stick for giggles.
Which part of the test did it fail on? Is it on a specific test?
I ran into an issue where one of my units would fail/reboot immediately once a specific test started. Even limiting the memory test to start on that test number it would happen.
What ended up resolving the issue for me was swapping to a different power supply. I assume that test caused a spike in power draw when starting which the provided power supply was not up to par. Once I ordered a more reputable brand power brick, the unit would pass the test easily.