Log S396 Streaming System

[hrh_ginsterbusch]

Option #4: Variation of the slide-in method - attach small 90° metal brackets to the panel, with pre-made sunk holes, and the pins / slides to the frame. Push the panel onto the slides, et voila: Mounted. And at best, leaving a few mm of space for improved accoustics and potentially improved airflow.

Option X: Raised cut-outs with neodym / magnetic tape covers. But then with all that effort involved, I could just raise the whole panel, so this seems counter-intuitive.

On topic of the air flow: If all fails, drilling some holes should improve that issue
The "mesh" of the panels remind me a lot of the GN saying about "more metal than holes" mesh.

cu, w0lf.

 

[hrh_ginsterbusch]

Aside of this, I'm currently trying to figure out how to replace the power extension cord, cause the Fractal Design PSU has its power connector in reversed position (compared to most other SFX PSUs). Ordered and already arrived are two adapters and one potential replacement cord.

cu, w0lf.

 

[hrh_ginsterbusch]

On topic of the air flow: If all fails, drilling some holes should improve that issue
The "mesh" of the panels remind me a lot of the GN saying about "more metal than holes" mesh.

Measured: Its at least 4.3 mm distance between fan and the top of the case.

• Fan frame =~ 2.3 mm
• Case frame "handle" =~ 2 mm

cu, w0lf.

 

[hrh_ginsterbusch]

Turns out: the Noctua fan duct stuff I thought I've already ordered .. a) isnt and b) its only for 92 mm fans anyway.

Created a make-shift fan duct using some foam padding I had sitting around, which was exactly 5 mm. Its a wee bit too thick, so 4.5mm or so would be a better goal. Currently testing it with a full CPU core load for 10 minutes.

The current measurements show at least a bit of a noise reduction, although temps are still as equally bad as before: 72, peaking out at 74. Might try adding a 92 mm bottom fan, and maybe adjust the fan curve to something more aggressive, too.

cu, w0lf.

 

[hrh_ginsterbusch]

Tested a bit: Panel closed - 74 C is maximum peak.

Panel slightly open (about a gap of 1 - 2 mm, plus the side bottom corner of like 3 mm): instant reduction to 71 C at max.

Also compared the "mesh" to other cases, ie. the Meshlicious and the AP201: Both have much thinner distances between the mesh holes, and the hole diameters seems to be larger.

What the makeshift fan duct helps is: Improve noise level. From before around 50 db its now down to 40 - 41 (measured from 60ish cm).

cu, w0lf.

 

[hrh_ginsterbusch]

Hotspots are mostly at the VRM and the m2 spot. With removed side panel, temps drop to 68, with short peaks to 70 C.
The real disadvantage seems to be the location of the slim fan: its moved slightly away from the mainboard, so the heat cannot be properly dissipated. But there would be no spot to mount the fan anyway.

Possible helper: Enlarge the line of slits that are located on the mainboard side of the back panel. And also figuring out where to place other smaller fans. I've got a range of small fans ranging from 80 to 40 mm, so there should be options. Also still got a fan hub and 3-way PWM split cable sitting around.

cu, w0lf.

 

[hrh_ginsterbusch]

Next test: Removed the 25 mm fan from the split cable and am now testing with a 40 mm and the slim 120.

So far, temps seem to be similar.

And looks like as if the noctua Y-splitter is broken - only the main fan works, the 40 mm doesnt spin (although its already switched to DC). Means: the 25 mm probably didnt work either. Will switching out connectors and see if is indeed broken.

cu, w0lf.

 

[hrh_ginsterbusch]

Changed the Y-Splitter. Doesnt seem to make much difference, despite the fact that the second fan is running now. Difference is, at best, 1 C. Meaning the second fan at the end of the case is mostly useless. It might help when using a long GPU, but it doesnt help any for improving CPU temps.

My next try probably either is adding a small fan on the front (40 mm) as intake or replacing the 120 mm slim with two slim 80 mm to cover more surface above the mainboard. Might try both.

cu, w0lf.

 

[hrh_ginsterbusch]

Test run with 60 mm intake: No change in temps, but the VRM / NVME temps were slightly reduced.
Test run with 40 mm exhaust: No change (74 C peak).

Why I am so fixated on this is: The thermals could be quite a bit better, and also, the issue with the second fan resulting in exactly zero difference ..

Mostly it feels like a construction issue: The "mesh" holes are much smaller than any other case I've taken a look at. It may look cooler like this, but it certainly doesnt STAY cooler.

Doing a last test with the CPU side panel removed.

cu, w0lf.

 

[hrh_ginsterbusch]

Preparing to test different fans on the Thermalright AXP120-x67.
What I should have done BEFORE installing the mainboard: Add extension cables to ALL fan connectors - one cannot get any of the connectors out without like .. breaking a few fingers.

Update: Managed REMOVE the fan connector from the board using flat nose pliers. NOW the big question is, whether or not I can get them back in with the same method

cu, w0lf.

 

[hrh_ginsterbusch]

Next update: Managed to put the CPU fan connector back in. Used a fan extension and then plugged in the cable of the bigger fan (SilverStone Airblazer 120RW).

Plans for moving the whole fan collection to as much to the front as possible is delayed for now: Trying to mounting things to the top of the frame using magnetic tape did fail. It stuck for a few seconds and then detached

So my next option is ordering very strong neodym magnets of 4 mm height, and hoping they might work out. Else, its drilling a few holes, and then mounting the fan frame with its existing screw inserts to said newly drilled holes.

Did run an open frame test again: Really .. this is SO disappointing. With closed frame: peak at 74 C. With open frame: peak at 71 C (and next to inaudible fan noise ~= 36 db).

Which means the top fans amount next to nothing. Well, the one single fan above the mainboard. The other one might help improve GPU thermals if one has a very long GPU (which I dont).

Other option might be to install the 60 mm fan between the front and the regular fan frame mounting points, and then hope it somehow reduces temps. Gonna have a test session on Saturday, after the neodym magnets I've just ordered have arrived (hopefully, there is strikes all over the week).

cu, w0lf.

 

[hrh_ginsterbusch]

Got the neodym magnets, which allowed to reliable mount the fan frame to the top of the case. So, no more flimsy screws to forcefully fumble into badly aligning holes.

Also, another bunch of 1 x 1 x 1 cm neodym cube magnets keep the CPU side panel firmly stuck to the case (2 per side, so 2 x 1 x 1 cm). There is about a gap of maybe 5ish mm now, but no difference in thermals or noise. Overall space available within the CPU side has increased to approx 97 mm - its ACTUALLY 85 mm (its late and I havent slept much today).

Modified the fan curve for the top fans a bit more aggressively, and lowered the 80 and 100% throttle of the CPU fan a bit, but .. all three measurements have maybe amounted for 1 C difference on the long run (seldom peaks to 74, mostly hovering between 72 and 73). So my next try is flipping the CPU fan to exhaust and see if this gonna help.

I'm increasingly convinced the Silverstone Airblazer fan is primarly optimized for radiators, not for CPU fan grills and heatpipes.

P12 Max is scheduled next for the CPU cooler (testing both with classic intake and in reverse = exhaust).

After that, its time for the Thermalright SI-100, which will be tested without the CPU side panel both with the default fan (25 mm depth) and the Silverstone Air Slimmer (15 mm depth). If I get expectionally cranky, I also gonna test with the P12 Max.

cu, w0lf.

 

[hrh_ginsterbusch]

On another note: Managed to reroute the power cord extension through the GPU / PCIe slots. Just on the outer third slot. Its sticking out a bit, similar like with the Fractal Design Ridge, but there is just not enough space left to stuff it properly back into the case. At least its now not blocking most of the cooling area above the mainboard.

The now empty cut-out for the power extension cord could potentially be used to act as a small intake, with a 40 mm fan attached on the inside.

cu, w0lf.

 

[hrh_ginsterbusch]

Currently doing a test run with the CPU fan (still on AXP-120x67, using Silverstone Airblazer 120RW) set to exhaust.
Already after 1 minute, the temps climb above 75, peaking at 76 C. All the numbers, both P- and E-Cores, are about 1 - 2 C higher.
So, highest P-Core is at 75/76, highest E-Core is at 56° C (vs. 54 on intake).

Only the NVMe numbers are very decent, staying below 35° C (currently at 33.9 C), and Wifi dropped from 33 to now 30° C (fluctuating wildly between this number every other second).

Edit: After 8 minutes, the CPU peak is at 77° C.

After 11 minutes of full CPU load, the thermals are:

- Max P-Core: 77°C
- Max E-Core: 57° C
- Wifi: 35° C
- NVMe: 34.9° C

Clearly, the advantage of slightly better cooling the NVMe and Wifi is dimished by the rest of the system running around 3 - 4° C hotter. Oh, and the noise level is also quite something (48 - 49 db).

cu, w0lf.

 

[hrh_ginsterbusch]

BTW: All my tests with the S400 have been done at ~ 21° C room temperature. I'm very sensitive to sudden temperature differences, so I try keeping my room at a constant level.

cu, w0lf.

 

[hrh_ginsterbusch]

Adding this here before I forget it again: Idle temps with the AXP-120x67 were at 24 max (and 29 when fan used as exhaust).

 

[hrh_ginsterbusch]

Installed the Thermalright SI-100, with original fan, and did a test run (open case, because thats 100 mm, max rise I managed was 85ish).

1. Idle temps:
- CPU 26 °C / 23 (P / E-Cores)
- NVME: 34.9 °C
- WIFI: 35 °C

2. 100% CPU load:
- CPU load initialy: 59 °C
- after 20 minutes of 100%: 63 °C max
- NVME temps at 100%: 36.9 °C
- WIFI: 39° C

Might have to reinstall it, because its sitting over the RAM right now, and nearly touching the mainboard ATX cable. Did install it in this direction because it looked like it would be touching the heatsink, but now, after having it installed, it does look as if there wouldnt be any issues (MSI B760I Edge Wifi).

In terms of noise: The loudest part in the case are now the top fans, while the CPU fan stays very silent even at 70% speed. Its set to start at 25 °C, but at a rather low speed level, crank up at 45, then 60 and finally, 90% speed at 80 °C. Thats the "more aggressive" fan curve. The one before was optimized for liquid cooling, but did perfom quite decent with air cooling, too, so I kept it - UNTIL I did my "move the top fans to the front as most as possible" mod. After that, I did change both the CPU and the top fan curve.

cu, w0lf.

 

[hrh_ginsterbusch]

Got the second S400 (in black) yesterday. Quality-wise, it feels worse than the first S400 I got at the end of October 2023. Painting is done in a lackluster way and the used materials are all a bit thinner (2.2 mm front in the first S400 vs. 1.79 mm front in the black newer one etc, fan bracket is only 1.6 mm and feels much less sturdy).


So the plan is to use the black one for testing out the modifications, and then move on to the better case in white and replicate the modifications. The current streaming system is then going to be moved into a different case, either the Silverstone Sugo SG16 (in white), or if I'm lucky, and come across another smaller one, something like 8 or 7 L, also only in white or another color, then that.

cu, w0lf.

 

[hrh_ginsterbusch]

So the plan is to use the black one for testing out the modifications, and then move on to the better case in white and replicate the modifications. The current streaming system is then going to be moved into a different case, either the Silverstone Sugo SG16 (in white), or if I'm lucky, and come across another smaller one, something like 8 or 7 L, also only in white or another color, then that.

As mentioned elsewhere, I've abandoned this approach, and have the black shoddy S400 case repackaged, waiting for a day this week to be dropped off at the next DHL office.

The Silverstone SG 16 is scheduled to arrive this Thursday, but I'm still going to test a few things, including fans on top of the case, plus the NH-L12S with its various configurations, before I finally decide on whether I gonna keep the streaming system in the S400 and use the SG16 for my main system (including an AIO mod for the Silverstone Vida 240), or move it into the SG16 and use the S400 for the main system ..

cu, w0lf.

 

[hrh_ginsterbusch]

SG 16 has arrived, and is looking rather fine. Would also allow for liquid cooling with the Silverstone Vida 240, but space-wise, a regular sized radiator up to 25, maybe even 27 mm, should fit.

cu, w0lf.