S4 Mini w/ Ryzen 1700 and 1080 TI

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Thanks.

Ok, felt good enough to get 2 more fans tested, the NZXT and the Coolermaster. NZXT did as well as some of the others, but the cooler master was no better than the Noctua fans and couldn't keep the CPU under 78C.

Last but not least is the Gentle Typhoon, which I'll test tonight hopefully. Then move onto the GPU fans.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Ok I have finished the CPU tests. I'll have a video with the audio from all the fans once I finish the GPU tests.

I have graphed the following per fan using Audacity and the ACX plugin. All measurements are in dB.

  • Idle peak
  • Idle average (using RMS)
  • Metro peak
  • Metro average (RMS)
  • Max peak
  • Max average (RMS)
My setup is a MXL 770 Cardioid Condenser Microphone (XLR) run into a TASCAM DR-40 with a metal pop filter in front of it. The Mic is positioned about 6 inches from the chassis and level with the CPU fan (chassis is in horizontal orientation). The DR-40 is calibrated at -12dB for voice level, so while you can't hear the fans you can at least get a view on the scale.

What's interesting is that the Phantek sounds great at max, but is actually louder below max. It's not outrageously louder, just 3 dB.

I can't pick a winner (for myself) yet because I haven't finished calculating the effectiveness across the various tests I ran. I expect I'll do those calculations Sunday.


 
Last edited:
  • Like
Reactions: alexep7

AleksandarK

/dev/null
May 14, 2017
703
774
Fingers crossed in my precious gentle typhoon..:)
That is also my favorite fan. Works so good on both the radiator and the heatsink. Quiet and nice looking.
Ok I have finished the CPU tests. I'll have a video with the audio from all the fans once I finish the GPU tests.

I have graphed the following per fan using Audacity and the ACX plugin. All measurements are in dB.

  • Idle peak
  • Idle average (using RMS)
  • Metro peak
  • Metro average (RMS)
  • Max peak
  • Max average (RMS)
My setup is a MXL 770 Cardioid Condenser Microphone (XLR) run into a TASCAM DR-40 with a metal pop filter in front of it. The Mic is positioned about 6 inches from the chassis and level with the CPU fan (chassis is in horizontal orientation). The DR-40 is calibrated at -12dB for voice level, so while you can't hear the fans you can at least get a view on the scale.

What's interesting is that the Phantek sounds great at max, but is actually louder below max. It's not outrageously louder, just 3 dB.

I can't pick a winner (for myself) yet because I haven't finished calculating the effectiveness across the various tests I ran. I expect I'll do those calculations Sunday.


I just love how detailed this build log is. Fan set ups, temps results, fan modding, noise outputs.... You could start a website.
 

WH-team

SFF Lingo Aficionado
Oct 5, 2017
102
186
case is too thin so that isn`t good for CPU temp. I think single FPU the 1700 will up to 90℃....is bad for working:(
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Small note, fixed the scale in the fan noise graph, the top scale was wrong (was showing -50 to 0, should have been -50 to -12 like the bottom).
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Here are 2 graphs that I am using for evaluating the right fan for me. After the graphs I have some info on how I have been using them to evaluate the right fan for me, and why. Note that you may not make the same conclusions as I have, which is fine. While I've been sharing the data as I go, ultimately I'm looking for what will work for me and my expected workload on the machine.

First off, noise and temps from the max synthetic tests (AIDA64 for CPU, Kombuster for GPU).



And here is the same data for the Metro benchmark test (3 runs each).



If I was using just the synthetic results, then the Phantek fan would look great from a noise perspective, but it likes to hover around 75C for temps. The NZXT would be better for me, well below 75C in both peak and average, while only a bit noisier. The Gentle Typhoon would also be a good pick but it has the 2nd worst noise levels (max and average).

When we factor in the Metro results though, things get interesting. The Gentle Typhoon looks fantastic until you notice that the noise level is almost the same in Metro as the Max synthetics. What that really tells me is I could do a better job dialing in that fan (having it work less, but get comparable performance in Metro as some of the others).

So I plan to play with the GT some more, see if I can get a better balance of noise to performance (slightly less performance for less noise). If I can then the GT is the CPU winner. If I can't get it, or I don't like the final fan sound then I'll fall back on the NZXT. I've used GTs before, and I am not of fan of their harmonics above 75% fan speed.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
case is too thin so that isn`t good for CPU temp. I think single FPU the 1700 will up to 90℃....is bad for working:(

With stock fan I agree that the CPU and GPU will get too hot, but with the fan mods I'm making I am getting excellent performance in thermals on the CPU; I'll be working on the GPU next.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Well I couldn't get the fan controller on the motherboard to cap at a certain max level so I put an external fan controller on it (the small Noctua controller). I used double sided table to stick it to the inside of the motherboard case panel so that the knob sticks out through one of the vent holes. I have then dialed in 1800RPM manually and am running synthetics on it now.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
1800 RPM on the GT resulted in 78C on the CPU, so redoing the test at 1900RPM. If I have to go much higher then I'll probably switch to the NZXT; I don't have PWM control anymore on the GT but the motherboard fan controller did a pretty good job managing the NZXT speeds.

I'll finish the testing tonight and post one way or another. I want to move onto the GPU testing.
 

MarcParis

Spatial Philosopher
Apr 1, 2016
3,663
2,771
Strange about gentle typhoon because on mines (version 3 pins, max rpm 1850) they arr fairly quiet/discrete till 1400rpm..after i'm using corsair commander mini to handle them precisely and smoothly.

After noise outputs also depends on what is before and after fan.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Strange about gentle typhoon because on mines (version 3 pins, max rpm 1850) they arr fairly quiet/discrete till 1400rpm..after i'm using corsair commander mini to handle them precisely and smoothly.

After noise outputs also depends on what is before and after fan.

Don't forget the PWM versions are 2150RPM. I'll see if I can find a PWM version at 1850.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Thanks, unfortunately PPCs is out of stock, going for Amazon but it will take 2 weeks to get.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Ok order placed. I think I have the non PWM 1850 version, so I can at least try that from a max capacity, noise perspective. Time to get out the dremel and knock 2 corners off a fan.
 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Ok I put in the 3pin 1850 GT and it was horrible. So I started thinking about why that might be, and I realized that the GT needs a lot of air to be effective, and with those small vent holes maybe the fan is getting starved of air to push.

I took off the cover and the temp dropped like a rock, see the graph below. I left it for another 10 mins after the graph and it went as low as 65C. So it's definitely the lack of access to air.

The NZXT has fewer fins, so it can grab the air more effectively. This test has finalized my personal selection - the NZXT.

 

Reglar

SFF Lingo Aficionado
Original poster
Jul 16, 2017
111
129
Well I have started testing GPU fan configs. Tried the 2150 RPM GT and it worked well as a GPU fan; I think there is more room between the blades and the chasis so It can grab air better.

Then I tried a 92x92x25 + 92x92x14; it didn't perform very well but I did the whole suite of tests.

I'm on test 3, the 120 Noctua F12 and 2 50mm uxcell fans. The uxcell fans are not PWM, and load as can get out, so test 4 will be same Noctua and no extra fans.

As is though I now have a new problem, the Noctua is so good at moving air I can keep the GPU down to 75C; the problem is the heat it's clearing is adding another 5C to the CPU; I may need to try inverting the system to see if that helps at all.

I expect it will take most of the week and into the weekend to complete the tests and calculate all the data. I'll be using similar data as I did on the CPU, only I'll be adding information on core clock level and benchmark scores. I saw the system lost some real ground in the benchmarks with the 2 92mm fans.