SPCR

SilenX "11dBA" 92mm fan, Arctic Cooling AF9225 fan, GlobalWin NCB 92mm fan, Nexus Silent 92mm fan. All recently purchased. 30dBA SLM range used for noise measurements, ambient noise <30dBA, 1/4" microphone distance. Airflow sensor propeller RPM 695 for all fans (equal airflow). The Nexus fan is an old design now. I don't know if that's why it's noisier than the similar-RPM new NCB fan or if I just have a poor sample. No matter, the point is it and the NCB require 34.7% more RPM than the SilenX for a given airflow, and that's why the SilenX is a lot quieter. [Similar to the new Scythe SlipStream 120mm fans that are really quiet for the same reason.]

Here's the stuff needed for these measurements:


First row the Extech SLM, the 3.5-digit multimeter used to record the fan voltage, and the Radio Shack timer used for the 15-minute fan warmup. Next up is the 92mm test fixture with the red airflow sensor propeller and the "clamps" that hold the fan in place, and the Nexus fan. The black mark on the fan hub is for the Extech stroboscope that's right behind the 92mm test fixture. At the top right you can see the corner of the 220mm fan on my #1 computer, which runs (very slowly) during these tests; it does not measurably affect the ambient noise level.

Felger Carbon wrote:

Code: Select all

new SilenX"11" 8.61V 46.1dBA 695RPM 1153RPM
A-C AF9225     6.60V 48.8dBA 695RPM 1390RPM
GW NCB         9.97V 49.6dBA 695RPM 1596RPM
Nexus Silent   12.0V 52.1dBA 695RPM 1553RPM

my rig contains 3 silenx fans. one 120mm in the back and 2 92mm in the front. they usually run at 3 volt and are quiet. at higher rpm's broadband noise and a lot of airflow.

what strikes me with these fans is, that the volt-rpm relation is not linear. at 3 volt the 120mm still runs at about 550rpm, while max rpm is about 1450rpm.

do you know why that is?

sanse wrote: at 3 volt the 120mm still runs at about 550rpm, while max rpm is about 1450rpm. do you know why that is?

Not only do I not know why, but I don't know how you know it's 550RPM at 3V. The computer usually stops reading RPM somewhere in the 6-7.5V range. Do you also have a tachometer? That works, of course.

Hello FC,

Where did you get the SilenX fan, and what do they look like? (Pics, please?)

NeilBlanchard wrote:Where did you get the SilenX fan, and what do they look like? (Pics, please?)

I got the 92mm X fan here (w/pic).

I got the 60mm X fan I recently tested along with some other 60mm fans here (w/pics).

My next order, I'm gonna see if they have one of the new 80mm X fans in stock. The X fan I tested recently with some other 80mm fans was bought 2.5 years ago, so isn't the new model. [edit: actually, I already bought the new-model 80mm X fans from Newegg and have reported data taken with it. Sorry; a lot of fans, too few brain cells...]

Felger Carbon wrote:

sanse wrote: at 3 volt the 120mm still runs at about 550rpm, while max rpm is about 1450rpm. do you know why that is?

Not only do I not know why, but I don't know how you know it's 550RPM at 3V. The computer usually stops reading RPM somewhere in the 6-7.5V range. Do you also have a tachometer? That works, of course.

no, i do not have a dedicated tacho-meter. de rpm i mentioned is the average of the rpm's provided by the bigng fancontroller and the rpm fed back to the motherboard.

when i decrease the voltage below 3 volt both rpm's decrease and when i increase the voltage both rpm's increase. i do not know how accurate both figures are, but i think they are pretty indicative and not far from the truth.

the mobo and the bigng keep reading those rpm's also at voltages below 6 volts. no problem.

sanse wrote:the mobo and the bigng keep reading those rpm's also at voltages below 6 volts.

Ah! I know it's late, but I think I know (with cmthomson's help) why the RPM-voltage is not linear, and why you can read the RPM below 6V.

You're using PWM-controlled mobo fan headers! (The mobo manual will claim that the DC level is what's controlled but that's incorrect; PWM is used.)

cmt has reported that at low PWM ratios (i.e. at ~3V-equivalent) the mobo periodically pulses the voltage higher so the mobo can read the fan's tach signal to get RPM. This happens 4 times a second. So at 3V, the fan does not run at 3V full-time. It runs at a higher voltage (12V?) part of the time. And so the fan RPM is higher than one would expect at 3V.

Sorry for the late reply, but I'm just learning this from following cmt's posts.

Felger Carbon wrote:

sanse wrote:the mobo and the bigng keep reading those rpm's also at voltages below 6 volts.

Ah! I know it's late, but I think I know (with cmthomson's help) why the RPM-voltage is not linear, and why you can read the RPM below 6V.

You're using PWM-controlled mobo fan headers! (The mobo manual will claim that the DC level is what's controlled but that's incorrect; PWM is used.)

i can assure you, that the mcubed bigng fancontroller does not use pwm in it's analog mode. neither does the gigabyte mobo i'm using have pwm fanheaders.

Sanse, I just read your Bigng review thread. The replies by Mongokitten and Mario were especially interesting. It seems that in analog mode, Bigng uses the same trick that mobo 3-pin PWM headers (which are often not acknowledged as PWM in the mobo manual) to read the RPM at low voltages - did you note the "click 4 times a second"?

And Mario points out that in PWM mode, Bigng does not measure RPM at all; it simply takes a proportional guess.

If your mobo has a 3-pin fan header that can be controlled, it in fact uses PWM no matter how hard they deny it. Fact of life.

Felger Carbon wrote:Sanse, I just read your Bigng review thread.

i hear no clicks from the fans. they run smoothly and continuously. i also do not believe the bigng still pwm's the fans, when in analog mode. the silenx fans stop, when voltage is lowered to 2.5 volt and do not start out of themselves before voltage is above 3.5 volts. that is not pwm behaviour.

it is true that the bigng does not continuously measure rpm's, when in pwm mode. it is not possible with pwm. max rpm is measured during startup and then rpm is calculated from the voltage. for the value shown in the bigng gui then. there is a separate connector to feed tach-signals of all 4 channels back to the mobo.

do you think the latter is a fake signal? even, when in analog mode? it behaves very similar to normal fan tach signals. the 600 rpm value at 25% voltage is similar to the value i got, when the bigng was not present in my rig yet and the fans were controlled by speedfan (at 25%).

Your reply that the fan RPM is sampled during the brief "start" pulse of 12V to the fan makes sense with a 1000RPM fan indicating 800RPM at 100% voltage. Many - most - fans do not reach their final RPM instantly. I can well believe that a 1000RPM fan samples at 800RPM during the brief start pulse.

It is beginning to seem that fan clicking is not a simple matter:

1) Most users report no clicking. A few users definitely report clicking.

2) One posting recently reported that of four (apparently) identical fans, one clicked and the other 3 did not. The fan that clicked remained constant as the fans were attached to different mobo headers.

3) Many reports of clicking now are asserted to occur at "about a 4 per second" rate. The operating system interrupt period for servicing the RPM (and some other functions) is close to 4.8 seconds (4.78?).

4) Therefore, I sat up and took notice at reports that bigNG's analog mode produced clicks at about 4 times per second.

5) To decrease the workload on the CPU, a PWM control chip is now used on mobos, including on mobo 3-pin fan headers with controllable "voltage". Although the Intel PWM control specification calls for ~25KHz as the PWM frequency, the actual frequency used by this mobo chip is programmable. One SPCR user recently discovered that clicking could be eliminated by programming using the "advanced tab" (?) and setting this frequency to 25KHz (where it should have been all along).

6) I think there's more to be learned about this new "clicking", which does not seem to be the same as the clicking that was heard years ago, and solved by fan motor control ICs with "soft edges" (and alas, no tach signal 'way back then). For sure, I have a lot to learn on the subject - which is hard because I haven't used controllable 3-pin headers yet (or bigNG) and so have never heard this new form of clicking.

Felger Carbon wrote:Your reply that the fan RPM is sampled during the brief "start" pulse of 12V to the fan makes sense with a 1000RPM fan indicating 800RPM at 100% voltage. Many - most - fans do not reach their final RPM instantly. I can well believe that a 1000RPM fan samples at 800RPM during the brief start pulse.

don't think it has to do with the startup rpm sensing of the bigng. maybe it will stay in the dark why these silenx fans show a non-linear rpm-voltage relation.

btw, you were the one, who mentioned this non-linear behaviour of some fans first in the 120mm fan thread:

viewtopic.php?t=40636&postdays=0&postorder=asc&start=30

FC....concerning your problem reading low RPM. It's the MB. All the Intel boards I have used over the last few years, read rpms well below 5V. The D865PESO board I'm currently working with, picks up rpms as low as 200 when my 40x10 Sunon PSU fan sometimes starts up that slow.

In the morning when it's cold in my test room, my 1600rpm 120mm S-Flex runs for a while under 500rpms. It gets a 12V burst to start, but the controller goes well below 5V before things start to heat up. It idled all morning about 550rpms.

It is true many boards do not read low rpms......I don't think it's the fan, rather the board itself that's to blame.

The only way to know (aside from having access to the schematics) whether a fan controller uses PWM or linear voltage is to look at its output on an oscilloscope. For sure you should not depend on the UI or the manual or especially the spec sheet.

There is a very strong propensity in the PC world to copy designs that work. Many motherboards that support both 4-pin and 3-pin fans use the same controller chip, one that uses PWM in both modes, albeit with subtle variations. Based on the posts above (and their references), I suspect the BigNG controller uses the same chip as most motherboards.

The most common fan controller chip lets software configure an average output voltage ("DC mode") that is actually a high-frequency PWM signal (25-95 KHz) with a superimposed 4 Hz full-voltage pulse that makes the tachometer work with ordinary 3-pin fans.

Felger Carbon wrote:Airflow sensor propeller RPM 695 for all fans (equal airflow).

Code: Select all

new SilenX"11" 8.61V 46.1dBA 695RPM 1153RPM
A-C AF9225     6.60V 48.8dBA 695RPM 1390RPM
GW NCB         9.97V 49.6dBA 695RPM 1596RPM
Nexus Silent   12.0V 52.1dBA 695RPM 1553RPM

No matter, the point is it and the NCB require 34.7% more RPM than the SilenX for a given airflow, and that's why the SilenX is a lot quieter. [Similar to the new Scythe SlipStream 120mm fans that are really quiet for the same reason.]

Next up is the 92mm test fixture with the red airflow sensor propeller and the "clamps" that hold the fan in place, and the Nexus fan. The black mark on the fan hub is for the Extech stroboscope that's right behind the 92mm test fixture.

Your results, or at least the way you've presented them, seem to say that the new SilenX "11" requires 8.61V and 1153RPM to achieve... 695RPM, or that the GW NCB requires 9.97V and 1596RPM to achieve... 695RPM, etc. Reading that aloud does not make any sense.

RPM != CFM if the terms are used in abstract. Different fans spinning at the same RPM can have different shapes and number of fan blades, some pushing more CFM them others. So in the context of your tests, how do the "red airflow sensor propeller" and the "Extech stroboscope" measure CFM?

C. Zoui wrote:Your results, or at least the way you've presented them, seem to say that the new SilenX "11" requires 8.61V and 1153RPM to achieve... 695RPM, or that the GW NCB requires 9.97V and 1596RPM to achieve... 695RPM, etc. Reading that aloud does not make any sense.

You missed reading the very first line of mine that you quoted. The "695RPM" is the RPM of the sensor propellor. This is repeated for each fan to emphasize that the airflow is the same for each fan. After that comes a second RPM figure, which is the RPM of the fan under test.

C. Zoui wrote:RPM != CFM if the terms are used in abstract. Different fans spinning at the same RPM can have different shapes and number of fan blades, some pushing more CFM them others. So in the context of your tests, how do the "red airflow sensor propeller" and the "Extech stroboscope" measure CFM?

The airflow from each fan (one at a time as the fan is under test) is monitored by the sensor propeller. I adjusted the voltage on each fan under test until the RPM of the sensor propeller was 695. Therefore, each fan (all of which are 92mmx25mm) was producing equal airflow. I do not claim to be measuring CFM. I do claim to be setting each fan under test to equal airflow. Then I use the SLM to measure the noise at equal airflow.

The Extech strobe is used to determine the sensor propeller RPM. In practice, I keep the strobe set (in this case) to 695RPM and adjust the voltage to the fan until the sensor propeller "stands still". A comfortable warmup time is allowed for each fan under test.

Again: I do not claim to measure CFM. I do claim to be able to set different makes and models of 92mmx25mm fans to the same airflow.

Was it SilenX that used to be a bad word around here, or am I thinking of something else?

I think it wasn't so much their products, but one of their 'reps' which had a bad apple roaming here.

Felger Carbon wrote:You missed reading the very first line of mine that you quoted.

I did indeed; my apologies.