Nexus voltage, rpm & mounting

JVM · Post by **JVM** » Tue Jan 31, 2006 12:15 pm

How do I translate the voltage on Nexus 120 and 92 fans to rpm speeds at 12, 9, and 7 volts?

Also, can the Directron rubber sticks work with Nexus 120 and 92?

Aleksi · Post by **Aleksi** » Tue Jan 31, 2006 12:23 pm

Hi,

basicly RPM follows voltage.

(desired voltage / fan's rated voltage) * fan's top RPM = RPM at desired voltage.

A fan with a top speed of 1000RPM at 8V is:

(8V / 12V) * 1000RPM ==> 667RPM

EDIT: Off to bed, too many typos...

Firetech · Post by **Firetech** » Tue Jan 31, 2006 12:30 pm

Aleksi wrote:Hi,

basicly RPM follows voltage.

(desired voltage / fan's rated voltage) * fan's top RPM = RPM at desired voltage.

A fan with a top speed of 1000RPM at 8V is:

(8V / 12V) * 1000RPM ==> 667RPM

EDIT: Off to bed, too many typos...

Thanks for that, have always wondered about it. Sleep well

Aleksi · Post by **Aleksi** » Tue Jan 31, 2006 12:38 pm

Good to know I'm posting something useful sometimes... Basicly RPM == voltage, but with powerful fans it flattens out at high rpm because the fan venturi and blades themselves contribute to stall.

JVM · Post by **JVM** » Tue Jan 31, 2006 12:45 pm

I've seen the Nexus 120cm fan rated at 9 volts=800 rpm, 7 volts=600rpm.

92cm fan rated at 9 volts=900 rpm and 7 volts=700 rpm.

Are those accurate translations?

Also, can the Nexus 120cm and 92cm fans be mounted using Directron rubber sticks?

jaganath · Post by **jaganath** » Tue Jan 31, 2006 12:47 pm

with powerful fans it flattens out at high rpm because the fan venturi and blades themselves contribute to stall.

Hi Aleksi,

how does the venturi effect apply to PC fans?

Thanks,

J.

JVM · Post by **JVM** » Tue Jan 31, 2006 2:19 pm

Excuse me, but I do have some questions here

Firetech · Post by **Firetech** » Tue Jan 31, 2006 2:50 pm

JVM wrote:Excuse me, but I do have some questions here

I'm sure Aleksi will get back to you after he gets out of bed

Tibors · Post by **Tibors** » Tue Jan 31, 2006 9:23 pm

JVM wrote:Also, can the Nexus 120cm and 92cm fans be mounted using Directron rubber sticks?

The Directron rubber sticks are to be used with fans with closed corners. So you can mount the Nexus 120mm fan with them, but not the Nexus 92mm fan.

In a forum post last weekend I showed one of my Nexus fans running 654RPM @ 6.07V. So it actually turns a bit faster than you would expect from the rule of thump Aleksi mentioned.

Shadowknight · Post by **Shadowknight** » Tue Jan 31, 2006 10:49 pm

The Directron rubber sticks are the same ones veraxfans.com sell in their accessories section.

The Nexus box kit comes with a version of the "closed flange" sticks, as well as a selection of EAR style grommets zip ties, and open flange mounts, for about $20. The kit sucks with the exception of the closed flange sticks and the zip ties.

Felger Carbon · Post by **Felger Carbon** » Wed Feb 01, 2006 12:04 am

Aleksi wrote:basicly RPM follows voltage.

(desired voltage / fan's rated voltage) * fan's top RPM = RPM at desired voltage.

A fan with a top speed of 1000RPM at 8V is:

(8V / 12V) * 1000RPM ==> 667RPM.

(Hesitating briefly before sticking leg down the lion's throat.) Alexi, I'm sorry to report that that's wrong. It's a useful rule of thumb, a first order approximation, but any plot of RPM vs. voltage would reveal the curve - and it is a curve - is not linear.

Suppose you apply a voltage to a resistor. Double it and you double the current, resulting in a 4X increase in power. OK? Now apply 6V to a 12V fan, check the RPM, and then double the voltage. If the fan were a resistor, then the current would double and the power would increase 4X. But the fan is not a resistor.

What is true is that CFM is substantially proportional to RPM. So if the RPM doubled when the fan voltage went from 6V to 12V, the CFM would have doubled too. And that's where the problem arises, because there are no perpetual motion machines. There is no free energy. I say this because the energy increase required to double a fan's CFM is a factor of 8. Not 4, 8.

Yes, the energy of a given column of air is proportional to the cube (not the square) of its velocity. But the electrical energy available by doubling the voltage is only the square of the voltage, not the cube.

If we had fans wound with zero resistance wire (imagine superconducting motor windings for a moment) and zero resistance motor switches, then doubling the voltage would indeed double the RPM and the CFM, but it would require 4X the current, not 2X, so that the total energy input meets the 8X required to push the air. 2X the voltage times 4X the current meets the 8X required.

Of course, we don't have zero resistance wire or switches.

This means that the electrical "load" on the motor is reflected back to the voltage source non-linearly. More current is being "asked" for, but then more voltage is lost in the resistance of the winding. The imperfect fan motor, its wiring having parasitic real-world resistance, draws more and more current (more than a simple resistor), dropping more and more voltage across the actual wire winding resistance.

Now, a side trip to the noise of the fan:

The simple fact is that the purely acoustic noise of an ideal fan (and that's the only noise an ideal fan has) is directly proportional to the energy of the moving column of air the fan produces. Since the energy is proportional to the cube of the CFM, the noise of an ideal fan increases by 60 times the log to the base 10 (60LOG10) of the CFM ratio. Double the CFM, get 18.06dB more noise.

And with that last statement I have irrevocably thrust my leg all the way into the throat of the lion. Because everybody "knows" that the noise in dB is 50 times the log to the base 10 of the CFM ratio. The problem is that what everybody "knows" is wrong.

The 50, not 60, figure is engineers' way of expressing that a fan is not ideal because its motor windings have resistance. The actual increase of noise with doubled voltage is less than 18dB because of this resistance. The widely used 50LOG10 (15dB) figure is an engineering fudge used to approximate the real-world effect of a fan motor's winding resistance. Repeat, it is an approximation. Not a law of nature/physics.

What the 50LOG10 figure says is that if the fan voltage is increased from 8V to 12V then the noise increase is ~8.8dB, while an ideal fan, whose noise is proportional to the cube of the voltage, would have a noise increase of ~10.6dB. In fact, that noise increase has been carefully measured and reported (note 1). And four very good fans all increased on the close order of 9.7dB. Not 10.6dB, because very good fans exist in the real world but are not ideal. But not the engineering approximation of 8.8dB either.

So, I've kicked over two anthills here. RPM is not directly proportional to voltage. And noise does not increase at 50LOG10(voltage ratio). The first item is proved by figures given earlier in this thread, and can be reconfirmed by anybody who can measure voltage and RPM. The second item is harder; after all the engineering compromise was originally chosen to be useful. But it was chosen a long time ago and fans have improved since that time.
--------------------------------------

We can come close to approximating an ideal fan. Suppose that we operate a fan at constant voltage. To change the RPM, we use fewer turns of wire in the motor, but of larger size wire, so that the weight and volume of the wire does not change. This would result in an increase of both CFM and noise. What would the increase of noise be vs CFM ratio? Why, we know that number. Honest. We know it very well.

The above is exactly, precisely what a fan manufacturer does when creating a line of fans at varying RPM points. Take the Panaflo FBA line for example. There's the L, M, H, and U models, all based on changing the wire turns and size. The 92L is 2100RPM at 27dBA and the H is 2850RPM at 35dBA. That's an RPM ratio of ~1.357 and a noise increase of 8dBA. Suppose we take the LOG10 of that ratio and multiply by 60. The result is ~7.96, which is mighty close to 8dBA. Well, Panaflo makes the 4 models and the specs are well known. Perform this calculation for any combination you wish. You'll see that 60LOG10 (not 50LOG10) accurately predicts the specified noise levels. Panasonic publishes good numbers (all taken using the same methodology).

Oh, boy. I report accurately that a computer with left-panel CAGs is quieter when placed to the left of the monitor and three people accuse me of being drunk! I hate to think what'll happen now. Do I get drummed out of SPCR, with my epaulets ripped off?

-------------------------------

Bottom line: the energy of moving air is proportional to the cube of its velocity. This is a law of nature/physics. All of the above flows from this simple fact, which can be confirmed by consulting the right physics textbook.

Note 1: (from a private email) On a German site, I ran across some really good HSF noise data. The best I've seen anywhere. For starters, these people have (access to) a 15dBA ambient testing facility! That site has published a lot of good data on fan (HSF) noise at 12V and 8V. For instance:

Zalman 7700-cu 30.7, 21.2, delta 9.5dBA
Akasa silent fan 28.3, 18.6, delta 9.7dBA
AVC Z7UB 30.2, 20.5, delta 9.7dBA
A-C Silencer 64 Ultra 27.4, 17.7, delta 9.7dBA

Edit: 7.96, not 7.596; added Note 1

Aleksi · Post by **Aleksi** » Wed Feb 01, 2006 2:16 am

This thread has grown without any topic reply notification coming to me!

OK, I've slept and now at work. Will get back to you honey bunnies regarding the fan laws later tonight!

Firetech · Post by **Firetech** » Wed Feb 01, 2006 3:34 am

Felger Carbon wrote:Oh, boy. I report accurately that a computer with left-panel CAGs is quieter when placed to the left of the monitor and three people accuse me of being drunk!

In fact, only one person accused you (vaguely) of being drunk, the other two only admitted to being drunk!

PS Nice article FC

JVM · Post by **JVM** » Wed Feb 01, 2006 5:56 am

Yes, great article, but how do I translate voltage to rpm with Nexus 92cm and 120cm fans? Or is that impossible to do accurately?

Specifically, I am seeking the rpm at 7, 9, and 12 volts for the Nexus 92 and 120cm fans.

jaganath · Post by **jaganath** » Wed Feb 01, 2006 6:16 am

how do I translate voltage to rpm with Nexus 92cm and 120cm fans? Or is that impossible to do accurately?

Specifically, I am seeking the rpm at 7, 9, and 12 volts for the Nexus 92 and 120mm fans.

Um, doesn't the 92mm have rpm sensing? I know the 80mm and 120mm do. So all you have to do is use a voltmeter to find the voltage at a given rpm.

The rpm at 12 volts is already given in the technical specs: 1000rpm for the 120mm, 1500rpm for the 92mm.

JVM · Post by **JVM** » Wed Feb 01, 2006 6:31 am

jaganath wrote:
how do I translate voltage to rpm with Nexus 92cm and 120cm fans? Or is that impossible to do accurately?

Specifically, I am seeking the rpm at 7, 9, and 12 volts for the Nexus 92 and 120mm fans.
Um, doesn't the 92mm have rpm sensing? I know the 80mm and 120mm do. So all you have to do is use a voltmeter to find the voltage at a given rpm.

The rpm at 12 volts is already given in the technical specs: 1000rpm for the 120mm, 1500rpm for the 92mm.

No, I don't have a voltmeter. Since there are many people here using the Nexus fans at various voltages, I assumed this would be an easy question.

Tibors · Post by **Tibors** » Wed Feb 01, 2006 9:49 am

Most people don't pay that much attention to exact fan speed and voltages for various reasons. You build a system, adjust the fans till the temps and noise are right. Then who cares at what voltage and RPM that is.

I did some quick measurements. Note this is in free air, so in a real system the fans will move slower. I also did not wait ten/fifteen minutes for the RPM readout to stabilise between each measurement. Since I'm going from slow to fast that means my measurements could be too low too. Hopefully both effects cancel each other out

Code: Select all

Nexus fans powered by a Zalman fanmate from a MSI RS480M2-IL system fan header via a 50cm extension wire. All measurements with the fan in free air. The 120mm fan is in use cooling my HD. The 92mm and 80mm came out of the parts bin, so after being broken in they will spin a little faster.

   120mm
  V     RPM
 5.01   510
 6.15   617
 7.07   690
 8.07   793
 9.04   865
10.00   945
10.52   960
11.94  1033 (no fanmate)

    92mm
  V     RPM
 5.01   577
 6.07   731
 6.95   842
 8.03   983
 8.97  1117
10.02  1241
10.52  1293
11.95  1446 (no fanmate)

    80mm
  V     RPM
 5.20   605 (Does not run on fanmate minimum setting, so this voltage was used in stead.)
 5.98   731
 7.00   871
 8.10  1007
 9.04  1138
10.02  1307
10.54  1370
11.96  1536 (no fanmate)

Some other measurements I had on file:

Code: Select all

Nexus 120mm fan directly powered by a MSI RS480M2-IL system fan header.
Listening during the morning hours (~11:00AM). Residential area with mostly 5 level high apartment blocks. Double glazed windows. Winter, so no kids playing outside, but the upstairs neighbour was moving around.
%PWM   RPM   sound
100   1050   audible
 90   1015   audible
 80    924   audible
 70    842   almost inaudible
 60    749   inaudible
 50    630   inaudible
 40    508   inaudible
 30     ??   inaudible
 20      0   silent

JVM · Post by **JVM** » Wed Feb 01, 2006 10:29 am

Tibors,

You deserve a reward for your extensive efforts

Many thanks for all the work you put into this topic. No question in my mind you deserve the "post of the year" award

Absolutely amazing! Thank you!

cpemma · Post by **cpemma** » Wed Feb 01, 2006 11:25 am

Felger Carbon wrote:
Aleksi wrote:basicly RPM follows voltage.

(desired voltage / fan's rated voltage) * fan's top RPM = RPM at desired voltage.

A fan with a top speed of 1000RPM at 8V is:

(8V / 12V) * 1000RPM ==> 667RPM.
(Hesitating briefly before sticking leg down the lion's throat.) Alexi, I'm sorry to report that that's wrong. It's a useful rule of thumb...

Which is the main point, I think. If you take it to extremes, at around 3V-4V the fan will grind to a halt, so the 'law' obviously isn't working there. But to get a ball-park figure, it's a useful approximation.

Here's a trace I did using Motherboard Monitor to get the RPM figures, averaging readings over about a minute at each voltage (this was a 60mm fan in use on a heat-sink - no free air conditions) :

The curvature isn't the real problem in the 'useful' range, it's the fact that the line doesn't extrapolate back to 0,0. Linear(ish) but not proportional. If speed is 5100rpm at 12V, at 8V you might expect 3400RPM; MBM5 showed a tad over 3800RPM, about 12% higher. A bit like driving, mpg goes down at higher speed. But not too shabby for a ball-park figure, and if you can measure the RPM at say 7V and 12V you'll be able to get close for voltages in between.

Other, non-mechanical, things can conspire to upset our calculations. Here's the guts of a simple fan,

The brown black-banded object between the red (12V) input and the Hall sensor chip is a rectifier diode, sometimes fitted so an accidental reverse power connection doesn't blow the electronics. A Panaflo L1 I dissected also had the diode (but along with a 16-pin IC and other extra clever bits). But it means applying 12V to the fan only sends about 11.2V to the motor, apply 6V it gets 5.2V, and 6/12 != 5.2/11.2.

Use the fan law if you want a guesstimate to go out and buy a suitable resistor or zener diode to drop the voltage, otherwise see Tibor's post.

Aleksi · Post by **Aleksi** » Wed Feb 01, 2006 12:36 pm

Interesting discussions in a Nexus RPM thread

I had to go through mails from Dorothy and people alike to get enough examples and also data to address both issues.

Hi Jaganath,

your question about the venturi effect applying to PC fans?

Yes it does apply, but you really can't answer it in a simple and short way (as I will) as it consists of two levels (blade and the housing).

A fan venturi covers three specific stages:

External environment around the fan intake
- to create a smooth path for air to correct feed into the blades
- to manage vortices off the blades against perimeter intake airflow

Internal environment around fan the intake, tube and exhaust
- to create a smooth path managing vortices off blade tips and for optimising blade efficiency with respect to housing

External environment around the fan exhaust
- optimising blade efficiency particularly at exhaust resistance.

So it's not just through the fan, but also in the surrounding environment. A 120mm fan intake is on a 130mm circle and exhaust is on a 132mm circle. A 80mm fan intake is on a 86mm circle and exhaust is on a 92mm circle.

What I wrote about airflow dropping at high speeds? Partly a matter of the flow choking due to the restriction of the venturi and partly due to the the blades, because at high RPM the blade efficiency falls as a blade profile has an optimum operating range in the P-Q curve.

A venturi shaped housing is beneficial for lowest noise. It does result in a physically weaker fan, so closed corners are often used to stiffen it. A pure tube shape produces a more rigid fan, however it can produce turbulence at high flow (you will hear the air shear on the intake side)

Basicly, you need someone with very good knowledge about fluid dynamics, fans and preferably with modelling software to explain it to you. That isn't me.

and hello Felger Carbon,

Where to start this... Cpemma was very kind to post the idea with the fan laws. You're basicly on track with your physics. Lets start by listing some of the fan laws:

Fan Laws from the Comair Rotron website's tech notes

Fan Laws are not laws (like the laws of physics) because they do not generally model reality. Fan laws are curve fit to data produced by computational fluid dynamics (CFD) equations which are validated from real world testing. Fan laws are a rule of thumb limited in application, as a subject to curve fit and logarithmic laws.

So where can you use them? You wrote about the Panaflo L, M, H and U spec, I'll use them as an example.

You can predict H spec from M (or vice versa) as they are within the curve fit range. For example:

- You can determine whether to use M at 12V or H at ~10V
- You can check if the M spec is in line with the figures gotten with the fan laws from the H spec.

The Panaflo H spec is 32dB(A) according to the manufacturer. The Panaflo M is 28dB(A) according to the manufacturer. If you count the M noise from the given H noise with the fan laws you get a projected number of 28dB(A).

You can NOT predict U spec from L or L spec from U, because that is outside the curve fit range. For example:

- you can NOT count the noise of U spec from L. Why?

Panaflo L spec is 21dB(A). Panaflo U spec is 38,2dB(A). If you count based on the L spec you would get a projection of 34dB(A). What's wrong? It's outside the curve fit range, because the noise of the L spec consists mainly of motor noise. The noise of the U spec consists of mainly motor noise, some airflow noise, some turbulence noise and some cavitation noise.

- You can NOT count the noise of a Panaflo L at 6V from the noise at 12V. Why?

If you count the noise at 6V you would get ~6dB(A). That is not true, because even at 6V you still have a baseline motor noise of, for example, 16dB(A).

In short, fan laws are "false laws", made by engineers applying curve fit. They are meant to ONLY model reality over the curve-fit-range. They give you an approximation of fan attributes, not the exact scientific data. You have to understand where and how to use it. I had to ask several questions about the fan laws before I realised what diffferent aspects you have to consider when looking at the different projections.

Yes, RPM vs voltage is a curve. But it is directly proportional to voltage over the small range in which fan laws are to be used.

Noise does increase at 50LOG10(voltage ratio) over the small range over which the fan laws are to be used. Fan Laws are a prediction by curve fit on a SMALL part of the curve. From 1982 to 1989 fans were modelled well by the 60 figure, from 1990 fans were better modelled over a smaller range by the 50 figure. The improvements were in the housing and rotor, for most fan makers it was the housing (going venturi over simple tube), for some also the rotor.

OK, that's enough used time on my part making this topic off-topic.

PS. Felger, can you please tell a finn what the hell are epaulets?

Felger Carbon · Post by **Felger Carbon** » Wed Feb 01, 2006 12:54 pm

cpemma wrote:Other, non-mechanical, things can conspire to upset our calculations. Here's the guts of a simple fan,

The brown black-banded object between the red (12V) input and the Hall sensor chip is a rectifier diode, sometimes fitted so an accidental reverse power connection doesn't blow the electronics. A Panaflo L1 I dissected also had the diode (but along with a 16-pin IC and other extra clever bits). But it means applying 12V to the fan only sends about 11.2V to the motor, apply 6V it gets 5.2V, and 6/12 != 5.2/11.2.

A very interesting post. First pic of a fan controller I've seen. Howsomeever: if I understand what's happening on short review, the diode effect counters the square-cube non-linearity effect, making the fan volt-RPM curve closer to ideal than would otherwise be the case. Let me know if I got that backward, please?

Felger Carbon · Post by **Felger Carbon** » Wed Feb 01, 2006 1:05 pm

Aleksi wrote:PS. Felger, can you please tell a finn what the hell are epaulets?

In the old days, when one was drummed out of a military service, all military insignias, esp. those indicating rank, were stripped from the miscreant's uniform. A military shirt has a button-down strap on each shoulder, intended to hold an officer's rank. This strap is the epaulet. The epaulets - both shoulders - were stripped.

Full disclosure: I was honorably discharged from the USAF after serving 4 years.

cpemma · Post by **cpemma** » Wed Feb 01, 2006 1:20 pm

Felger Carbon wrote: Howsomeever: if I understand what's happening on short review, the diode effect counters the square-cube non-linearity effect, making the fan volt-RPM curve closer to ideal than would otherwise be the case. Let me know if I got that backward, please?

In that the effect of any diode is more pronounced at lower voltages where the fan is more "efficient", I think you're right. But I've no idea whether the 60mm fan tested had a diode or not.

To explain the simple fan, this post shows the internals of that hall switch. The only other parts are a transistor centre-left (with its base resistor nearby) that acts as a switch for the speed signal. Plus the motor coils and permanent magnet rotor.

Just to add Tibor's Nexus figures in graph form,

80mm and 92mm linearish (though not proportional), the 120mm more of a curve, but join two points on any and you'll not be far out with values in between.

jaganath · Post by **jaganath** » Wed Feb 01, 2006 1:52 pm

The noise of the U spec consists of mainly motor noise, some airflow noise, some turbulence noise and some cavitation noise.

I thought cavitation only happened underwater?

A military shirt has a button-down strap on each shoulder, intended to hold an officer's rank. This strap is the epaulet.

http://en.wikipedia.org/wiki/Epaulette

Epaulette pronunciation: ĕp'-ǝ-lĕt, a French word meaning "little shoulders" (epaule, meaning "shoulder"), originally meant only one type of ornamental shoulder piece or decoration used as insignia or rank by military or other organizations. Epaulettes were fastened to the shoulder by a shoulder strap, parallel to the shoulder seam, and a button near the collar. The placement of the epaulette, its color and the length and diameter of its bullion fringe were used to siginify the wearer's rank.

at high RPM the blade efficiency falls as a blade profile has an optimum operating range in the P-Q curve.

By the way, what is the P-Q curve? I remember Dorothy mentioning this as well. I'm assuming P is pressure, Q is fan speed?

Tibors · Post by **Tibors** » Wed Feb 01, 2006 10:24 pm

P is pressure, Q is flow.

P.S. Last time I checked my english grammar the possessive form of Tibors was Tibors'.

cpemma · Post by **cpemma** » Sat Feb 04, 2006 6:57 am

Tibors wrote:P.S. Last time I checked my english grammar the possessive form of Tibors was Tibors'.

For such a singular noun ending in 's', my grammar gives it as "Tibors's"; in the case of plural nouns a trailing apostrophe - so "A Nexus fan's noise level is..." but "The two Nexus fans' noise levels are...".

My bad memory for names.

It's an education, this thread.

JVM · Post by **JVM** » Sat Feb 04, 2006 12:08 pm

I found this article on SPCR with ratings for Nexus 120 and 92mm fans:

http://www.silentpcreview.com/article30-page1.html

Nexus voltage, rpm & mounting

Nexus voltage, rpm & mounting

Sooooo Sleeeeeeeeeeeepy!