What type of resistors do the Noctua in-line resistors use? I was thinking of making some (over other available options). They are placed in-line with the 12V lead, correct? Just need to know the resistor value and the rated power. Thanks.

Zalman has one of those, some talk about it here
http://www.silentpcreview.com/article85-page1.html

Personally I think it's a "dirty" solution. I prefer to feed the fan 5v (or even 7v using the 7v trick) rather than to use a resistor.

From the SPCR review of the Noctua NH-D14 :

"Ultra-Low-Noise Adaptors: The blue one has an 81 ohm resistor
, while the black one measures 50 ohms. The first brings the speed of the 12cm fan down to ~900, and the second does the same to the 14cm fan."

Not too sure on power ratings, but the larger of the two fans is rated at 1.2 Watts @ 12 Volts, so you'd be looking at less than 1 Watt if using a similar fan.

EDIT: Admittedly the 5v or 7v trick is easier and cheaper (free, in fact!), but conversely I think this is the "dirty" option. Possibly due to me destroying a hard drive by plugging it into the wrong adapter...

To avoid things like that, all tweaking must take place on the fan side, not the PSU side.

Thanks, looks like it's dependent on how much current the fan consumes, so it'll take some calculating to do. I was going to do this using just the 3-pin connectors, so I don't see how a hard drive would be destroyed there =/

If I work with the molex connectors, then I lose the ability to sense the fanspeed, correct? Since it's just feeding the fan either 5V, 7V or 12V. I've done the 5V trick, but need a little bit more, and the 7V seems to be not a good idea for the power supply? I remember reading this somewhere.

Yes yes, I know, momentary retardation on my part. The fan in question was a 4-pin molex type, I made an adapter so that I wouldn't have to modify either the PSU or the fan wires/molex connectors. Then at some point I rebuilt the PC, not paying too much attention, plugged the molex-SATA power connector into the modified molex adapter, hence sending 12v through the ground rail into the hard drive. Needless to say, all the magic smoke escaped.
The smell of 300Gb of films burning up is one I shall never forget, oh the horrors of war...

You can retain the speed sensing of a 3-pin fan by just running the yellow speed signal wire alone to the motherboard header, and getting power from the molex.

The 7V trick is only dodgy if you are drawing more power from the fans on 7V than you are drawing elsewhere from the 5V rail, which you would need an awful lot of fans to do, 1 or 2 fans at 7V will be fine.

You can also use rectifier diodes (1N400x series, eg 1N4001, 1N4004 etc). They drop the voltage by about 0.75V, so put a few in series until you get the required fan speed. They cost about 10c each from almost any electronics shop.

I might give that a try, also. I was reading one of the tutorial stickies up to that used this method, though it didn't make sense, no diagram to follow. So I guess you just splice into the 12V line, line these diodes up in series, and then see if it works? If I wanted to measure the resulting voltage after applying this, what's the best way to do this (either mathematically or with a multimeter)?

Revoltec sells some that feed a constant voltage to any fan attached, no matter its power rating. They use Zener diodes, which are harder to test on a DIY project, but are really cheap and well made. I use one that gives 9,5v to a Slipstream 800 RPM, dialing it down to absolute quietness. They also don't dissipate heat like a resistor, and are not as dangerous as the dreadful 7v trick...

Another reason the resistors are not that good a solution is that the resistance needed for a certain voltage drop varies with fan -- ie, 50 ohms will not give you the same voltage drop with every fan. So you really need to play around with the fan and different resistors before you get to soldering. Zener diodes are better; just don't forget to heatshrink them.

Thanks for the tips. I've never seen Zener diodes being used to control fan speed before. Is this the typical setup - http://www.reuk.co.uk/Zener-Diode-Voltage-Regulator.htm or is there another guide I can look at?

From the now deceased cpemma:

Ahh, neat. Putting it in series makes it even more simple.

I would recommend using rectifier diodes rather than a Zener. Zener diodes are more expensive, and you need to know in advance the actual voltage required. Rectifier diodes, being so cheap, you can buy a handful of them and keep stringing them together (in series, with the silver bands pointing in the same direction, towards negative) until the desired fan speed is obtained. No need to worry about the rectifier diode's power rating either, they are rated 1A, more than enough for any fans you're likely to use.

Here's something I posted a while back which may be of interest to you.

Thanks. I like that fanspeed knob you made out of the diodes. If I were to reduce fanspeed based on voltage only, how would I measure the voltage, just between the fan +/- leads?

+1 on the regular diodes.

Cheap and bulletproof. Zeners are mostly useful as a voltage reference.

What I used to do:

The rectifier diode adapter works well. I use a Zalman Fanmate ( $CAN 10) to determine the fan-rpm-airflow combo that does the job and then measure the voltage that the Fanmate is delivering to the fan. Then I build an adapter with the required number of diodes to get the voltage drop I need.

Now:

I run the chassis and/or CPU fans from the CPU PWM fan connecter with a PWM y-adapter and get quiet operation at idle and lots of airflow on demand when recoding a video.

I've asked multiple times already in the thread, but how do I measure the voltage across the fans, just between the +/- leads?

You can't run multiple fans off that one PWM connector, right?

To find out the voltage being applied to the fan, you first need to know what the supply voltage is (I assume you'll be using a 12V lead from the PSU). Then use your voltmeter to measure the voltage drop across the diodes, ie put one lead on one side of the diodes and the other lead on the other side of the diodes. Subtract this value from the supply voltage.

Example: let's say you've got 6 diodes in series, then you should see a voltage drop of about 4.5V across the diodes. Subtract this from 12V and you get 7.5V being supplied to the fan.

Cool. I guess these are the diodes everyone's talking about: http://www.radioshack.com/product/index ... Id=2036268

So just to know how this is working, diodes are supposed to allow current to flow in just one direction, so we're placing these diodes to oppose the current (which results in the voltage drop somehow), but it still lets some current through because it doesn't completely oppose all the current, yet it's okay and the diode does not burn up. Hmmm?

Yes, that is the correct diode, although it's much more expensive than what I pay for them. Anyway, for the usage being discussed here, the diodes do not oppose the current. The diode will only oppose current when its polarity is reversed (reverse biased). We are using the diodes with a forward bias. The voltage drop is caused by the need to "push through" the diode's internal junction to switch it on before current can flow. As such, what we are doing here is exploiting a side-effect of the diode's function; these diodes are not normally used for this purpose.

Just to reiterate, these rectifier diodes need to be operated with a forward bias, which means that the cathode (the end with the band on it) must point towards negative, and the anode must point towards positive. If you get it wrong, it will not "burn up", but it will simply block the current flow and your fan won't spin.

1N4001
=> On ebay: 1N4001, 100 pieces, US$2.99, Free Shipping from Hong Kong China

"multiple fans off that one PWM connector"
=> I've never had a problem running 2 or 3 fans on the CPU fan header. Try searching for "fan power splitter" on ebay

Awesome, thanks for the great explanation!



I *think* at Radioshack it is 50pcs for $1, I think, I'll just check at the local Radioshack since it's 5min away.

So what is the safe loading level for one connector? Because you cannot pull too much current from one connector, right? But then we are undervolting these fans, which results in less current being pulled anyway.

"what is the safe loading level for one connector?"

I've never had any trouble running 3 low current fans. Fans use approximately 1/2 the startup current when running. The stated current requirement on the fan's label is the worst case startup current draw. Mainboard manufacturers aren't making a commitment about the capability of the cpu fan header, but current conjecture is that there is at least a 500ma capability. Some evidence of this capability is evidenced in that the factory supplied fan on the Pentium 840EE is rated at 420ma (which would be the startup current).