new fan standard -> i'm giddy!

fancontrol · Post by **fancontrol** » Tue Jul 13, 2004 12:45 am

Anyone fan-nerdy enough to read about the new 4-wire fan (no, not the big clunky molex variety) standard? Intel is behind this one, too. a taste of the future.

The bottom line is they seperate PWM from tach, so the PWM frequency can be much (1000x) higher while still getting good tach signals.

Also:
They selected a connector such that the fan will work like a normal 3-wire fan provided you have a few mm of free space where pin 4 would be.
They require the fan to run -- and start -- at 30% pwm.
The PWM input is logic-level: controllers no longer have to switch power to the fan (that would have cut big space and $$ from my design).

In short this is how it should be.

The downside will invariably be cost and availability. Which leads to my first question: Anyone know if they exist in nature yet? I couldn't find any off hand. I'd love to get one and compare the noise to the different PWM configurations I have for existing 2- and 3- wire fans.

Also, anoyone know of MBs or controllers that support them? I know Analog Devices and some others are making controller ICs that (will, someday) support the fans.

And, yes, I did search first.

silvervarg · Post by **silvervarg** » Tue Jul 13, 2004 2:10 am

I see a few problems with this proposed standard. The first is that the fan becomes more expensive. Most people and manufacturers pick rather cheap fans, so it will have a hard time to make a major break through.
This should make it cheaper to manufacture motherboards and fan controllers that use PWM, but since the lifetime of a fan is usually less than the lifetime of a motherboard it makes sence to put the extra circuits on the motherboard rather than on the fan.
Extra stuff in the fan should translate to a bigger fan hub, and that means bigger deadspot.

So, while the standard does have a few advantages it also has some drawbacks. Not many users care about reading the fan tach, so it is not a big advantage that you can read it even at low PWM duty cycle.

Jan Kivar · Post by **Jan Kivar** » Tue Jul 13, 2004 7:43 am

Hey, it's Intel's spec. They'd better start shipping them with the new BMW-model-number CPUs. If Intel is willing to do so, the prices should drop, at least a bit.

Oh, and samples of these were provided to the reviewers with the new i915/i925 press kits; like here (see the last pic).

Cheers,

Jan

sthayashi · Post by **sthayashi** » Tue Jul 13, 2004 8:31 am

At this risk of sounding like a complete moron, what advantage does PWM provide over current linear methods of fan control? The best I can think of right now is finer control. But just how much fine control does one need?

bomba · Post by **bomba** » Tue Jul 13, 2004 9:33 am

sthayashi wrote:At this risk of sounding like a complete moron, what advantage does PWM provide over current linear methods of fan control? The best I can think of right now is finer control. But just how much fine control does one need?

I did a very quick skim of the document. Some key points that I picked up:

The primary purpose for the new fan standard is reduced accoustics.
These are 4-pin motherboard speed controlled fans.
The 4-wires are 12Vdc, ground, tach (Sense) and pwm (Control).
The pwm signal is TTL logic.
Fans must start at 30% pwm.
Minimum fan speed is controlled by the fan and may not be overridden by the motherboard/controller.
Fan vendor must specify the minimum rpm and corresponding pwm duty cycle.
Minimum RPM must be no more than 30% of max speed.
If fan does not start at minimum speed, it is acceptable to provide a brief 30% pwm startup pulse.
Depending on implementation, fans may be run at speeds lower than their startup speed as well as turned off completely.
Presumably control shall be automatic based on thermals, implemented either in software or BIOS.

My first take is that it's all good!

fancontrol · Post by **fancontrol** » Tue Jul 13, 2004 2:42 pm

silvervarg wrote:I see a few problems with this proposed standard. The first is that the fan becomes more expensive.

I'll bet most of the cost will be in the wire; the new style just reshuffles the components in existing fans. Fan mfgr's blew it when they didn't put the header on their part and forced someone else to pay for the cable.

Extra stuff in the fan should translate to a bigger fan hub, and that means bigger deadspot.

Again, no extra stuff. I'm thinking of taking a panaflo apart to see if I can do it myself.

Not many users care about reading the fan tach, so it is not a big advantage that you can read it even at low PWM duty cycle.

Not many people at home. Business machines and servers that care about monitoring fan health do quite a bit.

Tibors · Post by **Tibors** » Fri Jul 23, 2004 5:19 am

fancontrol wrote:Anyone fan-nerdy enough to read about the new 4-wire fan (no, not the big clunky molex variety) standard? Intel is behind this one, too. a taste of the future. [...]

Looks like the future is already here Spire SPA775 QuadroFlow II™ Socket 775 Cooling kit.

asapin · Post by **asapin** » Fri Jul 23, 2004 8:41 am

More like CoolWave II

http://www.spire.nl/fcc.asp?ProdID=197

If you don't like the 4-pin pwm you can plug the fan into a normal 3-pin header. The fan's temperature control takes over in 3-pin operation.

new fan standard -> i'm giddy!

new fan standard -> i'm giddy!

Re: new fan standard -> i'm giddy!