1 of 1 people found this helpful
The DZ68BC has 4 PWM fan headers (CPU, rear, front, aux). I consider PWM fans superior than the voltage controlled or always-on fans.
My case (a fractal design R3) came with two 3-pin fans. I just replaced them with two PWM fans.
Figure 26 in the Intel Desktop Board DZ68BC Product Guide shows that it is possible to connect a 3pin fan to the 4 pin fan header, but I think that the fan will then always run on max RPM .
The CPU fan header can only support 4-wire PWM-controlled fans. If you use a 3-wire fan on this header, it will run at full speed always.
The Front, Rear and (if present) Aux fan headers can support either 4-wire PWM-controlled fans or 3-wire standard fans. We have a circuit that can detect the type of fan present. If a 4-wire PWM-controlled fan is detected, we utilize its inherent capability for managing is own speed based upon the PWM duty cycle delivcered to it. If a 3-wire standard fan is detected, however, we control its speed using a circuit that scales the voltage delivered to the fan (again based upon the PWM duty cycle).
vbaderks is correct in one aspect: 4-wire PWM-controlled fans *are* superior. They are more expensive too (but, IMHO, worth the expense), though prices have been slowly working towards parity.
Now, back to the original question. It is indeed possible to connect two 3-wire fans to a single fan header using a "Y" cable (available from many after-market component and Internet stores) - but you have to be very careful doing this!! You have to be careful to not exceed the current limits for the fan control circuit; if you do, you will damage the circuit permanently. The limit is typically 1.5A (read your board's specs to be sure!) but I strongly recommend that you not exceed 1A. You also need to understand that the fan speed that will be seen from the two fans will not be accurate. Each fan sends (typically) 2 tachometer pulses per revolution (via the 3rd pin). In the simplest of terms, the speed is determined by counting these pulses over a sampling period. When two fans are connected, they are both going to be sending pulses on this wire simultaneously. Where the individual pulses overlap, one pulse will be seen (and counted); where they do not, the separate pulses will be seen. Since it is almost impossible for the two fans to spin at the exact same speed and/or have their tachometer pulses overlap perfectly, the fan speed measured will fall somewhere between the speed of the slower of the two fans and the speed of the two fans combined. Secondly, a fan stall can only be detected when both fans stall...
Is it possible to connect two pwm fans to the cpu fan header with a Y cable if the amps doesn't exceed the circuit limits? Will there be a conflict in signals as described with 3 pin fans?
It should, in theory, work the same. It will be more difficult to accomplish this off the CPU header, however; CPU fans typically have higher current requirements. I have also never seen a 4-pin "Y" cable available commercially, but I suppose you could make one. There would still be a conflict created by the wire or'ing of the tachometer signals, however...
Thanks for the clear explanation about the fan headers. Would this not be something for the technical manual of the board?
In the BIOS it is possible to assign the 'purpose' to a fan (CPU \ inlet \ outlet \ system \ memory \ etc). Is this just a text string or are these values actually used?
(something also not documented in the manual...)
I would like that the system increases the RPM of the outlet fan but I don't see that when running a stress test. I have changed the purpose of the outlet fan to 'CPU' but it doesn't seem to make any difference.
When the chassis fan header detects a 3 wire fan and uses the circuit that scales the voltage delivered to the fan what voltage range is available? Does it have preset voltages output such as low, med and high or is it like a fan controller capable of any voltage output? By adjusting the pwm duty cycle in the bios is it possible to get a specific voltage desired and fine tune fan rpm? Haveing that kind of control on 3 wire fans would be nice.
I have a 4 wire " Y" cable purchased with the thought to connect two 4 wire pwm radiator fans to the DZ68BC cpu fan header. Glad I found this thread and now realize the complexity. There are pwm splitter cables available that connect one pwm fan to a header and split the pwm control signal with two additional fans that are molex powered . The pwm fan header only has an rpm input from its attached fan , powers only the attached fan but all three fans are controlled by a common pwm signal. It works for some ,but every mobo is different. Sept. 2005, 4 wire pwm spec. revision 1.3 by Intel states " If driving multiple fans with a single pwm output , an open-drain/open collector output buffer circuit is required." Would the DZ68BC meet that spec and could multiple fans be controlled by a single header if powered by molex? Also are the pwm signals strong enough to split, read they can be checked with a multimeter? Not being an EE this is hard to understand , imagine there are a few out there like me wondering what the possibilities are for fan control on this board.
@everyone who helped answer my question - thanks!
I guess I will have to replace my chassis fans with PWM ones for the best control. I am very interested to hear Intel's response to the fan control Qs and suggestions for updating the documentation that have come up since the original post.
The Usage indicator that you can change in the BIOS is strictly a labelling capability. The parameter is picked up by applications like Intel(R) Desktop Utilities, which will use it in displays and alerts. Changing the Usage indicator has no affect on the fan control capability.
The chassis inlet (front), outlet (rear) and (if present) auxiliary fan headers are currently set to respond to changes in the PCH and Memory temperatures. Likely, these will not see significant change when you stress just the processor, however. Unfortunately, we did not include a capability to change the assignments in the BIOS. I have proposed an addition to the BIOS to provide this capability but it is not approved as yet...
The voltage scaling circuit generates what we call an "offset" voltage. The voltage delivered varies linearly from 5V to 12V over the 20% to 100% duty cycle range.
I am the software architect and not an EE and thus not the right person to be trying to explain the nitty-gritty aspects of the hardware design. I am pretty sure we use open-drain output buffers on all of our boards, but this is not the whole story. The fan's ability to "understand" the PWM signal is dependent upon pull-ups implemented within the fan (see requirements in the 4-wire fan specification). If you power the fans separately (off molex), I am not sure whether it will work (again, I am not an EE). When two fans are connected, there may also be issues caused by the combining of the pull-ups and this could result in the fans not responding properly to the PWM signal (resulting in restricted control range, fixed speed etc.). I will discuss this with one of our hardware experts and see what they think...
In response to morabo, an issue with 3-wire fan control is that the minimum voltage that you use must be sufficient for the fan to overcome its inertia and start spinning. Depending upon the fans used, this can have the side-affect of resulting in higher-than-desired minimum fan speeds. This is not an issue for 4-wire fans, however; you can use far lower minimum duty cycle values (20% typically) and achieve far slower (yet stable) minimum fan speeds. There is no start-up issues in this case because the fan is getting 12V constantly and can use this to easily overcome inertia. Secondly, the response of many 3-wire fans to the linearly-scaled voltage is, well, not very linear. In many cases (and relatively speaking), small increases in voltage at lower voltage levels can result in large increases in fan speed and small increases in voltage at higher voltage levels can result in only small increases in fan speed. Not so with high-quality 4-wire fans; they respond much more linearly. But, you need to understand that there *are* 4-wire fans out there that use a cheap, cheap, cheap voltage scaling circuit and thus are no better than the 3-wire fans in the linearity of their response; remember the credo: you get what you pay for...
Scott, thank you for your informative input on fan control. You have given me alot to think about , much more complex than I thought. I now see why the corsair link for fan control first announced in 2010 is still not perfected .