PWM vs. DC fans - which are generally better?

[uplink-svk]

Greetings,

I was using Vardar PWM fans till today, changed them for Aerocool Dead Silence. These are DC fans. They can be regulated pretty well.

Are there any down sides and up sides of PWM vs. DC at all? [besides the digital/analogue aspect, and 3 vs 4 pin header].

Thank You!

Regards

uplink

 

[Paperdoc]

Some practical points: because of the difference in how ports operate, a 3-pin fan plugged into a true 4-pin (PWM Mode) port can only work at full speed all the time. You have NO control of that fan's speed this way. You can only control the speed of a 3-pin fan with a true 3-pin port OR (on some mobo's IF the BIOS Setup screens allow you to reconfigure that 4-pin port to behave like a true 3-pin port in Voltage Control Mode).

Mis-matching the other way - that is, plugging a 4-pin fan into a 3-pin port - WILL work with fan speed control. The difference in this case is that the 4-pin fan actually is working as a 3-pin fan, and hence losing the advantages of 4-pin design.

What advantages? 4-pin fans under PWM control mode can be run down to lower speeds without stalling. They also can be started up at lower speeds that a 3-pin fan can. For this reason, when a mobo is controlling a 3-pin fan it will always feed it full 12 VDC at first to start it up reliably. Then, in a few seconds after it gets a temperature reading it can reduce the voltage to the fan to whatever is right for that cooling requirement. Start-up of a 4-pin fan in PWM mode does not require this.

 

[uplink-svk]

Paperdoc - thank You for Your comprehensive answer! Most likely the adept of "solution".

Allow me a few more questions.

Whilst using all 6 + 2 headers of my motherboard, with 3 x Noctua Industrial + 7 x EK Vardar 1150 rpm version, all PWM, I encountered [using pwm splitters], I encounter each time the motherboard starts up full speed of fans. I guess that's the feature of motherboard, not the limitation of the PWM, right?

Secondly. I observe identical behaviour with 3 pin fans [yet, 3 x Corsair AF Quiet Edition series [led] + 4 x Aerocool Dead Silence + 3 pwm Vardars [about to be changed].

In general, the only difference I noticed is tuning. Whilst tuning DC, it does funny stuff, turns on some voltage and turns if off and goes to another, whilst tuning PWM, it goes seamlessly.

My final question is. Are there any real disadvantages for me, using DC instead of PWM? Dead Silence have 3V starting voltage, I cherry picked them for this purpose, and they can go from up to 350 rpm, which is really silent, though, I use them on 600-800 rpm for performance. Corsair-wise, can't do much in there, since they're not really that special, but also, can go as low as 430 rpm.

So, should I re-think it and go after PWM, or will DC do just as fine job for me as PWM?

And one totally last. How come, that all splitters, no matter whether 3 pin or 4 pin have control pin only on one connector, and others are 1 1 0, or 1 1 0 1? [1 - pin, 0 - no pin] - are missing the control pin. Are the other 1-xy fans being controlled as well? Is the control from one fan being enough?

Thank You! You seem really knowledgable!

 

[Paperdoc]

1. Starting up at full speed (or not) is set in the way the BIOS was written. It's a choice the programmers made.
2. I'm not sure what you mean by "tuning". But if a fan stops, it MAY be because the voltage to it is low, not off.
3. Real disadvantages switching to 3-pin? YES, for your equipment. Never mind the fine details of lowest speeds, etc. If you plug a 3-pin fan into a true 4-pin header (that is, it uses only PWM control, and not voltage control) the port CANNOT control the 3-pin fan's speed - it will always run full speed. Unless you have enough 3-pin fan ports to plug in all your 3-pin fans, you will have this problem.
4. On both 3- and 4-pin fans, connector Pin #3 is the speed signal. It is generated in the fan motor as two pulses per revolution and sent back on this line to the mobo for counting. If you were to use a splitter that fed BOTH pulse train signals back to the mobo, it would be counting twice as many pulses and give you a weird speed! Even worse, those two pulse signals actually would be from fans running at just slightly different speeds, so the spacing between pulses would change constantly, making pulse counting very erratic! For this reason you should only send back to a mobo port the pulse signal from ONE fan. That's why splitters only connect ONE fan's speed signal through to the mobo. All the other fans on a splitter run just fine, but there is no way to read their speed signals.

To help understand, it is useful to note that control of a fan's speed does NOT need or use the speed signal at all! That speed signal is used for display of info, and in many cases it also is used to verify that there are some pulses indicating that the fan is still turning and has not failed. But CONTROL of the fan's speed is done entirely by manipulating either the voltage (3-pin fan) or the PWM signal (4-pin fan) sent to it from the mobo port. To decide how fast the fan should run the mobo control system depends entirely on a TEMPERATURE measured by a sensor and compared to a target. For the CPU cooler the temperature sensor is built into the CPU chip itself and sent out to the mobo on one pin of the CPU. For the case ventilation fan(s) the temperature sensor is built into the mobo by its manufacturer. Speed control isn't really that: it is TEMPERATURE CONTROL, and the particular thing manipulated to achieve that is the speed of the fans.

 

[uplink-svk]

Paperdoc - now that is an excellent answer. In general I have no further questions. Due to fact my motherboard does have full 3 and 4 pin control [checked the chip that takes care of it], I will stay with the 3 pin, since they're super silent, and are more reliable in speed control.

E.g. when I used Noctuas Industrial, they went haywire a lot. They went full speed, because of bad interpretation of PWM signal.

Thank You once more!

P.S.: If I understand right, there are no real downsides of 3 pin, when my motherboard is fully hybrid, right?

 

[latedev]

In actual fact the only difference between the fans is that one (PWM) has a pulse out which is normally set to 2 pulses per revolution, and the other is a similar fan with no output pulse. Both fans can be controlled by PWM.

PWM is just a variable width high frequency pulse, where the average Voltage is proportional to the on time of the PWM pulses. Your just using the remanence of the coil to iron out any possible fluctuations.

I will point out here that you don't need 4 wires for PWM.
Most PWM fans are 3 wires gnd, sensor out and power, the only reason for the 4th wire 12V is to have a pull-up resistor attached, to stop the PWM pulse from reducing in Voltage level. You could of course attach a normal 2 wire fan and let the Voltage from the PWM reduce or still have the same arrangement and just not have the sensor wire attached. Any external temperature sensor can then be used to control the fan.
Fan control is no different for any DC fan, unless you are talking about stepper motors.

 

[Paperdoc]

My understanding is quite different.

3-pin fans only operate in Voltage Control Mode. The signals to them are Ground on Pin #1 (Black), varying +DCV on Pin #2 (Red), and Speed pulse signal on Pin #3 (Yellow). The voltage supplied on Pin #2 can range from 0 to 12 VDC, although fan headers rarely reduce this voltage below 5 VDC because there is a real possibility the fan could stall and fail to re-start until voltage is increased significantly. This varying voltage supplied to the fan is the means of changing the fan's speed. The fan motor itself generates a speed signal as a series of pulses (2 per revolution) That is sent out on Pin #3 to the mobo header. The mobo counts those pulses to generate a speed readout for information. Most mobos also monitor that signal for fan failure - a missing signal will trigger an alarm of failure. In the special case of CPU_FAN headers, some mobos go further and will shut down the system after a very brief alarm warning to prevent possible CPU overheating, without even waiting for the CPU's internal temperature sensor to report such a major problem.

4-Pin fans work differently using PWM Mode. For computer fans this is not done the same way as PWM speed control of common DC motors in industry. A computer PWM fan is supplied with these signals: Ground on Pin #1, +12 VDC fixed on Pin #2, Speed pulses on Pin #3, and the PWM signal on Pin #4. Note two differences here: Pin #2 always has +12 VDC, not varying; and Pin #4 has the new PWM signal. The PWM signal is something like a "square wave" in that it is either full on or full off. BUT it is not fixed at 50% On as a square wave is - the PWM signal's "% ON" value varies. The signal frequency is 20 to 25 kHz. Inside the fan motor case there is a small printed circuit board with a few components, and some of these modify the current flowing through the motor windings from the +12 VDC supply from Pin #2. The current flow is turned on and off by the PWM signal, so that the average torque produced in the motor rotor, and thus the fan's speed, is altered.

The signals on the first 3 pins of each fan type are similar for compatibility. For example, the Speed pulse signal on Pin #3 is the same in both. This similarity and the mechanical design allows both 3- and 4-pin fans to plug into both 3- and 4-pin mobo headers. BUT there are limits on compatibility.

If you plug a 3-pin fan into a 4-pin header using PWM Mode, the fan gets a fixed +12 VDC on Pin #2 and never receives the PWM signal from Pin #4. It has no circuitry to use that signal anyway. So, that fan always operates at full speed with its fixed voltage supply.

If you plug a 4-pin fan into a 3-pin header operating in Voltage Control Mode, the fan receives no PWM signal, so its modulation circuitry cannot do that job - it merely feeds the power supplied on Pin #2 to the motor windings. However, the power supplied to the fan on Pin #2 is NOT fixed at +12 VDC - it varies, so the motor speed varies according to that voltage, just as a 3-pin fan would do. This arrangement eliminates two small advantages of 4-pin (PWM Mode) fans over 3-pin: PWM fans can be started from a stopped state with less than 100% "On" of the PWM signal, and they can be operated at lower minimum speeds than can 3-pin fans without stalling.

Computer PWM fans are not the same as common DC motors used with PWM speed control systems. They do require 4 wires because they must have a smooth clean 12 VDC supply and a separate PWM signal. Their makers specifically tell you NOT to try to use them with the types of PWM speed control systems normally used with common DC motors, because a PWM-modulated DC supply can damage the computer fan's internal circuitry.

If you connect a 2-wire fan such as those supplied for connection to a PSU Molex output to a mobo fan header's Pins 1 and 2, it will act exactly like a 3-pin fan but without feeding any speed pulse signal back to the header. On a Voltage Control Mode header its speed will be under mobo control; on a PWM Mode header it will run at full speed all the time. In either case the mobo may warn of fan failure since it will receive no speed signal.

 

[uplink-svk]

Shame I can't pick two same solutions from two identical people