[SOLVED] 2 x PWM Fans on one header. PWM signal sharing, speed from just 1.

[Boltar]

Hi

I'm setting up a minecraft server PC. It's an AMD FX 8 core 4GHz CPU but the motherboard is somewhat basic, I only have 2 fan headers, one for the CPU and one for the system fan. I bought a push/pull type cooler from eBay but the problem is that the 2 fans on the cooler are wired together to a single 3 pin plug and as a result they always just run at top speed as there's no PWM. I want to get 2 PWM fans to replace the ones the cooler came with but I'm worried about the current draw from the single header (although I am reasonably sure it is OK up to 4 fans) and if the PWM signal will work if it's shared between two fans. My mental concept is as follows:

Fan1-12V -> Fan2-12V -> PSU Molex-12V
Fan1-Gnd -> Fan2-Gnd -> PSU Molex-Gnd
Fan1-Spd -> Motherboard Header-Spd
Fan2-Spd -> No Connection
Fan1-PWM -> Fan2-PWM -> Motherboard Header-PWM

Connecting the fans to the PSU should eliminate any worry about current draw from the header. I was concerned that wiring the speed together for both fans would confuse the motherboard so just wiring one should be okay, they are the same type of fan so their speed should be similar for a given duty cycle anyway right? Of course I won't know if the undetected fan fails, but I can live with that. My main worry is if the PWM will be okay for both fans, my intuition is that it will be, I'm assuming the fans use a pulled down p-type/pnp setup for this and so should be a high resistance on the base/gate of the driver? But I'd just like to confirm all this with folk more in the know about this type of thing. I appreciate any help you can give.

Cheers
Marc

EDIT: Done a bit more research and I think I can make do with the existing fans. Looks like I was wrong about the p-type setup, they seem to use an n-type. That to me seems a bad idea, switching the gnd instead of switching the +ve is sketchy. I could, i expect use a mosfet and a resistor to turn both fans into PWM capable ones. Although I'd probably want to invert the motherboard signal with a BJT and use a p-channel fet so the fan is constantly grounded. Although that would be dicey at 25kHz, I'd need a pretty low value rail pull resistor on the inverter tranny. Hmmmm.

EDIT2: Wait, from the diagram I'm looking at, the n-fet is pulled up, so the fan is on full if the PWM is not connected I assume. So the motherboard signal would already be inverted right? But that is at 3.3V which is not enough to turn off a p-fet so I'd need a driver as well. Then I'm back to the low value resistor because of the 25kHz. I'm actually curious at to what n-fet these fans use that can be turned on at 3.3V? Even signal level fets are dicey at that voltage.

 

[Paperdoc]

You are doing it the long hard way. I really suspect that the root of the problem is this: the mobo CPU_FAN header is a 4-pin male and it IS using the new PWM Mode to control its fan's speed. If you plug a 3-pin fan into a header using PWM Mode for control, it always runs full speed, just as you report.

Virtually all mobo fan headers can supply up to 1.0 A max current to the total load. So look at the specs for the fans you are considering and check for their max current at full speed. Most these days are 0.10 to 0.25 A, so there is NO problem connecting 2 (even up to 4 for some) fans to one header. To do this, just get a simple 2-output 4-pin Splitter. That is a simple device that merely connects the Ground and +12 VDC and PWM lines in parallel from both fans to the mobo header pins 1,2 and 4 , respectively. Only ONE of the output connectors will have a Pin #3 so it can return the speed signal of that one fan to the header and ignore the other. Do NOT get a fan HUB. It is a different device that also has a third type of "arm" that must plug into a power output from the PSU to get power for all its fans, avoiding any draw from the header +12 VDC pin. You do not need that unless you get some REALLY powerful fans that draw unusually high current.

In choosing your fans, look for high air flow max, and also check the max pressure rating. It should be over 2mm, preferably over 2.5 mm, for use on a CPU heatsink with fins.

The label "PWM fan" may have confused you, because what that means for a computer case fan is NOT what most motor control people call a PWM drive system. Any other place, a PWM drive supplies to the motor a DC supply that has been modulated with a PWM signal so it is a pulsed voltage producing rapidly pulsing current flow though the motor windings. In computer case fans they've altered the details to achieve the same end. The signals from the mobo header are Pin #1 Ground, Pin #2 constant +12 VDC, Pin #3 fan speed signal (5 V pulses, 2 per revolution sent back to the header), Pin #4 PWM signal (5V p-p, about 20 KHz frequency). Inside the motor case there's a small circuit board that, among other things, provides the electronic switching of power to the several stationary windings in the frame. (The permanent magnet is on the rotating shaft.) This replaces the older system of stationary magnets and stationary brushes in the frame contacting moving commutator segments on the armature - its the brushless motor design. In the new 4-pin PWM computer fans, that circuit board also contains the small chip that uses the PWM signal supplied on Pin #4 to modulate the power supply from Pin #2 to modify current flow through the windings. Thus the current modulation is done inside the motor case, not outside by a motor power supply module. In fact, makers of these fans specifically tell you NOT to try to supply them with a modulated power source because they need the constant +12 VDC supply for circuits other than the windings, and a modulated source containing voltage pulse transients might actually damage some components.

I don't know circuit designs. But you are right about how that PWM signal is used, and this goes back to making it as compatible as possible with the older 3-pin fan design. (That had brushless DC fans with no special modulator chip, and thus the VOLTAGE supplied on Pin #2 was altered by the mobo header from 12 V (full speed) down to about 5 V (avoid stalling if voltage is lower.)) So the new PWM design was set up so that if it is plugged into an older 3-pin fan header that can only do that Voltage Control Mode and cannot supply a PWM signal, the PWM fan deprived of any signal on Pin #4 will still have its speed controlled by the fact it receives reduced voltage for slower speed, and it does NOT modulate the voltage supplied.