0

I play around with an old 3-wire CPU fan from a dell server PC. It is this thing: 3612KL-04W-B66 https://www.coolingsurplus.com/index.php?option=com_content&view=article&id=86:3612kl-04w-b66&catid=6:92mm-fans&Itemid=7

It was mounted at the end of heat pipe over the CPU. It was pulling the air from the CPU to the outside of the case.

I found a datasheet here (https://www.electronicsdatasheets.com/manufacturers/nmb-technologies/parts/3612kl04wb66) but this particular model B66 is not in it, only B50 and B60 and other round numbers and they are all 2-wire fans.

According to some info I found about dell fan wiring, black wire should be GND, red wire is +12V and white wire is TACH feedback.

As I understand the basics of 3-wire brushless fans, I can control the speed by varying the voltage (10.5 - 13.2 V according to datasheet) or low frequency PWM the 12 V input.

As a less experienced fan user I would expect the PWM method to cause the fan to slow down while running at full speed at 12 V. But the fan acts in the opposite way, at least for me.

It spins up to speed for a short period after connecting the positive wire to 12V PSU (made from ATX PSU), then it slows down to very low speed.

When I connect and disconnect the positive wire manually in short succession (0.5 Hz by my fingers judgment) the fan speeds up significantly.

I ask you guys for some advise how to actually control the speed of this fan, roughly and manually.

Edit: I managed to resolve the mistery behind the low RPM at 12 V continous. There is some speed controll built into the fans board. I can see a small part coming out from the side of the stator case. Might be a temperature sensor. I pointet a heat gun at it from behind the air flow. The fan revived to significant speed :-) I am stil curious, what is the white wire for (seems not to be TACH, only bias on the analog scope) and how to get rid of the internal speed control

Edit: Shorting the leads of the termistor (youtube.com/watch?v=AVovy7UWDQ0, is exactly the same fan by the way) lets the fan run at full speed and I can control RPM now by lowering the voltage. The question about the function of the white wire stays in the room :-)

Artur Cichosz
  • 125
  • 1
  • 7
  • @DKNguyen You can't PWM the power input on a fan because it goes through the electronic controller that is driving the brushless motor itself. Actually you can. Today's PCs control the speed of fans by driving them with low-frequency (or even sometimes high-frequency - >20kHz) PWM. Measuring the speed can be problematic though, but I'm sure they drive with PWM. – Rohat Kılıç Jun 08 '20 at 19:43
  • @RohatKılıç Oh really? I did not know that. Are you sure they aren't using a buck of some kind? How does the integrated driver handle constantly losing power like that? – DKNguyen Jun 08 '20 at 19:44
  • All of the server PCs I have looked at use fans with a direct attached voltage and ground connection. The fan speed is controlled by a third wire that receives a 0 to 5V PWM signal which is normally modulated at over 20kHz. For failure detection a fourth wire may be present on the fan to be able to monitor a 0-12V signal that shows a TACH signal that follows actual fan speed by some factor of 1X, 2X or maybe even 4X. – Michael Karas Jun 08 '20 at 19:55
  • @DKNguyen I had read an article about that but couldn't find it now. I'll drop the link here once I found. I don't know the details about the internal circuitry but I'm sure they can be driven via a low-speed PWM. IIRC, there are a few problems with LF PWM like audible noise and reading the tacho signal (it's present if and only if the power is present). – Rohat Kılıç Jun 08 '20 at 20:04
  • @RohatKılıç Was it this article? https://www.analog.com/en/analog-dialogue/articles/how-to-control-fan-speed.html – DKNguyen Jun 08 '20 at 20:11
  • @DKNguyen Ah, yes it is! – Rohat Kılıç Jun 08 '20 at 20:17
  • @DKNguyen funny, I read this article to get the basics. That is why it is confucing to me, why the fan behaves this way. It is 3-wire fan, so the 3rd (white) wire should be the TACH feedback and not the PWM input. the fan should be at near full speed with continuous 12V applied. – Artur Cichosz Jun 08 '20 at 21:14
  • @MichaelKaras I believe you mixed something up. 3-pin fans have 12V, GND and a tacho out and 4-pin fans have 12V, GND, tacho out and pwm in. As far as I know, 3-pin fans are controlled by analog voltage as the OP wrote and applying a pwm to the 12V line does not automatically work. At least for me and the fans I tried, a pwm of 25kHz (which I believe is the standard you're refering to) did not work – Sim Son Jun 09 '20 at 13:09
  • @MichaelKaras also, the 12V signal is not 0-12V, it's a oulsed open drain signal. You wrote all server pcs... I don't know if there are special fans for server racks, but for commonly used pc fans I'm pretty sure – Sim Son Jun 09 '20 at 13:11
  • @ArturCichosz you could try to put a capacitor to the 12V line and apply a high frequency pwm. The reason the fan speeds up when driven with 0.5Hz pwm probably is the fan's internal circuit. Many fans appear the ramp up to max speed on power up so they never stall (which could happen when starting with very low rpm). So, each time you plug it in, the fan will be in its "boot sequence"... – Sim Son Jun 09 '20 at 13:20
  • @SimSon - You missed a key part of the quote. I actually wrote "All of the server PCs I have looked at...". That is not the full universe of servers! You are correct, some fans would have an open drain type TACHO output. When we build servers we typically bias TACHO with a pullup to +12V right at the fan connector and then two more lower resistors to adapt the total swing output to 0-5V or 0-3.3V depending upon the logic level input range of the SOC that monitors the fan speeds. – Michael Karas Jun 09 '20 at 14:40
  • @MichaelKaras hmm, I don't undestand that. If the tacho signal is a 0-Vcc analog signal there is no need for a pull-up. If it is an open-drain signal, there is no need for a voltage divider, you just pull the tacho line to the mcu's logic level. I once developed a fan controller and all the fans I tested (also not the whole universe ;) used open-drain pulses for the tacho. Do you know fans with an analog tacho signal. – Sim Son Jun 09 '20 at 15:17
  • @MichaelKaras I just tried to read the tach signal with a scope (analog discovery 2 device from digilent) 5V/div, PSU minus wire to Scope minus input and TACH signal to scope plus input. There is nothing but bias around 0V. 12 V continous applied to the fan running quite slow. The fan starts spinning at a round 9V and the RPM rises slowly until 12 V but is stil quite slow for my taste. Should I try a pull-up resistor between TACH and +12V from PSU to read the TACH signal? – Artur Cichosz Jun 09 '20 at 15:46
  • I managed to resolve the mistery behind the low RPM at 12 V continous. There is some speed controll built into the fans board. I can see a small part coming out from the side of the stator case. Might be a temperature sensor. I pointet a heat gun at it from behind the air flow. The fan revived to significant speed :-) I am stil curious, what is the white wire for and how to get rid of the internal speed control – Artur Cichosz Jun 09 '20 at 16:01
  • Shorting the leads of the termistor (https://www.youtube.com/watch?v=AVovy7UWDQ0, is exactly the same fan by the way) lets the fan run at full speed and I can control RPM now by lowering the voltage. The question about the function of the white wire stays in the room :-) – Artur Cichosz Jun 09 '20 at 16:36

1 Answers1

2

I figured it out myself.

There is some speed control built into the fans board. I can see a small part coming out from the side of the stator case. Might be a temperature sensor. I pointed a heat gun at it from behind the air flow. The fan revived to significant speed :-)

Shorting the leads of the termistor (youtube.com/watch?v=AVovy7UWDQ0, is exactly the same fan by the way) lets the fan run at full speed and I can control RPM now by lowering the voltage.

Artur Cichosz
  • 125
  • 1
  • 7