Building a 300W Fan Controller from an ATX PSU

Howto by geoffrey @ 2008-05-24

What do you if you have more than 20x high speed fans in your PC and want to control them through an easy front-end? Build your own fan controller of course! In this tutorial we show you how to convert that old 300W ATX PSU into the most powerful rheobus you have seen.
  • prev
  • next

Circuit analysis

Circuit analysis

Electricity is all around us, since the industrial revolution and Edison’s lamp we now have a world that is based on the consumption of electric energy. Over all the years we saw the electric network grow and become more important, all because of the higher efficiency that was necessary to please the growing population. The electricity network was being standardized, all you need to do these days to get your bread toasted is plug in the mains connector and wait until the toasts jump out; hot and all.

Next to the industrial revolution was the digital revolution, with the introduction of the computer as a driving force. We no longer need to write a letter and bring it to the post office so that it can be sent to your dearest beloveds, no... all we need to do is open our digital/virtual mailbox and hit the "send message" button, few seconds later the message is all ready received by who ever it is being send to, wherever he lives on our planet. Modern computers differ much from the ancestors who were build in the late 1950's, today we have standards and everything is being made plug-and-play; again because of the efficiency and the demand for it.

The progress in efficiency introduced the transistor and with it low voltage digital circuits. These circuits weren't able to run at the 110V/220V voltage that was available on the net so the game of power conversion became a big industry with great manufacturers like Texas Instruments, National Semiconductor, Philips, ... coming out of it. For today’s computers we need three common voltage sources being 12V, 5V and 3,3V. All of these sources are converted from the 110V/220V voltage available on the net; it’s the power supply which transforms/converts the high voltage electricity to lower levels.

Power conversion in ATX PSU's hasn't really changed over the past 10 years, even the modern pc power supply is somehow based on the same TL494 driver IC which Texas Instruments released a decade ago. This type of circuit is called a 'half bridge converter' and makes use of fast switching high-side transistors to make the power supply highly efficient compared with linear power conversion. In the power supply I have the manufacturer choose to use the SG6105 regulator produced by System General (SG).

Madshrimps (c)


Let us have a look at how the SG6105 is put into use for converting electronic power; here is an overview of a typical half bridge converter circuit:

Madshrimps (c)
Original picture courtesy of Elektor International Media B.V.
Elektor n° 476: "DC-power nodig? Neem een PC-AT-voeding"


In modern computer supplies we can clearly notice two parts in the PCB design, one side being the primary part of the power supply which can be quite dangerous to work on since it's a high voltage circuit and the secondary part which contains the various low voltage circuits and control unit. In our picture above the high side primary part is found at the left. Both sides of the PSU are separated by the T1 transformer which makes both parts galvanic separated (improving safety). In our example we have a 220V AC mains voltage, instead of transforming it straight to low voltage levels we find a rectifying circuit. Yes that's right, the AC net voltage is first being rectified through the bridge rectifier and C1/C2 capacitors. If you ever have opened your PSU you may clearly notice these two capacitors since they are about the largest components you will ever see inside a PSU. The 320V DC is then being switched on and off by the high side Q1 and Q2 FET's (transistors used as a switch). The current on the primary side of the T1 transformer will then switch polarity and thus create a current on the secondary side equivalent to the type of transformer being used. You may question why we don't go straight to lower level voltage instead of rectifying the high voltage first? Well, the output voltage of the PSU is being controlled by the switching signal of the primary FET's, the longer the FET's conduct the higher the output voltage will be, but controlling the output voltage can only work correctly when the voltage remains at the same level, so that's why we convert the 220V AC into 320V DC first.

On the secondary side of our PSU, after the T1 transformer we get low level voltage which is again being rectified, for the 12V line for example diodes D1 and D2 will conduct depending on how the transformer is polarized. The DC voltage is further stabilized and smoothed out by the linear coils L1(a,b,..) and the filter capacitors. You now have a full functioning power supply, though applying different loads will make the voltage levels swing and thus there is a need to control the PSU via a control unit, being the SG6105 in our example. You know that the Q1 and Q2 FET's duty cycle (duty cycle = time on versus time off) decide how much the output voltage will be in our PSU. The SG6105 decides how long the FET's will be active and inactive, the FET's are being driven through the low power T2 transformer, although that an opto-coupler may also be used. Again, a transformer is being used to make sure that there is galvanic isolation between high and low voltage sides of the PSU, this because of the higher safety. The SG6105 also reads the output voltage of the PSU, if a high load is applied then the output voltage may tend to drop, but the SG6105 will then increase the duty cycle of the high side FET's in order to keep the output voltage at constant level. Okay theoretically, in practice there is always small transient noise and ripple voltage.
  • prev
  • next
Comment from geoffrey @ 2008/11/01
On demand, a picture which shows how the Velleman digital panel meter is configured, P3 = open, P1 = open, P2 = closed

Comment from thenextgeek @ 2008/11/05
Hi there, it was me that made that request, and THANKS for helping me out, i really appreciate it. .

I have tried to hook the vIN/vOUT (your brown&white) wires to have a reading,
and I am unclear as to the proper way to do this.

(as with a Ammeter hook up, one runs a +Positive lead to one ammeter and then back out the other ammeter lead - to your device, and that works)

I have not had luck reading a steady voltage, or one that i would count as being accurate when trying to hook this LEDPM/5 Voltmeter up the same way.

Am I trying this correctly? or do i got something mixed up.. *Thanks!

(maybe if you could just tell me where you have your brown & white leads coming from your LEDPM/5 (vIN & vOUT) to/from your device? power? or ground? or both? ? ?

Thank you for taking your time to show this. .

Cheers,
thenextgeek
Comment from geoffrey @ 2008/11/05
Don't forget you have to add a resistor on the circuit, you can easily spot what I'm talking about, it is on the right side of the LCD controller

Brown/white is used to read voltages on my picture. What you do is connect the two in between a potential difference. For instance you can try reading a 9V battery by connecting the VIN pin to one of the battery terminals and the VOUT pin to the other battery terminal.
You can also read the supply voltage source, you can connect brown to red and white to white and read the voltage source voltage. In my PSU I connected brown with +12V (ATX: yellow) and white with ground (ATX: black).
Comment from thenextgeek @ 2008/11/05
AhhhhhHA! I bet that's the problem, I don't have that resistor on there!

hehe, sorry so newb at this, the unit i received (locally) had no manual, (or at least one that I could read so it has been kind of a trial and error,

until now..

THANKS a lot for taking the time, I will put a resistor on as your picture indicates it should be...



Cheers!
thenextgeek
Comment from thenextgeek @ 2008/11/06
Did you just remove (desolder) the small black square part that you put that resistor to? (#RA?) I circled it in red with an arrow for referral

Attachment 2176


Sorry I don't know the name of what that is.. I have that Resistor, but it *looks* like you removed the original part then put that resistor on?




Thanks again for all your help.



Cheers,
thenextgeek
Comment from thenextgeek @ 2008/11/11
Hey, sorry if i'm cluttering your forum here, you can feel free to delete the posts if it looks bad, not a problem...


The place you put the resistor (RA / RB) as described in the manual. .

I have put the resistor on there as yours shows, but it *looks* like you may have also cut the other circuit resistor (RB? - the other one). .


Is that so? I dont want to cut those little things, as i know i would NEVER be able to put it back on, way to small.. hehe

Ok, well thanks if you can make any suggestions, i appreciate it !

I am asking this detail, because i am getting "unstable", and usually what would appear to be "inaccurate" Voltage readings, ie; even from a fully charged 9volt batt., a DC Wall adapter (universal), etc... it dosen't seem to read, usually about 1-2 volts different, although, I can use a voltage adjuster and see the changes, they don't seem accurate enough...

afaik, i also used the correct resistor, but i'm willing to double check.. but pretty sure.


THANKS SO MUCH for your site, and your suggestions...

Cheers,
thenextgeek
Comment from geoffrey @ 2008/11/12
Quote:
Originally Posted by thenextgeek View Post
Did you just remove (desolder) the small black square part that you put that resistor to? (#RA?) I circled it in red with an arrow for referral
Yes that one is removed.

Quote:
Originally Posted by thenextgeek View Post
I have put the resistor on there as yours shows, but it *looks* like you may have also cut the other circuit resistor (RB? - the other one). .
Yes there is one, label: 106. I don't know if that is the original one or not.
Comment from thenextgeek @ 2008/11/19
wow, ok, somehow. . lucky you.

I talked to velleman, via email, they sent a diagram showing that RB is the one to be removed, and RA should *not* be removed (the one you have your resistor on). .

and of course the resistor/s "should" be inserted into the wholes located top and bottom on there, but I often have done it differently as well.

Anyways, I am seeing how to fix mine now.. heh I need to put the RA resistor (the little black one) back on it... *i think*, unless there's an easier way... hopefully Radio Shack has this part or something... i dunno.

Any ideas?

it almost *seems* that your RB resistor WAS the resistor that USE to be on RA.. #[106], because RB is [D] (or [O]), and RA on mine was [106]

Jfyi?..
did you buy yours first hand/new? Mine was a new unit.

Thanks,
thenextgeek
Comment from geoffrey @ 2008/11/19
New one here too, yes I remember resoldering one of these SMD's, use tweezers, it's not that hard.

 

reply