• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Making high power fan controllers - guide

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Ven0m

Member
Joined
Sep 7, 2003
Location
Warsaw, Poland
== Making high-power fan controllers ==

--------------------------------------------
L337 M33P's controllers
--------------------------------------------


LAST UPDATE: 28.05.2005
- linked L337 M33P's thread

UPDATE: 23.01.2005
- added Speedfan-driven high-power controller
UPDATE: 19.01.2005
- some more data added
- added eye candies section
UPDATE: 15.01.2005
- added dual transistor stable circuit with calibration and kickstart

To be added:
- circuit for my uber-controller project (delayed due to major problems)
- any suggestions? (it would be good if someone who has experience with non-stabilized PWM circuits contected with me)

--------------------------------------------
Transistor & pot controllers
--------------------------------------------

Introduction:
Rheostats are usually not powerful enough to use them alone as fan controllers. Combining 12v, 5v and GND lines gives 0/5/7/12v options, what doesn't satisfy most computer enthusiasts. In this case, semiconductors may help.
High power bipolar transistor combined with rheostat gives many possibilities.
For further understanding: power = current*voltage, current = power/voltage.

--------------------------------------------

How does it work?:
Transistor has 3 lines - Collector, Base and Emitter. When electricity is applied to Collector, giving proper amount of electricity to base, we decide how much electricity will go from Collector to Emitter.
Bipolar transistors have many parameters:
- they can be PNP or NPN (we will consider NPN in this article, PNPs will be added later)
- Ptot - maximum power output - it should be greater than added power of all fans connected to this transistor
- UCB - maximum collector-base voltage - in this case 12V+ would be ok
- UCE - maximum collector-emitter voltage - also 12V+
- IC - maximum current - should be greater than added current of all fans connected to this transistor
- B(beta) - gain - I will call it "boosting factor" - multiplier - how much electricity goes from Collector to Emitter compared with electricity that goes to base.
- max frequency - max input signal frequency for transistor - not important for pot-based fan controller

Transistors, similar as any resistive part, produce heat. Small metal sheet should be enough for radiator. Notice, that metal on back of transistor has current equal to Emitter (it is Emitter in fact). So don't connect all transistors to 1 radiator without insulation (or use separate smaller radiators) and don't let electricity go to your case.

What if I have fans with 3-pin plug and I want to read RPM?:
You will need separate female 3-pin connector for each fan. Yellow cable should go to motherboard using additional connector.

--------------------------------------------

Transistors that you may be looking for

NPN:
BD135, BD137, BD139 - 12.5W *
BD433, BD435, BD437 - 36W
MJE521 - 40W
D44H8, D44H11 - 50W
2N3771, 2N3772 - 150W, metal case

PNP:
BD136, BD138, BD140 - 12.5W **
BD434, BD436, BD438 - 36W

* - tested
** - simulated

Thanks to Shroomer for info about 2N3771, 2N3772, BD433, BD434, BD435, BD436, BD437, BD438, D44H8, D44H11, MJE521 and more.

If you have found some datasheets for transistors that might be useful, PM me, so I add them to list. Also, if you tested some transistors, please make me know, plus if you have possibility, also send voltage table like above with transistor number, rated fan power, rheostat used and voltages at specific rheostat settings, the best if in 10% (or less) intervals, eventually min and max voltages.

--------------------------------------------

Design:
Draw how your controller should look like. Mark rheostats and connectors placement. Panels for 5,25" and 3,5" bays can be easily made from nickel coated metal sheet, wood or plexi (plexi bends nicely when heated to proper temp). As well, you can use modded computer case parts. Prepare holes for screws.

--------------------------------------------
--------------------------------------------
Where to buy:
--------------------------------------------


Any local electronics shop or e-shop:
Mouser Electronics, thanks larva
Futurlec, thanks archilochus
Notice that these parts are cheap - roughly $0.20 for transistor and $0.20-$2 for rheostat depending on its shape and look.

--------------------------------------------

If you need the most basic circuit, the first is for you, but if you need top quality - skip it and check the second one.

--------------------------------------------
--------------------------------------------
Simple 1 transistor controller (use oly if you lack confidency):
--------------------------------------------


Characteristics:
+ very cheap
+ very easy to make (only a few parts on each line)
+ high power
+ low voltage drop
- non-linear control
- requires choosing proper pot for specific fans power

--------------------------------------------

Shopping:
We make controller with n separate lines, m Molex outputs and f fan outputs (you can connect more than 1 fan/Molex output to line)

What should land in our shopping cart?:
1 x male Molex connector - for powering our fan controller
n x 10kOhm rheostat (we choose what we like)
n x bipolar transistor (I used BD137 and BD139 - rated 12W)
m x female Molex connector
f x male fan connector
(eventually additional fan connectors for RPM readout - read above)
some cable (shouldn't be too thin for high power)
all parts you have on your list from design phase
soldering equipment

--------------------------------------------

Ways of connecting:
= <fan "after" transistor>, basically it's:
12v - transistor - fan - 0v
+ there should be no problems with fan RPM readout
- transistor's metal part has voltage equal to given to fan - don't make it contact case

= <fan "before" transistor>, basically it's:
12v - fan - transistor - 0v
+ transistor's metal part has 0v - potentially safe to screw to case without iinsulation
- if case grounding is on 3rd plug pin, it's not recommended to screw to case without insulation
- there might be problems with fan RPM readout


--------------------------------------------

Connections NPN Circuit with fans "after" transistors:
circuit.gif

Transistor, rheostat, fan power, rheostat settings table for circuit above:
Code:
Transistor:	Rheo.:	Fan:	0%	20%	40%	60%	80%	100%
NPN		------	---	----	----	----	----	----	-----
A*		10kOhm	2W	5.3v	5.9v	6.6v	7.7v	9.1v	11.2v
A*		10kOhm	10W	1.7v	2.0v	2.5v	3.2v	4.8v	11.1v
A*		20kOhm	2W	3.5v	4.1v	4.8v	5.9v	7.7v	11.2v
A*		20kOhm	10W	0.9v	1.1v	1.4v	2.0v	3.2v	11.1v
PNP		------	---	----	----	----	----	----	-----
B*		10kOhm	2W	6.7v	7.1v	7.7v	8.4v	9.5v	11.2v
B*		10kOhm	10W	2.5v	2.8v	3.2v	3.9v	5.1v	11.0v
B*		20kOhm	2W	5.1v	5.6v	6.3v	7.1v	8.4v	11.2v
B*		20kOhm	10W	1.6v	1.9v	2.2v	2.8v	3.9v	11.1v

A* - BD135, BD137, BD139
B* - BD136, BD138, BD140
Data comes from simulation and might be inaccurate.

--------------------------------------------

Soldering time (NPN with fan "after" transistor):
Before you start - clean all parts with alcohol or anything that cleans and doesn't leave stains. I don't recommend soap.
If your rheostats are hard to mount, do it before soldering.
Solder:
- 12V rail (yellow) from male Molex you bought to transistors Collectors and rheostats.
- proper transistor Bases with proper rheostats (rheostat has 3 pins, use one from centre and one from side)
- ground rail (black, the best if one close to 12v) to ground rails from all connectors to which you will connect your fans
- proper transistor Emitters to proper connectors (12v rails)
(- eventually prepare extensions for RPM sensors)
- mount radiators on transistors
- make whole thing not to fall apart, it's ready to use.

--------------------------------------------
--------------------------------------------
Stable 2-transistor controller with calibration and kickstart (recommended):
--------------------------------------------


circuit2.gif
Characteristics:
+ very cheap
+ easy to make
+ high power
+ stable
+ low voltage drop
+ linear control
+ kickstart
+ minimum voltage calibration
- more parts than in previous circuit

--------------------------------------------

R1 is primary pot - it's what you will use
R2 will be used for calibration and it can be a cheapo PCB-use pot - set R1 to minimum, and decrease R2 untill your fan rotates at minimum speed you want. This way you can't set fan speed below minimum you used earlier - this way of controlling speed is good especially when you use different fans with different characteristics and purposes.
Capacitor C1 gives kickstart for around 100msec for 10W fan. Using bigger capacitor will increase kickstart time. Don't use too big ones, as they will make speed adjusting harder (response time to speed change will be similiar to kickstart time). Removing that line with C1 will result in no kickstart (not recommended), however voltages will remain very stable.
Q1 and Q2 are bipolar transistors. Q1 is NPN and Q2 PNP. Using BD135 / BD137 / BD139 as Q1 and BD136 / BD138 / BD140 as Q2 allows theoretical maximum power of 12.5W, but it's good to leave 10% margin. They can be replaced with other transistors. For example you can use as Q2 - BD434 / BD436 / BD438. This will give theoretical maximum of 36W on each line. Using proper transistors, you can get even more power. And Shroomer said "As long as Q2's gain is over 25 and Q1's gain is over 80, then you can use a 200ma transistor for Q1" (big thanks for tests and reports). As result, you can use cheaper transistor for Q1 and there's no need to use heatsink with it.

To have more lines, use this scheme multiple times. There can be many fans connected to each lines in parallel. In this case total fans power for each line should not exceed its maximum power output. In this controller, fan voltage change is linear with R1 change. Tranasistor Q2 should have radiator attached and they should be insulated or should not touch case, as there is voltage on metal parts of transistors. You may use thermal paste between transistors and radiators, but this is not a must. When you plan to draw lots of power through controller, it's good to have a little more complex radiators than simple small metal sheets + a bit of airflow. Most of heat is produced on Q2, while load on Q1 is close to none, and with fan at full speed maxes still remaining low. Pots and capacitor should remain cool.

--------------------------------------------
--------------------------------------------
Speedfan controllert (experimental, not tested):
--------------------------------------------


circuit3.gif
Characteristics:
+ high power
+ PWM driven - may be stabilized or not
+ uses Speedfan to adjust fan speed, so it inherits all Speedfan properties
+ very cheap
+ very simple
- not tested
- needs Speedfan running to adjust power (no speefan - max power)
- some motherboards don't support Speedfan
- maximum of around 11-11.3v depending on fan

So what's the deal?
Speedfan allows adjusting of speed of fans connected to mobo. Some fans need too much power, to be connected, so there's a solution. Mobo uses NPN transistor (here BD139, but can be other) to deliver power from molex to fan.
As you see, there are more parts than seem to be needed, but this soultion will work for both 3-pin connector powered fans, as well as with molex-powered ones. If you don't need some lines or connectors, don't use them.
C1 - this capacitor is for one purpose - stabilization. Some fans produce pretty irritating noise, when powered from simple PWM controller, this capacitor will remove this effect. But if your fan doesn't have this property, it might be better not to use C1 (and lines conencted with it) - you will be able to lower fan power even more without stopping it. So C1 in fact depends on your fan. Eventually you can add some switch in line with this capacitor, to have possibility to change way this controller works.

This controller potentially allows you to power any fan or array of fans, having them controlled from mobo.
I have not tested it and take no resposnibility for using it. I don't write it to scare you, but rest of circuits were tested and proven to work. This one should work too, but I cannot guarantee that it will in proper way.

--------------------------------------------
--------------------------------------------
Eye candies:
--------------------------------------------


If not defined, visual addons should run in parallel to fan. If you want to use multiple visual sets, by analogy, connect them in parallel.

LEDs:
You may add some LEDs. Most of blue leds run around 3v. Using proper resistors, you could as well make these LEDs indicate voltage on fans. You can use 4 ~3v LEDs in series as well (recommended), with tiny resistor if total voltage drop from leds exceeds 11.3v. White leds will act in similiar way, but remember that leds in different colors have different voltage drop. You can surround pot knob with acrylic and illuminate with these LEDs.

Bargraph display:
These look definietly nice. Usually they consist of set of green, orange and red flat leds. There are decoded and undecoded bargraph displays - undecoded needs some more wiring, but decoded is simply plug & play. Get one operating around 11v and have fun - when connected in parallel with fan will indicate its voltage pretty accurately + it will look cool.

Analog voltage display:
If you for some reason don't like leds, and still want indicator, you can get analog voltage probe operating at 12v (or a little more). These are usually more expensive than leds, but add unique stylish look with needle indicating voltage accurately. You can add some lighting to it, for example with blue or white LEDs.

A digital circuit with led blinking rate dependant on voltage:
So far I haven't got circuit for it, but I think it can be done pretty easily at low cost. If there's demand, I'll try to add it.

Any ideas?
Do you want ot have some other indicator or other controller-related eye candie? Let me know, and I'll try to design it and post here, if only it's going to look cool.

--------------------------------------------
--------------------------------------------

Cheapo ghetto controller in 8 steps - photos
Some of my modding pics, that you may find useful.

--------------------------------------------
--------------------------------------------
A great guide by L337 M33P, aka Jonathan Bell about making PWM controllers.
--------------------------------------------
--------------------------------------------
Credits (alphabetically):
L337 M33P - controllers stuff
larva - for all his support and advices
Shroomer - for all his help, advices and testing

--------------------------------------------
--------------------------------------------
If you found some bugs in text or you have some nice ideas on what could be added or changed - send me PM, as it would be better to keep this thread clean. And if you know some good electronics e-shops - I'd add them to list.
 
Last edited:
A big thanks to Venom for authoring this guide. Basic electronic circuits are rarely explained on the level we fumblers need. Excellent work and a true benefit to this community and its members.
 
I dont know if u actually built one or not, but do you happen to have a picture of one that you/someone else custom built.
 
I have some ghetto-style controllers, but they're all in some enclosures, so you could hardly see anything but small rheostat and aluminium radiator. I was going to make one soon, with 4 lines and both Molex and 3-pin fan output, bought parts already. If I get camera, there will be some photo documentation - step-by-step. Till then, perhaps I'll find photos of my ghetto controller, when I was making it.
 
If someone is really bored out there, maybe compiling a compatible transistor list would be a good idea. It'd be good to see some that can handle 30-40 watts with cooling. I was trying earlier to find some and had no luck. Perhaps even a compilation of stores to get parts from would be good. Also, a pictoral representation of a PNP circuit would be great. It is my understanding that the heatsinks would not have to be isolated in a PNP circuit, but I don't really remember.
 
There's only one store you need if you are in shipping range of these guys:

http://www.mouser.com/

As far as a list of suitable transistors, Venom is the expert here. How bout it Venom?
 
So far I suggested BD137 and BD139 as I used them and they work. They have nice case and can power most of fans, and when used 2, pararelly any fan I've seen including monster 120mm Deltas.

I'll try to make some kind of "recommended transistors" table with given max power rating, but so far for fan controllers I used only these 2. I haven't tried PNP tansistors so far - I'll get some and check them, cause how-tos shouldn't be based only on theory.

Super-high power controllers:
They can be made in similiar way to common high-power ones, but along with more powerful transistors, you'd need to use transistors with higher boost or more powerful rheostats (around 1W), not to fry them.
 
I didn't mean to offend anyone here... You've made a great how-to, I just think that there is more info to be had. Is Hoot still around the forums? I think he was the one who showed me this circuit. Should work the same, right?

Edit:

It looks like the BD136 is pretty much a PNP version of the BD137. I might give this an order and see what happens. Link

Also... Pay more attention to Ven0m's posts because he has a lot more experience than I do at this. I made 1 speed controller once and it didn't work very well. Probably because of this "beta" value that I don't understand.
 
Last edited:
I haven't felt offended in any way. I just realize that this guide is done in max 20% and I'm grateful for your replies.

About analogy between these PNP and NPN transistors - you are right, with a small "difference" BD136 has params similiar to BD135, 138 to 137, 140 to 139. Differences are type - NPN and PNP, and maximum supported frequency - 75mhz for PNP and 250 for NPN, but that's not important.

BD 135 and 136 have Beta factor of 40-250 and 137 to 140 have 40-160, so I suggested them because Beta is more "stable".

Big thanks for PNP circuit. I'll get some PNPs and try. I'm only not assured if it's ok to use case as normal grounding. If it works, some mondern electricial installations might be "not too happy" with it and do some electricity shutdown. It's only my guess and would depend on way of grounding case.
Plus PNP circuit seems to require less attention than NPN due to operating on ground instead of + rail. I'll check behavoiurs of both PNP and NPN controllers mounted both "before" and "after" fan.

Scheme looks ok, but I'd like not to post this solution as official untill I (or someone else) test it. I had electronics classes like 2 years ago and I'm not 100% sure of circuits, so first test - then guide update. I'd just want to do everything not to make this guide mislead someone and result in some problems.
 
Yeah, we've already got a great start on the most definitive guide on the subject. Great work so far Venom, and thanks for everyone else's input. I can't wait to see the expanded versions as time allows.
 
Ven0m said:
Big thanks for PNP circuit. I'll get some PNPs and try. I'm only not assured if it's ok to use case as normal grounding.
But it would allow us to mount the transistors on a common heatsink at the least. I like this approach better, as we can use something beefy for it and as powerful as transistors as required for more extreme loads.
 
But it would allow us to mount the transistors on a common heatsink at the least. I like this approach better, as we can use something beefy for it and as powerful as transistors as required for more extreme loads.

That was my thought exactly. I have an old aluminum thermaltake heatsink laying around... I could get some kind of perferated plastic to make a box for it and put it in there.
 
Almost everyone lives near a Radio Shack. Most stock the TIP42C PNP transistor for about $1.60. It would be a good candidate for the chasis mounted approach. There are better deals on them mail order, but unless you're ordering a lot, the shipping will cost you about as much as buying 2 or 3 from the local Radio Shack.

Hoot
 
Hoot, what size rheo or pot do you use with that transistor and does it use the whole range of the arc to change the speed?

EG... Would we just used the 10k rheo that Ven0m has suggested.
 
Last edited:
I got some proggie for circuit simulations, as there are no electronics shop near me.

Today there will be PNP circuit (I hope) in guide + transistor/rheostat table.

About grounding PNP transistors - according to new formfactors, at least in Europe, electric installation in home needs outflow protection. It checks how much electricity goes in and out and if they're not equal, it performs shutdown. It increases safety a lot, but if case grounding is connected to 3rd pin in plug, a few watts going to it will trigger fuses. So after checking it and some simulations, there will be guide update, planned on today.
 
Ven0m said:
I got some proggie for circuit simulations, as there are no electronics shop near me.

Today there will be PNP circuit (I hope) in guide + transistor/rheostat table.

About grounding PNP transistors - according to new formfactors, at least in Europe, electric installation in home needs outflow protection. It checks how much electricity goes in and out and if they're not equal, it performs shutdown. It increases safety a lot, but if case grounding is connected to 3rd pin in plug, a few watts going to it will trigger fuses. So after checking it and some simulations, there will be guide update, planned on today.

AKA GFI (Ground Fault Interruptor)

As far as an initial potentiometer value, try 10k linear taper.

Hoot
 
Back