Making high power fan controllers - guide

[AlabamaCajun]

Was experimenting with my own variable fan speed controller about 2 years ago, using a Basic Stamp microcontroller as the heart of this system, I hooked up several radioshack thermistors (one for cpu, hd, case temp) I can calculate what the temp of each thermistor by running 5v through it and then polling to see how long it took to charge a cap of know value.

After determining the temp at various locations, I output PWM to a power transistor which controls the fan speed of the various fans, and also reports system temps via a serial LCD module. Works fairly well, I don't have any circuit diagrams as I kludged this together. This system is able to control the fan speeds depending on the actual temp at the thermistor in real time, no twisting pot's, it does all the work.

For 60 to 100 bucks, the stamp make an excellant controller. Have you though of connecting it to serial port/usb or adding buttons to change the thresholds?
I was thinking you could use an up/down control to select from quiet & warm to louder & cooler. If would'nt take to much changes to the program if the threasholds are stored in EEprom and loaded to register at boot & reset.

 

[JTanczos]

Can you guys post a part number or link to a capacitor? I have been looking for one to smooth out the PWM lines and I dont know which to get. 22uF-Pol doesnt seem to be a part number.

JT

 

[Ven0m]

Can you guys post a part number or link to a capacitor? I have been looking for one to smooth out the PWM lines and I dont know which to get. 22uF-Pol doesnt seem to be a part number.

JT

As far as I know, capacitors have only given capacity, maximum voltage, if they're electrolitical (polarisation is given then) or not and capacity tolerance.

 

[Shroomer]

Just noting that the speedfan amplifier circuit might be good for automatically controlling a DC pump in a watercooling system.

 

[Ven0m]

Just noting that the speedfan amplifier circuit might be good for automatically controlling a DC pump in a watercooling system.

You have really great ideas

 

[Amarkarian]

Im confused oh well

 

[Stalin223]

Nice work on the second circuit there mate. I'm going to one-up you on that and use a NTC thermistor and POT in place of R1. This will allow me to run each fan based on the temperature of the component that it will be cooling. The thermistor should have a low enough resistance at the max safe temperature of your probed item, that it will turn the fans all the way up to ~11v. A smaller trim POT in series with the thermistor would allow you to adjust the speed:temperature relationship, allowing you to make very accurate changes to the fanspeed:temp ratio.

This will be used for the GPU, Northbridge, and CPU fans, with the regular circuit for the remaining case fans.

If I can find my circuit simulating program, I'll draw up what it would look like. $2 per channel in added parts for a controller solution you can simply leave inside the case is pretty good. Say it with me, Autoregulating Temperatures.


Edit, sortof:
The thermistors are normally 10KOhm resistors, and at 60C, they drop to about 13-25% of thier rated value at 25C. This means an OP AMP may be required to boost the current values to useful levels to drive the power transistors.

I must sleep, work in 7 hours. Night all.

 

[Ven0m]

Thanks. And if you make some cool DIY controller, it would be great if you shared circuit and results, so more could benefit from it.

Your idea sounds pretty cool. I anticipate moment when you make it and post results.

 

[Stalin223]

How do I add transistor details in Circuit Maker? And how do I read the numbers off the catalog items to understand what the gain is?

 

[pejcao]

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

Great post! I'm planing on doing 2 lines x 36W, now:

Q1: BD437 + BD438 (each line) will do the job? (I dont really care 'bout cheaper combinations, I Just want to go the safe way)

Q2: Both PNPs (BD438 one from each line) can be together on the same heatsink? or should I use 2 heatsinks (one 4 each)?

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)...

Q3: What 'bout ~20Ws ? Should I use a ~36V 9.4 microF (each line)?

Thxs on advance!

Edit:

Just found inside my broken Epson S-740 these babies! (heatsink included)

2SA1469 PNP/60V/5A
2SC3746 NPN/80V/5A

those r gona be my test bed

 

[phantom punisher]

ive been using some quike and dirty thermister controlled inline ones i made two years ago. all the stuff came from radio shack but i dont have any of the part numbers.

they are for loud fans like my old 80mm tornado or my 120x38 sunon. good thing about them is they work great at the ranges those typs of fans need and are small enough to live wrapped in electrical tape in line with the molex connectors. there adjustable and when the thermister is placed behind the cpu, they are spot on for controling temps.
pm me if anyones interested and ill take one apart and get the info for you

 

[pejcao]

OKA!!!
Made two HP-FC's using 2 BD434 & 2 BD437 (could'nt found matching pairs) and they work extremely WELL!!!

With a 10.8W fan ,heatsinks on the 434 can be pasivelly cooled, but using 2 of those fans (21.9W) they MUST BE active cooled! (even driving the "min" voltage to the "max" {~7 V} )

Not using the capacitor, tho', I can't hear the fans sticking my ear to them around two volts (6V) over the minimun operative voltage (so I guessI dont need'em).

My 1K (primary) pots have 6 pins (3 for the "pot" function, and 3 more as a 2_by_1 selector), so I use the pot pins as the diagram, and 2 pins from the selector function to switch "off" the ground line, so my circuit has an "off" function just after the "min" (counterclock)

THX a LOT ven0m!!!

PS: use of 2SA1469 & 2SC3746 from my epson stylus 740 pending, I'll let u guys know how do they behave.

 

[AlucardCasull]

Ok, I want to make the second fan controler for running 2 or 4 fans rated at 29.40 watts, 2.45 amps each. I started to read the little write up and I was lost instantly. Any way to barney that up a bit for me? From what I understand, if I were to run all four fans off of the single controler, all I would need to do is replace the Q1, Q2, BD139 and BD140 with the 2N3771, 2N3772 transistors you mentioned? Or will it be much more complicated than that. Again, I want to run all 4 (or 2) fans off of one knob. Thanks for any help.

 

[Ven0m]

They don't have to be these exactly. You just need high power NPN transistor for Q1 place and PNP for Q2. 2N3771 and 2N3772 are both NPN - you'd need both NPN and PNP. Eventually, if you couldn't find transitors powerful enough, you may connect them pararelly - I mean all NPN grouped / all PNP grouped. This way such group should have maximum power tolerance as all transistors in it added. If you decide for grouping, whole group should consist of transistors of 1 kind, for example 2x BD433 and 2x BD434. Notice that with power of fans you want to use, these transistors will need some cooling - they should be connected to some metal sheet / heatsink, perhaps with fan if there's not enough airflow around them.

For high power devices, PWM method is rather used, due to less power loss on controller. Voltage characteristic isn't that nice as with linear controller (scheme 2), but they ca handle more power and will produce less heat.

 

[AlucardCasull]

So what your saying is i would need a 120w NPN and 120 PNP in place of the BD139 and BD140 respectivley? That way i would be able to control 4 30w fans from one controler knob? Thank you!

 

[MassiveOverkill]

Help: I've got a bunch of regulators, I don't even know if I can use them: Here are links to specs:

http://www.onsemi.com.cn/pub/Collateral/CS5207-1-D.PDF

http://www.singsun.com.cn/databook/alpha/AS2850.pdf

http://www.fairchildsemi.com/ds/KA/KA7912.pdf (don't think this will work......works off negative voltages)

http://www.fairchildsemiconductor.com/ds/FD/FDP7030BL.pdf

Thanks for any help.

 

[Ven0m]

That looks realy interesting and promising (especially 1st link), but unfortunatly I have no experience with such devices.

Could someone test them? They might bepretty good looking at properties and they'd allow stable voltage with very simple circuit.

 

[MassiveOverkill]

I don't know which legs are which.......any help with that?

 

[The Big One]

how can i make a 12/var/5 controller ?

 

[drunkn.bear]

Very detailed post. Very thorough, but not hard to follow