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

Fan Controller / PWM Booster / Regulator / Converter / Etc. Thread!

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

Bobnova

Senior Member
Joined
May 10, 2009
Please note: I am no longer producing these controllers.
I'll answer questions (slowly), and might do a one-off commission type thing if the money is there, but it'd have to be a decent pile of money.


I think it's time for me to stop nuking the DIY PWM Fan Controller Thread and make my own for my projects and such.

This thread is the place to discuss the following, plus anything you'd like to see me build, are interested in building, anything you'd like me to add to my stuff, etc. Also fielded will be questions on PCB design and so on. Also included will be links to purchase the various bits from the classies.
 
Last edited:
Custom fan controller

rev1-1-top.jpg
rev1-1-bottom.jpg

Provides power and control for PWM fans up to ~5 amps.
Has a jumper to invert the PWM output.
Has a jumper to select between low (~1333MHz) and high (~25kHz) PWM frequencies. Custom frequencies between ~100Hz and ~100kHz are available!

Current sale thread: http://www.overclockers.com/forums/showthread.php?t=746053

Please note that I will NOT accept orders via this thread. If you want to buy one you MUST post in my current classies thread. If you can't post there, you can't buy one.

If you have thoughts on this fan controller, questions, suggestions, or anything elses, post in this thread! I want to know!
 
Active PWM Booster / Splitter

rev1-top.jpg
rev1-bottom.jpg

This board takes a PWM signal (from a motherboard, fan controller, GPU, anything really) and amplifies it, then splits it to ten powered PWM fan ports on the board. Power comes via Molex, so the board is limited to ~10 amps of fans. Individual fan headers tend to fry between 3a and 5a (the fan side is what fries), so keep that in mind.

Pictured is revision one, coming soon is rev 1.1! 1.1 will have mounting holes, as well as two short MOSFETs rather than one short and one tall. Other than that it'll be roughly the same.



Current sale thread: http://www.overclockers.com/forums/s...d.php?t=746053

Please note that I will NOT accept orders via this thread. If you want to buy one you MUST post in my current classies thread. If you can't post there, you can't buy one.
 
Last edited:
Voltage to PWM Converter, Stackable

stack1.jpg
stack2.jpg
stack3.jpg

Ignore the purple-brown NCC cap hanging off the side, that was part of testing and will NOT be on the release version.
Also please note that unlike the previous posts, this post is about a prototype product, it is not finished and hence does not look finished.


Takes a voltage fan control signal (control for 3 pin fans, or for obscure server fans) and turns it into an amplified PWM control signal.
Self calibrates for the input voltages, so you get 0% to 100% even if your controller caps at 5v and 11v. The data is preserved through power cycles! No re-calibration until you want it. There is a clear_calibration jumper to reset that calibration data.

Also has a low/high speed frequency (~1333Hz and 25kHz respectively) select jumper and an Inverter PWM output jumper.

These are stackable by using 4-40 bolts/nuts, so you can have a few channels converted/boosted/powered. 12V is provided via one of the end bolts, ground via the others. Ring connectors (crimp or solder) with a Molex cable on it is the suggested method of providing power.

For stacking you need to use right-angle fan headers and jumper headers. If you don't want to stack, feel free to use vertical headers.



These are not for sale yet, the prototypes work though.


More information here: Voltage-PWM, tested!
 
Last edited:
just a thought about the pwm booster for the power to the fans. lose the 4pin molex connector switch it to a 6pin pcie connector, 3 yellow, 3 black, then split it off to three 4pins. that would give you a theoretical max of 33amps if the pins can handle 11amps each. even going up to an 8 pin would allow more scaling, since you said its possible for up to 200 fans. though all based on power required. other wise, you would have to at the fan side split the pwm and power wires. if you were exceeding the power able to be handled by the pwm booster with that many fans.

also what bout angled mounts for the 4pin pwm headers to allow for a stackable version. you have the main board with the pwm input then a header on the other boards to allow the connection between them for a stackable version. ie, the 4pin(black one on the voltage to pwm converter for the fan) header connection not the 4pin pwm header(standard connection used on motherboards for pwm fans).
 
The PCIe plugs are rated for 7 amps/pin, but that assumes a perfect connection. As far as PCIe SIG is concerned the connectors are rated for 150w (12.5a) if you have three power and three ground.
The difficulty is that I need a 5V source to make the circuit work. I could add a regulator of course, but I'd like to keep the parts bill low.

On the angled headers, are you thinking put them on the edges like the stackable so they stick out? That's certainly doable. There'd be fewer ports, but it'd be quite stackable if combined with an angled Molex plug.
 
Please note that I will NOT accept orders via this thread. If you want to buy one you MUST post in my current classies thread. If you can't post there, you can't buy one.

But but but :rofl: I'll post pics of my board soon as I get them.
 
if my guess is right, yea you would lose some output ports. im thinking you would lose two, still giving you 8 outputs. not sure if this would work but follow me, could you use the smaller 4pin from say like a floppy power connector to the board angled. then still use a angled 4pin molex but instead of 5v on the other wire do 12v. you would then have a power just for the fans alone on the molex. with one 12v/gnd going to 4 output ports and the same for the others. this would then, not to max out the pins for current, give you roughly a safe 2.5amps per fan on 4 output ports. gives a 1amp safefy margin since you said it maxes out at 11amp/pin. then each "passive" add-on just needs the signal to be passed through and then have its own dedicated 4pin molex for power to the fans.

for some reason i thought the pins used in the pciE power connector were rated higher then that.
 
The PCIe plugs are rated for 7 amps/pin, but that assumes a perfect connection. As far as PCIe SIG is concerned the connectors are rated for 150w (12.5a) if you have three power and three ground.
The difficulty is that I need a 5V source to make the circuit work. I could add a regulator of course, but I'd like to keep the parts bill low.

On the angled headers, are you thinking put them on the edges like the stackable so they stick out? That's certainly doable. There'd be fewer ports, but it'd be quite stackable if combined with an angled Molex plug.

Use right angle headers.
 
The spec on PCIe connectors is a complex beast.

Right angle headers would be a must for what Evil's describing, definitely. Now finding 90° angled PWM headers with the proper tab, that's proven to be difficult. I haven't found any yet.

I don't like the idea of rewiring a Molex to have two 12V pins, while it's something I certainly could do, if someone plugged a normal Molex into it that would be the end of their PSU and possibly the end of any device attached to the 5V rail as well. Messy.
PCIe+floppy power would work.

At some point having a separate power delivery setup for the fans becomes the best idea, that's what BruceUSA is doing. These will just be providing PWM to his fans, 12V will come from elsewhere.
Stackable for even more fans is doable too.
 
well i my idea for the molex would be, that a custom power single 12v/gnd/gnd/12v, gnd/12v/gnd/12v, gnd/12v/12v/gnd in, to two 4pin wired to grab only the 12v line. the custom cable wouldnt need to be very long, maybe 3in if that.
 
For that sort of thing it might be better for me to leave the Molex off the board, and just solder the converter cable to the board. Then you can draw off two Molex plugs without having to have a suicide plug on the PCB itself.
That could be done with the current boards actually, plug one Molex in, and then solder another on via wires.
 
would the traces be able to handle that kind of current coming in. assuming your pulling 10amps per, being 20ampsX12v=240watts total. might you need to go thicker on the traces for the input to the split out each 4pin header.

For that sort of thing it might be better for me to leave the Molex off the board, and just solder the converter cable to the board. Then you can draw off two Molex plugs without having to have a suicide plug on the PCB itself.
That could be done with the current boards actually, plug one Molex in, and then solder another on via wires.

would you not need another 12v/gnd input point on the pcb? asking mainly due to what i brought up above about the traces and power at the input side before the break out to the 4pin headers.


*edit*
im not sure or looked up what the pin spacing is for the pwm fan connectors, but i found these on ebay, they look to be what you need.
http://www.ebay.com/itm/20pc-2510-4...n-Socket-/111318388432?_trksid=p2054897.l4275
http://www.ebay.com/itm/100-x-Right...-Housing-/181111472719?_trksid=p2054897.l4275
http://www.ebay.com/itm/10x-KF2510-...ht-Angle-/330957445568?_trksid=p2054897.l4275
http://www.ebay.com/itm/20pc-2510-4...n-Socket-/111318388432?_trksid=p2054897.l4275
 
Last edited:
I used copper planes on the PCB for 12v/gnd, 12V is the entire top copper slap, GND is the entire bottom (or vice versa, I don't remember). Eventually it'll fry, but the Molex will go blam first I expect. The trace width calculator I usually use says it could do it with a ~12°c temp rise in the PCB. Not bad.

For input points you can always just scratch the solder mask off and solder straight to the planes too.
 
Bonus post!
The linked connectors are all I've managed to find too. Their tabs are too wide, the proper headers have a thinner tab that is offset to one side.

Unrelated: Is there anything that anybody would like me to make? I'm thinking that a LED controller of some sort might be fun.
 
Does the PWM buffer have a pullup on the input? Some motherboards look for that to detect if a fan uses the PWM signal or not.

Another useful gadget might be a buck converter driven from a PWM signal for controlling regular fans.

LED controller might be interesting, but there are already many projects of that sort on the web. Basically just an Arduino (or other microcontroller) and a bunch of driver transistors.
 
Voltage to PWM Converter, Stackable

View attachment 143505
View attachment 143506
View attachment 143507

Ignore the purple-brown NCC cap hanging off the side, that was part of testing and will NOT be on the release version.
Also please note that unlike the previous posts, this post is about a prototype product, it is not finished and hence does not look finished.


Takes a voltage fan control signal (control for 3 pin fans, or for obscure server fans) and turns it into an amplified PWM control signal.
Self calibrates for the input voltages, so you get 0% to 100% even if your controller caps at 5v and 11v. The data is preserved through power cycles! No re-calibration until you want it. There is a clear_calibration jumper to reset that calibration data.

Also has a low/high speed frequency (~1333Hz and 25kHz respectively) select jumper and an Inverter PWM output jumper.

These are stackable by using 4-40 bolts/nuts, so you can have a few channels converted/boosted/powered. 12V is provided via one of the end bolts, ground via the others. Ring connectors (crimp or solder) with a Molex cable on it is the suggested method of providing power.

For stacking you need to use right-angle fan headers and jumper headers. If you don't want to stack, feel free to use vertical headers.



These are not for sale yet, the prototypes work, but need more testing.

Are you sure it's a good idea to have your email on the boards? :)
 
Does the PWM buffer have a pullup on the input? Some motherboards look for that to detect if a fan uses the PWM signal or not.

Another useful gadget might be a buck converter driven from a PWM signal for controlling regular fans.

LED controller might be interesting, but there are already many projects of that sort on the web. Basically just an Arduino (or other microcontroller) and a bunch of driver transistors.

Yeah, the motherboard does not have a pullup on it, so without one on the board nothing would happen. It shouldn't, anyway. The Intel spec calls for a pullup in the fan, the motherboard uses an NPN or N-Channel to drive it low to send the signal.

Buck controller project is in the works, I had I prototype, but blew it up. Need to replace the dead bits and play with it more.

LED controller is pretty generic, I haven't really looked into it. I did convince an Atmega328P to speak USB directly, which opens some doors that a normal Arduino doesn't have open.
Definitely a busy market though.


Email is one that I only use for PCB stuff, it was created for just that purpose :D
 
just a thought but with the pwm controller or booster. could you not make a led indicator to show fan speed in a way. say like with 6-8 leds, green being low yellow for mid and red for high, as in showing fan speed. since i dont fully know how the pwm works, could the micro controller look at the pwm signal to indicate which leds to light up, like showing sound level on a mixing board/eq. could have a segment for each fan, that way it could give an indication if the fan is working or not. the micro controller would need to see some kind of feed back from the fan as which segment to light up(ie leds for fan1 or fan 2 and so on. is what i mean by segment)
 
The booster is analog, there's no brain there. Now I could add a brain and some LEDs to indicate what's going on, that's definitely doable.
The controller is smart, but short on output pins (like, it has one spare after the two jumpers). There's a "OneWire" bus I could use to talk to a second controller though.
Alternatively I could make a completely separate board that could be wired into any PWM setup/source, and run some LEDs.

Fan status (RPM / no RPM) would be easy to do with just passive parts. Very easy, actually.
If the fan failed in exactly the right way it would confuse a passive system. An intelligent system would need two microcontroller input pins per fan (one input, one output). Not undoable by any means, but expensive in PCB space. It would, however, be more resistance to being fooled by a cleverly dead fan.
 
Back