how can I make my own fanamp?

S

silverphoenix

Gawd
Joined
Feb 15, 2006
Messages
780
I saw the mcubed fanamp, and the problem is that it doesn't supply enough power to do what I want to do and the only US retailer that sells them is OOS. So anyone who's an electronics whizz please help me out on circuit design of one? I don't need manual standby controls like the mcube it just needs to have the output voltage = the input voltage, but the output just needs more amperage.
 
I'm a mechie and have no idea where to start. So far I'm reading up on http://en.wikipedia.org/wiki/LM317

The LM317

the description states
Variable output voltage is achieved by using a potentiometer or a variable voltage from another source to apply a control voltage to the control terminal.
Click to expand...

which sounds like what I want to do, but now I need to know how to precisely regulate the control voltage to achieve the same output. Basically I need to know how to make it so that whatever the control voltage, output voltage = control voltage. The motherboard will provide a certain voltage probably ranging from 6-12V, but fed directly into the control lead the output from the LM317 probably won't equal that. From my common sense and logic I think you'd need a transformer, or something to provide some sort of conversion factor that will give the necessary control voltage. Oh god I hope I'm making sense to someone. And anyone who's an expert please feel free to chime in anytime and correct me if I'm babbling about the wrong thing.
 
I did a quick Google search for 'mcubed fanamp' and it just seems to be... a fancy fan controller? It just has potentiometers that control output voltage...

it just needs to have the output voltage = the input voltage, but the output just needs more amperage.
Click to expand...

This simply sounds like a unity-gain voltage follower with high output current. You can achieve this using an op-amp in a unity-gain configuration. What output current spec do you need?
 
the fanamp can power up to 25watts, I was hoping to do closer to 40watts. so about 6-7amps total. If it's not feasible I don't mind just 25-30watts.
 
So I had to make sure I knew what the mcube was first. At first glance it looks like a PWM fan controller. The simplest way to do this is a 555 timer driving a power FET. If you really want something like the LM317 to just follow an input voltage it's a bit easier. It wants to control the output pin such that it is always 1.2 volts above the adj pin. Connect the +12VDC supply to Vin, the fans to output pin, a fairly high resistance (>1K) from Vout to Vadj. Drive the adj pin with the control voltage through two diodes. This should give approximately the control voltage on the output of the LM317. You may need multiple LM317's and heat sinks. You can tie the adj pins together.
 
Does this look about right? I'm assuming the diodes are so there is no feedback to the mobo header, and the high resistor is to prevent feedback down that line as well.

fanampnc5.jpg
 
My first thought would be this, but I have no clue what dark magic this "mcubed" doodad is supposed to be unleashing. Also I'm a bit sleepy.

05106nw8.png


So, yeah, why not just slap a standard variable voltage divider on it and be done? If anything, make a quick modification so that it presents a very high input impedance and call it a day.
 
silverphoenix said:
Does this look about right? I'm assuming the diodes are so there is no feedback to the mobo header, and the high resistor is to prevent feedback down that line as well.
Click to expand...

The anode of the lowest diode would not connect to ground as shown. It would connect to your controlling voltage. Put the 2k resistor from the adj pin ground. Also know the output voltage won't follow all the up to the input voltage. This is referred to as drop out voltage for linear reg's. Get one with the lowest LDO that will still do the job. You also show the mobo header connecting to the same point. Is this the speed control of a 4-pin fan header? If so then this is likely PWM controlled and this is not the circuit you want, and you want PWM fans. You might be fine connecting the fan speed control pin of each fan to the same pin, but you may need a buffer for each fan. The speed sense would be from one fan only and the others would just follow at roughly the same speed.
 
The fan header on the mobo is 4pin but can run in analog 3 pin mode if only 3 pins are connected. I'm confused on the placement of the components, and I have no idea how to build a buffer, I'm sure building this would only cost me about $20 at most but from how confused I am I'm thinking it's worth it for me to pay $80-$90 to ship the mcube over here and hope customs doesn't impound it...

I never thought something that seems simple could be so complicated... well to me anyways. I only got my picture from the actual LM317 PDF and relabled the Vin and Vadj and added 2 diodes. I have no idea how the LM317 actually works so I have no idea how hooking it up a certain way provides whatever outcome. I really only know how these components work : Battery, resistor, capacitor, leds, diodes that's about it. I know transistors are electronic switches but beyond that I don't know how they work or any of the actual math involved with them. Thanks for trying to help mtrupi your the only person who's come close to explaining on how to build one of these things. Seems like mcube has some huge industry secret cuz from other forums no one seems to have a clue on how to build something similar, and most have no idea of the concept I'm proposing, no offense ashmedai but I have a spare baybus, but I'm trying to utilize the thermal controls built into the motherboard so I can set a certain temp that the fans can kick into a higher rpm at.

if it helps at all here is a picture of the mcubed fanamp

fanamp1bf1.jpg
 
silverphoenix said:
The fan header on the mobo is 4pin but can run in analog 3 pin mode if only 3 pins are connected.
Click to expand...

Well, yeah. That's because one of the pins is at Vcc+ and another is at GND. After that, it's just electrons going from one to the other through the fan. Nothing fancy.

silverphoenix said:
and I have no idea how to build a buffer
Click to expand...

If you're talking about what I think you're talking about:

280px-Opampvoltagefollower.svg.png
 
ashmedai said:
280px-Opampvoltagefollower.svg.png
Click to expand...

yeah to me that is just a diagram of a triangle and lines and means nothing to me in terms of components. Sorry for being such a noob at this...

I don't know what that is, or where I'm supposed to put it or what specs or equation I should be using to figure that out
 
It's an op-amp.

A follower/buffer is a general purpose op-amp circuit that makes the input impedance look really high and the output impedance look really low. This makes it so you can easily connect two circuits while regarding them as black boxes, i.e. you don't have to redesign both circuits with the connection in mind. Thus the name.

It's a really simple yet really useful trick that...well...this little baby is *everywhere*.


Horowitz's Art of Electronics is a fairly standard recommendation if you want to learn more. I personally think it could use a bit of a rewrite, but it's still probably the best option around for a basic education in the art.
 
I don't mind learning a few new things... but right now is this wired up correctly? everyone is telling me very vague things like random generic components to just throw on I have no idea where to put these things since the majority of them (Op-amps, transistors) I've never dealt with before. I've only done simple calculations in Physics for components like Capacitors I've never taken circuit design.

I'm thinking I can just take a list of all the components and grabbing some random op-amp and caps from ebay a bunch of them and playing musical chairs with them until one setup works. Can you guys give me a bigger picture (feel free to edit this one as pix are better than words) on where the hell I put the diodes and op amps and where I hook up PSU 12V line, the mobo's 12V line the ground etc?

fanampnc5.jpg


In case it helps this is the original design from the LM317 PDF file, I just modified it according to whatever I could make sense from the posts. From what my friend has told me most likely I'll just end up blowing something up.

LM317.gif
 
remmeh is tantalizingly close to an even simpler solution: A unity gain amplifier can be as simple as a single BJT (don't worry, it's just a type of transistor). Here's a simple diagram:
Code: 
       +12V
         |
       |/
input--|   <-that's the BJT
       |\
         |
         |
        FAN(s)
         |
         |
        GND
Base at the left, Collector on top, Emitter at the bottom. There's one slight gotcha: the voltage your fans see will be about 0.6-0.7V lower than the voltage going into the transistor. In practice, you won't be able to tell the difference. You just need to find the right transistor. mouser, newark, and digikey all have pretty good search functions to help. Assuming the 40W means 40W @ 12V, that's about 3.3A. To be safe, let's choose a transistor for 5A or more. yada, yada, yada, you end up with a search like this one. Take your pick.

You'll need a heatsink, like the one in the original product. If you have a busted PSU, that's a great place to scavenge one, since it'll have holes drilled and tapped already to attach to the transistor.
 
Mohonri said:
remmeh is tantalizingly close to an even simpler solution: A unity gain amplifier can be as simple as a single BJT (don't worry, it's just a type of transistor). Here's a simple diagram:
Code: 
       +12V
         |
       |/
input--|   <-that's the BJT
       |\
         |
         |
        FAN(s)
         |
         |
        GND
Base at the left, Collector on top, Emitter at the bottom. There's one slight gotcha: the voltage your fans see will be about 0.6-0.7V lower than the voltage going into the transistor. In practice, you won't be able to tell the difference. You just need to find the right transistor. mouser, newark, and digikey all have pretty good search functions to help. Assuming the 40W means 40W @ 12V, that's about 3.3A. To be safe, let's choose a transistor for 5A or more. yada, yada, yada, you end up with a search like this one. Take your pick.

You'll need a heatsink, like the one in the original product. If you have a busted PSU, that's a great place to scavenge one, since it'll have holes drilled and tapped already to attach to the transistor.
Click to expand...

Aha, I had thought about a BJT emitter-follower or a FET source-follower, but thought that the Vbe drop might be a bit much (~1V+) for such a high current FET. I'm not sure, I haven't pulled up any datasheets.
 
remmeh said:
Aha, I had thought about a BJT emitter-follower or a FET source-follower, but thought that the Vbe drop might be a bit much (~1V+) for such a high current FET. I'm not sure, I haven't pulled up any datasheets.
Click to expand...
the Vbe will be in the .6-.7V range. The base current will depend on the current gain and the load current going through the transistor.
 
silverphoenix said:
I saw the mcubed fanamp, and the problem is that it doesn't supply enough power to do what I want to do and the only US retailer that sells them is OOS. So anyone who's an electronics whizz please help me out on circuit design of one? I don't need manual standby controls like the mcube it just needs to have the output voltage = the input voltage, but the output just needs more amperage.
Click to expand...
So basically, this has little to do with the the Fanamp at all (because you're not scaling in v out, using a thermostat or the other features), and you just want a plain jane diy buffer-amp based on the mobo's fan speed signal? Mohonri's solution looks to be on the money for that.

While this may have the advantage of construction simplicity (granted a big bonus for those with limited electronics experience), I've got to say I've never been a big fan [pun intended] of fan controllers that produce gads of heat in order to operate - seems a bit counter-intuitive to me. Especially as they will generate the most heat attempting to run the fans at the slowest speed - an evil circle for sure.

The mobo fan speed signal is most likely pwm generated to begin with... too bad it's probably not easy to tap into that digital signal before it's filtered, because then you could just amplifiy the digital signal, which would be much more efficient.

The LM317 should have been written off at the get-go due to it's 1.5A capacity.
 
There seems to a misunderstanding of the fan speed signal from the motherboard and whatever it is silverphoenix is trying to do.
 
mtrupi said:
There seems to a misunderstanding of the fan speed signal from the motherboard and whatever it is silverphoenix is trying to do.
Click to expand...
If I understand correctly, he is attempting to use the mobo controlled fan speed output to operate more devices than the 1 originally intended smal fan, and needs a buffer (amp) to provide the additional current.

This is unrelated to the mobo fan speed monitoring input signal that has been the most common up to the newer models offerring the speed output capability.

However, I also see this as being very close to simply using a psu that offers thermal fan speed control functionality, or some other solution based on fans with their own internal thermal control.
 
Mohonri said:
Would you mind informing us why the emitter-follower amp diagrammed above would not work?
Click to expand...

I don't want to share the solution publicly, but the emitter follower amp is harder to keep cool and it's not just simply a voltage follower design.
 
mtrupi said:
I don't want to share the solution publicly, but the emitter follower amp is harder to keep cool and it's not just simply a voltage follower design.
Click to expand...
With all due respect, that doesn't mean the voltage follower method suggested above wouldn't work - it's just not going to be efficient. But I'd suggest that is more a compromise towards the technical ability of the OP than the best solution possible.

Also, am I to infer that you work for or otherwise have some connection to Mcubed?

The bigger Mcubed bretheren like the T-Balancer are quite obviously microcontroller based pwm solutions, albeit a higher level of complexity in their closed-loop feedback designs. I can spot an Atmel Mega AVR at 50 paces ;)

Though I am unable to discern what the 8 pin chip is on the Fanamp, it would not be out of the question for it to also be some small microcontroller that generates pwm signals, which I still think is the preferrable method for efficiently controlling loads of this size.
 
agent420 said:
With all due respect, that doesn't mean the voltage follower method suggested above wouldn't work - it's just not going to be efficient. But I'd suggest that is more a compromise towards the technical ability of the OP than the best solution possible.

Also, am I to infer that you work for or otherwise have some connection to Mcubed?

The bigger Mcubed bretheren like the T-Balancer are quite obviously microcontroller based pwm solutions, albeit a higher level of complexity in their closed-loop feedback designs. I can spot an Atmel Mega AVR at 50 paces ;)

Though I am unable to discern what the 8 pin chip is on the Fanamp, it would not be out of the question for it to also be some small microcontroller that generates pwm signals, which I still think is the preferrable method for efficiently controlling loads of this size.
Click to expand...

Okay, yes the voltager follower can "work" but it is not without problems to overcome.
edit: Just a bit more on the voltage follower with BJT output. Driving the base with an op-amp as inferred from the various post will not provide much emitter current. There needs to be a pre-amplifier stage.

I think he wants to do something like fanamp. It is hard to tell what the chip is in the picture. The fanamp page says it takes a 7-12 volt level input and spreads it adjustably across the entire 12 volt range. It also has a high temp sensor which over-rides everything and sets the fan to full speed. I'm not sure there are any motherboards that control fans this way, but with 3 pin fan headers only this must be what is expected. I would think PWM fan control is more likely.

I believe I have a design that has some significant improvements and will cost less. It would take the PWM signal directly and convert it to the a fan voltage. In this way the motherboard PWM control could work on non-PWM fans. I am contemplating making and selling these myself. Do you think anyone would buy it?
 
If the current capacity of the motherboard's PWM signal is insufficient for a single-transistor package, just throw a darlington pair in there instead--sure, you'll lose a little more voltage, but you'll get much higher current gain. I agree that it isn't the most efficient design, but it's simple and it'll work, given enough heatsinking.

silverphoenix, can you tell us exactly what your intentions are? To this point, I've been operating under the impression that you want to take a single output from the motherboard and use it to control the speed of several fans, nothing more. Is that correct?
 
Mohonri said:
If the current capacity of the motherboard's PWM signal is insufficient for a single-transistor package, just throw a darlington pair in there instead--sure, you'll lose a little more voltage, but you'll get much higher current gain. I agree that it isn't the most efficient design, but it's simple and it'll work, given enough heatsinking.
Click to expand...

Yes, driving the PWM signal to multiple fans is easier. Power for the fans could come directly from the larger Molex connector so there would be no need for active devices in line with the fan power.

silverphoenix, can you tell us exactly what your intentions are? To this point, I've been operating under the impression that you want to take a single output from the motherboard and use it to control the speed of several fans, nothing more. Is that correct?
Click to expand...

This would help immensely, but with the lack of this I went down the path of what I thought would be a more generally useful idea.
 
mtrupi said:
I believe I have a design that has some significant improvements and will cost less. It would take the PWM signal directly and convert it to the a fan voltage. In this way the motherboard PWM control could work on non-PWM fans. I am contemplating making and selling these myself. Do you think anyone would buy it?
Click to expand...
Unless I am missing something, the only difference here would be filtering of the pwm output to create the (potentially current amplified) analog voltage?
 
agent420 said:
Unless I am missing something, the only difference here would be filtering of the pwm output to create the (potentially current amplified) analog voltage?
Click to expand...

That can give a voltage which proportional to the fan speed, and is only one aspect of making it useful.
 
mtrupi said:
I believe I have a design that has some significant improvements and will cost less. It would take the PWM signal directly and convert it to the a fan voltage. In this way the motherboard PWM control could work on non-PWM fans. I am contemplating making and selling these myself. Do you think anyone would buy it?
Click to expand...

Isn't this just... an analog lowpass filter?...
 
Cool--that looks like the opposite of the MIC502 mentioned in the other thread.
 
The MIC502 is an easy circuit and will do exactly what you want. I have a tube of 25 sitting on my desk. it works and it is easy. It even incorporates temp control.
 
Currently building a fan controller for all 8 fans in my P182, including the PSU fan :p It's gonna be controlled by both the PSU fan output voltage and the CPU PWM output. Should have it built within the next week, so I'll post what I did here and see if you can get any ideas from it :)
 
MisterSparkle said:
Currently building a fan controller for all 8 fans in my P182, including the PSU fan :p It's gonna be controlled by both the PSU fan output voltage and the CPU PWM output. Should have it built within the next week, so I'll post what I did here and see if you can get any ideas from it :)
Click to expand...

I'm interested in seeing what you build.
 
Bleh, got a large audio installation setup at a bar which is taking up all my damn time, haven't even gotten around to ordering parts for it yet :(
 
agent420 said:
If I understand correctly, he is attempting to use the mobo controlled fan speed output to operate more devices than the 1 originally intended smal fan, and needs a buffer (amp) to provide the additional current.
Click to expand...
That's not straightforward, the mobo signal isn't a nice DC level, it's very load-dependent so geared to a "typical" fan power. Increase the load and the voltage will drop.

Using a single "typical" fan plugged into the mobo and following its voltage to pass to the other fans seems one way.
 
You must log in or register to reply here.
Back
Top