Mosfet Circuit

tylerplowright · 23rd April 2011, 9:16 PM

Hey,
Just trying something out with a few mosfets so that I can get more power out of my fan controller. Just want to make sure I have the wiring right. The fan controller adjusts the 12v rail to suit, so hopefully this will allow me to get more power out of it. i.e. as much as the power supply will let me in theory, but I only need a couple of watts.

Top mosfet is a P channel, and bottom one is a N channel mosfet.
Top plug goes to the fan controller, the others go to the fans themselves.
The green negative track is connected to all the negatives, you just can't see it as its under each of the red tracks that join the 4 sets of 3 (3pin) fan headers.

Click to view full size!

Thanks in advance.
Any comments are welcome by the way as to improving it.
EDIT: mosfets are TO-220 standard/design

mtma · 23rd April 2011, 9:29 PM

Is there an original circuit diagram? I don't think that this will do what you want it to do.

As for the layout, how are you going to be making this circuit board? It looks like two layers so why not use a ground/power pour?

tylerplowright · 23rd April 2011, 10:03 PM

Designed myself, no circuit diagram. And have a mate who has all the gismo's to make like 4 layer boards. I'll just use a double sided one which is a piece of cake.

As for not doing it what do you mean? 12v is the signal, goes to negative, then back to positive. That's the idea anyway.

What do you mean by ground/power Pour? I know 4 layer boards are for like 2 signal layers top and bottom and then power layers in the middle. But they're too expensive and too hard to make.

mtma · 23rd April 2011, 11:05 PM

Pours here probably won't make a difference as there's no major decoupling or interference to worry about, but as I see it you pay for the copper that gets taken away to make traces anyway so you may as well make the traces nice and fat. Typically you use the pours for power distribution so that the lowest resistance is obtained for all devices taking power from that pour. The reality is of course more complicated than this, and structuring the pour and the component layout also bears on how effective the scheme is.

As for the circuit...

If you're using enhancement MOSFETS of the most common type in TO-220 size and that lower row of MOSFETs are Nch, you will get conduction through the body diode straight through to your fans. In other words, the fans will receive power but no control will occur.

How the devices are connected is also suspicious, I would suggest you to draw it up and review the function before making any boards. The way I see it, you're missing some resistors and/or actually want to use FETs in source follower and not what the drawing currently indicates.

klightspeed · 24th April 2011, 3:23 AM

Quote:

Originally Posted by tylerplowright

Top mosfet is a P channel, and bottom one is a N channel mosfet.

From your board, that's the wrong way around - the top row should be N-channel, and the bottom row P-channel.

You want the top row to be high-speed low-power enhancement mode N-channel FETs, and the bottom row moderate power (i.e. >1A) enhancement mode P-channel MOSFETs.

Code:

 +12V o-------------+
                    |
           PFET G|--+S
              +--|-</
              |  |--+D
     NFET G|--+D    |
  IN+ o----|->\     +---o OUT+
           |--+S    
              |
  GND o-------+---------o GND

SENSE o-----------------o SENSE

paulvk · 24th April 2011, 4:14 PM

Quote:

Originally Posted by tylerplowright

Hey,
so hopefully this will allow me to get more power out of it. i.e. as much as the power supply will let me in theory, but I only need a couple of watts.

Thanks in advance.
Any comments are welcome by the way as to improving it.
EDIT: mosfets are TO-220 standard/design

I have a 12V fan controller that uses a thermistor to control 540watts of fans (45amps) it runs 2 motors 4 fets per motor, uses a lm494 to control the speed, simple circuit if you want it?

tylerplowright · 24th April 2011, 4:18 PM

Click to view full size!

That looking a little better? Hope that is what you meant.

tylerplowright · 24th April 2011, 4:21 PM

Quote:

Originally Posted by paulvk

I have a 12V fan controller that uses a thermistor to control 540watts of fans (45amps) it runs 2 motors 4 fets per motor, uses a lm494 to control the speed, simple circuit if you want it?

Not sure thats what I'm after. Or if it will work with the fan controller I've got. If you're offering I wouldn't mind having a look at it though. Got any pictures? Might be able to get ideas or something from it.

Goth · 24th April 2011, 4:42 PM

Why do you need two FETs anyway? Why can't you just drive the fans using a single FET?

tylerplowright · 24th April 2011, 5:26 PM

Quote:

Originally Posted by Goth

Why do you need two FETs anyway? Why can't you just drive the fans using a single FET?

Made sense to do it with 2. I don't know. The fan controller uses the +12 to regulate voltage/speed, and the fans are to be controlled via the +12v line so it made sense to make it with 2 instead of with 1 and controlling it off the negative wire (which is what I'm assuming you mean).

paulvk · 24th April 2011, 9:50 PM

Here is a link to a circuit of 2 controllers running 3 motors http://www.vk2pl.com/uploads/Motor-speed-con.pdf this circuit can be scaled up or down to suit the size of the motor by adding or subtracting fets.

klightspeed · 24th April 2011, 10:05 PM

Quote:

Originally Posted by Goth

Why do you need two FETs anyway? Why can't you just drive the fans using a single FET?

As in e.g. emitter-follower with NPN transistors?

Or are you thinking negative switching?

It looks like he wants to use the PWM from the motherboard to control multiple fans.

Quote:

Originally Posted by tylerplowright

board

The circuit described by your board (which is the same circuit described by your revision) was correct, it was your description of which FETs went where that would not have worked.

tylerplowright · 24th April 2011, 10:31 PM

Quote:

Originally Posted by klightspeed

As in e.g. emitter-follower with NPN transistors?

Or are you thinking negative switching?

It looks like he wants to use the PWM from the motherboard to control multiple fans.

Not the motherboard but this fan controller: Aerocool Touch 2000
The output on that is a 3 pin fan plug and it regulates fan speed via the +12v

Quote:

Originally Posted by klightspeed

The circuit described by your board (which is the same circuit described by your revision) was correct, it was your description of which FETs went where that would not have worked.

Dammit it is to, didn't think about it when you put your diagram up. Will revert back to the old design but will update what each mosfet is (I think it had better spacing and such)

klightspeed · 24th April 2011, 11:37 PM

Quote:

Originally Posted by tylerplowright

Not the motherboard but this fan controller: Aerocool Touch 2000
The output on that is a 3 pin fan plug and it regulates fan speed via the +12v

Do the fans on each channel total less than 6W. If so, you could drive the fans directly from the controller, with the sense wire from the first one going to the controller.

Quote:

Originally Posted by tylerplowright

Dammit it is to, didn't think about it when you put your diagram up. Will revert back to the old design but will update what each mosfet is (I think it had better spacing and such)

One thing all of us forgot is gate discharge resistors and gate charge current limiting resistors. The gate discharge resistors should be between 200 ohms and 1kohm, and the gate charge current limiting resistors between 25 ohms and 100 ohms.

Without the gate discharge resistors, the FET gates take about 50ms to discharge. This is slower than the fan controller's likely PWM frequency of over 100kHz.

Without the gate charge limiting resistors, the gates act like 500pF capacitors. Assuming the FETs have an R_DS(on) of about 1 ohm, it would cause 12A 500ps surges on the rising edge of every PWM cycle. This would make your computer's power somewhat noisier.
100 ohm gate charge current limiting resistors would reduce the surges to 120mA 50ns surges on the rising edge of every PWM cycle, which can be dealt with by a small decoupling capacitor.

tylerplowright · 25th April 2011, 12:12 AM

Quote:

Originally Posted by klightspeed

Do the fans on each channel total less than 6W. If so, you could drive the fans directly from the controller, with the sense wire from the first one going to the controller.

Sadly, no, they're 0.3A each at 12v and there are 3 of them. So its about double. Else I wouldn't have bothered doing anything this complicated.

Quote:

Originally Posted by klightspeed

One thing all of us forgot is gate discharge resistors and gate charge current limiting resistors. The gate discharge resistors should be between 200 ohms and 1kohm, and the gate charge current limiting resistors between 25 ohms and 100 ohms.

Without the gate discharge resistors, the FET gates take about 50ms to discharge. This is slower than the fan controller's likely PWM frequency of over 100kHz.

Without the gate charge limiting resistors, the gates act like 500pF capacitors. Assuming the FETs have an R_DS(on) of about 1 ohm, it would cause 12A 500ps surges on the rising edge of every PWM cycle. This would make your computer's power somewhat noisier.
100 ohm gate charge current limiting resistors would reduce the surges to 120mA 50ns surges on the rising edge of every PWM cycle, which can be dealt with by a small decoupling capacitor.

Ok now that's over my head. Care to draw a diagram and possibly give me some values of capacitors/resistors to get? If you could that would be the final piece to the puzzle and I would be very grateful. I have some 1k and 100ohm resistors lying around. Unsure about the wattage though.

23rd April 2011, 9:16 PM	#1
tylerplowright Member Join Date: Jun 2010 Location: Newcastle Posts: 598	Mosfet Circuit Hey, Just trying something out with a few mosfets so that I can get more power out of my fan controller. Just want to make sure I have the wiring right. The fan controller adjusts the 12v rail to suit, so hopefully this will allow me to get more power out of it. i.e. as much as the power supply will let me in theory, but I only need a couple of watts. Top mosfet is a P channel, and bottom one is a N channel mosfet. Top plug goes to the fan controller, the others go to the fans themselves. The green negative track is connected to all the negatives, you just can't see it as its under each of the red tracks that join the 4 sets of 3 (3pin) fan headers. Click to view full size! Thanks in advance. Any comments are welcome by the way as to improving it. EDIT: mosfets are TO-220 standard/design __________________ A race car is like a woman, its always taking your money, wasting your time, but man its fun to drive.

23rd April 2011, 10:03 PM	#3
tylerplowright Member Join Date: Jun 2010 Location: Newcastle Posts: 598	Designed myself, no circuit diagram. And have a mate who has all the gismo's to make like 4 layer boards. I'll just use a double sided one which is a piece of cake. As for not doing it what do you mean? 12v is the signal, goes to negative, then back to positive. That's the idea anyway. What do you mean by ground/power Pour? I know 4 layer boards are for like 2 signal layers top and bottom and then power layers in the middle. But they're too expensive and too hard to make. __________________ A race car is like a woman, its always taking your money, wasting your time, but man its fun to drive.

24th April 2011, 4:18 PM	#7
tylerplowright Member Join Date: Jun 2010 Location: Newcastle Posts: 598	Click to view full size! That looking a little better? Hope that is what you meant. __________________ A race car is like a woman, its always taking your money, wasting your time, but man its fun to drive.

24th April 2011, 4:42 PM	#9
Goth Grumpy Member Join Date: Jun 2002 Location: /dev/null Posts: 9,303	Why do you need two FETs anyway? Why can't you just drive the fans using a single FET? __________________ "How is anyone supposed to know that this isn't just a bunch of crap?" - Richard Feynman.

23rd April 2011, 9:29 PM	#2
mtma Member Join Date: Aug 2009 Posts: 2,648	Is there an original circuit diagram? I don't think that this will do what you want it to do. As for the layout, how are you going to be making this circuit board? It looks like two layers so why not use a ground/power pour?

23rd April 2011, 11:05 PM	#4
mtma Member Join Date: Aug 2009 Posts: 2,648	Pours here probably won't make a difference as there's no major decoupling or interference to worry about, but as I see it you pay for the copper that gets taken away to make traces anyway so you may as well make the traces nice and fat. Typically you use the pours for power distribution so that the lowest resistance is obtained for all devices taking power from that pour. The reality is of course more complicated than this, and structuring the pour and the component layout also bears on how effective the scheme is. As for the circuit... If you're using enhancement MOSFETS of the most common type in TO-220 size and that lower row of MOSFETs are Nch, you will get conduction through the body diode straight through to your fans. In other words, the fans will receive power but no control will occur. How the devices are connected is also suspicious, I would suggest you to draw it up and review the function before making any boards. The way I see it, you're missing some resistors and/or actually want to use FETs in source follower and not what the drawing currently indicates.

24th April 2011, 9:50 PM	#11
paulvk Member Join Date: Mar 2010 Posts: 698	Here is a link to a circuit of 2 controllers running 3 motors http://www.vk2pl.com/uploads/Motor-speed-con.pdf this circuit can be scaled up or down to suit the size of the motor by adding or subtracting fets.