Circuit critique

[m4rv]

Hi,

I'm trying to design a simple circuit which will vary the voltage to some brushless DC fans used in a project.
The fans are rated at 12v and all together draw around 120mA.
The control is coming from a PWM output from a micro and through a smoothing RC filter. I don't really want to control the fans with PWM directly due to acoustic noise and (so I'm lead to believe) reduced lifespan of the fans.

The fans will probably mostly be running at a low speed (or off) but need to be able to run at full if the unit gets too hot. The maximum voltage available in the project is 12V so I need to overcome any voltage drop issues.

So I've designed the attached circuit. The P-Channel MOSFET will turn on fully when the control voltage is above 4.7v or so, supplying the full 12v to the fans. as the control voltage falls, the P-channel turns off and the n-channel takes over to supply a varying DC voltage, all the way to 0v.

This is my first attempt at designing something like this. It works as expected in the simulator but I haven't had a chance to breadboard it yet. The only thing missing from the sim that I would probably put in on the breadboard is a bypass cap for the op-amp supply.

I'm basically looking for any critique on the design. Should I have some extra components to aid voltage stability? Have I missed anything that could cause problems long-term? etc...

Thanks.

[dannyf]

if it is a brushless motor, it isn't driven the typical way a brushed motor is driven.

[m4rv]

I'm talking about the equivalent of a PC fan. It runs fine from a varying voltage from the bench power supply.

[mij59]

You can simplify the circuit by deleting the right side of the schematic ( Q2..)
Replace Q1 with Q2, swap the + and - input pins of the opamp, remove C1.

I am using a LM358 opamp, total fan current is 300 mA.
In order to get the circuit stable I used a 470 uF output capacitor, a 100 pF capacitor across the output of the opamp and the - input of the opamp, and a 22k resistor is series with the input control voltage.
 

[dannyf]

If you want to drive the motor at full power (12v), you can simply use a p-ch mosfet or a pnp. Kill the npn + associated opamp.

[m4rv]

Thanks for the replies.

The only reason I didnt go for the single P-Channel is because I'd seen information that those topologies were somewhat unstable. When I modeled it in the sim, the circuit oscillated quite easily.

I'll try the suggestions that mij59 made.

Thanks again.

[mij59]

Keep in mind that the quiescent current of a fan is an average current, even a small fan e.g. 12V 2W, will have current spikes up to 3 A. 

[dannyf]

those topologies were somewhat unstable.

That's true, particularly for the load you have here - it has electronics in the motor so how it behaves when driven by your circuit is a complete unknown.

I would experiment more.

[m4rv]

Ok, thanks.

Do you have any advice for choosing a suitable mosfet? Apart from one which will handle the current. When I look at suppliers websites, there are SO many!
Previously, I have only really used mosfets as switches, so I look for the current handling capability and a low RDSon. Possibly also a Vgs that isnt too high for my application.
I don't know enough about using them in this configuration.

[dannyf]

Rds(on) has no meaning for what you are trying to do. I would get a sense of the current when running at full speed, and figure out th equivalent resistance on the motor, and pick a mosfet - assuming a typical fan, any to220 should work.

The opamp should be picked to be able to go r2r on the output: it needs to go to the ground within Vgs of the n-channel, and to the 12v rail within Vgs of the p-channel.

Again, I would experiment first - how th emotor behaves in a current limited environment is entirely unknown - as it is specific to the electronics embedded in the motor.

[mij59]

Ok, thanks.

Do you have any advice for choosing a suitable mosfet? Apart from one which will handle the current. When I look at suppliers websites, there are SO many!
Previously, I have only really used mosfets as switches, so I look for the current handling capability and a low RDSon. Possibly also a Vgs that isnt too high for my application.
I don't know enough about using them in this configuration.

I am using a IRF9Z34N with a small heat sink.