need some help charging some mosfets

[52613 sfw356645eh5ah6j75]

hi, i'm finishing my fan controller and i'm having problems controling the current to the fans.
As the specs says, the standard intel made for the pwm fans used in atx motherboards allows a maximum of 50mA for the pwm signal
(50 sourcing, and 50 sinking (80recomanded) )

So I need to control 24A (8x 3A 12V fans), switching on and of the 12V that goes to the fans.
The pwm signal works at 25Khz, , that's 40µs each cycle, so it has to, at most, charge the gate in less than 0.4µs (for 2% to 98% cycle duty precision)

**broken link removed**
I have this circuit in the simulator but the ocsiloscope still shows some noise in the 12 volt signal.
(with just one pair of TIP104, 105, i still have the same problem)

Is there any better way to do this? Because even with all the amplification I still cant charge it fast enough as to have a straigt edge instead of a curve in the gate voltage.

 

[MikeMl]

You have three sets of emitter followers (5Vbe drops) between the 5V control signal and the gate of the fet, so the gate signal can only be pulled to 5-5*(0.65) = 5-3.25 = 1.75V which will never turn on the Fet. Use only one inverting push-pull driver, or go buy and IC specifically made for this job; one that will drive the gates with a full 5V swing.

I haven't looked it up. Is the Fet a "logic-level gate"?

 

[ronv]

Totem Pole

Maybe something like this. Little old fashon but fairly fast.

 

[Thunderchild]

That "driver" is a mess. 25KHz is not like a terribly high frequency, a single pair of drivers will do fine

 

[52613 sfw356645eh5ah6j75]

@mike
yes it's logic level, (with 12Vds, 4A at 3Vgs, 20A at 4Vgs, and 70A (out of specs) at 6Vgs, so at 5Vgs i suppose it can do 50A)
IRL3303 datasheet(1/8 Pages) IRF | HEXFET POWER MOSFET (see page 3)

@ron
ill take a look at it, seems good enough, thanks.

@thunderchild
the problem is that the voltage line from the osciloscope doesn't go up straight away, but with a little curve, so if i set the pwm at less than ~10% cycle duty, it starts to interfere, same at ~90%.
(ill post a screenshot in just a minute)

 

[Thunderchild]

I achieved fan control with a 555 timer and a general purpose mosfet, perhaps you should build it ? could be your simulator is being a bit over the top

 

[52613 sfw356645eh5ah6j75]

**broken link removed**
lower one is input, upper one is Vgs


even if i build i won't be able to know if it works properly because i don't have an osciloscope.
I have to see about using my sound card as an osciloscope via software as some friend advised me to not have to spend money on the osciloscope till i really need it.
and i'll try the one ron used in the screenshot.

 

[Thunderchild]

sounds like you really need one right now.

At the end of the day your aim is to not overheat and/or waste too much power in the mosfet, so build it and see how 'ot it gets. Then you will have a true answer. Simulators are not always right as i once discovered

 

[52613 sfw356645eh5ah6j75]

yes, i don't believe the accuracy of the simulators much neither, as something as simple as adding a protection diode in a circuit that wouldn't have any effect sometimes throwed errors that even the built-in assistant wouldn't be able to reproduce xD

this is the one i was talking about (xoscope) up to 44.1KHz
**broken link removed**