What is a better solution to drive a IRF3205 with an NE555?

[Chriss]

Hi!
I'm interested to drive an IRF3205 with an NE555 with a 30KHz pwm signal.
I read a lot about a FET driver ic, push-pull totem pole driver etc.
I think the most badly option is to drive a FET directly from the output pin of the NE555.

So, I decide to make thinks better because in the future this FET should drive a fan which will
consume a current around 10A when it is full on.
Actually the whole think should work on 12v.

My question is:
Should I just drive the mentioned IRF3205 connect direstly through a resistor on the NE555 pin3 (output)
or is a better way to use a totem pole driver with two transistors like on the attached picture?
Please don's take care about the NE555 configuration, this is not configured to run in pwm mode I know,
just pay attention to the output circuit pls.

Thank you very much for any suggestion.
My best regards.

Thanks.

[Paul Price]

With a 12V supply and 32KHz, the 555 timer output would be just fine to drive the MOSFET. The MOSFET would be almost completely fully on when the rising edge of the 555 output is  >5V and completely on when the output reaches 11+V.
As far as turn off, the fall time of the 555 output would even be faster than the turn on time and quickly turn off the FET.

[Chriss]

So, you vote to use the FET just direct to the NE555 output?

[Paul Price]

With a 20/22 ohm resistor in series.

[Chriss]

Yea, of course.

Thank you.

[ogden]

With a 20/22 ohm resistor in series.

That 100A MOSFET have beefy gate, but hopefully push-pull 200mA capability of ne555 will be enough for 30KHz. It shall be. Thou I would check gate voltage waveform with scope.

[rs20]

I'm not so sure -- doing a (admittedly very simplistic liberal calculation) assuming:
-- Switiching time = IRF3205 total gate charge / NE555 output current
-- MOSFET dissipates 10 amps * 12 volts while it is switching

Gives a dissipation of 2.8W. While I realise that my calculation is vastly overestimating in many regards*, I'd like to know if there were any calculations behind your statements Paul? Even if it turns out to be true that this particular combination of parameters is OK, it might be more educational to explain where the cutoff lies.

In any case, a more optimized choice of transistor can reduce dissipation in the FET -- just picking the lowest on-resistance is a bad idea because low on-resistance comes with high gate charges --> long switich times --> higher switching losses. If you really want to have some fun, you can download digikey's transistor catalogue and use spreadsheet formulae to find the best transistor for your application, perfectly trading off switching losses vs resistive losses.

(Also, the decision pedantically depends on how willing the OP is to trade off wasted power in heat dissipation vs solution size vs BOM cost.)

* 2.8W would require heatsinking, but given that the real figure might be about 10 times smaller (0.28W), I'd guess that the MOSFET would not need a heatsink and just get a tiny bit warm.

[Paul Price]

You're welcome!

You might be mindful that the fan you are driving could create inductive turn-off spikes that could exceed the 55V rating of the MOSFET. In this case a protective zener diode or a snubber resistor-cap combo might be needed.
Fans that have internal transistor commutating for motor drive will not produce spikes.

RS20: I would expect MOSFET switching loss to be closer to the .28W level. Turn on time I would guess to be in the low microseconds.

Having a slower turn-off just using a 555 as a drivers means less inductive spike amplitude, if any.

I don't know what fan you are using, but I would try out this fan driver first with a small heat sink and check dissipation with use of your "digital" temperature sensor(quick touch with a finger!).

[Paul Price]

Note: The FET spec sheet shows 62Amp 101nS turn on time with 10V gate drive voltage with 4.5 ohm series gate resistor, but this is with a high current, fast-risetime gate drive voltage. The more-limited 555 high state output drive current will give a turn-on time >101nS but  you are only needing to achieve ~10Amps. An oscilloscope will quickly show how quick the FET is turning on and how big are any inductive spikes.

[Chriss]

Ok people here are all the stuff with measuring etc.
I put in series a 10 Ohm resistor with the FET Gate pin.

Made measuring with my scope and here are the results, I think it is really good.
I have only one/two questions.
Please pay attention on the pictures where are spikes and ripples on the first channel the yellow one.

When I connected to the FET a light buld 12v/25w I got spikes on the D pin of the FAT.
When I connected the small fan from the PC which is 12v and Imax=0.15A.

I think something is to do to avoid that kind of signal distortion.

Btw. the FET is cool even if the light bulb is attached and dimmed up and down.
No heat is sensing on the FET.

Here are the pictures and my schematic how I made it:

[ogden]

Gate pin waveform shows ne555 is excellent driver for that MOSFET, at least for given frequency. PC fans can act like inductive load, that ringing/overshoot seems clamped (inside fan) and safe for mosfet. Congrats!

[Paul Price]

The gate drive waveform looks excellent with the 555. The only complicaton of the output distorted waveform would be very nearby AM radio interference. A snubber (.1uf series with 22ohm) D-S should clean this up.

To be sure, you need to test this circuit with the real 10-amp fan as a load.

It would be nice to post a 1us/div scope scope pictures of the gate voltage and drain voltage to clearly see both gate and turn on/off rise and fall times.

[ArdWar]

Of course with 0.1A the losses are insignificant.

The low switching frequency also helps mitigate the weak drive. Though, at 10A load and 200mA gate drive the MOSFET switching losses might be about 4 watts, and additional 1 watt from Rds(on). That might be fine, or not, depends on how your MOSFET is heatsinked.

[Chriss]

Here are the pic's about the heat sink on the FET.
It is a small one but I hope it will hold, if not I can put a bigger one if needed...
The FET is cool like a dead man after 48h. 

I think the whole think is set up correctly what the FET configurations means.
Am I correct?

I will check later with the R/C components connected as you described to
shut down the ringing if it is possible.

Are there any suggestion to modify in any direction maybe to make think better but not over complicated? 



Thank's

[Paul Price]

Rds(on) is spec'd at .008-ohm at 107Amps, but at 10-amps, when operating at 100% duty cycle only, then Pd <=.8W due to Rds(on).

 A small heatsink may be required. Yours looks much bigger than might be needed operating at room temperature, but the size is dependent on air flow, air flow temperature and operating max. ambient temperature.

You must view the output switching waveform with the actual fan, unless you are controlling 100 .1A fans in parallel. In the case of the ringing I see here the distorted waveform would not cause any interference problems unless the fan controller is extremely close to some sensitive device. No need to add additional components.

[ArdWar]

That should be more than enough, It's not that hard to get 15K/W R_th(j-a).

By the way, for hard switching application MOSFET turn-on/turn-off losses are almost always higher than conduction losses. Also R_ds(on) usually is specified at 25^oC, it'll be higher when the die gets hotter. Not that much, but keeps in mind if you don't have much thermal headroom.

[Paul Price]

Keep in mind that the wire lead length to the fan to be controlled will create some inductance, and any inductance will create ringing and spikes on turn-off. If long wires are involved (>5cm) then you might need a snubber to reduce RFI/spike voltage.

The slower turnon/turnoff times when driving the FET with the 555 ensure you will have less RFI and spike amplitudes.

[Chriss]

Paul:
Can you pls tel me how you got the Pd<=.8w so quickly?

Is there a diagram in the datasheet what I can use for a quicker reading of this parameter?
Or I have always calculating it.

The wire is from the fan and the FAT is not longer than 15cm.

My best regards.

[Chriss]

Ok, here my result from today when I connected the real fan to my pwm controller.

I_fan= 10-12A
U_fan=12V

The FAT is totally cool on a minimum duty and a maximum duty too.
Also when I was testing on a longer period of time approximately 10min on a duty of 30% on time.
FAT still cool.

But I have some problem with the spikes which killed my D3 on my schematic in my previous post above.
D3 is the dieode between the D-S of the FAT.
The spikes are very high, as you can see on a duty of 50% the spikes rise almost to 40V.
When I gone higher duty around 80% the spikes was over 100v and killed the D3.

Any idea how to prevent this to happening and hot to eliminate the spikes?
I know, I used a general purpose rectifier diode, but at the moment I don't have any schottky diode.
I also know does the diode was killed because the diode is not rated for that high voltages....

Would a schottky do the job?
Check the picture about the scoping.

Thank's
My best regards.

[rs20]

Did you install the diode at all? Did it make proper contact? The whole point of the diode is that it clamps the voltage spikes -- if the voltage is reaching 100V, then the diode has already failed to do its job; whether it subsequently gets damaged by that voltage spike is of little interest.

Now if the diode is getting damaged by excess current, then failing to act as a diode, then the symptom of that failure would be the voltage spiking to 100V. Given that you're putting 12A through it, but the diode is only rated for 1A average continuous / 30A non-repetitive, it could be excess current that's the problem (but I'm not sure about this). I'd be wanting to try a diode that's rated for 10A.

There may also be details about the switching performance of diodes that might be relevant here, but I've never heard of diodes getting damaged by excessively fast transition.

[Chriss]

I think the problem is the specification of my diode what I use.
The diode should handle minimum the current of the motor or maybe more.

I think the spikes are drive the diode in a short because an over voltage generated by the spikes,
and so the FET is actually delivering a big amount of current to the diode but for a short of period of time.
After several min the diode reach his maximums and crack.

But lets see what other peoples also say.

UPDATED

As I wrote.
I found a fast switching power schottky diode in a PC PSU,
this is a  MBR1645DT and replaced my old 1N4001 with this one.
Now is everything fine and cool.

Here is my signal with the MBR1645DT:
I think this signal and this pwm controller succeeded.
What you think?

My best regards.

[Damianos]

So, what you have to do now is to test it in the entire range of PWM duty cycle.
Then it is ready to start production! 

-------------------------------------------------------------------------------------
Some points on your conclusions:
The spikes are produced due to to the "leakage" inductance of the motor (I'm not sure if the term is correct). These are in the foreword direction of the diode, so it is (was) not the overvoltage that break the diode but the overcurrent. The reason that there were spikes with the diode there was that the conventional rectifier diodes "need" some more time to start conducting.
Yes, the diode must be specified for the maximum current of the motor. This because when the motor just "tries" to start rotating or is stalled is a pure inductor, especially with low duty cycle.

[Chriss]

Damianos:
I was smiling until I read your part of post "Then it is ready to start production!" 
Maybe it is a good idea, I don't know, never thought about it.

I tested it in the whole range of pwm in this meaner, I let the fan running for 30min continuously
and changed time by time the speed.
Everything was fine, the fet and the new schottky diode was also cool all the time.

Is there any other way to make better tests?

Thanks.

[T3sl4co1l]

Oh hey, those schematic symbols I drew so many years ago.

Tim

[Chriss]

Paul Price:
Sorry I don't sent the requested scoping picture from your replay #11

Here I share the 1uS/div scoping on the G and D.
The D is fully opened after 200nS compared to the signal of the NE555.
The output voltage rise to 9.2V on the NE555 after 200nS.
This measurement was taken on a full load of the FAN under the max power.

What are you think?