Driving P channel Mosfet

Wouldn't it be easier to use a gate driver chip, like the TC4420.

Still not clear why you want to do this.

1 Like

That doesn't seem to go up to 24v, max of 20 recommended is 18.

I really did try to look for a driver, ust couldnt find one that fits.

Please answer post #11.

What is your application?

Thanks... Tom.. :smiley: :+1: :coffee: :australia:

as i mentioned ill be driving LEDs, LED strips to be precise. 5v, 12v, and 24v version the 5v version dont need to be DIMMED so their switch frequency would be <1Hz. 12v and 24v on the other hand will be dimmed via PWM, i have read articles that switching above the aduio frequency is best. so my targert is 100kHz frequency PWM, 100 dimming levels is fine with me so atleast 1MHz switch frequency on the PMOS, I would really like to make only 1 board for the 3 voltage source needed.

I have been looking into complementary components where they are packaged in 1 IC, is shoot through still a problem for those components

Please give links to that BS.

I already told you that PWM dimming is a trade-off between PWM frequency and resolution.
If you choose 100kHz, then you might only have five equal dim steps, not 100.

I have to disagree with you since it has already been done, the zero can achieve 250Khz freuquency at 6bit resolution (63 steps) even at 63 steps ill be fine with it, and if we go by what you said at half that frequency the step would only increase


That is basically applicable to motor drives, anything mechanical that is being controlled by PWM.
LEDs do not produce any audio.

In fact going to 1MHz will probably produce radio RF interference.

So get back down to the basic PWM that controllers produce like the UNO etc.
Push pull will not be needed in my opinion.

Tom... :smiley: :+1: :coffee: :australia:

I dont think a manufacturer as well known as Analog Devices would post something bs


If I was making a high power, audio visual, TV studio, 10,000 spectator sports arena, then some of that data in the link would matter.

From that link, they don't actually give an example of audio noise from LED lighting.
High power LEDs and High Power supplies, the noise would be from the inductors in the SMPS, not the LEDs.

The human ear perceives vibrations up to about 20kHz, which in some applications can become the important factor in determining PWM frequency. The versatile LT3755 and LT3756 are members of an elite group of LED controllers that can support very high PWM dimming ratios, as much as 50:1, at 20kHz. These controllers support a variety of topologies, including buck mode, boost and buck-boost at various power levels.

They don't give an example.
They say;

which in some applications can become the important factor in determining PWM frequency

Which some applications?

They just go into pushing their product and how it overcomes the problems of high frequency PWM.

Tom... :smiley: :+1: :coffee: :australia:
PS, Have you built any hardware yet, breadboarded your project?
PPS, If I was writing that link for my lecturer, it would be thrown back and told to cite examples, basic report practice.

64 PWM steps is not 63 equal brightness steps.

100kHz PWM @ 6-bit could be a 156ns pulse.
Can your mosfet/driver switch in <= 156/2= 78ns?

That's because standard MOSFETs want 12V of drive. Perhaps upto 15V at a pinch.

Is it possible to ask the forum participants, if anyone has heard how the LEDs "sing", let them tell us.
I've never heard of it before.

The ceramic capacitors on the LED chips could make noise.
But for that you could try to lower the PWM frequency.

I PWM my LED downlights (PCA9685) currently at 400Hz (12-bit dimming),
because the supply is too noisy at >800Hz.
I only can see some strobing if I drop well below 200Hz.

Yes that is actually the thing that i did not want, as the power supply would be in the same room as the led strips. I did not mean the LEDs "singing" as i have never heared of that happening too.

Not yet, im trying to make a circuit that on paper will work first so atleast the first prototype has a chance of working. although parts are relatively cheap , shipping is not.

I could not remember what was the value i came up with that mosfet, since then i have found a better mosfet which has a turn on-off delay of 10ns and 30ns respectivly. That should be plenty of time for 100KHz PWM

you gave me an idea, the arduino board zero i am using alone cant drive the pmos aside from its 3.3v 5ma GPIO a level translator is a needed.

if my target is 1MHz the time between rise and fall is 500ns, I think the mosfet should be able to transistion from high to low of atleast 1/10 of that time so assuming a gate current of 3A it should be doable as the datasheet of the new PMOS has a propagation time of at wors 40ns. There are many 3.3v logic NMOS than can handle more than 3A. The new circuit would be like this i guess

Ps: R1 has the wrong value should be much lower to achieve 3A (8 ohms for 24v).

Well lowering the frequency would be much simpler if the circuit can do 100kHz :grinning_face_with_smiling_eyes:

If you set PWM to 50% (~70% of max brightness), then you have HIGH/LOW times of 500ns.
The fet must of course switch much faster than that to preserve a square wave.
Shorter times (HIGH or LOW) is when PWM is not 50%.
6-bit dimming demands 64 times shorter switching times.
8-bit dimming is already poor. 6-bit dimming could be almost useless. You can't dim very low.
That's why I use 12-bit dimming (with an 8-bit lookup table).

I think there is a misunderstanding here i still am going for 100kHz PWM, with 10 dimming levels (not 100* my bad for saying 100 earlier) so the fastest i need to switch is during the 10% where after going high, its immediately followed by a LOW 1000ns after (1MHz),

and since audible freuquency is upto 20kHz i have a very large buffer to slow the frequency down if it turns out the mosfet is very slow.

Hm, unfortunately those 3A will flow through M1 when M1 is turned on, and doing nothing except degrading efficiency of your circuit.

Finally I understand what you mean by 1MHz. The PWM frequency is always 100kHz, but you want to vary the duty cycle from 10% a steps of 10%.
Have you considered a voltage controlled current source (via PWM of lower frequency and low pass filter). No noise, no EMI, simple constant current. Matter of fact, the principle of constant current driven LEDs is also used in professional LED dimmers, also because they do not flicker at all. Disadvantage is higher effort in electronics - or also efficency when done with linear circuits.

I will be driving digital leds strip too constant current will be a problem for them since the voltage will vary by a lot , and i need them to be switched off, because on my testing with the strips i have they consume 200mA even when programmed to be off , hence this 1 circuit catch all solution.

This thread is getting weirder.
No wonder that you were hiding this info until now.

You are going to try to PWM the supply of addressable LED strips with 1Mhz ?
Are you forgetting that they likely have 100n decoupling caps (each) across the supply.
They also could loose their settings when you PWM them (== turn them off).


You will be switching at a PWM Frequency of 100Khz, and want down get down to 10% duty cycle.

Tom.... :smiley: :coffee: :coffee: :coffee: :coffee: :coffee: :australia: