Can not switch Mosfet faster than 10KHz ???

[SH@RK]

Hi

I am working on a project in the university.

The project is to make a battery charger.

The charger is going to use current feedback to adjust the PWM on the switching mosfet (going to use buck converter).

The circuit is expected to work on voltage range (18-36VDC from solar panels) and output current (up to 5A DC).

So I designed the whole the circuit and when started making it I faced a problem with the mosfet is not switching as intended to.

So up to 10KHz the switching is acceptable, but after that the mosfet takes more time to turn off and at 100KHz it does not close.

The mosfets datasheets
IRF9630
http://www.vishay.com/docs/91084/91084.pdf
IRL520
http://www.vishay.com/docs/91298/91298.pdf

This is the switching circuit alone.



And here is the input signal (from the function generator) and output signal at point (Vo).
The input signal is the upper one in the pictures and the bottom one is the output
 
First the 10KHz



The 50KHz



The 100KHz



As a start,I traced the problem and found that the N-Mosfet is behaving the same alone without the P-Mosfet.

From what I understand from the datasheet the N-Mosfet should have (td(off)+tf)=(21+27)nS=48nS
SO it should withstand switching up to 20MHz, I am not greedy I want 250KHz only.

SO what I am doing wrong ?

Thanks

[tszaboo]

SO what I am doing wrong ?

You are not using a proper MOSFET driver circuit. A driver can do as much as 2-3A gate drive, while your 3KOhm diver will do fraction of that. More current = faster switching.
Also, running lots of current through a P channel FET is not good. N channel has less on resistance, just be aware with the implications of doing that.

[SH@RK]

SO what I am doing wrong ?

You are not using a proper MOSFET driver circuit. A driver can do as much as 2-3A gate drive, while your 3KOhm diver will do fraction of that. More current = faster switching.
Also, running lots of current through a P channel FET is not good. N channel has less on resistance, just be aware with the implications of doing that.

If you are talking about the load (R3),  I only wanted to see what was happening not load the system .
About the mosfet driving circuit, I am will use arduino (or PIC) to drive the mosfet , isn't that good ?

For this P-mosfet it should withstand 6.5A (my expected load is max 4A but I will design for 5A to be safe)
Or should I put two in parallel ?

I am going to make a buck converter like this, so using N-mosfet to switch from up is hard to achieve . 



Thanks

[suicidaleggroll]

About the mosfet driving circuit, I am will use arduino (or PIC) to drive the mosfet , isn't that good ?

Look up the gate capacitance for your mosfet, and the drive current for your MCU.  You should be able to calculate the switching time, and ultimately the maximum switching speed.  A low Rds(on) MOSFET typically has a high gate capacitance, and when you're feeding it with the piddly 10-20mA that an MCU can supply, it's going to take a while to charge/discharge.  A dedicated MOSFET driver would be able to supply a couple orders of magnitude more current than that, drastically reducing switching time.

[relic38]

You will not be able to drive power MOSFETs using GPIO at higher frequencies.
Switching at high speeds with relatively high current through the load, you will need a proper gate driver circuit, or just use gate driver IC.
Gate capacitance is the main culprit, and it is made worse with higher load currents.

[Audioguru]

Here is a very simple and inexpensive Mosfet driver:

[poorchava]

U need a proper mosfet driver. Generally, the you can use almost any driver Ic, just in some cases you will have to reference it to positive rail, rather than ground. You need to remember though, I that switching is essentially exchanging charge of the gate, so the quicker you need it to happen, the higher peak current you're gonna need. And high frequency + high peak currents = huge EMI.

Sent from my HTC One M8s using Tapatalk.

[bktemp]

Using a highside driver for N-channel mosfets can introduce new problems:
If the output is connected to a battery or another voltage source, the bootstrap circuit can not charge and therefore the highside driver does not start up. So you need either a highside driver with a self clocked charge pump, or you could use the circuit posted by Audioguru. It is not ideal, but it will work if you replace Q1 with a smaller mosfet with much less gate charge so that it can be driven directly by a microcontroller. You should also reduce R1 to about 1k and increase R2 to at least 10k.

[Mephitus]

Pardon me if this is a newb connection and I am misunderstanding the implementation, but in this circumstance, would a mosfet driver like the ADP3654 (or similar) be an appropriate IC in this case?

http://www.analog.com/media/en/technical-documentation/data-sheets/ADP3654.pdf

If I am reading the application notes correctly, would adding a low value cap after the mosfet improve the gate discharge rate? This might also be a newbie question, but for even better efficiency couldn't this be setup as 2 parallel circuits on one PWM driver using both P and N channel mosfets? (very rough schematic attached)

[mobbarley]

Pardon me if this is a newb connection and I am misunderstanding the implementation, but in this circumstance, would a mosfet driver like the ADP3654 (or similar) be an appropriate IC in this case?

It looks to be. there are lots of single channel devices too if you only need 1 x power mosfet.

http://www.analog.com/media/en/technical-documentation/data-sheets/ADP3654.pdf

If I am reading the application notes correctly, would adding a low value cap after the mosfet improve the gate discharge rate?

Where did you see this? I couldn't see it in the document. Adding capacitance would increase the gate discharge time and lead to higher power dissipation in the mosfet. The supply to the IC will need some dedicated capacitance for decoupling and to handle fast switching power transients.

[Mephitus]

Where did you see this? I couldn't see it in the document. Adding capacitance would increase the gate discharge time and lead to higher power dissipation in the mosfet. The supply to the IC will need some dedicated capacitance for decoupling and to handle fast switching power transients.

I think it was in this application guide: http://www.radio-sensors.se/download/gate-driver2.pdf
But I cant seem to find where it is now. I admit I could have misread it. Time to hit the books again I think.

[tszaboo]

SO what I am doing wrong ?

You are not using a proper MOSFET driver circuit. A driver can do as much as 2-3A gate drive, while your 3KOhm diver will do fraction of that. More current = faster switching.
Also, running lots of current through a P channel FET is not good. N channel has less on resistance, just be aware with the implications of doing that.

If you are talking about the load (R3),  I only wanted to see what was happening not load the system .
About the mosfet driving circuit, I am will use arduino (or PIC) to drive the mosfet , isn't that good ?

For this P-mosfet it should withstand 6.5A (my expected load is max 4A but I will design for 5A to be safe)
Or should I put two in parallel ?

I am going to make a buck converter like this, so using N-mosfet to switch from up is hard to achieve . 

Not R3, R1. R1 is the driving of the FET in one direction, Q1 is the other. Q1 is OK-ish, and one way your waveform is OK-ish.

The DC maximum current is not the only design parameter for any buck converter. You have extra losses for the FET, so it will heat up because of your switching.

Using N FET is not that hard. Just look up some FET driver datasheet and application notes.

[exe]

R2 "closes" mosfet to slow. That's why on your waveform it is fast to switch on, but slow to switch off.  Try using "push-pull" scheme with two BJTs. Also, there are dedicated ICs for doing that.

[SH@RK]

Sorry for being late on reply, I had very busy weak (Exams)

Thanks for everyone for replying, I really appreciate it.

Ok SO I tried to apply the the ToTem Pole driver as suggested by Audioguru but did not give any improvement.

I worked with an other P-Mosfet IRF9540 (I do not have the first one now) with a resister as a load as in this coming circuit.



But the wave form is not that good. I used 7.82KHz because Arduino will give me that for pwm (can get up  to 31.37KHz or 62.5KHz)

Do not worry about the shape as the signals are not squares due to my old oscilloscope (but the starting and ending time should be correct)

This is the input signal on the best Chanel

 

The top signal is coming from the Function generator and the second one is at Vo

At 3KOhm load



At 220Ohm Load



At 4.7Ohm load




So I opened my power electronics book and there were a small section (page and half) about gate driving.

It simply show the same Totem Pole .




So I think that I am doing something wrong but I do not know what.

I will reassemble the components again later hoping it was only wiring mistake.

I looked in the local market for IC driver but there were nothing at all.

So, what do you suggest me to do ?
 
Thanks

[suicidaleggroll]

That's not the same totem pole.  How you drive the transistors matters.  In that example they're being driven directly by the output of an opamp, in yours they're being driven by a weak little 3k pull-up.  You need to drive them better, a start would be reducing the value of R1 significantly.  The values proposed by bktemp would be a good start, but you may want to go even lower on R1.

[danadak]

These might help -


http://www.irf.com/technical-info/appnotes/an-937.pdf

http://www.st.com/web/en/resource/technical/document/application_note/CD00003900.pdf

https://www.fairchildsemi.com/application-notes/AN/AN-6069.pdf


Regards, Dana.

[wiss]

Max Vgs is usually 20V, many of your drivers will blow the gate oxide at higher supply voltage.

[Porch]

At that frequency, you need to slam that Mosfet open and closed. A pull up resistor will not cut it.
I use a MIC4452ZT
http://datasheetz.com/data/Integrated%20Circuits%20%28ICs%29/MOSFET,%20IGBT%20Drivers/576-1212-datasheetz.html

The IN connects to your micro controller, the OUT goes to the Mosfet, and give it 5V and it will switch that Mosfet up to 2Mz. All in a 4 piin TO-220 package.  Simple and works.

[Alex Trofimov]

I think you just don't get it.

Input capacitance of IRF9540 is 1300pF.
R1 is 3k.
Effectively, these two make an RC-filter. It's cutoff frequency is Fc = 1 / (2 * Pi * R *C) = 40kHz. So grossly it will pass all the signals below 40kHz and block the signals above it. And this is almost exactly what you experience.
So, what you need to do is to change Fc.
Simply, try to change R1 to 300 Ohm.
Generally it's not a good practice because R1 will dissipate some additional power while Q1 is open, but is should work.

[Zero999]

I think you just don't get it.

Input capacitance of IRF9540 is 1300pF.

The static capacitance may only be 1300pF but the capacitance at the instant the gate voltage exceeds the threshold and the MOSFET starts to turn on will be higher, thanks to the Miller effect.

[Zero999]

Plenty of good circuits have been powered in the following thread: https://www.eevblog.com/forum/beginners/bjt-totem-pole-questions/msg888381/#msg888381

[SH@RK]

Thanks for you all

Here is the circuit after changing the resistors values and reassembling.



Using 10V supply and 50% duty input signal from function generator at 62.5KHz (the max frequency Arduino would give me for PWM)

And the signal should be squares but due to old scope it does not show perfect squares.

The top wave is coming from the function generator and the bottom one is taking at Vo

At 1Kohm load



At 33ohm load



At 9.4ohm load



At 4.7ohm load




So what do u think , Is it good enough ?

side note, the R1 heats up (rated 0.25W), would that be a problem or I just get higher wattage one ?

I will try to get smaller gate charge mosfet later this weak but for now I will work with what is in hand.

And for more information about the project goal,
the plan is to run the system from 36V solar panel and the battery to be charged would be 24V and charged at 1.5A.
The charging will be in constant current mode maintained by changing PWM from Arduino.

So if you have any suggestion or things to be consider I will be very glad to hear it.

And now about the replies

My MOSFET switches at 10MHz, at, ugh, 30A current .


You do need a gate driver to drive a power MOSFET to anywhere above 1kHz.

That is too much current, can you maintain good switching at very small load like 50mA ?

That's not the same totem pole.  How you drive the transistors matters.  In that example they're being driven directly by the output of an opamp, in yours they're being driven by a weak little 3k pull-up.  You need to drive them better, a start would be reducing the value of R1 significantly.  The values proposed by bktemp would be a good start, but you may want to go even lower on R1.

I did try that but it will heat up, I am now on 470ohm but I tried 220ohm , it is faster but consumes power and make heat.
And about R2, is increasing it like bktemp suggested would help ?
From what I understand , the smaller it is the faster the mosfet closing.

These might help -


http://www.irf.com/technical-info/appnotes/an-937.pdf

http://www.st.com/web/en/resource/technical/document/application_note/CD00003900.pdf

https://www.fairchildsemi.com/application-notes/AN/AN-6069.pdf


Regards, Dana.

Very good reading (actually fast scanning), I understood many new things , for sure it is helping.

Max Vgs is usually 20V, many of your drivers will blow the gate oxide at higher supply voltage.

Would not that kill the mosfet ?
And getting higher supply will need voltage doubler, I think that would complicate the work.

At that frequency, you need to slam that Mosfet open and closed. A pull up resistor will not cut it.
I use a MIC4452ZT
http://datasheetz.com/data/Integrated%20Circuits%20%28ICs%29/MOSFET,%20IGBT%20Drivers/576-1212-datasheetz.html

The IN connects to your micro controller, the OUT goes to the Mosfet, and give it 5V and it will switch that Mosfet up to 2Mz. All in a 4 piin TO-220 package.  Simple and works.

Unfortunately I looked in the market but there were no drivers available.

I think you just don't get it.

Input capacitance of IRF9540 is 1300pF.
R1 is 3k.
Effectively, these two make an RC-filter. It's cutoff frequency is Fc = 1 / (2 * Pi * R *C) = 40kHz. So grossly it will pass all the signals below 40kHz and block the signals above it. And this is almost exactly what you experience.
So, what you need to do is to change Fc.
Simply, try to change R1 to 300 Ohm.
Generally it's not a good practice because R1 will dissipate some additional power while Q1 is open, but is should work.

Now that is an other way of looking at it I never thought about.
I made R1=470ohm and I will get IRF9630 later this weak which has capacitance = 700pF which is half.
These changes would make me in better position if I understand you correctly.

I think you just don't get it.

Input capacitance of IRF9540 is 1300pF.

The static capacitance may only be 1300pF but the capacitance at the instant the gate voltage exceeds the threshold and the MOSFET starts to turn on will be higher, thanks to the Miller effect.

But making it half (700) the original (1400) would help for sure.

You don't really need a totem pole output, you just need to improve turn-off, your IRL520 already shorts the gate to ground for fast turn-on. When you try to switch at higher kHz like what you are doing propagation and transtion times add up. You should use a small low gate charge fet instead of the irl520 and try just a bjt or n-fet to improve turn-on time. I don't really think its worth it if you want 250kHz just get a driver like the fan3100 they are only about 50 cents and you will never be able to match its performance.
 

I can not say anything now, I will need to study your circuit and try it.
But thanks for your work.

Plenty of good circuits have been powered in the following thread: https://www.eevblog.com/forum/beginners/bjt-totem-pole-questions/msg888381/#msg888381

Thanks, I will take a look and see what I can do.


Thanks

[Seekonk]

My Arduino solar systems uses opto isolator drives into 820 ohm and I drive double FET, so like 2500pF.  At 490 Hz it really doesn't matter that an opto is four times slower in turn off.  That is likely a good thing for most Arduino users. I don't have heating issues. Many wall warts will work at lower voltage, enough to power a high side driver.  Be sure to operate panel at power point, even if fixed voltage.  Assume you will be using a buck converter.

[retrolefty]

I really think you need to calibrate your probe.

 Ah, but calibrate is not adjustment, Dave said.   

[krish2487]

Why did you decrease R2??
Only R1 needed a reduction to 470E. You really don't understand how a transistor drive works..
You have in effective shifted the weak drive from the P channel to the N channel FET.


By decreasing R2 to 470 E and assuming a 5V PWM drive you are robbing the N channel gate of 5V / 0.5K = 100mA of current.
Which might not seem significant but is good enough to cause a slow rising edge / ramp on the FET during turn on and turn off.
(It is not just the frequency of switching but also the rate of rise that also causes FETs to heat up and blow.)


What you need to do is increase R2 from 470E to something like 4K7, so that it presents a reasonable stiff load to the PWM drive and then use a 10E or a 22E resistance in series with the N channel FET gate. just to take the edge off the rise and fall times of the N Channel FET.


You can dispense with the Totem pole output. and use just the FETS to switch on and off properly. There is no additional advantage to using a  totem pole output in your particular application.