MOSFET Driver Problem

[forgoreth]

Hello everyone,

I was going to build a buck converter today but -knowing the difficulties of high-side driving- I thought it is better to just drive the MOSFET in the low-side with just a resistive load for exercise. But I failed...
I am producing my PWM signal by an Arduino microcontroller and the duty cycle can be adjusted by a potentiometer. PWM signal goes to the MOSFET gate via an opto-isolator and a MOSFET driver. Everything about the gate signal looks fine. But unfortunately, MOSFET did not turn "off". After I took the photo, I connected a RCD snubber between the source and the drain but it did not affect at all (maybe because I don't know how to design a proper snubber. R=150 Ohms, C=10nF and the diode is 1N4148); I am attaching my circuit schematic, my gate signal and the voltage on the load oscilloscope screens. I used IXKP20N MOSFET; Vdd is 15V; Vi is 20V and Rload is a variable resistor. The current was between 500mA-1A.

I have three questions:
1-) What may be the reason for staying in "on" state?
2-) 4n25 optoisolator has bad rise and fall times and this leads to phase shift between generated PWM and gate signal. This can be tolerated but it affects the duty cycle a little bit. Is there any solution to reduce -at least- the fall time?
3-) I saw applications using the MOSFET at the low-side where the source of the n-channel MOSFET is connected to the ground of buck converter. Now, I am thinking about possible disadvantages, but couldn't find any. The advantage is obvious but is there any disadvantage of replacing the switch to the low-side?

Thanks in advance!

 

[cntube]

1) Circuit construction error, is driver return and source really connected? or MOSFET may be blown. MOSFETs almost always fail short, in my experience.
2) Reduce the collector resistance or drive LED harder, more power -- but greater bandwidth.
3) Buck with switch on low side... usually circuits are designed with fixed ground reference, so a switch on low side of load would not allow for that.

 

[forgoreth]

Thank you first of all...
1) The driver return and source are really connected...
2) I am driving the LED with its rated current. At first I started with a much lower collector resistance but the result was 7 Vpp PWM with +7Vdc offset. Then I increased the collector resistance and 1kOhm is the resistance that hit the "low state" to the ground.
3) Last summer I also did the low-side switch buck converter but it was low power. Before I try it again, I just wanted to make sure if there are any disadvantages like component stresses, etc. The source of the mosfet is connected to the ground of the Vi and the drain is connected to the cathode of the diode.

 

[mtwieg]

You should actually identify what's being shown in the oscilloscope traces. Do any of them show drain voltage?

 

[forgoreth]

Actually I tried to identify the traces by the file name but it is not appeared as I thought I don't have Vds screen but I have the load voltage.

1st channel is load voltage and second one is Vgs.

 

[Doom2pro]

Try putting a 10K resistor from gate to source, the internal miller capacitance might be turning on the floating gate.

 

[mtwieg]

Actually I tried to identify the traces by the file name but it is not appeared as I thought I don't have Vds screen but I have the load voltage.

1st channel is load voltage and second one is Vgs.

When you say load voltage, you mean the voltage across Vi? That's not really load voltage; you need to probe the drain to see the load voltage. But we do see ripple on Vi, which indicates that the load is, in fact, switching properly. None of the waveforms you posted really indicate any problem. So why exactly do you think the FET isn't turning off?

 

[forgoreth]

Try putting a 10K resistor from gate to source, the internal miller capacitance might be turning on the floating gate.

Didn't work but thanks.

When you say load voltage, you mean the voltage across Vi? That's not really load voltage; you need to probe the drain to see the load voltage. But we do see ripple on Vi, which indicates that the load is, in fact, switching properly. None of the waveforms you posted really indicate any problem. So why exactly do you think the FET isn't turning off?

When I say load voltage, I mean load voltage; the voltage between the terminals of the resistor. The probe is between the positive terminal of the Vi and the drain of the MOSFET. It is not turning off because the screen must have shown a signal with a 20V peak voltage and 0V bottom voltage that changes in every 15 microseconds when the duty cycle is 50%. It should have been the the same but amplified version of the PWM signal.

Edit: The voltage between the drain and the source is -not surprisingly- "0". And equal to "V(i) - Vload(voltage on the resistor)".

 

[mtwieg]

When I say load voltage, I mean load voltage; the voltage between the terminals of the resistor. The probe is between the positive terminal of the Vi and the drain of the MOSFET.

Oh, so you have a differential probe, or the oscilloscope ground is completely isolated from the circuit, right?

 

[forgoreth]

Yes, I use differential probes. I did not mention it before, sorry. You made me think about oscilloscope channels... Probably the channels are not isolated; I hope they are not! I am going to test it next week with two oscilloscopes that supplied from different isoaltion transformers (just in case!)

 

[forgoreth]

oh, wait... the mosfet is on the low-side. channels of the oscilloscope don't need to be isolated since their reference probes should be connected to the same node. the source is already connected to the ground of the circuit... it is not working...

 

[picgak]

HI forgoreth,
TC 426 takes a voltage above .8V as high , so is your low at the input of TC426 below .8V chek with scope and reply

 

[mtwieg]

He already showed his gate drive waveforms, which look fine.

Assuming what you're describing is true, we can only assume that either the MOSFET is damaged, or it's connected improperly somehow. There's really no other explanations.

 

[forgoreth]

I start to supply the driver by a totally different power source that includes a transformer. Hence, I thought that using an opto-isolator is unnecessary and bought TC4427 which is directly connected between the arduino PWM output and MOSFET gate. This time I placed the MOSFET on the high-side. The ground of the isolated supply is connected to the source of the MOSFET. This time it worked but the voltage shape on the resistor was too "messy". I will try it again with a snubber circuit.
A TC4427 is burned when I try to observe the effect of the drain current to the voltage on the load. The signal was in a better condition but there was a great instant reverse voltage on the switch-off times. I think I can solve this problem with a snubber circuit. What do you think about this incident? May it be because of the operation without snubber circuit or are they unrelevant?