Buck converter P-channel MOSFET switching on problem

[bhathi123]

As one constraint in my assignment I have to use P-channel mosfet, as the switching device in my SMPS. Switching frequency is 100kHz.

As illustrated in the image, during switching on, there is a ripple (which cripples everything). what can i do to overcome this?
Quick replies are very very valuable as I'm in a hurry.

 

[goldsmith]

Hello bhathi
At first step i should tell you , your circuit has some problems : the series resistor with gate of mosfet should not exceed from 39 ohm for your application and i suggest you , put a 4.7 ohm in series with gate .(1.5 k is very high . in the other hand , you should use a series resistor , with the base of bjt . it will destroy if you prevent to use that resistor ( in practice ). the calculation of the out put filter has many problem . can you tell me , how much current and voltage do you need in out put , please? thus i can help you more than this .
and the other problem is , that your circuit has not feed back ! .
BTW , those oscillations are not ripples . those are ringing effect .
Try to doing things that i said .
Best Regards
Goldsmith

---------- Post added at 13:22 ---------- Previous post was at 13:20 ----------

By the way , you should use the schotky diodes . ( very fast and very low TRR )

 

[bhathi123]

Thank you very much for the quick reply goldsmith.:wink: I will right away try what you have said.

My outputs are 74v,60mA and SMPS must go into discontinuous mode at 50%load(critical point must be at 30ms Idc).
Input is 325.26v 50Hz AC (I used just 50v DC, because I couldn't find an appropriate p-channel MOSFET that has Break-over voltage over 325v).
And about the feedback circuit, I have tried to use TL494. But Orcad simulation doesn't work very well I think. I thought of UCC3570 also, But unfortunately there are no Pspice models.(Can u suggest me with something?).I must do a simulation using any kind of working software.


I'm very very thankful to you.

 

[goldsmith]

Hello bhathi
Is it possible , that you tell me why you want use pmosfets , please? i think you can use Nmosfets , easily and with lowest prices and with lower Rdson . for example : IRFP450 LC : it's properties are about : ID= about 15 A . VDS max = 1500v .
Why you don't try them ?
Good luck
Goldsmith

 

[mtwieg]

To answer the original question, that ringing is typical in converters when they operate in discontinuous current mode. There is nothing that can be done about them (except prevent the converter from operating in DCM), but they're nothing to be concerned about anyways.

 

[goldsmith]

Dear mtwieg
Hi
At buck converters , those ringings can decrease the efficiency . and it is better to reduce them .
Respectfully
Goldsmith

 

[mtwieg]

Dear mtwieg
Hi
At buck converters , those ringings can decrease the efficiency . and it is better to reduce them .

I don't see how. They're just the result of the switching node returning to Vout, in a very undamped oscillation. It does not increase dissipation. The only way to "reduce them" would be to short out the inductor after the ringing starts, or to damp the oscillation with resistance (which would obviously decrease efficiency).

 

[goldsmith]

Dear mtwieg
again Hi
Hes can use a simple snubber network to reduce them ( non dissipative RCD snubber ) . and it will not decrease the efficiency .
Best Regards
Goldsmith

 

[mtwieg]

Dear mtwieg
again Hi
Hes can use a simple snubber network to reduce them ( non dissipative RCD snubber ) . and it will not decrease the efficiency .
Best Regards
Goldsmith

I've never heard of applying a RCD snubber to DCM ringing (and RCD snubbers are by definition dissipative). Can you provide an example?

 

[goldsmith]

if you read some expert books about SMPS design ( like pressman's book ) you will see the Application of the snubber networks and at that season , ( that book has a season about snubber ) , you will see non dissipative RCD snubber networks . ( the dissipation is insignificant ) .
All the best
Goldsmith

 

[mtwieg]

if you read some expert books about SMPS design ( like pressman's book ) you will see the Application of the snubber networks and at that season , ( that book has a season about snubber ) , you will see non dissipative RCD snubber networks . ( the dissipation is insignificant ) .
All the best
Goldsmith

Yes, but I'm wondering how you could apply it to DCM ringing. I'm not going to buy a reference book just to see that.

 

[goldsmith]

What thing is not clear? probably : how the snubber can reduce the ringings ? or how ringings will increase the dissipation here? which one is not clear ?
Respect
Goldsmith

 

[mtwieg]

What thing is not clear? probably : how the snubber can reduce the ringings ? or how ringings will increase the dissipation here? which one is not clear ?
Respect
Goldsmith

Both. I can't see how you could clamp it or dampen it without introducing dissipation. And I don't see how allowing it to ring introduces dissipation.

 

[goldsmith]

Hello mtwieg
If those spikes has large amplitude ( as you can see at his attachment ) , thus the mosfet have to try to switch with that frequency and that wave form will have effect on the current wave form . ok ? thus with this amplitude changing , the internal capacitors of mosfet will change ( according to the miller effect ) . ok , now , do you like to write the dissipation equation ??!!! it will has effect on dissipation . because the dissipation at best situation is
1/T integral from 0- ton Rdson^2*ID dt . ok ? thus you can see that ton increased ( and you didn't want it . ) . and the other disadvantage is that you can not control the duty cycle exactly .
Regards
Goldsmith

 

[mtwieg]

Hello mtwieg
If those spikes has large amplitude ( as you can see at his attachment ) , thus the mosfet have to try to switch with that frequency and that wave form will have effect on the current wave form . ok ?

No, the FET isn't switching at that frequency. Nothing is. It's just an oscillation between the inductance and parasitic capacitance at the drain node.

thus with this amplitude changing , the internal capacitors of mosfet will change ( according to the miller effect ) .

The changing capacitance of the FET is irrelevant, and has nothing to do with the miller affect.

ok , now , do you like to write the dissipation equation ??!!!

Yes.

it will has effect on dissipation .

I'm pretty sure you're wrong, but I'll give you the opportunity to show otherwise.

because the dissipation at best situation is 1/T integral from 0- ton Rdson^2*ID dt . ok ?

This looks like an equation for conduction losses (though it's wrong), which has nothing to do with that ringing.

thus you can see that ton increased ( and you didn't want it . ) . and the other disadvantage is that you can not control the duty cycle exactly .

No, the ringing does not change the effective duty cycle. It does not affect the current drawn from the source or delivered to the load (unless you consider exactly when the start of the next switching cycle begins, which may either improve or worsen efficiency, similar to quasiresonant converters. But in general it has no effect overall).

 

[goldsmith]

Dear mtwieg
My mean by that mosftet have to switch ...... was not the voltage . as i told at top the current will be some thing like that . and i know that the didn't provided by mosfet . my mean was it's current.

And about the equation of dissipation : it we have to take it's average because it is a discontinuous wave ( square ) . and the average of that will be : 1/T integral rdson*ID^2 dt ( from zero up to ton ) isn't it ? and the ton increased by those spikes ( the mosfet should be turn off at that time . and you can not say that the operation of regulator is in discontinuous region ( wave is discontinuous of course ) because for discontinuous behavior the wave will clip from right not left .
Ok , ton increased and thus the behavior of circuit is not like that we want . and at that time the loss power will be higher .
Regards
Goldsmith

 

[mtwieg]

Dear mtwieg
My mean by that mosftet have to switch ...... was not the voltage . as i told at top the current will be some thing like that . and i know that the didn't provided by mosfet . my mean was it's current.

What.

And about the equation of dissipation : it we have to take it's average because it is a discontinuous wave ( square ) . and the average of that will be : 1/T integral rdson*ID^2 dt ( from zero up to ton ) isn't it ?

That's the equation for dissipation caused by conduction through the PFET. But the PFET is off during that ringing. It's not conducting, so it has no conduction losses during the ringing.

and the ton increased by those spikes ( the mosfet should be turn off at that time . and you can not say that the operation of regulator is in discontinuous region ( wave is discontinuous of course ) because for discontinuous behavior the wave will clip from right not left .

The FET clearly turns off properly, and quickly. The ringing only begins after the catch diode is no longer forward biased, and persists until the FET turn on again.

In those waveforms, the FET is on when the drain voltage is equal to 50V. When the drain voltage is ~0V, that's the catch diode conducting. During the ringing, neither is conducting, as they should be. I can't see how you could possibly think that the ringing affects the conduction of the PFET.

 

[goldsmith]

"That's the equation for dissipation caused by conduction through the PFET. But the PFET is off during that ringing. It's not conducting, so it has no conduction losses during the ringing."



I'm completely disagree with you . it is Pmosfet right . but it is enhancement not depletion ! see it's circuit diagram carefully.


And i can suggest you that you try reading some professional book about this area , or search about SMPS basics at google .
Good luck

 

[mtwieg]

So you're saying you think that the PFET is conducting during the time interval circled in the OP's image?
View attachment 68656
It's not.

 

[goldsmith]

Dear Friend , my mean is that the P mosfets are alike with N mosfets but with this difference that if the voltage across the GS of them become -15 volt the will be at saturation region .
so with this considerations , the efficiency equation and loss power equation will be equal with NMOSFET . isn't it ?