Questions about the DCM condition of PWM buck converter

[rockycheng]

Dear all,

When the load current of the PWM converter is low enough, the inductor current will below zero at some time, and the converter goes into DCM condition.

My one question is that when calculating the efficiency, should I take the negative part of input current into account? If not, should I take that period of input current as quiesent current or just zero?

Another problem is that, when the load becomes very light (below 5mA), my converter starts to be unstable. I can't figure out why. I think that when load current decreases, the gain of the inner loop shall decrease too, and the unity gain frequency will be lower. There seems no reason to cause this problem. But it really starts to be unstable (looks like sub-harmonic oscillation, repeating every two cycles)!

Anyone know why?? Thanks a lot!

 

[v_c]

I don't think your input current should be going negative. If your input is indeed going negative, it would mean that you are putting power back into your source!

What kind of topology are you using? If you are using a buck converter, for example, the input current should be zero when the switching transistor is off.

What kind of compensation are you using? Buck converters are fairly easy to compensate. Boost and Buck-Boost (inverting) converters have right-hand plane zeros. As you can imagine, the small-signal models for a given topology is different depending on if you are in CCM or DCM - so your controller should be stable under both conditions. If you can post a jpeg schematic, we might be able to help.

You should also take a look at this topic **broken link removed** buck converter design issue which is somewhat related to your question.

Best regards,
\[v_c\]

 

[rockycheng]

Thanks v_c! The simulation really shows that both the inductor current and the input current can be negative for some time every cycle. In DCM condition, the inductor current is below zero when the P-SW is turned on, so the P-SW picks up the negative inductor current, and the current should do flow toward the input source. It looks strange since this current is "charging" the battery. I also have confusion about it.

I've posted the input current wave.

 

[v_c]

I am not sure why it is going negative. Do you have an input filter (consisting of inductors and capacitors) at the input of your system?

Can you post the schematic or the netlist that you are simulating?

Best regards,


\[v_c\]

 

[rockycheng]

Hi V_C,

In the schematic, I only use a voltage source (vdc) as the input. There is no input inductor.

For that negative part of input current, I've simply taken it as zero, and the calculated efficiency seems reasonable. I think most of this part is wasted. Maybe that's why the efficiency is low when the PWM buck converter works in DCM state.

Added after 3 hours 17 minutes:

Please look at the inductor current. It starts to be like this when the load is below about 5mA. Did any one meet this problem before? Please help me!

 

[narccizzo]

YOU SHOULD POST THE SCHEMATIC

 

[v_c]

There is no way for me to specifically figure out why you are getting this without seeing the actual controller topology and information about loop gain, etc.

I guess it is possible to get negative current in the MOSFET but it would have to be conducted through the body diode of the MOSFET in the opposite direction of normal current flow.

Best regards,
v_c

 

[rockycheng]

v_c, I agree with you, the negative current is dissipated on the body diode of P-MOSFET. Thank you for your replies!