buck converter design issue

[sumitj]

dcm buck converter stability

Hi All

Suppose the requirement of load current is such that, theoretically the buck converter can go into both discontinuous and continuous mode over the range of required load current. But the equations governing in cont mode and discont mode are different.

Now when it comes to designing, than either we can design of cont. mode or discont. mode

How do i make sure that a single design works well for both cases? One way is to use synch converters .... can someone tell me some different way of designing ??

Thanks

 

[gevy]

design buck converter+closed loop

Modern buck converter ICs are used in continous or discontinous mode equally. These converters have PWM in continous mode and PFM in discontinous mode.
See buck converter ICs from Maxim, Linear Technology etc.

 

[v_c]

buck converter low current continuous mode

You can also use a digital controller in the feedback loop with two sets of controller parameters - one for DCM and one for CCM -- and switch between the sets depending on the mode of operation.

Check out the Texas Instruments web site Analog, Embedded Processing, Semiconductor Company, Texas Instruments for their
UCD7xxxx, UCD8xxxx and UCD9xxxx series of integrated circuits.

Best regards,
\[v_c\]

 

[VVV]

pfm issue buck converter

Any buck converter will go into DCM at some pint.
But even though the equations are different, the output will be maintained constant due to the feedback.
Therefore, you do not need to do anything about the design, just make sure stability is ensured under all conditons.

Generally, you want CCM for low ripple, but maintaining CCM down to very low currents means large inductors, a thing to avoid. So you compromise and let the converter go into DCM when the current drops to 10~15% of the nominal. That means an inductor current ripple of 20~30% of the nominal. So that is how you select the inductor and then you know when DCM will happen.
Last, verify stability for both CCM and DCM.

 

[sumitj]

if a system can't give the required output in open loop than is it possible for system to give output in closed loop ??

I did'nt got the point that how to make the loop in discont mode also ?

 

[VVV]

If you cannot achieve the required output in open-loop there is no way you can do that in closed loop. Closing the loop does not fix any deficiencies, it merely regulates the output, as long as it can be brought under control by changing the duty-cycle.
For instance, if your input voltage is insufficient to obtain the correct output even using the maximum possible duty-cycle, what could the feeedback do to remedy that?

You need to compensate the EA in such a way that the loop is stable even in DCM. This is usually not too difficult. The EA is actually compensated for optimum performance at nominal output and then stability is just checked at low currents. If inadequate margin is found, make some compromise that results in still acceptable performance at nominal current and satisfactory stability margins at low current (DCM).

On paper, draw the Bode plots for the minimum current and check stability (gain+ phase margin).
In the lab, just use the gain-phase analyzer to check the phase and gain margins. Simply reduce the current and take another reading.