DC-DC converters in DCM

Let us assume that we have built a particular PID for a buck DC-DC converter considering the continuous conduction mode(CCM) of operation. In this case, the combined response of switch and the filter of the buck is a second order system and thus we need PID.

Consider a situation where the load current drastically reduces and take a case where buck enters discontinuous mode of operation.

In this scenario, the combined response of switch and the filter of the buck will be a single order system. So, we had designed the PID for CCM operation and that PID is still in place along with a single order system, instead of intended second order system of CCM.

In this scenario, will the loop become unstable?

A question bothering me from long time.:thinker:

If ccm current mode buck is stable in ccm, then it will always be stable when it reverts to dcm in light load.

treez said:

If ccm current mode buck is stable in ccm, then it will always be stable when it reverts to dcm in light load.

Click to expand...

Need not be. Gain margin might make the loop unstable.

there is many refs saying it is always stable in dcm if it had been already stable in ccm.

treez said:

there is many refs saying it is always stable in dcm if it had been already stable in ccm.

Click to expand...

Can you please quote those references here?

page 12 of the attached tells it...i know it is for flyback, but you get the idea, and the same holds for the buck...you make out a buck open loop calculation and then do it for dcm and you will not be able to get it to go unstable if it had been stable in ccm

treez said:

page 12 of the attached tells it...i know it is for flyback, but you get the idea, and the same holds for the buck...you make out a buck open loop calculation and then do it for dcm and you will not be able to get it to go unstable if it had been stable in ccm

Click to expand...

As a system, DCM will make the buck as a single pole system and thus we can expect it to be unconditionally stable.But please consider my question statement.

In my question I have clearly mentioned that, we are using a PID control for the CCM buck and the same buck moves to DCM, with PID in place. In this condition, DCM buck's pole and pole from the PID can make the loop unstable. This was my concern.
Correct me if I'm wrong.

Hello,
Notice that in the attached Buck converter (28vin , vout=1.5V, iout max = 10A, Fsw=200khz), it is always stable in DCM as long as it was stable in CCM.
DCM/CCM boundary is at 3.6A
First adjust the feedback compensation components for phase margin of greater than 45 degrees at max load (CCM), and ensure crossover frequency of less than 10% of switching frequency….choose any feedback compensation components that you like. Also Ensure gain margin at least 10dB.
-You will notice that it is then IMPOSSIBLE to then get it to be unstable at any lighter load where it is in DCM. This is partly because its current mode and partly because DCM has less power stage gain than CCM.
Attached is the schematic, LTspice simulation and LTPowerCAD file which you can use to adjust the feedback compensation components and see the resultant Bode Plots.

There is another unusual problem sometimes seen with buck regulators, that a buck regulator can only increase the output voltage, not actively decrease it.

All the feedback can do is reduce the PWM down to zero, but if the load is a battery, or a truly huge storage capacitance with very low dc rate of discharge, the feedback loop will open and lose control.

All this drama can occur at very light load, and DCM, so DCM is not guaranteed to be always unconditionally stable and under full control.

One possible solution for this problem might be the synchronous buck regulator where the buck regulator becomes bi directional, and the output and duty cycle stays within reasonably close limits, and the feedback loop stays active even under zero load.

Thanks Warpspeed, granted those interesting cases.