I am designing a peak current mode, boost DC-DC with a boost factor around 9 and typical load current close to the DCM-CCM boundary.
But the standard state space averaged models available (I used Erikson/Maksimovic book) give a discontinuity at the DCM/CCM boundary. Specifically, when looking at control voltage to output transfer function, the DC gain falls when going from CCM to DCM by around a factor of five while the dominant pole is fairly constant meaning the UGB falls by a factor of 5. Is this a modelling artifact or the DC gain/UGB actually falls at the boundary?
To see if this is indeed true, I did an experiment with an ideal DCDC. I closed the loop with a (intended) poor phase margin. Then I looked at the transient response while giving a load transient. I chose the Iload =Icrit (boundary) and gave a positive and negative step. positive step means the DCDC will go into CCM and negative step means the DCDC goes to DCM. I noticed an appreciable change in the frequency of ringing. The ringing frequency was indeed around a factor of 5-6 different in the two cases.

Thanks in advance,
Arun