Does the observed behavior of DCM and BCM mode conform to the internal programming model ??

Hi Ryan, 

The above set up works pretty well. 160Vin (115-120VAC) reasonable efficiency and stability and good transients. 

Since this is an industrial application I am providing a series of front end and back end options . Higher efficiency Back-ends w synch rectifier and LC filter for the 100W version or simply Back-End with parallel power schottky and caps (plus oring and Pout controllers). 

On the front end similar choices. Either a standard passive PI front end w EMI chokes, inductors, and rectifiers plus 200-500uF caps or a PFC 2 to 400KHz boost to 380V with much smaller caps. This is to accomodate large Vin swings and reduce input capacitance by creating an active high frequency energy pool. 

The 380V=Vin case is of interest since it allows the main converter to be optimized toward a fixed input voltage. It allows my lab work to focus on functional blocks while having limited interdependencies between blocks (plug and play) etc. It also reduces front end EMI sensitivity to back end loads/frequencies. PFCs like the LT8312 or the UCC28050/1.

When I model/calculate the LT8316 (with SSR, Synch rectifier 8309, and LC) using 380V= Vin, the model remains in DCM mode pretty much from 0 to higher load. But it will not go beyond 25W or so and remain at 12-15V max. I had experimented with varying loads and the Rsen settings all the way from 20m to 50mOhm. Obviously the di/dt (Vin/L) a is nearly 4x of the 120VAC case. so Ton is extremely short given the 160uH.

However, If I calculate the energy balance for DCM at full load it appears feasible.

Assume fsw=120KHz as the target frequency for practical purposes. 

Assume an .85% efficiency target. Vout=24V Iout=4A

Pin = 112.94 w .85%

Ecyc=Pin/fsw=941mJ per cycle

Ipeak = sqrt((((24*4/.85)/120000)*2)/160e-6) = Sqrt(2*Ecyc/Lprim)= 3.43A

Given that di/dt=Vin/L ... Ton=Ipeak*L/Vin=3.43*160e-6/380=1.44usec (D=0.173)

Given that Toff = Ipeak*L/Vor(24V*N(4)=5.72us (Vor is the reflected voltage*N(4))

Obviously the remaining deadtime is considerable at 120kHz, but this shows that the system [given these conditions] is balancing in DCM mode.

So question one is why is the model not able to scale to full 24V by simply remaining in DCM. (i neglected the remaining voltage drop across the secondary MOSFET etc.). 

I then use the same approach to calculate the ideal frequencies at full load and BCM.

Ipeak= (2*Pin*(vin+Vor)/(Vin*Vor)= 2.95A 

Ton~1.24usec

Toff~4.92usec which transforms into 162.4kHz. 

Obviously neither possible considering the 142KHz limit nor healthy for the transformer. So, I either toss the transformer and shift the design to a larger 250-500uH type, or use a chip which utilizes a different frequency dutycycle control. (allows to force Voltage and Current/frequency control). like the UCC28740 which is basically a DCM chip but allows for limited frequency modulation. 

Let me know if there is a way to get the LT8316 cooperate w the Vin PFC requirements. Or perhaps suggest an ADI chip that does. 

Thank you