Duty cycle is not different. Why?

It is a boost PWM dcdc. When Vin<3V, F=500KHz, Iload>300mA, the duty cycle will change to be different. The switch device ON time is 60% in a cycle, and the another cycle ON time is more than 85%. But Vin increase, or Iload is small, or F is high. The waveform is standard. What happen is in the circuits? Which thing can lead to this difference? Thanks. I have change the compensation value(R value or C value), but it is no use.

My opinion: There is some conduction loss which makes the convertion ratio M(D)=V/Vg no longer close to 1/(1-D) when D is close to 1.

exp: boost with inductor series res Ri
M(D)=V/Vg=[1/(1-D)][1/(1+Ri/Rl/(1-D)^2)]

If you plot M vs D, there are two duty ratio solutions to make V/Vg as what you want. If your controller has high gain/ bandwidth, it might oscillate between these two solutions. If you ramp the load up slowly or decrease the input voltage Vg slowly, this oscillation will be gone.

have you tried checking the output frequency from the mosfet??? in such a device the main contributor to the duty cycle is the mosfet... check the mosfet if it is of the same model and vendor... each component has a different reaction as to per electrical characteristics.... if all the components are identical, try swaping the mosfet of the first to the second... observe the reaction of the circuit... if swaping the mosfet makes no significant change, try evaluating your opto coupler then your inductor then your pwm... i think that the failure occurs in component level... also try checking for the feed back. if you are using SMD devices, try checking for cracked components... in component level testing, there are many factors that can affect your output.

best regards,

anun

I have found when Duty cycle is large with high load current and low line voltage. The duty cycle of inductor current waveform will be different. When output voltage is set smaller, the waveform is normal. I cannot explain it. I have ever think the slope compensation is not enough. But if the slope compensation is not enough, the light load current would be occur the abnormal thing. Now, the light load current is also normal. So there may be other answers. Are anyone can tell me?
And I want to know the relation between slope compensation and load current.