SEPIC with very very low duty cycle.?

[grizedale]

Hello

I am doing a 70KHz SEPIC converter.

Vin = 100V
Vout = 5V, 10W

As you can imagine , the duty cycle is very low..........0.047

This means a fet on-time of just 680ns at maximum load.

Will this be ok?

....the switching transitions are going to be about 120ns rise and fall, so the FET is in the linear region for 240ns....

so 35% of the time, the fet will be ion the linear region

Will this have an affect on the small signal transfer function of the converter and give it a very poor transient response?

 

[lm2577]

... Will this have an affect on the small signal transfer function of the converter ...

I had a similar problem. In my CCM-SEPIC with Vin/Vo>5 was observed instability. In this case the measured value of the phase margin corresponds to the calculated value, but the gain margin was about 10 dB less. It looked as if the loop has an additional pole increases the negative slope of the phase characteristics at frequencies higher than unity-gain frequency. I think it was associated with a decrease in pulse duration at which the delay in switching becomes significant.
I think, CCM-SEPIC with Vin/Vo=20 will be difficult to implement, but maybe it will work in DCM.

 

[FvM]

Please enlighten me! I would never think of using a SEPIC topology for this voltage ratio.

You can use a basic buck converter instead. Efficiency will be bad anyway due to the large switched power of > 200 W in CCM respectively > 400 W in DCM. I would suggest a transformer flyback converter.

 

[grizedale]

SEPIC was done for cheapness...because it can use the very cheap UC3845 pwm controller, and theres no transformer needed, and no high side fet drive.

The output power is 10W Max...switch frequency = 40KHz.

Surley the low duty cycle is not a problem? do you agree?..........it does mean a low FET on-time, ....but there are loads of SMPS's around with 500KHz-1MHz switching frequencies......even at there max duty cycle, the on-time of there fets is less than my 100V-5V SEPIC.