When should I use soft-start on the LTC4269-1?

When should I use soft-start on the LTC4269-1? I am using the LTC4269-1 on a circuit that has a lot of capacitance on the secondary. It's about 1.5 mF, but could be more in future designs. What I have noticed is the Vcc oscillates for small values of Ctr capacitance. If I reduce Rtr to 12.1 kΩ and increase Ctr to 60 uF, I don't get any voltage sag below 12 V and the part starts up fine. The problem then is the time it takes for Ctr to charge up initially can be too long for the PSE. It should be less than 300 ms, but I'm taking 414 ms.

Can I use soft start to reduce my Ctr value and still avoid a sag on the Vcc during power bootstrapping because the output current is limited? What I'm thinking is change Ctr to 22 uF like on the demo board, and set some reasonable soft start capacitor value. My goal is to keep the time it takes for Ctr to charge below 300 ms and still have no sag on the Vcc below 12 V during startup.

Here is what I see when I add 1 uF to the SFST pin (approximately 70 ms of soft-start time).

It appears soft-start does not help me to get rid of the voltage sag and still maintain a short ramp of the Ctr voltage during start up.

Here is the functionality I want. Ctr voltage rise time (Vcc) of less than 250 ms. No voltage sag during the power bootstrapping period (Vcc > 12 V). When the LTC4269-1 powers on, limit the output current until my bulk capacitance on the output of the secondary have filled up. Then operate as normal. How can I accomplish this?

I also tried 100 nF for soft-start. Here's what I saw.

I repeated the measurement about 10 times. Sometimes it looked like this too.



Added plots for 100 nF STST cap
[edited by: current view at 8:42 PM (GMT 0) on 13 Dec 2019]
Parents
  • +1
    •  Analog Employees 
    on Jan 7, 2020 8:20 PM over 1 year ago

    Using Soft-start is optional but recommended when there is a large output capacitance. Soft-start will not allow you to reduce your VCC capacitor and avoid VCC from drooping during startup. Soft-start limits the current in the primary switch but it does not change the current drawn by the VCC pin. What you’re seeing with these waveforms is the VCC turn-on / turn-off hysteresis, which is between 4V and 6.5V.

    When a flyback starts switching, the uncharged output capacitors look like a short circuit. At this point the feedback voltage is low and the output of the feedback amplifier, Vc, is also low. Vc is the input to the current comparator and determines when the primary MOSFET turns off. The peak switch current increases while the feedback pin is below it’s internal reference. This continues until Vc reaches its 2.56V clamp which corresponds to a 98mV Vsense level.

    Softstart is achieved with the SFST pin, which injects a current into a capacitor and then prevents the Vc pin voltage from exceeding the SFST pin voltage. This allows you to limit the maximum switch current during startup beyond the normal Vc compensation, which reduces stresses in the primary MOSFET and smooths the output voltage ramp up. However, using soft-start causes the output voltage to rise slower than without soft-start. In your case, the output voltage is taking too long to reach regulation so the third winding cannot properly bias the IC before VCC reaches its turn-off voltage.

    I would size the VCC capacitor and its trickle charge resistor such that it both charges within 250ms and holds VCC above its turn-off threshold until the output voltage is in regulation. Soft-start can be reduced or disabled if it doesn’t work with your circuit configuration. Another option would be to use an BJT + Zener to provide VCC instead of the trickle charge resistor. However, this would change short circuit behavior, as the VCC hysteresis from the trickle charge resistor reduces stress during short circuit. A more drastic option would be to use the newer LT4276B, which has an internal buck controller for providing VCC. The LT4276 also has an overcurrent fault timer for an improved short circuit response.

    Best Regards,

    Eric

Reply
  • +1
    •  Analog Employees 
    on Jan 7, 2020 8:20 PM over 1 year ago

    Using Soft-start is optional but recommended when there is a large output capacitance. Soft-start will not allow you to reduce your VCC capacitor and avoid VCC from drooping during startup. Soft-start limits the current in the primary switch but it does not change the current drawn by the VCC pin. What you’re seeing with these waveforms is the VCC turn-on / turn-off hysteresis, which is between 4V and 6.5V.

    When a flyback starts switching, the uncharged output capacitors look like a short circuit. At this point the feedback voltage is low and the output of the feedback amplifier, Vc, is also low. Vc is the input to the current comparator and determines when the primary MOSFET turns off. The peak switch current increases while the feedback pin is below it’s internal reference. This continues until Vc reaches its 2.56V clamp which corresponds to a 98mV Vsense level.

    Softstart is achieved with the SFST pin, which injects a current into a capacitor and then prevents the Vc pin voltage from exceeding the SFST pin voltage. This allows you to limit the maximum switch current during startup beyond the normal Vc compensation, which reduces stresses in the primary MOSFET and smooths the output voltage ramp up. However, using soft-start causes the output voltage to rise slower than without soft-start. In your case, the output voltage is taking too long to reach regulation so the third winding cannot properly bias the IC before VCC reaches its turn-off voltage.

    I would size the VCC capacitor and its trickle charge resistor such that it both charges within 250ms and holds VCC above its turn-off threshold until the output voltage is in regulation. Soft-start can be reduced or disabled if it doesn’t work with your circuit configuration. Another option would be to use an BJT + Zener to provide VCC instead of the trickle charge resistor. However, this would change short circuit behavior, as the VCC hysteresis from the trickle charge resistor reduces stress during short circuit. A more drastic option would be to use the newer LT4276B, which has an internal buck controller for providing VCC. The LT4276 also has an overcurrent fault timer for an improved short circuit response.

    Best Regards,

    Eric

Children
No Data