LT3751 flyback converter doesnt work in LTspice when leakage inductance added...why?

[Flyback]

Hello,

We are trying to design a Boundary Conduction Mode Flyback with an LT3751 controller.
We wish to charge a 4uF output capacitor up to 300V within 5ms. Vin = 12V.

The attached is the LTspice simulation of the LT3751 based flyback. For some reason it doesn’t work when the coupling in the flyback coupled inductor is made less than one. Eg, if we make K=0.995, then the flyback doesn’t work properly. Why is this?

The attached simulation works absolutely fine when the coupling factor, k, is put as one. Why doesn’t it work when we add the leakage inductance into the transformer?

(On page 20 to 21 of the LT3751 datasheet, it says some strange things about mitigating the leakage inductance in flyback converters. It only mentions using an RC snubber across the primary to mitigate leakage, whereas a flyback with several Amps of peak primary current would always need an RCD clamp. It also says that the leakage inductance could make the drain node ring below ground and thence damage the NFET. –This is nonsense, after all, the NFET has an anti-parallel diode inside it. The datasheet states that a simple RC snubber should be placed across the primary to mitigate the leakage inductance voltage spike…..but says that the time constant of the RC used should be no more than 1us. That’s far too low for any flyback with more than a few Amps of primary current. Do you think that the LT3751 is a bit of a dodgy part?)

LT3751 datasheet:
http://cds.linear.com/docs/en/datasheet/3751fc.pdf

 

[ronsimpson]

Do you think that the LT3751 is a bit of a dodgy part?

No the part is not the problem.
I added a snubber. (also changed C6)

 

[Flyback]

Thanks Ron, but you're simulation of LT3751 is still not working. -In the beginning, the FET current is going up to 55 Amps instead of limiting to 19 Amps peak like it does when transformer coupling is unity.

 

[alec_t]

D2 is inadequately rated for the job.
The sim runs as is, with k=0.995, but there is a spurious high frequency oscillation causing the lack of current regulation. You can eliminate this HF oscillation (apart from a burst of a few uS initially) by a kludge which involves putting three D2 in series, reducing R11 to 2.2 Ohms and reducing C5 to 1nF !! There must be a better way .

 

[ronsimpson]

Thanks Ron, but you're simulation of LT3751 is still not working. -In the beginning, the FET current is going up to 55 Amps instead of limiting to 19 Amps peak like it does when transformer coupling is unity.

I noticed the voltage on the CSP pin is about 3x bigger than it should be for the time it takes to get the output to 130V. (270uS in my sym)

 

[ronsimpson]

reducing R11 to 2.2 Ohms and reducing C5 to 1nF !! There must be a better way

I have used other ICs that use the transformer for timeing. (cannot have much ringing)
What I noticed from the first .ASC is that when things are working well the frequency is about 70khz. But when ringing is causing problems the frequency goes to 1.5mhz or more. So that may be a way to tell if the smubber is working.

At 1mhz the current limit does not have time to work.