800W resonant converter using UCC25600


Working on 800W resonant mode converter with UCC25600 LLC controller.
Build up as the typical application diagram and by use of the design tool.
It work perfect in steady state (same output current, same load) and Eff. shows up to 95 % close to max load.
But with transient loads something strange happens.The Upper fet is blowing  even at very light loads. So changing
from load to no load which means that the converter goes into burst mode damage the fet. (FCPF22N60)

Can this be caused by the high drain current when gate turns on again after the burst break?

I did'nt quite hit the Lr calculated in the spreadsheet 31uH but ended up with a
Leakage induction Lr at about 22UH.

Here is the input parameters used in the spreadsheet.


Vin Min: 375V

Vin Max: 410V

Fsw norm.: 134khz

Fsw Max.: 350khz

Fsw Min : 65khz

Max. Power Limit: 800W

Max. Output Power: 775W

Full load eff: .92

Vo: 155V

M: 9



Td: 200nsec







38 Replies

  • I've designed a LLC with almost the same specs ;-)

    Power: 800W output
    Non-isolated: 85-264VAC (PFC UC28061) to 400VDC.
    Isolated: 400VDC to +/- 82VDC (165VDC from + to -) (LLC). Unregulated!
    fsw = 100kHz.
    I managed to get 510W out due to the lack of load resistors. I got around 97% efficiency for the LLC alone and total above 93% with the PFC in front at 500W out and 230VAC in.

    1) Use a IR2110 gate driver for the MOSFETs.

    2) Make sure you don't get a short through your MOSFETs when switching! Increase the dead time to 300ns by adjusting the resistor and measure the two gate signals at different loads.

    3) Measure the current in the Lr. Is it correct regarding your calculations?

    4) Is your output capacity high enough? Try increasing the uH.

    5) Make sure you're running on the resonans frequency when at max load since this is the opimum point.

  • It could also be your regulation.

    I used STW25NM500 MOSFETs.

    If you miss your leakage inductance your must insert a Lr of the rest value.
    E.g.: Total resonance inductance required at operation point (f_sw = f_resonance) = 10uH.

    If your leakage is measure to be 4uH, then your should insert a Lr of 6uH to total get 10uH resonance inductance.

    I coupled my circuit as shown: 

    With split resonance capacitors.

    Gate driver:

  • In reply to CWilson:

    Gate driver: 

  • In reply to CWilson:

    Hi CWilson

    Thanks for your reply.

    There's some good leads I will try to follow.

    I use a gate drive transformer, but it should't make the big difference as long as it doesn't cross conduct. I'll try to increase the dt to 300nsec. and see what happens.

    But the Lr value is probably the key to this problem. I'll do some test with this value and at the same time measure the current, pulse by pulse in the upper fet.






  • In reply to feb:

    Does it work?

  • In reply to CWilson:

    Hi CWilson

    Yes it works just fine now.

    I took out 830W yesterday without any problem,  instability or so.

    Burstmode is also working very well now. I change from transformer gate driving to FAN7390 high/low side driver, which seems to work better in burstmode.

    So everything is "cool" right now.

    Thanks for your feedback to my problem which was very usefull.






  • In reply to feb:

    You're welcome :-)

    What is your efficiency?

    By measuring the voltage in the half bridge node (between the two MOSFETs) and measuring the two gate at the same time, so can see if your Lm is OK and that you're in ZVS all the time --> Your deadtime must be larger than the time it takes to V_halfbridge to go to zero.

    Press "Verified Answer" if everything is ok now.

  • In reply to CWilson:

    Btw. I used a gate driver transformer and couldn't make the LLC work. Switched to IR2110 and everything was fine :-D

  • In reply to CWilson:

    Eff. is around 95% which is ok for lab model. I haven't selected diode and fets for Fvd and Rds_on.

    My Ln ratio (Lm/Lr) 4.2 might be to low, but I'll check the ZVS/ZCS. I still have the 350nsec deadtime to reduce if there is a margin.




  • In reply to feb:

    Very nice efficiency since your ratio is low. I assume you're regulating on your output voltage via an opto coupler?

    I had a ratio of about 90 because it wasn't regulating. As I recall from the TI seminar 2010/2011, they recommend around 8-9 is you're regulating.