Current limit in LLC ( CLL ) converters...
Colin J. Tuck ( Snr VP Global Corporate Engineering )
Power electronics IP at pwrtrnx.com
If you are operating near full power on your LLC converter and the load is suddenly shorted - there is very little to limit the current - and your primary side mosfets will see a very large increase in current until the control acts to raise the frequency and limit the output, and therefore, the pri side current too.
Why is the current not bounded? - when you short the sec side you are shorting the filter caps on the DC side ( so quite a pulse of current there - as these are low ESR/ESL caps ) and you are effectively now shorting - through the o/p rectifiers - the sec winding of the main transformer - thus the pri side now sees only the leakage part of the transformer as the load ( the separate Lres inductor if you have one ) in series with Cres.
If you are operating just a little above the resonance freq of Llk and Cres at the moment of short ( which you would be at full power ) then the net impedance, Z, seen by the primary fets, under a shorted load, is now very low indeed, ( as most of Llk is resonated out by Cres ) - and the currents in the pri fets can get very large indeed, > 10x normal - as can the peak volts on Cres - and the currents in the Tx and sec side rectifying devices - in just a cycle or two.
Failure to have a very fast acting current limit under these conditions - can lead to failure of output rectifiers, pri side fets, and over-volt failure of Cres - and Llk ( break down of windings ). Sometimes fusing of PCB tracks.
We have seen many attempts at current limiting circuits to try to prevent the above.
- CT on the sec side ( often centre tapped Tx ) windings of the transformer
- CT on the pri side in series with the Tx
- capacitor used as a current monitor on the pri side
- diodes used to limit peak capacitor voltages
- Sec side shunt
Some times two of the above are used for a belts and braces approach. Sec side shunts are great for steady state current limits - but too slow for what happens initially - a separate circuit on the shunt with a high speed op-amp or discrete circuit and high speed opto to the pri side - can reduce power ( raise operating freq ) and give time for the "normal" current limiting circuit to function.
Similarly CT's need two circuits - the steady state - and the fast responding.
Using Cres or splitting Cres and using a cap for current monitoring is generally only used in the half bridge where one side of Cres goes to gnd - again two circuits are required.
Lastly - there is nothing like a bit of oversized silicon to handle the transient over-currents until current limiting circuits take action - and this is often why large fets are seen in the LLC - not just for low losses - but to prevent blow ups when the o/p is shorted - of course bigger fets need better gate drive - but that is another story ...
Principal Design Engineer at Power Induced Design
3 年I have designed some CV-CC with dual opamps feeding the optocoupler, but I'm guessing that a hard output short pushes the LLC too far out of normal operation for the CC loop to control things.
General Manager at PRO SYS SRL
5 年When an output short circuit (SC) appears, I think it is much better to stop switching instead of increasing the frequency. Then, it can be restarted at a high frequency (possibly with low DutyCycle) to see if SC has disappeared and gradually return to normal mode.
Thank you so much for sharing this!!!
Power electronics IP at pwrtrnx.com
5 年Perhaps also a note that when you short the o/p the circuit res freq goes up compared to the control sw freq you are at? =>? hard switching?? =>? bang for some mosfets ...