Part Number: UCC256304
A customer has reached to us with spurious, but severe problems with their LLC application, (designed some years ago) with variable load scenarios.
The attached oscillogram (for full trace names: see its bottom part) shows a permanent hiccup condition, after very spurious events where the device enters (restart) hiccup behaviour. The device can be restarted correctly only after power cycling.
Apart from the potential cause of anomalies that stops the power conversion, the hiccup does not follow the 1 second pause constraint as outlined in datasheet and other papers.
Actually, the hiccup rate goes at approx. a repetitive period time of 380 ms.
Regarding the signals shown here: noise peaks on Vcc (yellow) and VCR (green) represent mere pickup of fast slopes of the PFC switching amplitude.
It is curious that VCR does not have its continuous basic DC level that is normally approx. 3,0 V. In a power-up scenario, VCR jumps from 0 to 3 V in approx. 50 µs after Vcc has exceeded approx. 8 V. With some delay until RVcc has becom 12 V, the full waveforms are generated, hence with power conversion active.
Second curiosity: gate of lower MOSFET is high over the full time that RVcc is active.
To me it seems that the residual supply voltages across this hiccup cycling does not get low enough to accomplish a POR event.
Further information (measured on the fly):
- FB voltage steady state approx. 8 V
- LL/SS ramps up to approx. 5 V as soon as RVcc is active, however like simple RC-timing, not linear
Additional information: spikes from switching events seen over a few seconds in an oscillogram (with 15 MSPS) appear to increase clearly upon lower power output, when burst mode is active. Upon zooming in, high-level spikes appear to occur only in burst mode, upon burst start.
But during normal operation where these (narrow) spikes can be seen on ISNS, operation is not affected. That shows that the failure trigger points for the three types of OCP failure seems robust enough, not to flag a failure.
Having said that, no design modification to use the newer UCC25640x seems necessary. Moreover, the transformer is designed to meet the recommended dV/dt of approx. 4 V/ns.
