UCC28910: Jumping from 12 to 19V

Hello Geraldo,

I think you are probably right that the E-field disturbance tricks the sensing into locking on the wrong point of the auxiliary waveform. I’m not sure if adding the snubber (shown across the output winding of the transformer in the datasheet example) will cure this, but it doesn’t hurt to try it out.

The purpose of that snubber is to limit and dampen the leakage spike voltage across the output diode, and is often placed directly across the diode instead of across the winding. It may or may not also attenuate any ringing on the auxiliary waveform, as illustrated in Figure 22 of the datasheet.

Primary-side regulation (PSR) relies on detecting the amplitude of the aux waveform as a reflection of the secondary output voltage during the demagnetization time. The contribution of the output diode forward drop is taken into account during the sensing interval, which is just before and at the end of the demag time, where the voltage falls off drastically. Detecting the fall-off edge is what triggers the VS sense detector to lock in the average voltage over the previous sample time interval. If there is significant ringing or distortion at that corner, the sampled voltage will not be accurate and the controller will change duty-cycle to make the improperly sampled voltage conform to the internal reference.

If the output voltage is to low, it indicates that the sample is taken at a false peak, however in your case, the output voltage goes too high, which indicates the sample interval is detecting too low. I recommend that you probe the auxiliary waveform with an oscilloscope and observe the shape of the falling corner. Since it appears to be sampling too low, this suggests that the fall-off corner is not very sharp and the detector cannot resolve the fall-off edge until much of the voltage has already dropped within the sampling window. Please avoid probing the VS pin directly since this can introduce additional distortion from the probe capacitance (unless you have available a special low-capacitance probe of < 2pF). Since the VS signal is the same as the AUX signal, only scaled by the resistor divider, probing the Aux usually gives you a good picture of what VS looks like.

Perhaps the E-field is altering the flux pattern within the transformer such that the aux winding waveform is rounded off. Relocating, reorienting or shielding the transformer may solve this. Please compare the Aux waveforms before and after the E-Field or relay is activated.

If, on the other hand, there is excessive high frequency ringing after the leakage inductance clamp reset time (as seen in Figure 22), then the focus could be shifted to the primary clamp circuit of Figure 40. I this is the case, follow the recommendations in Section 10.2.1.2.12 of the datasheet to see if dampening that ringing solves the problem. High ringing along the demag time usually indicates excessive leakage inductance in the transformer. It may be possible to restructure the windings to improve coupling between the primary, secondary and auxiliary to minimize or eliminate this ringing (and also reduce losses).

Regards,
Ulrich

Hello Geraldo,

One other point that pertains to this issue is the experience described on pages 7 and 8 of this application note for the UCC28722 in an E-meter design: www.ti.com/.../slua781.pdf . The UCC28910 uses the same PSR method as the UCC28722, so this voltage increase issue is applicable to your situation.

Regards,
Ulrich