UCC28780: About setting conditions of UCC28780

Hello User4369382,

Thank you for your interest in the UCC28780 ACF controller.

Unfortunately, this device cannot support such a high peak to average power ratio.  Based on the over-power protection (OPP) threshold curve compared to the primary peak limit, the controller can support 0.8V/0.6V (133%) at the very lowest input and up to 0.8V/0.425V (188%) at high line. 

So for a 30-W average power level, peak power is limited to 40W at 85Vac, and to 56.5W at 276Vac.  Conversely, for 200W peak power, the nominal power design point would have to be 150W, of which 30W is only 20% load.  Thermally, the power stage could be designed for steady-state losses from only 30W, but electrically, it would have to be designed as a 150-W converter.

Running at D > 0.5 is not a concern, because this device operates in Transition Mode (TM) only.  There is no duty-cycle related stability issue for TM flyback, regardless of input voltage.

For new ACF design, I recommend to obtain the latest Excel Calculator tool from our website, just recently updated: 

Suggested inputs for your application at cells D16-D25 are:
AC, 276, 230, 85, 80, 70, 47, 0.93

If D30 = "30", then for 200W peak, D32 = "500" in order to set the OPP threshold at 150W to get the peak limit at 200W.  I have never used such a high pk/avg ratio before, so I do not know if the tool will work properly or fail at some calculation step due to some unknown limitation.

If the UCC28780 cannot fulfil your needs and you must maintain 200W/30W, perhaps a controller capable of Continuous Conduction Mode (CCM) must be used, such as LM5021, to get the high peak power.  In the case of CCM, peak current-mode control with D > 0.5 will require slope compensation to maintain stability. LM5021 has that built-in.

Regards,
Ulrich

Hello Tak,

I'm glad the latest calculator tool is easier to use.

You mentioned "reducing the input capacitor" as part of the thermal design, but Cin must be sized for 200W, unfortunately, if the 200W must be delivered during the 160ms OPP interval.  Cin must still provide hold-up energy for 200W between each half-cycle peak of the lowest rated input voltage during this interval. Hundreds of switching cycles are processed at full power in each millisecond of operation and Cin cannot be undersized or Vout will fall out of regulation.       

It is not easy to determine the maximum switching frequency.  In this ACF operation, maximum frequency is a function of output power, primary inductance and the amount of negative current needed to achieve ZVS at high line, as seen in Equation 29 in the datasheet.  To use this, all of the factors must be chosen first.

Lm is determined based on the minimum switching frequency targeted by the user. The frequency at any other operating point cannot be determined until after the minimum is selected. The guideline for choosing fSW(min) is vague, though, based on balancing user goals with reality checks. For example, driving frequency higher is generally expected to drive transformer size down (preferably to zero).  However, the UCC28780 propagation delays will begin to affect operation above ~1Mhz, and ferrite core materials for higher frequencies become more difficult to get, Rac effects are larger, gate-drive bias currents are higher, Si-Fets can't support that rate, pcb layout becomes critical for low EMI and parasitics, etc.      
One may expect fSW(max) to typically be 2~3X fSW(min).  (It is not a fixed ratio.)

24uH seems low, but is not unreasonable considering that Lm is calculated to allow the peak current necessary for 200W to build up at the minimum bulk voltage and demagnetize at 30V within the 6.25us period. However, placing "200" in Cell D30 is too much overdesign.  You will get a 267W OPP level.
Instead, I suggest to set D30 to 136.4W and D30 to 110%, and 200W will automatically be the max power available, for 160ms.  This, plus setting Nps = 4, raises Lm to 39uH.

Concerning the SIMPLIS model: there are application notes on the model's schematic page that list the conditions for which the model is not valid. I believe fault-modes are modeled. The full AAM-ABM-LP-SBP operating range is modeled.  Start-up is abbreviated and input is DC to speed up simulation, but these simplifications can be changed by the user to include AC rectification and bulk capacitance and VDD charge-up.  However simulation time will be greatly increased.  

Regards,
Ulrich