LT3574 applications inForMation ERROR AMPLIFIER—DYNAMIC THEORYSelecting RFB and RREF Resistor Values Due to the sampling nature of the feedback loop, there The expression for VOUT, developed in the Operation sec- are several timing signals and other constraints that are tion, can be rearranged to yield the following expression required for proper LT3574 operation. for RFB: Minimum Current Limit RREF • NPS (VOUT + VF)a + V R TC FB = The LT3574 obtains output voltage information from the VBG SW pin when the secondary winding conducts current. where, The sampling circuitry needs a minimum amount of time to sample the output voltage. To guarantee enough time, VOUT = Output voltage a minimum inductance value must be maintained. The VF = Switching diode forward voltage primary side magnetizing inductance must be chosen a = Ratio of Q1, IC to IE, typically 0.986 above the following value: NPS = Effective primary-to-secondary turns ratio tMIN µ 2 H L VTC = 0.55V PRI ≥ VOUT • • NPS = VOUT •NPS • IMIN V The equation assumes the temperature coefficients of t the diode and VTC are equal, which is a good first-order MIN = minimum off-time, 350ns approximation. IMIN = minimum current limit, 175mA Strictly speaking, the above equation defines RFB not as an The minimum current limit is higher than that on the Elec- absolute value, but as a ratio of RREF. So, the next ques- trical Characteristics table due to the overshoot caused by tion is, “What is the proper value for RREF?” The answer the comparator delay. is that RREF should be approximately 6.04k. The LT3574 is trimmed and specified using this value of R Leakage Inductance Blanking REF. If the impedance of RREF varies considerably from 6.04k, ad- When the output switch first turns off, the flyback pulse ditional errors will result. However, a variation in RREF of appears. However, it takes a finite time until the transformer several percent is acceptable. This yields a bit of freedom primary-side voltage waveform approximately represents in selecting standard 1% resistor values to yield nominal the output voltage. This is partly due to the rise time on RFB/RREF ratios. The RFB resistor given by this equation the SW node, but more importantly due to the trans- should also be verified experimentally, and adjusted if former leakage inductance. The latter causes a very fast necessary for best output accuracy. voltage spike on the primary side of the transformer that Tables 1-4 are useful for selecting the resistor values for is not directly related to output voltage (some time is also R required for internal settling of the feedback amplifier REF and RFB with no equations. The tables provide RFB, R circuitry). The leakage inductance spike is largest when REF and RTC values for common output voltages and common winding ratios. the power switch current is highest. Table 1. Common Resistor Values for 1:1 Transformers In order to maintain immunity to these phenomena, a fixed VOUT (V)NPSRFB (kΩ)RREF (kΩ)RTC (kΩ) delay is introduced between the switch turn-off command 3.3 1.00 18.7 6.04 19.1 and the beginning of the sampling. The blanking is internal y 5 1.00 27.4 6.04 28 set to 150ns. In certain cases, the leakage inductance may 12 1.00 64.9 6.04 66.5 not be settled by the end of the blanking period, but will 15 1.00 80.6 6.04 80.6 not significantly affect output regulation. 20 1.00 107 6.04 105 3574f Document Outline Features Applications Description Typical Application Absolute Maximum Ratings Pin Configuration order information Electrical Characteristics Typical Performance Characteristics Pin Functions block diagram Operation Applications Information typical Applications Package Description Typical Application Related Parts