Ripple Rejection

Why are we interested in maximizing the open-loop gain T = GH anyway Because it can be shown that the effect of line and load variations on the output is decreased by virtually the same factor T.  

For example, looking at the earlier equation, we see that the line-to-output transfer function for the buck is the same as its control-to-output transfer function, except that the Vin/Vramp factor is replaced by D. So for example, if Vramp = 2.14 V, and D = 0.067 (as for 1 V output from a 15 V input), then the control-to-output gain at low frequencies is 

The latter represents attenuation, since the response at the output is less than the disturbance injected into the input. But both these are without feedback considered (or with the error amplifier set to a gain of 1, and with no capacitors used anywhere in the compensation). 

So when feedback is present ( loop closed ), it can be shown by control loop theory that the line-to-output transfer function changes to 

So the additional attenuation is 54 dB here. But we already had 20 x log (D) = 23.5 dB of attenuation. So by introducing feedback, the total attenuation of the 100 Hz input ripple component has increased to 54 + 23.5 = 77.5 dB. This is equivalent to a factor of io77 5/20 = 7500. So if, for example, the low-frequency ripple component at the input terminals was 15 V, then the output will see only 15/7500 = 2 mV of disturbance. 

---

Diodes D1, D2 limit the excursion of the amplifier outputs to being near the rails. Rfb, Cfb Set the comer frequency for roll-off of ripple rejection. 

Freg - comer frequency in ripple rejection (typical value about 600 Hz)... 

Advantages Low parts count, inexpensive, good accuracy, good ripple rejection Disadvantages Excessive power dissipation at higher currents, not as efficient as other topologies (owing to headroom requirements)... 

The rule-of-thumb is to pick an output capacitor with a ripple current rating equal to or greater than the worst-case RMS capacitor current calculated above. Its voltage rating is usually picked to be at least 20 to 50% higher than what it will see in the application (i.e. Vin.max for all topologies). The input voltage ripple of the converter is also usually a concern because a small part of it does get transmitted to the output. There can also be EMI considerations involved. In addition, every control IC has a certain (usually unspecified) amount of input noise and ripple rejection, and it may misbehave if the ripple is too much. Typically, the input ripple needs to be kept down to less than 5% to 10% of the input... 

Sometimes a bypass capacitor (10 /xF) will be connected between the adjustment terminal and ground to improve ripple rejection and noise. Generally solid tantalum capacitors are used because of their low impedance even at high frequencies. If the bypass capacitor is used, then the addition of protection diodes for the bypass and output capacitors is a good idea. The details are left for the databooks. 

Ripple rejection in decibels. This specihcation is for the line regulation, indicating the regulator s attenuation of the line voltage ripple. It is predicated on particular AC input signal levels and frequencies with a specified value of bypass capacitor. 

Ripple rejection A measure of how well the regulator attenuates the input ripple, usually specified in... 

The overall ability of a power supply to attenuate disturbances at its input is expressed as its PSRR (power supply rejection ratio). In graphs, PSRR is usually plotted as a function of frequency. We will invariably find that the rejection ratio is very low at higher frequencies. One reason for this is that the Bode plot cannot really help because the open-loop gain is very small at these frequencies. The other reason is, even a tiny stray parasitic capacitance (e.g., across the power switch and inductor) presents such a low impedance to noise frequencies (whatever their origin) that almost all the noise present at the input migrates to the output unimpeded. In other words, the power stage attenuation (which we had earlier declared to be Vo/Rin) is also nonexistent for noise (and maybe even ripple) frequencies. The only noise attenuation comes from the LC filter (hopefully). 

This description, of course, oversimplifies the communication process by assuming a imitary set of values into which the themes are absorbed like a pool of water absorbing the ripples from a dropped stone. The communication flow is much more complicated, and information is accepted or rejected and finally coded in terms of a pliuality of needs, values, membership and reference groups. 

---