A defective circuit board layout leads to unnecessarily high parasitic inductances, capacitances and resistances, which in turn lead to design disturbances or higher thermal loads.How can a poor layout be recognized and how can it be improved?In a buck converter, the input capacitor loop carries the highest switching currents, so any parasitic inductance in this loop will introduce unwanted oscillations.If the input capacitors are placed too far from the VIN and GND pins, and maybe even on the other side of the board, the situation shown on the left in Figure 1 will result.The figure illustrates this case using the example of the TPS56C215 synchronous buck regulator from Texas Instruments (TI).How noticeably the conditions improve as soon as the input capacitors are placed directly on the VIN and PGND pins can be seen in Figure 1 on the right: the oscillations at the switching node have disappeared and the output ripple has decreased to around 4 mV.The use of HF bypass capacitors at the input, which serve to decouple the high-frequency harmonics of the steep switching edges, also has a beneficial effect.The feedback pin of a switching regulator has a high impedance and is therefore very susceptible to interference.The layout example above in Figure 2 is therefore clearly to be classified as unfavorable.Although the output resistors (marked in green) are close to the VOUT pin, the connection to the feedback pin runs across the board and past the inductor, so noise is bound to be introduced into this line.How to do it better can be seen in picture 2 below.The improvements cannot be ignored.While the output voltage in the first case is subject to disturbances and ripples and the switching signal has jitter, the improved layout brings significantly better results (Figure 3).The TPS543C20 synchronous buck converter is able to provide a very tightly regulated output voltage (e.g. 0.85V at up to 20A) at high load currents.Nevertheless, a circuit equipped with this component can behave unfavorably, which manifests itself as follows: while the intended 0.85 V is present without a connected load, this voltage decreases noticeably as the load current increases.In fact, the accuracy with which the output voltage is regulated depends to a large extent on where on the circuit board the voltage fed back to the regulator is tapped.Particularly in the case of controllers with high output currents, significant voltage drops can occur along the printed conductors, which can lead to deviations from the target output voltage if they are not taken into account in the controller.Equation 1 gives in simplified form the resistance of a conductor whose length and width are identical:R = (ρ xl)/(T xl)=ρ/T (equation 1)You can model a long track as a series of such square conductor segments and use it to calculate the total resistance of the track in question, provided you know the thickness of the copper coating on the board and use the surface resistance from the relevant tables.In the present case, a total resistance of 1 mΩ could be calculated.This does not appear to be particularly high at first, but with a load current of 20 A it still causes a voltage drop of at least 20 mV.This problem can be eliminated by tapping the output voltage after the parasitic resistance, i.e. as close as possible to the consumer.In this way, any voltage drops that occur along the way are automatically compensated.However, not only the voltage drops just described are problematic for the feedback signal, but also interspersed disturbances.In the circuit in Figure 4, the resistor marked in green between the current measuring resistor (red) and the controller IC (orange) filters out any switching interference.This is a flyback circuit that offers a cost-effective solution whenever galvanic isolation between input and output is required.If the filter resistor is placed too close to the current measuring resistor here, the trace leading to the controller IC will be excessively long, which in turn entails the risk of interference variables interfering with this line and influencing the measurement.On the other hand, it is better to place the filter resistor directly on the IC's current measurement pin, because this also reduces the interference signals scattered along the conductor track.In the present case, the interspersed interference was so strong that the circuit did not reach its target output current, since the overcurrent protection had already responded beforehand.This is because as the output current increases, more energy is stored by the transformer's leakage inductance, which in turn increases the parasitic noise on the primary side.These disturbance variables can influence the current measurement and lead to the said response of the overcurrent protection even before the actual maximum load current is reached.Placing the filter resistor directly on the IC's current measurement pin helps here.Incidentally, the value of this resistor does not have to be changed for this purpose.Würth Elektronik eiSos GmbH & Co. KGWith many important topics such as GaN, micro modules, UPS, isolated controllers, GreenPak etc.© 2022 WEKA FACHMEDIEN GmbH.All rights reserved.