Power Distribution Network (PDN) and Target Impedance Calculation Method
INDUSTRIAL LCD DISPLAYS / IGBT MODULES DISTRIBUTOR

Infineon / Mitsubishi / Fuji / Semikron / Eupec / IXYS

# Power Distribution Network (PDN) and Target Impedance Calculation Method

Posted Date: 2024-01-26

Using target impedance to measure whether the simulated PDN impedance meets the standard is not a scientific approach. However, many times the IC selected may not provide the PDN impedance values ​​required for each frequency band, and even the entire Datasheet does not mention PDN. At this time, the target impedance method needs to be used to measure the simulation results. After all, there is an empirical formula. Better than messing around.

In design, PDN is also called power distribution network, and its full name is Power Distribution Network. It needs to provide a low-noise power supply for the load, a low-impedance return path for the signal, etc. It is an important indicator to measure the quality of the power supply design.

In a system, the frequency of PDN ranges from HZ to GHZ. Fortunately, PCB does not need to take care of such a wide range. The low-frequency band is left to VRM, and the high-frequency band is left to the on-chip capacitor of the IC. In fact, for PCB, what needs to be paid attention to is probably 100K~100MHZ. scope.

There is a horizontal line (Z target) in the picture above. This and the horizontal line are the target values. A very important indicator to judge whether the power supply design is reasonable is that the PDN impedance cannot exceed the target value.

This target value is clearly given in DATASHEET provided by some excellent IC manufacturers, and each frequency band corresponds to different target values. Usually low frequency requires a lower target value. As the frequency increases, this value will gradually increase (not to say high The frequency requirements are reduced, but a large part of this task is handed over to the on-chip capacitor).

But for IC manufacturers that do not provide this information, you can consider using the empirical formula: the target impedance method.

The formula is:

Where Ztarget is our target value, Vsupply represents the voltage value, %ripple is the percentage of ripple, and the denominator is the maximum current.

If the maximum current is used for calculation, the actual target impedance value will be more stringent. Under normal circumstances, 0.5 times the maximum current value can be used for calculation. Optimistically, 0.3 times the maximum current value can be used for calculation.

Assuming that the power supply voltage is 5V, the peak-to-peak value of the ripple is 3%, and the maximum current is 2A, then the target impedance obtained by the pessimist is: Ztarget=(5*0.03)/2=0.075;

The result obtained by a person with a normal mentality is: Ztarget=(5*0.03)/(2*0.5)=0.15;

The optimistic friend's result is: Ztarget=(5*0.03)/(2*0.3)=0.25.

Therefore, target impedance depends on mood to a certain extent.

No matter what your mood is, you have finally calculated a target impedance value, which can be used to determine whether the PDN impedance reaches the standard.

The same picture is like this. We assume that Ztarget is the target impedance we calculated. Then the actual PDN impedance we simulated does not meet the requirements in the range of about 200K~200MHZ.

There are many ways to optimize the PDN impedance value so that it is below the target value, such as changing the stacking. Among many methods, the most effective is capacitance.

The application of capacitors can be divided into two types in principle. One is to connect a large number of capacitors of the same type in parallel. In this case, the power supply impedance can be effectively reduced without changing the resonant frequency.

The other is to connect capacitors of different capacitances in parallel. This method can be used to optimize specific frequency bands, but it will introduce new anti-resonance points.

Capacitor adjustment is a very complicated matter. Not only do you need to consider the combination of various types of capacitors, but you also need to consider the impact of the actual installation location. In this case, it seems a bit inadequate to use pre-simulation for evaluation, and complex calculations still have to be left to post-simulation.

In the post-simulation, you can import the actual PCB board and the capacitor library we used, and let the software automatically optimize and give the optimal solution under limited conditions.