Sometimes a ferrite is used to remove ringing in some applications as in power electronics or digital circuits. Did you know that ferrites can create that unwanted ringing if selected in a wrong way?
Consider, for example, you have a digital (3V) 30MHz 50% duty cycle clock signal with rise and fall times of around 1ns (bandwidth for this signal is around 318MHz). The signal source is integrated circuit IC1 in Figure 1 (top). The receiver is integrated circuit IC2 with a 30pF input capacitance.
in series (middle), and with ferrite and damping resistor (bottom).
Then, someone recommends you add a small ferrite bead in series with the clock signal, shown in Figure 1 (middle).
The reason for that recommendation could be well reasoned or because of other unknown and historical reasons:
- Perhaps you are trying to avoid the possibility for ringing in the clock signal if the trace is long for the rise/fall times. This is a typical source termination for signal integrity designers. Remember a ferrite must add losses in high frequency to dampen the under-damped situation.
- Perhaps you are trying to reduce the high frequency contents for the clock currents so the global emissions from your PC board are reduced. Sometimes a small capacitor is an alternative to increasing the dv/dt of your clock (note this capacitor could be a problem for signal integrity or increasing the power supply transients).
- Perhaps that ferrite is put in place “just in case” some problem could appear.
- Perhaps that ferrite was “always” there in similar designs (sometimes nobody remembers why).
And note that, many times, the ferrite is selected from a kit in a very fast and “blind” way.
To your surprise, in any of those cases, the digital clock can appear in your scope with big ringing, shown in Figure 2 (left) when the ferrite is in place compared with a case without a ferrite. How is this possible? Ferrites kill ringing! Or not always?
Of course you must be sure that the ringing is not because your measurement setup (e.g. pigtail in your probe). Take care about this.
Sometimes I have seen a small resistor in series with the ferrite, shown in Figure 1 (bottom). If a small 45Ω Rd resistor is included in our example in series with the ferrite, you will obtain damping as in Figure 2 (right), shown in blue.
Note how rise time is degraded with this solution. In my opinion, this solution is not very good. You are using a ferrite to include resistive losses in high frequency and then you add more resistive effect in series for low, medium, and high frequencies so signal degradation is for sure. Can we avoid the ringing in other way?
To understand what is going wrong here let me analyze the ferrite we are using in our example. In Figure 3 (left), we have included the impedance versus frequency behavior of the ferrite used in previous examples (ferrite 1).
Note that ferrite 1 has more reactive than resistive impedance at frequencies below 400MHz. At 170MHz that ferrite offers 12Ω resistive versus 30Ω inductive. The inductive effect has a dominant effect in Z and an important resonating effect with the IC2 load capacitance appears.
When choosing a ferrite, we must try to add resistive losses, so let me show you what happens with a new ferrite with a similar package. Impedance of ferrite 2 is included in Figure 3 (right). Note that, at 170MHz, the new ferrite has a resistive effect of 20Ω compared with a smaller inductive effect. Now, full impedance Z is almost due to the resistive effect. That is because this ferrite creates the results in Figure 4, WITHOUT any damping resistor.
Note that now ringing is not created and signal degradation (increasing rise and fall times) is better than using a damping resistor.
So, be careful when including a ferrite in your designs if, at the frequency of interest, that ferrite has a big reactive (inductive) effect.
Try another reference with more losses.
