2 Parameters Affected when Cascading Multiplexers

On-Resistance (R_(ON))

The On-resistance(R_(ON)) of a multiplexer is the measured resistance from the input to the output of a multiplexer. In simulation, this can be modeled as simply a resistor in series. This resistance plays a role in many mux characteristics such as the insertion-loss, propagation delay and bandwidth. As R_(ON) increases, performance in these specifications decreases. The typical preference is to pick a multiplexer with a lower R_(ON) when cascading multiplexers since R_(ON) adds linearly as you increase the multiplexer count. Figure 2-1 shows how the insertion-loss increases linearly as you pass a signal through 2 multiplexers. Note how after the 1st mux you lose 1X of your 5V signal and after the 2nd mux 2X loss. This pattern continues through the n^th mux in series.

A key distinction to make here is the effect of the load on R_(ON). Regardless of the loading, losses accumulate linearly; however, when feeding the output of a multiplexer into a high impedance node the losses due to R_(ON) are suppressed by the load. This means when cascading into a high-impedance load, there is less impact from R_(ON) as the loss per cascade stage is minimized. Conversely, when the system has a smaller resistive load, a lower R_(ON) mux is preferred as you’ll accumulate more losses per mux the higher the R_(ON) of the multiplexer is. Figure 2-2 highlights the advantage of using a low R_(ON) multiplexer, such as the TMUX1574, in cascaded mux applications.

The center and right images above highlight the advantages of using a low R_(ON) multiplexer (right),such as the TMUX1574, versus a higher R_(ON) multiplexer (center) when the system resistive load is small. The far left image shows how the R_(ON) of the multiplexer has less impact when fed into a high impedance load. The large load suppresses the On-Resistance so the losses due to R_(ON) aren’t seen as significantly.

Bandwidth

The bandwidth of a multiplexer is the frequency range of signals that can pass through the multiplexer with no more than a -3dB loss relative to the DC gain. This equates to approximately 70% of the signal remaining, relative to the insertion loss at DC. The loss is associated with the frequency of the input signal. A key distinction needs to be made and understood here between the -3dB point and the -3dB loss point. Figure 2-4 below shows a couple examples to help with the understanding this definition. Here, the DC gain shifts where the -3dB loss point is measured from. This point won't always be at -3dB but rather 3dB less than the insertion loss at DC.

Notice how the bandwidth of example 2 (400MHz) is higher than example 1 (190MHz) but the overall loss (-3.8dB) is greater than example 1 (-3.1dB). See what the loss is at the given operating frequency since the bandwidth alone won’t tell you this. Typically, a multiplexer with a bandwidth close to or equal to the working frequency won’t be a good fit for minimizing losses in the system. A good strategy is to select a multiplexer with 1.5x – 2x the Nyquist frequency (twice your operating frequency) for sine waves and 5x-7x the fundamental (operating) frequency for square waves. This helps maintain that more of the signal gets retained when passing through the mux.

To help simulate the bandwidth, a multiplexer can be effectively modeled as a low-pass filter with some input capacitance by using On-Resistance and On-Capacitance as shown in Figure 2-5. These values can be pulled from the data sheet. For ease of modeling, the On-Capacitance is split in two on each side of the resistor representing On-Resistance.

As more multiplexer stages are added to create the cascading network, this creates a higher-order filter. As higher order filters are created, more higher frequency components are filtered out from the input signal. Figure 2-6 shows simulated results for how the -3dB bandwidth decreases when a 2nd and 3rd multiplexer stages are added. Since a high-impedance load is used we can estimate the DC loss is close to 0dB and can draw the -3dB loss point at approximately -3dB.

Figure 2-6 shows that the decay of the signal isn’t linearly proportionate to the number of multiplexers in the cascaded stage. Instead, the decay becomes less significant as more multiplexers are added. Figure 2-7 (high impedance load) and Figure 2-8 (low impedance load) show simulation results as we increase from a single mux up to 8 multiplexers in series. As previously mentioned, the multiplexers act as a filter and exhibit similar properties. Therefore, expect that as we approach higher-orders we see a sharper slope in the roll off. This in turn suppresses the higher frequencies and limits the bandwidth. While the -3dB bandwidths between low and high impedance loads are similar, the overall loss gets greater in low-impedance load systems as more multiplexers are added due to the On-Resistance contributions previously mentioned in the section covering On-Resistance (R_(ON)). This stands out more when operating near DC levels. For example, at 100kHz with a high-impedance load, by the 8th mux, the gain remains at nearly 0dB, equating to most of the signal being retained. At the same frequency with a low impedance nearly 20% of the signal is lost by the 8th multiplexer.