Comparison of THD in Analog Multiplexers

Hi Steve,

Currently, we are working on a THD calculator for our switches and multiplexers that should help people with similar questions.

You are correct. THD is directly related to RON Flatness and the device's on-resistance, circuit input resistance, and load resistance, as shown here: https://www.analog.com/media/en/technical-documentation/application-notes/AN-251.pdf 

To summarize what it says there, THD is proportional to Rflat / (Rin + Ron + Rload). Then, let's see the switches you are interested in:

• Rload = 110 Ω (Rin = 0 Ω)
  • ADG1407 -> proportion = 0.013389
  • ADG5208F -> proportion = 0.018056
• Rload = 10 kΩ (Rin = 0 Ω)
  • ADG1407 -> proportion = 0.00016
  • ADG5208F -> proportion = 0.000634

Looking at those proportions, even though the ADG5208F has a better (RON Flatness / RON) than the ADG1407, its RON is still very high and will worsen more the distortion than the ADG1407. Therefore, the ADG1407 will still have a better distortion performance than the ADG5208F.

That simple equation above and the AN-251 are a good way of comparing THD performance between devices. However, bear in mind that it is lacking any capacitance and frequency information that your circuit has. So, it should only be used as a guide.

Now, for the recommendation, you can have a look at the ADG1408 or ADG5408 (both 8:1 x 1, not 8:1 x 2), where the ADG1408 will have the best distortion performance overall.

• Rload = 110 Ω (Rin = 0 Ω)
  • ADG1408 -> proportion = 0.004386
  • ADG5408 -> proportion = 0.014575
• Rload = 10 kΩ (Rin = 0 Ω)
  • ADG1408 -> proportion = 4.998e-5
  • ADG5408 -> proportion = 0.00018

Please, let us know if any of these parts still do not meet your requirements.

Yours sincerely,

Igor Ono.

Thanks Igor. That's extremely useful information for me, both your excellent explanation and the app note. I will examine the various muxs in detail for my particular requirements.

Could  I ask a follow up question? Obviously as the signal amplitude approaches the power rails Ron increases and distortion will increase. I note that many of the datasheets have THD values where the signal is 50% of the rails (so say 15Vpp for +/-15V rails for example). Is there a general rule of thumb or recommended signal range for particular voltage rails? In my case I will have +/-13V power rails.

regards

Steve

Hi Steve,

If you are looking at extracting the best performance of the switch to minimize distortion, you should use voltage rails as high as possible (not the absolute maximum ratings), while keeping your signal range in a flatness region you can account for in your design.

Let's look at the +/-15 V supply case for the ADG1407. You can get a good RON flatness with an input signal of 15 Vpp (approx. 7.5 Ω to 8.25 Ω, Rflat = 0.75 Ω) when compared to 20 Vpp (approx. 7.5 Ω to 8.8 Ω, 1.3 Ω).

Then, you can see in this image below that reducing from 20 Vpp to 15 Vpp, gives you almost twice the THD+N improvement. (You can also see this in the equation above as well)

I am also adding the +/-5 V THD+N plot for comparison.

If you reduce the input voltage even further, you won't be reducing distortion as much. Also, you will be limiting your operating range even further.

The advantages of working with a fault protected part, such as the ADG5208F, is its RON flatness. Your signal can almost go to the supplies, while your THD will be kept roughly the same.

Regarding the 15 Vpp in our spec tables, this is standard for all of our parts. As you can see, some parts have a basically flat RON curve, that increasing the 15 Vpp would not reduce its THD+N performance.

You can use the input signal equal to 1/2 supply voltage as a quick way of calculating the "optimal" input voltage, or you share more info on your application so we try to find the best device and range. (Things that can help: input voltage range and its tolerance, frequency, load resistance and capacitance, input resistance, and schematic)

I hope this helps!

Yours sincerely,

Igor

Thanks Igor. After I replied I realized that i could i examine the plots that you show above. I have a lot to consider though because it's not a single mux application.

I would like to share the design with you and get your input on how best to optimize it. But it would need to be confidential. How best could I do this?

regards

Steve

Thanks Igor. After I replied I realized that i could i examine the plots that you show above. I have a lot to consider though because it's not a single mux application.

I would like to share the design with you and get your input on how best to optimize it. But it would need to be confidential. How best could I do this?

regards

Steve

Hi Steve,

I just sent you an email so we can continue discussing your circuit design.

Yours sincerely,

Igor