Non-linear behavior through switch in TIA loop

Hi,

Thanks for getting in touch. Because the non linearity is centered around 2.5V, which is also the center of the supply, my first instinct is the on resistance. Not the absolute value of the resistor (82k or 820k)  but the on resistance of the ADG1201. The ADG1201 has an On Resistance of 120 Ohms typically but that is at ±15V dual supply. As the supply voltage decreases, on resistance increases and you are using the very minimum supply voltage of 5V single supply.  The other parameter that increases significantly with lower supply voltage is the on resistance flatness, or how flat the switch resistance is across the signal voltage range. You can see from Figures 3, 4 & 5 in the datasheet the relationship between supply voltage, On resistance and On resistance flatness. There is not a plot for 5V single supply on resistance in the datasheet but it will  be higher than the Ron in Figure 4. Also as the supply voltage decreases the 2 peaks seed in figure 4 tend to merge to one peak giving a single point of max resistance in the center of the signal range.

Saying all that using some quick calculations even if the switch on resistance varied by 2k Ohms ~(which would be an exaggeration) it should not make as big of a difference as what you are seeing.

One quick way to rule it out though would be to boost the Vdd supply of the ADG1201 if that is possible in your design.

Let me know if this makes a difference.

David

Upping Vdd does eliminate the behavior. With Vdd=5V you get severe/high switch resistance when S,D are anywhere near mid-supply. I don't understand this really but the behavior and trend is clear enough.

In the attached I have three cyan traces with feedback resistors of 8.2K (most distorted), 82K, and 820K (least distorted) with Vdd=5V. I am sweeping the TIA input current such that the TIA output (voltage at S) is between 1.25..3.75

Then the final yellow trace is with 8.2K but when Vdd is raised to 12V which fixed the distortion.(i don't have the ability to easily try arbitrary value of supply voltage)

I think Analog should not include 5V as an allowed supply voltage.

Thank you for sharing your results.

You could have a look at some of our switches from the ADG7xx family for example the ADG741. These switches are better optimised for 5V single supply applications.

Regards,

David