When it comes to the topic of how to properly driver the front end of your SAR ADC, ADI provides quite a bit of guidance on the matter. For example, you'll often find a list of "Recommended Driver Amplifiers" in the datasheet for a SAR ADC - like this table on page 16 of the AD7980 datasheet:
Also in the datasheet, there is often guidance on a good starting point for the RC filter component values - as shown on page 15 of the AD7980 datasheet:
Other sources of ADC driver guidance can also be found in:
- The "Product Recommendations" section of the ADC product page on analog.com
- The SAR ADC and Driver Quick Match Guide
For a deeper dive on this topic, Front-End Amplifier and RC Filter Design for a Precision SAR Analog-to-Digital Converter is an excellent resource. This article will help you to understand the basic considerations of SAR ADC driving, and will help you to tailor your driver circuit design to better match your specific application.
We've been working to incorporate some of the knowledge and calculations we have across all of these resources into a new web tool - we have a "rough draft" version of this tool available for initial testing and feedback:
playground.analog.com/sardriverproto
We're going to keep working on this tool, making improvements to both the functionality and the look-and-feel. But the feedback we receive on this early version of the tool is going to be very helpful as we decide which features we're going to work on. Feedback and discussion on this tool can be provided in the comments section at the bottom of this post, or by clicking "Help us make this tool better" at the top right corner of the tool window.
To use this tool, you begin by providing your circuit design information - selected ADC and sample rate, selected amplifier/driver and circuit information, input signal frequency, and RC filter values.
The tool will generate illustrations/plots for these settings on four tabs. The "Circuit" tab will illustrate the circuit information that was provided in the input panel. For example, if we start with the rceommended RC value in the AD7980 datasheet, and use one of the recommended drivers, we'll have input settings and a circuit tab that looks like this:
Note the input values - AD7980 at 1MSPS, ADA4805 non-inverting, gain of one, 10kHz input frequency, RC filter values of 20 Ohms and 2.7nF.
Pro tip: as you're using this tool, the URL will continually update to reflect the configuration you have entered. If you want to "save" what you're doing in the tool, just copy-and-paste the URL, and use it to reopen the tool where you left off.
Once you have your circuit set up, you can take a look at the other tabs to view the performance. The "Noise and Distortion" tab will display and estimate of the combined performance of the ADC and driver. In this early version of the tool, you can see the ADC performance vs. input frequency is green, and the driver performance is blue. The combined performance is red, and is also summarized in the "Noise Performance" table at the bottom of the tool page.
In this example, you can see that the driver is not degrading the performance of the overall system very much. The system performance is dominated by the ADC. You can also compare the tool estimate to the datasheet SINAD plot:
Looking at the "Sine Response" tab, you can see that the sinusoidal error is very small (less than 1/2 LSB) - this means the RC values have been chosen such that the ADC input is able to properly settle after each kickback event, prior to the end of the acquisition cycle.
Taking a look at the "Step Response" tab - this tab simulates the settling behavior of the circuit assuming a multiplexed input, including the slew rate of the driver, and the effect of the RC filter.
With a multiplexed input, the input value can swing greatly between the mux input pins, and we can assume the worst case scenario is that the input voltage will swing the full range of the possible voltages. Looking at the "Step Response" tab, we can see that out current design isn't going to settle in time for a multiplexed setup. (See how the voltage is settling right at the end of the conversion cycle, but is not able to recover after the kickback in time for the end of the acquisition cycle?)
With this example, we can see that we don't quite settle from the kickback in time. If we tinker with the RC value and switch to an 18 ohm and 2.2nF value, the circuit will settle to 1/2 LSB in time.
Were you able to use the tool for your design? Did you find it helpful? What could we improve?