Blog Post created by tschmitt on Aug 15, 2017

Welcome back to the ADAQ798x ADC driver configuration blog series! In today’s entry, we’ll look at one of the configurations that can be used to interface the ADAQ798x to bipolar sensors and input sources. These types of signals are common in industrial and data acquisition applications. This configuration builds on the non-inverting configuration we discussed previously to convert a bipolar signal to a unipolar one for the integrated ADC.

The Non-inverting Summing Configuration

Bipolar signals swing above and below ground (0 V). Since the ADAQ798x’s integrated ADC can only convert signals between 0 V and VREF, bipolar signals need to be dc-biased and properly scaled for the ADC. The following configuration accomplishes this by adding two resistors (R1 and R2) to the standard non-inverting configuration.

This configuration performs bipolar to unipolar conversion by summing the input signal with a separate dc voltage to bias the ADC driver’s output to the ADC’s midscale input (VREF/2). Using the reference source (VREF) as the dc voltage is often practical, as it eliminates the need for additional circuitry (the ADAQ798x is always accompanied by a reference source anyway!). It also prevents deviations in VREF from adding offset error to the system, since the ADC driver’s dc bias will always be half of VREF. For these reasons, we will look specifically at this configuration utilizing VREF as the dc “shifting” voltage.

The transfer function for this configuration is:

Similarly to the regular non-inverting configuration, the ratio Rf and Rg determines the gain from IN+ to AMP_OUT, but this ratio now depends on the input amplitude of vIN as well. Note that vIN is bipolar, but the voltage on the non-inverting node is unipolar. That means that for the minimum value of vIN, the voltage on IN+ must be 0 V:

This relationship gives the ratio of R1 to R2:

Rf and Rg can be determined using the configuration’s transfer function and the condition that the output of the ADC driver (vAMP_OUT) is equal to VREF/2 when vIN is 0 V. Solving this equation for Rf and Rg gives:

We now have the ratios of R1 to R2 and Rf to Rg, but we still need to pick specific values. We addressed selecting Rf and Rg values in our previous post. R1 and R2 selection should be determined based on the application’s noise, accuracy, and input impedance requirements. Small resistances will improve noise and can reduce offset errors caused by its interaction with the ADC driver’s input bias current (see MT-038 and CN-0393), but large resistances are required to increase input impedance and reduce the output current of the reference source. The input impedance of this circuit is:

Note that for the specific case where the amplitude of vIN is ±VREF, the ratio of Rf to Rg is 0. In this case, the ADC driver gain is 1, meaning Rg is omitted and Rf can be 0 Ω.

Let’s look at an example where the ADAQ7980 needs to perform bipolar-to-unipolar conversion of a ±1 V input signal, with VREF = 5 V and using Rf = 2 kΩ. Using the above equations, R2 must be 5 times R1 and Rf must be 2 times Rg. Since Rf is 2 kΩ, Rg must be 1 kΩ. Specific values of R1 and R2 can be selected depending on the application’s requirements. For this example, we’ll aim to select a combination of R1 and R2 that negates the effects of the input bias current on offset error. MT-038 explains that R1||R2 should be equal to Rf||Rg to achieve this, which gives R1 = 800 Ω and R2 = 4 kΩ.

But let’s also consider an example where vIN = ±10 V with VREF = 5 V. In this case, we run into a problem where the ratio of Rf and Rg is a negative number, so we can’t actually achieve this input range with this configuration. In fact, the largest vIN that will work with this configuration is ±VREF, where the ADC driver gain is equal to 1. Luckily, we’ll be looking at two other configurations that will allow us to extend past this input range in future entries to this series!

Closing Thoughts

The above configuration can also be used for unipolar signals by connecting R2 to ground instead of VREF. This modification is useful for unipolar input signals that need to be attenuated for the ADC (with amplitude >VREF). In this case, the ADC driver will most likely be in unity gain, so Rf and Rg are not necessary.

As mentioned above, if the application requires a high input impedance, R1 and R2 must be large, which can increase the noise floor of the system. We can compensate for the increased noise with the addition of a shunt capacitor and/or by oversampling and decimating. Both of these options sacrifice input signal bandwidth to reduce the noise floor. For low-bandwidth or dc applications, however, the input bandwidth is not as important. For this reason, these configurations are better suited for low-bandwidth, high input impedance applications. We'll discuss this in more detail in our next post.

One problem this does not address, however, is the offset error caused by the ADC driver’s input bias current flowing across the resistors. Large resistances result in large dc errors. This error can be reduced at the expense of some input range by adjusting the ratio of R1 and R2 to compensate for the undesired voltage drop, or by selecting Rf and Rg values that cancel out the offset caused by R1 and R2. However, keep in mind that Rf must be small enough to ensure amplifier stability, so the second option is not always viable.

Thanks again for joining me in this blog series! Next week, we’ll look at another modification to the non-inverting configuration designed for use with bipolar input signals that are too large for the one we discussed today. Follow the EngineerZone Spotlight to be notified when the next addition to this series is available!

Have any questions? Feel free to ask in the comments section below!