In my last blog we looked at the noise entering the ADC from the power supplies and how we quantify the effects. We looked at how to measure the noise and see how the noise entering the ADC through the power supplies manifests itself in the output FFT.
Now let's move on to look at noise in the analog inputs and common mode voltage circuit of the ADC. Even though the common mode voltage circuit is more of a power supply type of circuit we will examine it along with the analog inputs as it is used to feed the common mode voltage for the analog inputs of the ADC.
Typically, when specifying the performance of an ADC, a very low noise input source is used along with a low bandwidth, high performance band pass filter. In this case, the noise from the input source can be neglected when measuring the SNR (which is one indicator of ADC noise) from the FFT of the output digital data. However, when utilizing an ADC in a real system the analog inputs are typically driven by an amplifier and an anti-alias filter (AAF) is employed. See Figure 1.
The driver amplifier will be a source that contributes noise to the system. The AAF, since it is made from surface mount passive components, will not have brick wall performance and will not have the rejection of an external high performance filter. The AAF will be tuned for a certain system bandwidth as well. All these factors must be considered. So let's now take a look at how we can calculate the impact from the amplifier noise on the overall SNR. There are a few things we need to know.
Let's look at an example situation. We'll take a pretty high level view of the overall process of calculations. If you'd like to see more details on the calculations please send me an email and I can pass along the spreadsheet that I used for this example. Consider the AD9643 driven by the ADL5202 at its maximum gain setting of 20dB. The band of interest is in the second Nyquist zone of the ADC (between 125MHz and 250MHz). The AAF is designed for a 1dB bandwidth of 75MHz from 145MHz up to 220MHz.
The 3dB bandwidth of the filter is 110MHz from 120MHz up to 230MHz. First, we need to know the amplifier output noise. This is typically stated in the data sheet for the device and is usually given in nV/√Hz. Sometimes only the noise figure of the amplifier is given and that must be converted to nV/√Hz. In the case of the ADL5202, the noise figure is provided in the datasheet and in the approximate band of interest, this value is 7.5dB. If we convert that over the nV/√Hz we will find that the amplifier output noise with a 150Ω output termination is 3.737nV/√Hz.
In part 2 of this blog, we will continue the analysis and calculations.
- ADC Noise: Where Does It Come From?
- ADC Noise: A Second Look, Part 1
- ADC Noise: A Second Look, Part 2
- ADC Power Supply Noise: PSRR & PSMR
- Interleaving Spurs: Offset Mismatches
- Interleaving Spurs: Gain Mismatches
- Interleaving Spurs: Timing Mismatches
- Interleaving Spurs: Bandwidth Mismatches
- Interleaving Spurs: Let’s Look at the Math
- Interleaving Spurs: More Math Details for Gain Mismatch
- Interleaving Spurs: The Mathmatics of Timing Mismatch