SPRADT0 February 2025 TMS320F280049C , TMS320F280049C-Q1
To reduce cost, some applications with slowly changing input signals, particularly DC voltage sampling applications, often eliminate the amplifier circuit. In such cases, Cs must be designed sufficiently large to ensure adequate sampling accuracy.
To achieve a settling error of 0.5LSB, the capacitance value of Cs can be calculated using the following formula:
For a 12-bit ADC, Cs should be 213 times CH. If CH is 7.5pF, then Cs should be at least 61nF.
At this point, Cs is large enough that its stored charge alone can charge and discharge CH, and the resulting voltage fluctuation remains within the 0.5LSB error range. No real-time charge replenishment from the external circuit is required. Regardless of the external filter resistor Rs value or the sample-and-hold time, the single-sample accuracy requirement can be met.
However, for repeated oversampling or very high sampling frequencies, excessive Rs values can still affect sampling accuracy. Therefore, the design must also consider sampling frequency and other factors.
In theory, the sample-and-hold time can be set to the minimum value specified in the datasheet. However, because of parasitic inductance in the circuit, oscillation may occur in the voltage across the internal ADC capacitor. Therefore, sufficient margin should be reserved by increasing the sample-and-hold time appropriately.
Experimental Verification:
A commonly used resistor-divider circuit is shown on the left side of Figure 4-3. According to Thevenin's theorem, its equivalent resistance can be calculated as:
The equivalent Rs is essentially determined by the resistor-divider circuit. The resistance value cannot be too small; otherwise, power consumption and heat generation will increase. Cs is the adjustable parameter. In the following experiment, different Cs values were selected while using the same AQCP value of 59 (approximately 600ns sample-and-hold time), producing the following sampling results.
| ADC Sampling Channel | Input Voltage | Rs | Cs | Sampling Frequency | Actual Sampled Value (LSB) | Theoretical Sampled Value (LSB) |
|---|---|---|---|---|---|---|
| A1 | 2.0V | 10K | 100nF | 10khz | 2480 | 2482 |
| A1 | 2.0V | 10K | 10nF | 10khz | 2479 | 2482 |
| A1 | 2.0V | 10K | 1nF | 10khz | 2462 | 2482 |
| A1 | 2.0V | 10K | 10pF | 10khz | 2417 | 2482 |
It can be seen that when the capacitor value is large, the sampled value is relatively accurate, whereas when the capacitor value is small, the error is significant, consistent with theoretical predictions.
Therefore, selecting an appropriate capacitor Cs has a substantial impact on sampling accuracy. In practice, the value of Cs must also be considered together with cutoff frequency, sampling frequency, and other factors.
If the sampling frequency is too high, sampling accuracy will also be affected. The following table shows sampled values obtained using the same RC parameters at different sampling frequencies.
| ADC Sampling Channel | Input Voltage | Rs | Cs | Sampling Frequency | Actual Sampled Value (LSB) | Theoretical Sampled Value (LSB) |
|---|---|---|---|---|---|---|
| A1 | 2.0V | 10K | 10nF | 10khz | 2479 | 2482 |
| A1 | 2.0V | 10K | 10nF | 100khz | 2459 | 2482 |
| A1 | 2.0V | 10K | 10nF | 200khz | 2434 | 2482 |
As shown above, higher sampling frequencies result in lower sampling accuracy. Therefore, for applications requiring high cutoff frequencies and high sampling rates, amplifier-based circuits should be considered to improve accuracy.