Solution to inaccurate AD sampling

Aguilera

Solution to inaccurate AD sampling [Copy link]

1: The reference voltage needs to be accurate enough. It is recommended to use an external high-precision reference voltage.
2: If the PGA is adjustable, the smaller the gain coefficient is, the lower the noise is.
3: It is generally best to use the full scale, so that the AD accuracy is not wasted.
4: If there is an offset, self-calibration is required.
5: Please note that when debugging with the DEMO board, PC noise will be introduced by the debug port and external noise will be introduced by the signal connection line. Therefore, it is recommended to use a shielded cable to transmit the signal.
6: Pay attention to the separation of analog power supply and digital power supply, as well as analog ground and digital ground on the board to reduce the coupling noise path.
7: Using differential input can reduce common-mode noise, but differential-mode noise will increase.
8: If it is an MCU with integrated AD on the chip, it supports high-speed clock. If it does not affect the performance, the lower the internal working clock, the less interference it causes to your AD sampling. If it is on the board, you need to pay attention to routing and partitioning.
9: The signal input is connected to the filter circuit before the stage. Generally, there are more first-order RC circuits. Note that Fc=1/1000~1/100 Pay attention to the sampling frequency, resistor and capacitor parameters. It is best to use sinc filtering when the signal is connected to the post-stage filter circuit. Note that the input bias current will limit the value of your external filter resistor.
R x Ib < 1LSB.
Some on-chip ADs also have integrated input buffers, which help suppress your noise. Generally, there are two types of ADs, depending on the relationship between the input signal range and the full scale.
ADs are divided into many types, SAR, FLASH, parallel comparison type, successive approximation type, Delta sigma type. Generally, the higher the speed and the higher the accuracy, the more expensive it is. Therefore, companies like ADI have always been so rich and made black money...
I personally think: you also have to pay attention to your layout. It is very important.

If you do the above points, your resolution will be improved by several digits.

When using AD chips, many issues must be considered, but I think the most important ones are as follows:
1. The conversion rate of the AD chip is generally measured in KSPS or MSPS.
2. Full-scale voltage. Currently, chips with 1V and 5V full-scale are commonly used, and there are also some other specifications. Some devices have variable full-scale, which usually changes according to the reference voltage. There is no such thing as tens of millivolts. In order to achieve satisfactory resolution (due to the influence of AD chip integral error, differential error and offset error, it is best to amplify the signal peak to more than 90% of the full scale before sending it to AD), the signal must be amplified.
3. The rise rate of the amplifier must meet the signal bandwidth requirements.
4. The power supply filtering must be good. The analog signal power supply must be separated from the digital signal power supply (there are many ways, mainly isolating the high-frequency component), and the ground wire is the same.
5. Pay attention to the analog signal feeding method, which generally has two types: single-ended and differential. Usually, differential feeding needs to be slightly optimized in SNR. However, the circuit is more complicated.
6. The reference voltage should be relatively clean, stable and accurate. It can reduce the error of AD conversion.
7. Some high-speed ADs consume a lot of power, so pay attention to the power supply capacity.