ADS1220EVM: How to reduce ADS1220 EVM noise

Hi Jeff,

Can you give us some more detail on the sensor? Are you using any RC input filtering? The noise tables on page 16 of the ADS1220 datasheet show typical numbers of 151.61 uVpp for gain of 1 at 600 sps, and 106.93 uVpp for 330 sps. In the table values we use a period of time (0.75 s) for collecting samples and not necessarily a fixed numbers of samples for the ADS1220. This will make a difference in your results in comparison to the tables. The reason for doing this is to avoid the effects of drift within a series of measurements. The values shown in the noise tables is the best the converter can accomplish using shorted inputs. This is the noise of the converter and conversion process. The spreadsheet is actually calculating not just 512 samples, but rather 2048 samples in the min/max values calculation. By reducing the sample size you will see a much closer number as given in the tables as typical values.

Further reduction in noise is possible by averaging, but this will affect the throughput rate. One way to make averaging more effective is to use a moving average. There are other methods of averaging as well. The moving average should yield results similar to the 175 sps data rate, by averaging 4 samples. The average is calculated with a FIFO buffer recalculating the average at the desired data rate. It would take 4 samples to complete the first complete average, so the first three samples would see higher noise, but after that the noise should be much lower.

You may have noticed that the input referred noise lowers with applied PGA gain. So using gain is also effective in lowering the noise. Slower data rates also lower the noise. There is one other method that can be used to lower noise and that is to use turbo mode followed by averaging. You can achieve better noise performance by using turbo mode at 1200 sps and averaging 2 samples (for 600 sps throughput rate) than you can from using 600 sps in normal mode. Turbo mode increases the PGA bandwidth (at the expense of more power consumption) that places the PGA in a lower noise mode.

It also appears that there may be some noise canceling similar to a ratiometric measurement when using the external 3.3 V source as compared to the internal reference. If this is the case, the noise may be improved slightly with a lower noise excitation source for the sensor.

Best regards,
Bob B