ADS1115 PGA

Wayne,


The inputs for the ADS1115 are not designed to take an AC coupled signal. The inputs will take the voltage it sees for the measurement.

If you connect the CH1+ input to 2.05V through a 10k resistor, the input impedance of the ADC is so high, that it will still look like 2.05V to the ADC. Figure 26 shows the basic equivalent circuit showing how it is sampled, and Table 2 shows the equivalent input impedance.

In scenario 1, the input would see the 2.05V, a PGA setting of 2 would likely put you at the full scale.

In scenario 2, you might be elminating the DC voltage, but the frequency response of the ADC (as shown in Figure 21) would limit the usable frequency range of the device. Still, I don't think you'd be able to couple in enough AC signal to measure anything.

I think it would be better to do some sort of signal conditioning on the front end to get you the input you need to measure with this device. If you let me know what it is that you are measuring. I might be able to help you better.


Joseph Wu

Thanks for the quick response. Please take a look at the attached image.

This circuit will convert a single-ended input to a pair of inverted signals that can drive the ADS1115.

The low-frequency signal is AC coupled to an unit-gain inverting op-amp. The capacitor C1 is required because the AC signal is originating from a sensor that has a static DC offset.

The inverted output of the first op-amp feeds the CH1- input to the ADS1115. This signal is also used to provide an input to the second unity-gain inverting op-amp. The output of the second op-amp then feeds the CH1+ input on the ADS1115.

Both op-amps are referenced to 2.5V. The DC offset voltage at both outputs will be 2.5V.

Will this work?

I am assuming that the gain of the PGA is a multiplier of the DIFFERENTIAL between CH1+ and CH1-. If this is the case, then all of the gains in the PGA should work properly because the offset voltage on CH1+ and CH1- is the same (2.5V).

If this circuit will work, would you recommend using a 1uF cap prior to the second op-amp.

Thanks again.

I wanted to provide an update. My circuit is working well. I use an op-amp in a cross-coupled configuration to convert the single-ended input into a differential output.

The sensor is AC-coupled to the op-am with a 1 uF cap. The Fc based on the components I am using is 7.95 Hz.

For my application, I wanted to eliminate the DC input component prior to feeding the signal to the op-amp and converter because DC offset is amplified by the PGA of the converter. I need to maximize my dynamic range and this technique seems to work well. With the values I have selected, the DC offset is 22 counts (out of 65535) for unity gain. I may try some 0.1% resistor values to see if I can trim the offset to be closer to 0.

I take 512 SPS with the converter. I then use a 512 point FFT. This gives me 1 Hz frequency resolution with a max frequency of 256 Hz. I am only interested in 10-100 Hz.

I don't know WHY TI selected free-running speeds such as 250 (instead of 256) and 860 (instead of 512). Using acquisition speeds that are not powers of 2 makes it a real pain to acquire and process via FFT. For example, instead of allowing the converter to run at 512 SPS (where I could monitor the RDY pin and then read the conversion value), I must now SEND a request to the device for a single conversion and then wait for the conversion 512 times per second.

I thought about running the device in continuous mode at 860 SPS and then reading the conversion results at 512 times per second. I am not sure how (if any) this would affect the FFT. I sure would like to do this if possible. Please let me know your thoughts on this.

Thanks!

Wayne,


I'm glad you were able to get your system running.

I believe that the ADS1115 should work the way you describe. It's basically operating the device at 512 SPS in single shot mode each time. To run in single shot, there's about a 20us needed to wake up the device before starting the conversion.

I don't think this would affect the FFT much. There may be some errors based on the internal oscillator (which would be more important if you're going over a wide temperature range).

Incidentally, I understand that the data rate selections are a bit unusual. This is due to the nature of the ADC. The operation of the ADC significantly changes while changing the data rate. There's different switching and digital overhead.


Joseph Wu

Joseph,

I have two options...

1. Start single-shot conversion and repeat 512 times per second.

>LOOP Start single conversion

>Wait .001953 seconds (1/512 seconds)

>LOOP Read conversion results and start over

2. Configure the device for continuous acquisition at 860 SPS and then read the latest conversion 512 times per second.

>Start Continuous Measurement at 860 SPS

>LOOP Wait .001953 seconds (1/512 seconds)

>LOOP Read conversion results and start over

The second method is easier because I do not have to start the conversion for each sample. I would think that the results would be almost identical between these two methods but I am not sure. What are your thoughts?

Joseph,

That is my thought as well. That is the method I will adopt.

Thanks for all of your assistance.