ADS1256: Seeing a lot of noise when reading small amplitude signals

Hi User,

I am looking into this issue and will respond in the next 1-2 days.

-Bryan

Hi User,

Just as a quick note, the signal you are trying to measure (0.287 µV) is below the level of the ADS1256 noise at G=1, 50 SPS, buffer off, which is 0.644µVRMS according to Table 4. Is there any reason you are not using a high gain given such a small input signal?

Also, the 4.7MΩ resistor will contribute 1.31µVRMS at T=25C and BW = 22.1 Hz (the BW of the digital filter at DR=50SPS in Table 12). So this is already adding a lot of noise to your system, more so than your actual signal.

Is there any reason you are dividing down your signal so much? This is certainly not helping your system's noise performance.

-Bryan

Hi Bryan,

For some reason, I have better performance (reading) whenever I use a gain 1. I also tried with buffer ON and it made everything even worse.

The reason I'm using this voltage divider is generate a very small signal. The smallest voltage pk-pk my function generator can generate is 135mV pk-pk but I need smaller amplitude voltage in order of micro volts. Is there a better way of generating micro volts pk-pk signals from a milli volts pk-pk without using a voltage divider? 

How did find the 1.31µVRMS noise at T=25C  that the 4.7MΩ resistor will contribute to the system?

Hi User,

I don't think using a resistor divider is a bad way to reduce the signal amplitude, but in this particular case you are reducing the signal below the system noise floor, so you will not be able to measure it. Can you try making the signal larger and see if everything works okay? Maybe see if you can measure the original signal (135mV) and then work your way down from there. This will at least confirm your system is operating properly, and that this is simply a noise issue. You can also try sampling slower (e.g. 2.5 SPS) and with a higher gain (e.g. 64 V/V)

The resistor noise equation is sqrt(4*Kb*T*R*F). You can find calculators on Google that will perform this calculation for you to determine the RMS voltage noise contributed by the resistor.

-Bryan

HI Bryan,

I've already tried larger voltages down to 10mV pk-pk and it seemed to work fine. Whenever I go smaller I get a lot of noise in the signal.

I've also tried with a DC power source (GW Intek Programmable Power Supply PSP-405 ) and surprisingly I got the same behavior (PGA =1 gives me the best performance with buffer OFF). I fed a 0.01 V DC voltage to the  AIN0 and ACOM of the ADS1256  and according to the datasheet with PGA = 64 I should be to read ±78.125mV pk-pk but with this 0.01 V DC voltage, I'm reading the max count which is 0x7fffff or 8388607 with buffer OFF. I get this max count with PGA = 32 and 64 with buffer OFF.

Is it normal that I have this kind of behavior when reading a DC value? 

Hi User,

For the 10mV DC signal, do you get the same result with the buffer on?

-Bryan

Hi Bryan,

For the 10mV DC signal, I have relatively better reading with the buffer on. All PGA values from 1 to 64 (with buffer on) give me a value close to 0.0106V which is better than what I get with buffer off. With buffer off, I tend to have better reading when the date rate is high. For example, with PGA = 1 (with buffer off) and date rate = 100SPS, I get 0.0201V for a 10mV signal whereas with data rate of 30000SPS, I get 0.0186V which is strange because I was expecting to see a lot more noise on the signal with higher data rate.

I  also noticed something strange. When I divide the 10mV DC signal with 10K and  and 4.7M (the output being taken on the the 10K resistor) I get the same reading which roughly 1.3010576367378235e-05 V (1397 in decimal or 0x575)(with buffer on). For this reading, I set PGA = 64 and data rate =10SPS (with buffer on). I was supposed to get 3.021148038e-5 I'm not sure what's the problem.

Hi Bryan,

I did measure the signal going to the ADS1256 with a DMM and the value I'm reading on the DMM corresponds to what I'm getting with the ADS1256 (with buffer on, ACAL enabled, common mode). But I can only check close few mV which is the maximum resolution for my DMM.

For all of my readings, I have the Auto-Calibration enabled. 

If I do level-shift the signal to mid-supply as you suggest am I still going to be able to use PGA = 8 to 64? Since their pk-pk amplitude is less than 2.5V or 1.5V as you suggested (±0.625V for PGA = 8 means the range is from 0V to 1.25V if reference voltage = 2.5V)

I also seem to have better performance when using common mode versus differential mode. I'm feeding a 0.01V DC common signal to the ADS1256 and I get 0.0105V using the common mode and 0.036V when using differential mode (PGA = 64 and Data rate = 10SPS, ACAL enabled and buffer ON, reference voltage = 2.5V). Is this normal? I thought differential reading should provide better accuracy than common mode. 

Hi User,

I saw that you clicked thread resolved, so I hope that you have figured out what you needed to make your system work. But I did want to address your last concerns:

Can you explain what you mean by common versus differential mode? The ADS1256 always takes measurements in a true differential manner (AINP - AINN), so this doesn't really apply to this ADC like it would with some SAR ADCs. Check out video 2.1 in the Precision Labs series that describes the different input signal types, in case my nomenclature is unclear. Measuring a ground-referenced input would be called a single-ended measurement - but the ADS1256 is a true differential input ADC, and its coding scheme reflects as such.

For the setup, I am suggesting something as shown below. You would need to add a 1.5V offset from your function generator to the 10mV signal, as shown below. Then apply 1.5 V to AINN on the ADS1256 (this assumes the buffer is on). So when you apply +10mV, AINP = 1.6V and AINN = 1.5V, so the differential signal seen by the ADC is +10mV. You can then see how -10mV would be similarly generated.

Again, you should generally try to stay away from the input limits on your devices, and it is best practice to keep your signals centered at the midpoint of the supply range (there are some exceptions to this, so you can consider this a general recommendation).

If you have any more questions on this topic, please let me know. If you run into additional challenges, please start a new thread and we will assist you.

-Bryan

Hi Bryan,

In the datasheet, it says to "use AINCOM as common input and AIN0 through AIN7 as single-ended inputs" for single-ended measurement and to "use AIN0 through AIN7, preferably adjacent inputs" for differential measurements. What I mean by single-ended mode is actually single-ended measurement that is I use AINCOM as common input.

The signal coming from the function generator is ground-referenced I assume that is why I have better performance when doing single-ended measurements versus differential measurements. 

My confusion in the previous message was that I thought if I use an offset of 1.5 V one terminal is going to be at 1.51V and the other terminal will be at 0V (Ground) since the function generator generates ground-referenced signals. But now that you explained how the ADS1256 works I don't think there will be a problem.

Thanks

Hi User,

Thanks for confirming, looks like we were talking about the same thing then: single-ended measurements. Note that you cannot enable the buffer and take single-ended measurements, since any variation in the GND potential could violate the ADS1256's datasheet specs for absolute input range.

If your signal coming from the function generator is ground-referenced then I agree you would not be able to level shift it as is. That is why you would need the 1.5V offset. Or, where adding a DC offset is not possible, you can use a level shift circuit such as this one using an inverting amp or this one using a non-inverting amp to bias the signal to a specific voltage. These circuits also attenuate the inputs as well e.g. for +/-10V inputs, that many ADCs cannot natively accept.

-Bryan

Hi Bryan,

I don't understand why I can't take single-ended measurements with the buffer enabled. Should the buffer be enabled only with differential measurements?

My input signal to the ADS1256 has a high input impedance and I won't be able to properly read the signal with the buffer off.  If I connect the ground terminal (probe) of the signal generator to AINCOM is there still going to be a violation on the Ground potential?

Thanks