• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 3 answers
  • Subscribers 46 subscribers
  • Views 1861 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ads1258 input noise ? how to improve?

Ivan Chou61
Intellectual 350 points
Other Parts Discussed in Thread: ADS1258, OPA2320, OPA320

We now have input noise problem. We want to use ADS1258 to measure a thermistor value (R3 value) The schematics is as follows.

Use R3 = 5K to test

And we use AIN6 to sample ADC_VREP,  AIN7 to sample THERMO1_B

The following is the adc cnt of AIN6 and AIN7

AIN6:

AIN7:

we found the variation of AIN6 is much smaller than AIN7.  Roughly, Adc cnts are 300 v.s. 3600.  

For impedance issue ( R3 & R4 values are big at thermistor circuit), we add opamp before input AIN7, but the noise still exist.

We dont know the source of the noise

Something I missed? any advice to improve noises?  

BTW, we get the average of 200 samples' values, and the average looks reasonable....

2. Another problem is that  there is a case that we get NEGATIVE adc count in data reading even if we use unipolar operation.

for example, from AIN7, we get adc cnt +7862690,+7862690, +7862690, -7262690, 7862690, 7862690

each channel possibly have this scenario..

anyone else report this issue?

thank you

regards,

                Ivan

  • 0
    TI__Mastermind 44550 points

    Hi Ivan,

    Have you tried testing the noise with the ADC inputs shorted (or connected to ground)?

    This will give you a reference point, from which you can compare the noise performance with a signal applied...

    The problem is noise can come from many different sources, with the main sources being the ADC, the reference voltage, the input signal, the power supply, and the layout. When you apply a signal, there is no way to discern where the noise is coming from. However, when you short the inputs to the ADC and measure the noise, you'll reduce the noise to only the noise coming from the the ADC, the power supply, and the layout. If your layout and power supply are clean, then you ought to see the noise performance as given by table 6 in the data sheet (between 3-12 uVrms, depending on the data rate). If the noise is still higher than datasheet values, than likely the power supply or layout are noisy.

    However, if you find that the noise is okay when the input are shorted, then the noise your seeing is from a combination of the input signal (sensor noise and amplifier noise), as well as the reference noise.

    Once, you've tested the noise performance with the inputs shorted, let me know:

    - What is the noise performance with the inputs shorted?

    - What data rate are you using?

    - Which opamp do you use for buffering/amplifying the thermistor?

    Best Regards,
    Chris

  • 0 in reply to Christopher Hall
    Intellectual 350 points

    - What is the noise performance with the inputs shorted?

    The above graph shows the distribution of 200 samples of ADS1258 OFFSET.

    max:106, min -242,  avg:-53 , DRATE[1:0]: 11 (default), DLY[2:0]: 010

    the result does not look good, and is it meant that we need to improve SUPPLY and LAYOUT?

     

    - What data rate are you using?

    DRATE[1:0]: 11, DLY[2:0]:010

    - Which opamp do you use for buffering/amplifying the thermistor?

    And the following is our input circuits.

    opamp: OPA2320

    R3:  Thermistor (testing target)

    R4: 200k ohm

    C1: 1nF

    C2: 1uF

    R5: 0 ohm

  • 0 in reply to Ivan Chou61
    TI__Mastermind 44550 points
    Hi Ivan,

    The ~300 codes peak-to-peak noise your observing is about 2x the noise of what you ought to see with the ADS1258 (for DRATE[1:0] = 11b). (The noise performance is typically 12 uVrms, or about 6.6*12uV = 79 uVpp, which is about 152 codes peak-to-peak with a 4.096V reference). ...So most of the noise is not coming from the input signal, which is good!

    From the schematic,
    I do recall you showed a schematic in another thread that had inductors on the supply pins. It's possible that these inductors are responsible for increasing the supply noise and degrading the ADC's noise performance since they form LC circuits (resonators) with the bypass capacitors. We usually recommend that ADC supply pins not be filtered by inductors, so you might try replacing the inductors with 0 Ohm resistors and seeing if that has an effect on the noise.

    The OPA320 is a very-low opamp, so that is a good amplifier choice to use as an input buffer!


    From a layout perspective, some common issues I've seen are:
    - Split analog and digital ground planes, or no ground plane at all (which creates inductive connections to the ADC).
    - Bypass capacitors not placed close enough to the ADC supply pins.
    - Analog and digital signal traces sharing common regions on the PCB (not partitioned, allowing for signal coupling).

    If you'd like, I be glad to review your layout and provide you with any feedback regarding commons issues that I've aware of.

    Best Regards,
    Chris
  • 0 in reply to Christopher Hall
    Intellectual 350 points

    Hi Chris,

    Chris

    12 uVrms, or about 6.6*12uV = 79 uVpp, which is about 152 codes peak-to-peak with a 4.096V reference

    1.  Vrms = 6.6 Vpp? 

    2. how to send you the layout file? email? And, the layout file type is .pcb file. Can you open it?

  • 0 in reply to Ivan Chou61
    TI__Mastermind 44550 points

    Hi Ivan,

    1. Vpp ~= 6.6 * Vrms. This is a statistical relationship: Vrms is the standard deviation of the noise and 6.6 standard deviations accounts for 99.99...% of all the noise peaks in the normal distribution.

    2. You can email the layout files to pa_deltasigma_apps@ti.com. I should be able to import most of the common PCB file types into Altium.

     

    Best Regards,
    Chris 

  • 0 in reply to Christopher Hall
    Intellectual 350 points

    removal of inductor seems to have some improvement on offset noise.

    Std of the offset value is smaller.

                                  Orinal              removal of inductor

    avg -53.18 -52.975
    dev 53.64238 26.29948

    max 106 17
    min -242 -175

    Is the noise normal compared with the spec?

  • 0 in reply to Ivan Chou61
    TI__Mastermind 44550 points

    Hi Ivan,

    That is a great improvement!

     

    Christopher Hall said:
    The noise performance is typically 12 uVrms, or about 6.6*12uV = 79 uVpp, which is about 152 codes peak-to-peak with a 4.096V reference

    You're pretty much on par with the ADS1258's noise performance:

        • 26 codes RMS = 13.5 uVrms 
        • 192 code PP = 100 uVpp

    I would call that good...When you're that close to the ADC's datasheet specs, any additional improvements will require significant effort for a fairly insignificant gain.

    Best Regards,
    Chris