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.

ADS8902B: What is the minimum input voltage changes is to be sensible

Part Number: ADS8902B
Other Parts Discussed in Thread: THS4551, ADS8695, INA240, TINA-TI, OPA189, OPA388
Hi sir/madam,
By referring the datasheet of ADS8902B, the Vref is 2.5V to 5V.
If I choose 2.5V as reference, the full scale is 2.5V, so the output should be FFFFF H for 2.5V analog input. Then the LSB is 2.5/220 = 2.384μV. 
If I give this much small voltage as input, will the device sense it and convert to give the proper output as 00001H? or is there any limitations/ offset/ tolerance?  Then what is the minimum sensitive voltage for the device to start conversion. And what is the minimum input voltage changes to be sensible(in microVolt)?
  • Hello Venkatesh,

    The ADS8902B is capable of resolving voltage changes in the 2-3uV range, but there are limitations.

    First, if you input a DC voltage, you will be limited by the internal noise of the ADC. This is specified as 2.3ppm at 500ksps, and is rms noise, which translates to 11.5uVrms, or about 76uVpp of noise. In order to see a single transition, you will need to do some type of digital filtering on the sampled data to see the changes. A simple averaging filter can work for testing purposes.

    Also, as you point out, there are offset and gain errors as well. If you try to measure near 0V, the offset error will dominate, and can be as high as 11.5ppm at 25C. This translates into 57.5uV, which can be positive or negative. If negative, you would need to input at least this much voltage before the ADC would transition from 00000H to 00001H. However, once the ADC starts responding, with appropriate filtering, you can detect changes in the 2.384uV range.

    There are also other errors to consider when measuring at other input voltages, including differential non-linearity, integral non-linearity, and gain. In addition, the reference accuracy will effect overall system gain and noise as well.

    If you have not already, I highly recommend you view our TI Precision Labs - ADCs training videos. These will walk you through all of the error analysis in detail, and help explain how to calculate and predict overall system performance.

    training.ti.com/ti-precision-labs-adcs

    Best Regards,
    Keith N.
    Precision ADC Applications
  • Hello Venkatesh,

    I need to make a correction; the ADS8902B is a fully differential input device. This means that the full scale input range is +/-Vref. In the case of 2.5V Vref, 1LSB will equal (2.5V - -2.5V)/2^20, or 4.768uV, not 2.384uV. This also impacts the ADC noise calculation; 23uVrms, or 152uVpp, as well as the offset at 25C. 11.5ppm of offset voltage translates to +/-115uV of offset.

    Also, when measuring a differential signal near 0V, the device can measure +/- differential voltages. If you want to measure single ended inputs from 0V to 2.5V, you will need a circuit similar to Figure 111 in the datasheet.

    I hope this helps clear things up.

    Regards,
    Keith N.
  • Thank you Kenith. I hope my design would be better with this part.

    Do you have any suggestions for alternate part with high precision? Because I have to monitor current in high side from 0 to few Amperes with 10uA resolution. 20bit ADC would be suffice to my design. so that I need to sense by current sense resistor and give to ADC with some amplification. But  the amplifier itself also has some offset right? for example of figure 111 in datasheet ADS8902B has the amplifier which posses 375uV offset.  so can you please suggest the suitable amplifier part? or suitable design?

  • Hello Venkatesh,

    You mention that you are measuring current on the high side.

    1. What is the maximum common mode voltage on the high side for the current sense resistor?
    2. What is the maximum current that you need to measure, 2A or 10A?
    3. What is the current signal bandwidth, or, what is the maximum sample rate that you need to measure?
    4. What is the current measurement accuracy that you need to achieve? 0.1%, 0.01%, etc.

    I ask, because this will help narrow down the choice for a front-end amplifier and the best fit ADC.

    Regarding amplifier offset, depending on your current measurement accuracy requirements, you may need to do a system calibration to eliminate the initial offset.

    Thanks,
    Keith N.
  • Hi Keith, Thanks for your response.
    Here is the key points you asked

    1. Nominal voltage to load is 3.3V but maximum is 5V (plan to use 250mOhm sense resistor, or lesser ohm)
    2. 3.5A
    3. 500kS/s would be ok
    4. 0.01% (or as much as good)
  • Hello Venkatesh,

    These are aggressive system specifications for this speed and bandwidth. To get 10uA resolution with 3.5A max, you will need at least 19b, so the ADS8902B is a good candidate. You will need a driver amplifier similar to Figure 111 in the datasheet. You will also need an instrumentation amp or current shunt amp to translate the shunt resistor voltage from 3.3-5V common mode down to ground.

    As you pointed out, the THS4551 in Figure 111 has a high offset; you will need to do a full system calibration to eliminate this initial offset, and then consider the effects over temperature.

    The INA240 is a high BW current shunt amplifier that may be a good option. If you can relax your resolution requirements, the ADS8695 is an 18b SAR ADC that includes a reference and input driver buffer. This part will simplify your design requirements since much of what you need is already integrated inside the device.

    Regards,
    Keith N.
  • Hi Keith,

    Thank you for your suggestion,

    Due to the concern the power dissipation of sense resistor, I plan to put the resistor value as 40mOhm. so for 10uA step the drop across resistor is only 0.4uV for one step, in this situation this 0.4uV change will be sensible by the ADC?

    I simulated the analog front end shown in Figure 111 of ADS8902B datasheet in separate with this 40mOhm resistor in the Tina-TI software. but the change of 0.4uV is not given any effective change in the output. the output changes atleast for 250uV drop. so what should I do?

    I also simulated the front end as shown in Figure 105 of ADS8902B Datasheet. It gives perfect change of voltage in 0.4uV range but still it is confusing that whether the ADS8902B will sense this much change? for full scale the voltage is only 140mV, but for map this to a good reference(ex:4 or 4.5V), I should amplify the signal. But i didn't find good amplifier. for map to 4V i need 28Gain, for 4.5v ref, i need 32gain. What should I do?

    the INA240 itself have fixed gains like 20,50,100,200 but the output is unstable in first 50mA(resistor drop 2mV). The changes also gives for 20uA once. But I need the changes for 10uA.

    and I have plan to amplify the drop with either OPA388 or OPA189 but the bandwidth of them are not satisfied. I require 500KHz. buth those are gives stable gain for few frequencies only. so can you please suggest anything?

  • Hi keith

    Please tell me that is it possible to change the gain of the analog front end shown in the figure 105 and figure 111 of ADS8902B. Or I should design a separate amplifier section before the analog front end? please suggest any application circuit regarding this.