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.

ADS1256: Linearity and INL calibration

Part Number: ADS1256
Other Parts Discussed in Thread: REF70, PGA280, ADS127L11, ADS1262

Hello,

1. Introduction

My application requires high linearity ADC. I've looked at TI product portfolio and I have chosen ADS1256. According to datasheets ADS126x series are not better and they cost more. The analog frontend consist of PGA280  connected to AnIn0&ANIn1   and discrete opamps circuits for other AnIn s. Circuit will operate in precision temperature controlled chamber @35degree.Humidity will be measured but not controlled. Vref is served by REF70 and crosschecked by LTZ1000A.

2. Questions

ADS1256 datasheet specify nonlinearity with single parameter "integral non linearity - INL"  3ppm typical 10ppm max.

   - Can I minimize INL error with calibration. E,g Using voltage source and highly linear HP3458 8 1/2 digit DMM.

   - What is the shape of of INL error vs output code? - Answer of this question will help me to apply linearization polynomial. 

   - are any TI devices with lower INL or easier way to characterize / linearize? 

   - How can I minimize INL od ADS1256. E.g. does two or more  ADS1256 running in parallel minimize INL?

   - what is the main source for INL errors inside IC, input SC, integrators comparators or something else?

Thanks in advance

Mirtcho

  • Hi Mirtcho,

    The INL for a specific ADC depends on process variations that result in shifts in the structures on the die. The system linearity depends on the ADC's INL as well as the performance of the driving circuitry e.g. the PGA280 and the "discrete op amps" you mentioned. In other words, linearity is a system-level design challenge and is not completely dependent on ADC INL.

    If you want to characterize the INL for a specific ADC, the following link details a procedure for doing so: https://e2e.ti.com/support/data-converters/f/data-converters-forum/944642/ads1262-inl-measurement-method/3490213#3490213

    The plots below are the INL curves from the ADS1256 datasheet. As you can see, the INL is temperature, input signal, and gain dependent. I will also mention that the shape of each curve will vary from device to device, though all devices are subject to the specifications in the datasheet. In other words, if you characterize the shape of the INL curve vs VIN for one ADC, you should not assume a second ADC's INL curve will have the same shape.

    Running two ADCs in parallel will not inherently improve INL. If 1-2ppm really makes a difference in this application, I would suggest characterizing the INL of each ADC and using this curve to calibrate out some of the error. It sounds like your system will be at a fixed-temperature, so this at least removes one of the variables that causes the INL curve to fluctuate.

    Without knowing more about your system needs, I can tell you that we are in the process of releasing a new 1-Ch delta-sigma ADC, the ADS127L11, that has been in class INL performance at 1ppm (typical). This is a pre-production device currently, but samples are available should you want to test out it out.

    -Bryan

  • Hello Bryan,

    Thank you very much. Your answer is comprehensive . It covers almost all aspects concerning my design.

    I have only two small questions, before I mark this thread as completely resolved.

    • Does ADC INL depend on humidity? – I measure humidity but I cannot control it. Equipment will run in Lab environment. And ADC+Vref will run in temperature chamber.
    • When will be available ADS127L11 samples? – specification looks exciting .It match DC performance of AD7177 but is has class leading AC performance.

     

     

    Thanks,

    Mirtcho

  • Hi Mirtcho,

    Here are the answers to your questions:

    1. I would guess that humidity has some effect on INL performance (indirectly by affecting the components on the ADC die). But I am not  sure how much of an effect it has, so I have reached out to one of our designers to see if he has any insight. I will let you know what he says.
    2. We are actually expecting to have samples available this week, though that is subject to change. You can hit the "Subscribe to updates" button that is circled in the image below, and this will notify you when anything changes on the product folder. EVMs are also available now

    Since you mentioned the AD7177, I will reiterate how the complete signal chain affects the overall linearity as this is pretty clear from the AD7177 datasheet. The AD7177 specifies INL with the buffers off (1ppm typ) and buffers on (3.5ppm typ). So you can see that once the amplifiers are switched into the signal path, the INL performance decreases. Comparatively, our 32-bit ADS1262 offers 3ppm typ INL with the PGA enabled, so the ADS1262 linearity is actually better than the AD7177 when you include the entire signal chain performance.

    Again, I would just keep this in mind moving forward, as your complete signal chain determines the linearity performance, not just the ADC.

    Finally, since you are working on what appears to be a high precision, high accuracy application, you might want to check out Keithley's low-level measurement handbook at the following link: https://download.tek.com/document/LowLevelHandbook_7Ed.pdf

    -Bryan

  • Hi Mirtcho,

    I just wanted to follow up on the first question you asked about if INL is affected by humidity. I spoke with our designer and he too was not sure, though he surmised that humidity would not have a strong affect on an ADC's linearity. However, he mentioned that without data we could not make a definitive claim either way.

    If you do find any correlation between INL and humidity, please feel free to share your results on the forum.

    -Bryan