• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 46 subscribers
  • Views 964 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Resolution of Vref reading

Norman
Prodigy 120 points

Other Parts Discussed in Thread: ADS1258, OPA350, OPA2365

Hello, 

this week I was facing the problem that I read a rather noisy sensor signal with my ADS1258. I wanted to verify that my reference voltage is not the problem, so i logged the Vref reading. On the Vref reading I also had a rather big noise (1mV on the 5V reference). Still, I could not find a reason why there should be so much noise in the system, so I started to doubt the Vref reading and connected Vref with a short wire to an analog input of the ADC where I then measured much less noise on Vref (something like 50µV, which is quite reasonable with my settling time and data rate settings). 

My question is: is it true that the Vref reading is not actually measured with the same resolution as the analog inputs, or do I possibly have some other issue in the system?

 

Regards.
Norman 

  • 0
    TI__Mastermind 43915 points

    Hi Norman,

    To me, it sounds like the root of your problem comes down to driving the ADC's reference pin. I assume you are driving the reference directly with a reference source without any external buffer. That may be alright depending on the drive capability of the reference source. However, a lot of reference sources are unable to drive the reference by themselves and require a driving buffer amplifier. As you can see if Figure 67, we use an OPA350 to drive the reference to ensure that the reference input is getting the current it needs to change any internal sampling capacitor as each sample is taken. Otherwise, you may see something that may look like a 'drooping' effect from reading to reading which may corrupt your readings on your output. I am going to recommend taking a look at this three part article series a few colleagues of mine wrote which explain how the reference affects the ADC performance. The articles are written with a SAR ADC in mind but a lot of the same reference theories hold true for delta sigma ADCs.

    Part 1 -http://focus.ti.com/lit/an/slyt330/slyt330.pdf

    Part 2 -http://focus.ti.com/lit/an/slyt337/slyt337.pdf

    Part 3 - http://focus.ti.com/lit/an/slyt354/slyt354.pdf

    Regards,

    Tony Calabria

  • 0
    TI__Mastermind 43915 points

    Norman,

    Two things I forgot to note

    - The reason that you see less noise when you wrap the reference back around and connect it to the inputs is because you are creating a ratio-metric measurement thus having the noise that is present from the reference line 'cancel-out' as it is also present at the ADC input. The problem sounds like it comes down to the reference source itself. Review those documents in the previous post and it should give you an idea of where you may want to start looking. I recommend to measure the noise in your system is by shorting the two ADC inputs together, tie them to a common point, and view the conversion results.

    - Second, the reference pin is very sensitive in the ADS1258 and it is very important that you have a 2.2uF bulk capacitor at the reference input pin.

    Regards,

    Tony Calabria

  • 0 in reply to Tony Calabria
    Prodigy 120 points

    Hi Tony,

     

    thanks for your input, that is all pretty interesting. Looking at part 3 of that article that you posted, I see in figure 5, that an ADC-Vref can cause up to 1mA drive current alternation and its true that theoretically this could cause trouble with our reference's load transient response, depending on how much absolute drive current is running (we're using some LT1021 here).

    But anyways, I also tried your suggestion to shorten the ADC inputs, tie them to a common point and view the conversion results, and that gives me a result of 15µV alternation with 5V reference, which means log2(5*2.13/1.5e-5) is approx. 19bit, which would be very much sufficient for me.

    So the very poor results seem to be due to the sensor that I am using and I will have to find something better there.

    What I do not understand 100% is, why the reference reading is so bad then.

    Regards

  • 0 in reply to Norman
    TI__Mastermind 43915 points

    Hi Norman,

    I am grad those articles could be helpful. It does sound like your noise is a result of the system rather than the ADC itself. Could you post a schematic of your setup? I am particularly interested in your input structure and what you have placed between MUXout and ADCin.

    Thanks,

    Tony Calabria

  • 0 in reply to Tony Calabria
    Prodigy 120 points

    Hi Tony, 

     

    I have an OPA2365 between MUXout and ADCin buffering both, the ADCINP and ADCINN from the Muxouts. Also with the 2.2nF between the ADCin pins and the 47Ohm resistors to protect the opamp from the capacitive load. Basically the same setup as in Figure 68 in the datasheet, just without amplification factor. The Vref input is also connected like in that figure with a 10µF and 0.1µF capacitor. We really didnt try anything fancy here.

    Norman

  • 0
    TI__Mastermind 43915 points

    Hi Norman,

    When I read your original first post again, I understand it as when you connected the Vref back into the ADC inputs, you had around 50uV of noise. The problems came when you went to read back your sensors, you had over a mV of noise. This tells me that you have something either in your signal path, layout, or maybe the way your sensor is connected that is causing the problem. If you could send a schematic of the system and maybe the layout files, I can take a look to see if this is where your noise is coming from.

    Regards,

    Tony Calabria