• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 775 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.

ADS1241 hang

Bogumil Pacan
Prodigy 120 points
Other Parts Discussed in Thread: ADS1241, TL431, STRIKE

Hi,

We're facing new problem: ADS1241 stops responding completely (for commands read/write registers and reading data).

We're using ADS1241 in different units in the past without problems. Few thousands of units are on market sharing similar hardware and code.

Now 2 independent customers reported in few units converter hangs few times a week.

Reset with reset pin doesn't help, a power reset is needed.

During 'hang' time Dout from ADS1241 is low.

What we already excluded:

- unit fail (by replacing all units)

- master firmware regression (by downgrading to previous version)

- temperature causing problems with crystal (temp. is monitored on site)

- reset pin to short (checked at office with e.g. r/w registers, shorting crystal for a while, etc.), every time goes on after reset correct at office.

We can't reproduce this problem at office, so probably it's EMI related. The site is HVAC control cabinet, usually quite noisy.

We have only online connection to the site, so we can't check everything.

Any tips would be welcome.

  • 0
    TI__Guru* 94305 points
    Bogumil


    There are only a few things I can think of that could cause the part to hang:

    1. Some sort of powerdown.
    2. A part being held in reset.
    3. The clock shutting down.

    However, I do have some questions that might lead to an answer. I don't know if you have access to testing it remotely to check on the status of the device, but I'll list some the the possible tests.

    1. Can data be read out at all? Even for a short amount of time?

    2. Can the registers be configured and read back? I'd also like to know what mode the part is running in. I'd like to know the contents of the configuration register (speed, PGA, etc).

    3. I'd also like to know the order of the configuration, and what setup is being done before reading from the device.

    4. If the part is running, check the current that is being drawn from the device. If the hangs, check the current again. With this measurement you can check to see if the the part is running in both cases (but the DRDYn has gone high), if the part has powered down, or if the part's digital section has shut down.

    5. What crystal are you using? Have you checked to see that the crystal is running? You may need an oscilloscope to check on either side of the crystal.

    It might help to have a schematic (or even a partial schematic) to help debug the problem.

    One final possibility is that the device had problems with the power supply and maybe there has been some sort of brownout condition. This might cause the digital section to become corrupted. However, since this device has a dedicated RESET pin, I would would have guessed that toggling the RESET pin would have cleared up all problems.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 120 points

    1,2 - we read always 0 (Dout low)
    3 - init: ACR = 0x30, SETUP = 0x00,
    periodic:  MUX = 0x08 to 0x78 (8 channels), CMD = 0xF0, some time, CMD = 0x01 twice

    4 - can't be done online
    5 - 4.9152M 18pF load, max.7pF shunt, ESR=150 with 0402 15pF C0G capacitors
    can't check if crystal is running online, but at office I checked that, if crystal is not running, Dout is high (pull-up in master enabled).

    As I wrote, toggling reset pin doesn't help, a power cycle is needed.

  • 0 in reply to Bogumil Pacan
    TI__Guru* 94305 points
    Bogumil,


    I would check to see if the crystal is sensitive to extra capacitance. If your layout has any extra parasitic capacitance, then maybe there crystal is being pulled out of operation.

    Also, with the supply do you have any inductors in the power supply line? In my experience, this can cause problems by spiking the supply and causing problems with the operation of the device. The inductance reacting with the digital current can cause voltage spikes with large L(di/dt).


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 120 points

    Joseph

    Thanks for answers.

    We are now able to easily recreate 'hang' at office after ESD pulse into input.

    In this state, crystal is running, reset pin doesn't reset device, Dout is 1.05V constant. After power cycle everything is ok.

    Digital part is supplied with DC/DC 3V3 without inductors in supply line, but analog part is supplied from 5V reference (TL431)

    via the ferrite bead (Wuerth 74279263). Both supply pins are decoupled with 0.1uF 0402. I'll try to replace ferrite and maybe input schematic.

  • 0 in reply to Bogumil Pacan
    Prodigy 120 points
    We found that the locked state of ADS was caused by spike between analog and digital grounds.
    Datasheet specify max. difference between grounds to be less than +- 0.3V.
    We separate ground planes and tracks where possible, but we're using also passive filter between grounds to filter HF.
    Shorting grounds outside PCB did not help a lot, but replacing filter with a short did the job.

    So our tip for grounds - do not place any components (R, L, C, Schottky) between grounds, because they might be too slow,
    and a spike between grounds will lock your ADS.
  • 0 in reply to Bogumil Pacan
    TI__Guru* 94305 points
    Bogumil,


    I'm glad you were able to find the issue. It's a tough one to find.

    You might see if adding extra series impedance with the inputs help reduce the effect of an ESD strike. However, if you do this, you may need to see if the extra series impedance combined with the input current gives an extra voltage error. An extra ESD protection device on the front end might help, but if that energy goes into the ground plane, you might be in the same position.

    I agree with you about the components between analog and digital grounds. There really shouldn't be any thing but a direct connection between them or you may have some sort of voltage drop that disrupts the operation. The layout of the ground should definitely be one of the first parts of the PCB design. In general, we've recommended using a single ground but separating the analog and digital sections to keep their return currents from interfering. If you do need to separate the analog and digital grounds, you can lay them out separately, but connect them under the ADC. A colleague wrote some comments on grounding in this note here:

    e2e.ti.com/.../1393


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 120 points

    Joseph

    Thank's for Your comments.

    The universal input circuit is quite complicated, capable of measuring voltages, currents and resistance from single ohm up to few megohms,

    also another faster, lower resolution ADC is used for fast voltage changes. So I wouldn't change it.

    Our solution with components between grounds was added, because we have some power tracks below analog ground (no other space for it).

    We didn't see problems with it for about 5 years, few thousands units installed all over the world.

    We're waiting for customers feedback if it solve their problems.

  • 0 in reply to Bogumil Pacan
    Prodigy 120 points

    All customers reported that above change did the job.

    Problem is 100% fixed now. Thank's for support.