• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 48 subscribers
  • Views 260 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.

INA3221: Optimal current monitoring

Kevin Kachikian1
Prodigy 140 points
Part Number: INA3221
Other Parts Discussed in Thread: INA4230, INA4235

Tool/software:

Using this part, sometimes it is difficult to place the monitor close to the current/voltage source you wish to monitor, especially with a 3 channel device such as this.

The data sheet recommends placing the current sense resistors close to the INA3221 (in my case 10m-ohm 1W), but what if the current needing monitoring is "further away" from the 3221? Wouldn't it be better to place the current sense resistor close to the thing being monitored and run the 2 traces (IN+ & IN-) back to the 3221 in order to reduce the high current resistance path so as to obtain best accuracy? 

Or might it be better to just keep the current sense resistor close to the 3221 as the data sheet recommends but then run thick traces from the thing being monitored back to the current sense resistor and 3221.

Thoughts?

Thanks!

  • 0
    Guru 285390 points

    Longer traces to the IN pins might pick up noise. But in general, having longer high-current traces would be worse.

  • 0 in reply to Clemens Ladisch
    Prodigy 140 points

    TI recommends two 10 ohm resistors and a cap for input filtering which I’ll implement. Hopefully that will help reduce noise in practice. Plus in my application, the current draw is pretty consistent over time in a low frequency, low noise situation, so I can accumulate several measurements in order to reduce noise.

    Any other suggestions on how to route 3 independent channels for 3 independent points of monitoring each of which may be far away from one another? This part is really amazing, but that issue isn't really addressed in the data sheet or corollary tech docs.
    FWIW, the EVM board DOES have the shunt resistors far away from the 3221 (about 4cm).

    Thanks
  • 0 in reply to Kevin Kachikian1
    TI__Mastermind 22315 points

    Hey Kevin,

    I agree with Clemens, just make sure you have the sensing input traces relatively matched. Here is a link to a video about routing that may help: https://www.ti.com/video/6076326896001

    Also, as an FYI in case you are interested, we recently released 2 four channel devices, the INA4235 and INA4230.

    Regards,

    Mitch

  • 0 in reply to Mitch M
    Prodigy 140 points

    I didn't know about these 2 new parts. I've switched to the 4235. Thanks!

    I think I'm going to try both - I'll try a channel where the shunt resistor is close to the current source and longer traces are run to the 4235. And I'll try the same but where the shunt is close to the 4235 and longer traces run to the current source.

    In my case, I want to obtain accurate current measurements so I think placing the shunt nearest the current source will yield the best results and as mentioned, the current charges will be slow so I have time to average out readings to reduce potential noise.

    How do you think other designers might be mitigating the issues of current accuracy versus noise in a 4 channel device like this? This must be a common challenge because the part is a 4 channel device and sometimes the current sources are far apart from one another. With a single channel device, this issue wouldn't be present BC the measurement chip & shunt & the current source would all be in proximity to one another.

    Thanks!

  • +1 in reply to Kevin Kachikian1
    TI__Mastermind 22315 points

    Hello Kevin,

    I'm glad that the INA4235 works out for you! I've asked the systems team if they know what customers are doing, and they said that most people rout longer IN+ and IN- traces to the current sources and then just use longer conversion times and averaging. Then, if that is not good enough, they add external input filter resistors. 

    Regards,

    Mitch

  • 0 in reply to Mitch M
    Prodigy 140 points

    Just a thought: Maybe pop that tidbit of information into the data sheet to address how to deal with 4 channels of measurement vs. a single channel where everything can be right at the source. Thanks!

  • 0 in reply to Kevin Kachikian1
    TI__Mastermind 22315 points

    Thanks for the feedback, I'll forward it to the systems team.