Connect with us

RLC Filter for Power Decoupling and Current Sensing

Discussion in 'Power Electronics' started by Farukh Khan, Oct 28, 2018.

  1. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Hi Guys,

    I am thinking of making a RLC Filter circuit for dc power decoupling and as well as current sensing through the resistor for three different voltages: 3.3V, 5V, 12V. What type of circuit schematic I should use to build this type of RLC Filter with Current Sensing by measuring the voltage drop across the shunt R. And what type of reverse voltage spikes I need to consider for my power supply and the powered devices? And how do I calculate the exact inductance, capacitance and resistance values for the RLC?

    Please let me know how to achieve this with RLC or any other system that might be more suitable.
    Thank You.
     
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    24,860
    2,571
    Jan 21, 2010
    It would be far better for you to tell us about the power source and the requirements for the load than for you to deny us this information an pose your idea of a solution.

    Firstly, inductors are rarely used in power supply filters these days, and for power supplies, a series resistor is even more rare as part of the filtering.

    A current sense resistor is another thing, and depending on the current you might pick a value between 0.01Ω and maybe 1Ω.

    Depending on whether you have any differential between the input voltage and the output voltage then you may be able to use a regulator, or you may be limited to just using a capacitor.
     
  3. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    The power source is a custom made SMPS capable of delivering 70A in the 12V rail, 25A at the 5V Rail and 25A at the 3.3V rail. The loads are rather small. Range of current consumption is from uA to 5A (therefore, I want to measure this range using the shunt). I want to use the Low ESR RLC to provide power decoupling for the different devices from the SMPS power rails. And there is no differentials between the power rails voltages and my devices requirements. I have sensors which will use exactly 3.3V. Then SBC which will use exactly 5V and then relay modules which are rated for 12V.
     
  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    24,860
    2,571
    Jan 21, 2010
    What's the problem (other than current measurement)? You can't measure the current in the negative rail because (although you don't state it) I bet you have a common negative. A current shunt placed between the output of the regulator and the voltage feedback circuit will allow you to measure the current without reducing the output voltage. Are all of the output voltages separately regulated, or is it a PC power supply that (often) only regulates one rail?

    Why do you need the additional filtering?
     
  5. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Yes I certainly have a -12V rail rated @ 0.3A. I cannot open up the PSU due to warranty issues, so basically I cannot use the PSU's voltage feedback. Are you asking me to put the shunt on a different ARM or ATMega MCU as a voltage feedback circuit? I don't think the voltages are separately regulated but I am not really sure about this. Although the PSU have only one 12V rail as the main rail (as far as I know). It is very similar to PC power supply indeed, just customly made. Upon contacting the manufacturer they provided me a consumer grade PSU specification which is almost similar to the PSU I currently have : https://seasonic.com/prime-ultra-platinum#specification . For some relays and sensors I will be routing the power cables a bit further from where the PSU is located. That's why I want to decouple the power for less Vloss through transmission, specially for the 3.3V.

    After that I thought why not use an inductor to make it an LC filter to reduce the ripple noise level further. Typically the PSU under good load gives average of ~13mV ripple noise on almost all the voltage levels. So, I was thinking that there's no harm in reducing this ripple further with a Low ESR Cap decoupling and an inductor.

    Afterwards, I thought why not add the shunt in the same circuitry and route a signal from the shunt to the power measurement MCU. In that way, I might be able to use an RLC circuit on different devices input and use it for multiple functions all together.
     
  6. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    24,860
    2,571
    Jan 21, 2010
    Decoupling won't do that.

    Do that locally where required rather than at the power supply. The long leads will add some resistance and inductance, so all you need add is more capacitance.
     
  7. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    BTW, the power supply has some internal feedback lines that comes out of the PSU to the cables but not really sure how the feedback system is suppose to work through the sense lines.

    Overall, what do you suggest for what I am trying to achieve? How to tackle the Vloss?
    Just adding some low esr caps on the device locally will reduce the ripple noise further? Hoping to get the noises under uV level.

    What about the current measurement? I want to collect the current and voltage data of each device inputs.
     
    Last edited: Oct 28, 2018
  8. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    24,860
    2,571
    Jan 21, 2010
    The sense lines are the voltage feedback. If you extend them to the load before connecting them to the output, the voltage will be regulated at the load end.
     
  9. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Got it. Thanks for the info.

    How about these:


    "Overall, what do you suggest for what I am trying to achieve? How to tackle the Vloss?
    Just adding some low esr caps on the device locally will reduce the ripple noise further? Hoping to get the noises under uV level.

    What about the current measurement? I want to collect the current and voltage data of each device inputs on uA-A and uV-V ranges."
     
  10. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    24,860
    2,571
    Jan 21, 2010
    For the first one I would recommend a linear power supply.

    For the current if try a hall effect current sensor.

    Then use a microcontroller with an ADC to convert the two voltages to some value. With a 10 bit ADC you can get mA to A and resolution of voltages to hundredths of volts (depending on a few factors)
     
  11. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Well, if I use linear regulators for the whole system then it would be really inefficient isn't it? how about a 3.3V LDO in series with he 3.3V SMPS rail?

    how many bits ADC I need to measure uA to A range? Is there any ADC IC I can use for the power measurements and feed the data from the IC to arduino using I2C or something? I want to collect current and voltage readings from more than one terminals.
     
  12. kellys_eye

    kellys_eye

    4,289
    1,143
    Jun 25, 2010
    Still over-thinking your design.....

    Is there really a need to measure current to μA levels?

    Any linear regulator - including LDO devices - require some overhead to work. It won't work using the same input voltage as the output.
     
  13. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Yes, I think I have to measure current in uA because I am planning to do some AI stuff with the collected data. So, I really need that accuracy for the AI plans I have in mind.

    How much overhead will I need for 12V, 5V and 3.3V LDO regulation? I want to keep the heat dissipation minimum.

    BTW what if I use a zener diode for each of the rails? will that decrease the ripple if the zener is in reverse bias?
     
  14. kellys_eye

    kellys_eye

    4,289
    1,143
    Jun 25, 2010
    You want to work with AI but can't figure out the power supply/monitoring issues? You are almost obsessing over them too...

    Are you sure you know what it is you want to achieve?

    Draft a full block diagram, indicate each active device (IC number/type) you want to form the central part of those blocks, get the data sheets for the devices, understand those devices specifications and power requirements and then tackle the issues that you come across.

    You don't (I suspect) need the PC power unit you've got but until you give us the FULL spec well never get to the end of these types of questions.
     
    (*steve*) likes this.
  15. Farukh Khan

    Farukh Khan

    123
    2
    Jun 12, 2015
    Roger that.
     
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Similar Threads
There are no similar threads yet.
Loading...