• State TI Thinks Resolved
  • Replies 5 replies
  • Answers 3 answers
  • Subscribers 15 subscribers
  • Views 48 views
  • Users 0 members are here
Support feedback
Options
Options
Related

ISOW1044: Power Supply decoupling

Andreas Nestler
Intellectual 400 points

Part Number: ISOW1044
Other Parts Discussed in Thread: LM317

Hello,

we are using 4x ISOW1044 in our design. We are facing noise issues on our +5V supply (80 to 100mVpp). Since the +5V supply rail is also used for other noise sensitive application we would like to minimize the noise. The ISOW application circuitry follows the recommendatations according datasheet so we have the ferrit beads and the capacitance on VDD pin:

grafik.png

Our +5V regulator is a buck converter with total 4x10uF output capacitance and a small ferrit bead at the output. I guess that the noise we observe are not from our 5V converter since the ripple signature is more or less random.

Q1: what are the recommendations for the driving supply? Do we have to add bulk capacitance and how much?

Q2: What is the frequency of chip internal isolated DCDC converter? Datasheet states in Feature section::

grafik.png

To be honest 25MHz seems to be a quiet high switching frequency.

Q3: Can you confirm the 25MHz switching frequency of internal DCDC converter?

Thanks in advance!

Best regards, Andreas

  • 0
    TI__Genius 16020 points

    Hi Andreas,

    Thanks for reaching out providing detailed inputs.

    VDD ripple from each ISOW is expected to be around 25mV-30mV and having 4 of them tied to same supply giving 80-100mV is expected from ISOW device.

    Now, you can reduce the ripple by increasing bulk capacitance at both out of buck converter as well as input of ISOW - Increase 10uF at each site to 47uF and we expect it to reduce the ripple.

    Yes, internal switching frequency integrated DC/DC is around 25MHz but the ripple that you measure at the common VDD supply will not have this frequency component as it gets filtered out by the local capacitors. If you measure the ripple frequency it should be in the range of 100s of kHz which corresponds to the regulation of the power supply.

    Regards
    Varun

  • 0 in reply to Varun Kumar
    Intellectual 400 points

    Hello Varun,

    Thanks for your response and the technical details regarding switching frequency and ripple.

    Are there application schematics available with switched mode power supply driving the +5V? In the EVM the classic linear regulator of type LM317 is used with 47uF output capacitance. Since we provide the +5V also to ADC converters we would like to get a ripple target of <50mVpp maximum.

    Thanks in advance!

    Best regards, Andreas 

  • 0 in reply to Andreas Nestler
    TI__Genius 16020 points

    Hi Andreas,

    We don't have any available schematics for SMPS. Generally, we do not recommend using SMPS to ISOW VDD as both of them are inherently noisy which is why EVM also has a LDO instead of SMPS.

    Also, it is recommended to have a LDO between common VDD supply from buck converter and ADC supply to prevent precision issue with ADC.

    Regards
    Varun

  • 0 in reply to Varun Kumar
    Intellectual 400 points

    Hi Varun,

    Thanks for feedback and recommendations. Unfortunately schematics design and PCB layout are fixed and we cannot introduce the LDO or another DCDC converter for ISOW VDD. Last questions to my issue...

    According datasheet and EVM there is VDD decouplked with 1x10uF, 1x 1uF and 1x 10nF. We can replace the 1uF with 10uF (same package).

    Q: Does this (2x10uF + 10nF) fit with decoupling requirements for ISOW VDD? According datasheet bulk capacitance >= 10uF and the 10nF is required. I guess that the 1uF in the middle is to provide broadband decoupling.

    Thanks and Best Regards,

    Andreas

  • 0 in reply to Andreas Nestler
    TI__Genius 16020 points

    Hi Andreas,

    We would not recommend removing 1uF as it can filter the higher frequency 25MHz noise more efficiently due to a higher SRF compared to 10uF.

    You can definitely experiment it out as well and see if it reduces to your requirement, but replacing 10uF with higher value such as 22uF or 47uF sounds more effective.

    Regards
    Varun