• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 63 subscribers
  • Views 1416 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.

TPS51200: VO oscillating

Steffen Martin
Prodigy 190 points
Other Parts Discussed in Thread: TPS51200

Hello,

I'm seeing significant oscillation on the output of a TPS51200 which seems to be part of the reasons why my board is having calibration issues (i.e. DDR2 initialization/calibration fails). The measurements show a 5 MHz oscillation (sine wave) with more than 600 mV peak-to-peak (see scope screenshot).

DC coupling:

AC coupling:

On top of that, the device gets extremely hot, even with the DDR2 memory (MT47H32M16HR-25E) inactive.

Now, I'm wondering whether this is a clear indication that the regulator is broken or whether there's another reason why it doesn't operate correctly. I've been using it on a prototype. Initially the supply voltage (VIN) had a significant ripple (about +/- 150mV; ~300 kHz; spikes) coming from DC-to-DC converters. This has been fixed by replacing them with the output of a linear (lab) power supply but it might have already caused some damage to the regulator.

This issue is consistent throughout all 3 prototypes (same revision) that I had built.

I've also checked whether this is an issue only under load. However, removing the ferrite bead (FB23) didn't make a difference (see schematic & scope screenshots). In fact, the device still gets extremely hot.

AC coupling:

Also, I checked all other inputs (REFIN 0.9V, VLDOIN 1.8V, VIN 2.5V)  and outputs (PGOOD -> ~0V indicating the failure; REFOUT 0.9V) which measure correctly and do not show any oscillation.

I really need to make sure that I'm in fact looking at 3 broken regulator as a rework is quite involved. Also, I'm wondering how sensitive these parts are with respect to supply voltage ripples.

I appreciate your help!

Regards,

Steffen

  • 0
    Prodigy 80 points

    Hi Steffen,

    I don't think that the chip is broken. Maybe this will help:

    - C104/105 have to be close to pin 3 (remove FB23 completely).

    - Check the capacitors C104, C105 - are they broken? Try 10uF.

    - C103 has to be close to pin 2.

    - C94 close to pin 1.

    - Place 1k or more on pin 5 (between C105 and pin 5). Disconnect pin 5 from pin 3.

    - Do you supply anything with the pin 6? Or it is just a test-point?

    - Put 1k or more on pin7.

  • 0 in reply to Paul Daniel Nachiu
    Prodigy 190 points

    Hello Paul,

    thanks for your quick response.

    As you can see from the layout image below, most of the things you recommended to check can be confirmed:

    • C104/105 are relatively close to pin 3
    • C103 is close to pin 2
    • C94 is close to pin 1

    I forgot to mention that I already removed FB23 completely and shorted its pads - the same oscillation (although with a lower amplitude and slightly different frequency) remained. As far as the connection between pins 3 & 5 goes: This essentially follows the design example 1 in the datasheet (DDR2 on 3.3 VIN) although I'm using 2.5V VIN here. I understand that you're suggesting to squeeze in a 1kOhm resistor between C105 & pin 5 which will be rather complicated to hack but I'll try (once I confirmed that the capacitors C104 & C105 are OK and replaced with 10uF as you suggested). Can you tell me the motivation for this experiment?

    As far as the output of pin 6 goes: It supplies the DDR2 memory's VREF input.

    Also, can you explain why you'd put a 1kOhm resistor on pin 7 (EN)?

    Just to give some more background information: The design I used is based on a reference design from a development board (which works). Here is the schematic and layout portion for comparison purposed:

     

    I appreciate your help!

    Regards,

    Steffen

  • 0 in reply to Steffen Martin
    Prodigy 190 points

    Paul,

    I guess the reason why you're suggesting to squeeze in a 1kOhm resistor between C105 & pin 5 is to improve loop-stability according to the recommendations in the datasheet with respect to the output capacitor (in case they don't have a sufficiently low ESR). Let me know if this is your motivation for trying this out.

    Thanks,

    Steffen

  • 0 in reply to Steffen Martin
    Prodigy 190 points

    Hi Paul,

    just as an update, the output capacitors on both, the reference design (C182 & C200) as well as my own design (C104 & C105) are the exact same ones (same manufacturer, same manufacturer P/N). That is, ESR and ESL characteristics (which seem to be somehow critical as far as the datasheet goes) are the same. Again, the reference design works just fine.

    Regards,

    Steffen

  • 0 in reply to Steffen Martin
    Prodigy 190 points

    Paul,

    it seems like it was indeed a problem with the output capacitors (C104 & C105). Unfortunately I didn't have any with 10uF, neither 22uF like the ones installed on the board. So I used two 33uF ones to replace the broken ones and all the oscillations went away. Also, the device's operating temperature went down significantly (it just gets warm a tiny bit as opposed to hot like before).

    Thanks for your recommendations. This saved me a lot of time (and money).

    -Steffen

     

  • 0 in reply to Steffen Martin
    Prodigy 80 points

    :-)

    Glad to help.

    Paul