• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 62 subscribers
  • Views 175 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.

LM5164: LM5164 Vpp to high (> 700mV)

Lada Lostak
Intellectual 330 points
Part Number: LM5164
Other Parts Discussed in Thread: LM76003, TLV172, TMCS1101

Hello, I have problem with LM5164. I used following schematics and PCB:

I had to use different layer for L (Bot) and IC (Top) with small overlapped area. So layout is not perfect (but on the other side not sooo bad):

Layerstack: Top/Int1/Int2/Bot

Top=IC and the rest
Int1=solid ground
Int2=only FB trace
Bot=L and connection between L2/R8

With 60mA load I am getting following result (AC coupled):

I have tried to remove C12+C14+C15 and solder short wire between R15 and U1.5 (FB trace). Trace is not ideal, but it is divided by solid plane and guarded by GND trace. Actual shot is from this state. There are only LDO and OP connected to the output of DCDC (no switching load). I am using 24V from LDO power source (30V/4A) w/o current limit. I also tried to skip input filter (no impact). Also there wasn't any visible impact for soldering short FB wire instead of trace on PCB (wire direct on IC/R15).

Inductor: Coilcraft MSS1210-224K (shielded, DCr=225mOhm)

I expected a bit worse result since inductor is direcly under IC/feedback etc. Before removing capacitors result was even worse. Under no load condition is Vpp>=1200mV. And voltage is oscillating with relative lowe frequency (~300 Hz).

From simulation @ Power Designer I expected Vpp to be around 80mV (f=33.2kHz then) and actual result is ~10 times more. Also DCDC converter is not much stable at all.

There is no another place for L. If this is caused by L at this place (I had to use similar layout few times before with LM76003 - and perfect results) - any sugestion how to make it better ?

Thank you !

R.

  • +1
    TI__Guru 71885 points

    Hi Lada,

    Thank you for designing with the LM5164.  It seems the injected voltage not adequate and the switching becomes less stable.  For a quick fix, could you try to increase C10 to a larger value like 47nF or 100nF?  

    Best Regards,

    Youhao

  • 0 in reply to Youhao Xi
    Intellectual 330 points

    Hello, thank you for suggestion. Here it is:

    C10=10nF - Vpp=520mV

    C10=100nF - Vpp=492mV

    C10=220nF - looks same Vpp=~460mV (somewhat lower)

    And finally - I tried to put back C12/C14/C15 (while C10=10nF) -> Vpp=98mV

    With small C10 and with C12/C14/C15 it was completely unstable. Now seems it "works" but why C10 need to be 200x more than Webench/simulation ? Also it requires 5x more Cout in compare to simulation. Webench was usually 'not so far' from reality. I wanted to try/compare excel calculator for this LM chip but I don't have MS Office and it doesn't work with Libre Office for some reason :(

    And if you will see amplitude is changing - oscilates. You can see nearly complete cycle. Is it safe to be used in production ? From my expierence any kind of this oscillation usually ends in some 'problem' later. On the other side, there is 'constant' current with no switching at all as load (only analog ICs)

    Vpp under 100mV should work for following AD circuits (bunch of TLV172 and TMCS1101 chips :) Low noise LDO will save the word and required precision should be achieved.

    Thank you!

    R.

  • 0 in reply to Lada Lostak
    Intellectual 330 points

    Sorry I found problem between chair and table - today was my scope probe set to 1x but view to 10x (from some other measurement) so after changing C10 Vpp decreses 10x more than you can see in last post. So, it helps completely.

    I removed FB wire so PCB is used as it is. I also changed volage divider to 820K/100K (less reels at the end). All capacitors are fitted.

    Here is final result (AC+DC):

    Vpp=10.5mV which is perfect. Vout=11.14V.

    So, the only problem was C10.

    Thank you !

  • 0 in reply to Lada Lostak
    TI__Guru 71885 points

    Hi Lada,

    Thank you for the updating!  I am glad that you resolved the problem. 

    May I close this thread?  If you have further questions, you are welcome to reopen it by adding a new post, or start a new thread.

    Again, thank you for using our solution, and good luck in your project. 

    Best Regards,

    Youhao