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.

LM25574 squealing noise

Other Parts Discussed in Thread: LM25574, LM46000

Hi, 

I'm testing a board with a couple of switchers on it (the only thing that is currently populated). The -8V inverting switcher is making a squealing nose. I have tracked it down to what looks like abnormal behaviour on the SW node of the switcher. The design and values were generated by webanch. I have attacked the schematic and 2 oscilographs of the SW node. 

-8V shematic.pdf   

What I believe is causing the squealing is the 2kHz toggling of the "normal looking" square wave.  Is this waveform normal for the SW node? If not what could be the problem. 

Best regards, 

Marko Antesic

  • The switch waveform is NOT normal. From a quick look at it, it seems like the inductor current is oscillating. The operating mode oscillates between continuous conduction and discondinuous conduction. It could be the loop is not stable. I will look into the schematic to get more details.
    -Yang
  • Hi, 

    any suggestion on how to fix it? Is it ok for me to run the converter like this with no load on it? I need to debug some code for the processor on board. 

    Best regards, 

    Marko 

  • I did a Webench design with LM25574 and your conditions. The BOM is mostly the same as your design except for the compensation. The Webench simulation shows 20kHz crossover frequency and 45 degree phase margin with its compensation network. Please try the compensation network shown below to see it is stablized or not. What is the load current you need? The LM25574 can supply 0.5A with a buck topology. With inverting design, the DC current capability is lower than that.

  • My current requirements are low, I think less then 10 mA but I can't accurately tell as the data sheet does not say. I need the -8V for a TFT display. When the Webench generated the topology for me the Ccomp2 was not there.
    I have made the compensation network the same as in your schematic and the problem is still there. What now?
  • There are few things we could try.

    Please probe the waveform of Vin and Vout.

    If Vin is oscillating, please add a electrolectic capacitor between Vin and ground to damp oscillation caused by ceramic capacitor and input power cables.

    If Vin is stable and Vout is oscillating, please add more Cin and a ceramic cap between Vin and -Vout for additional bypassing. The capacitor between Vin and -Vout should be rated >> 32V.

    Please also try if you have more load, say 200mA, at the output, is the oscillation still there?

    We can also review your layout if preferred.

    Let me know if any of these helps.

    Best,

    -Yang

  • Hi,

    1) Here are the wave forms of Vin and Vout and SW node. 

    2) Vin does not seem to oscillate. 

    3) Vout does not seem to oscillate. 

    4) I have added a 100R burden resistor so 80 mA and the noise got louder. Here is the sw wave form. 

    5) The layout

    The layer stack is top (picture), gnd plane, +3v3 plane on the bottom side there is only a via connecting pin 1 to the via near pin 7 and 8.

    Best regards,

    Marko  

  • Update: I wanted to check for bed solder joints. If I press on the board and components with an isolated stick then there is no change. But if I put my finger on the compensation network then the sound changes and if I press harder then the sw wave form becomes normal and the noise stops.
  • Thank you for the update. It could also mean the design is noisy and human finger changed the noise coupling, thus the behavior.

    From the layout, I can see the ground of the Cin caps are far from the ground of the Cout caps. High frequency noise will travel between these two grounds and may couple noise into the feedback components and causing unstable operation. It can be improved by rearrage the caps in the layout, and add 2.2uF bypass cap between Vin and -Vout.

    If you have the chance to redo the design, you can also consider LM46000 and an application note for inverting design and board layout: http://www.ti.com/lit/an/snva722b/snva722b.pdf the layout should be simpler.

    The components on the SS, SYNC and RT are optional. Webench should also be able to generate an inverting design with this part.

    -Yang

  • Adding the 2.2 uF cap between Vin and -Vout did nothing.

    I managed to solve it by adding a capacitor in parallel with R605. I have also removed the 10 pF capacitor over R609 and C620 as without it I have better results. I have also tested the stability of regulation and I get about a 100 mV overshoot for cca. 1ms on a load step response for a 70R resistor. (This is more then enough for my design.)

    Best regards, 

    Marko