• State Resolved
  • Locked Locked
  • Replies 14 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 1390 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.

LM73606: Design problems with DC/DC Converter.

user507251644
Prodigy 200 points
Part Number: LM73606
Other Parts Discussed in Thread: LMV358

858/5000

Hello, I have a circuit here with the LM73606. The problem is that I have a very large voltage change when load changes. I have set the voltage to 15V, I have this voltage charged with 3.8A. On the capacitor C155 I have a voltage change of 188mV. I think this voltage change is too big. At higher currents, the voltage change increases even further. At the moment, I have no idea how to get there. The problem with the circuit is the wide output voltage range of 4.5 to 23V.
Is there any other circuits for which the step size does not change over the setting range? Over the 256 steps, the step size changes from 20mV to 450mV. I would appreciate constructive suggestions.

The coil is a Coilcraft XAL1010-682MED.

  • 0
    TI__Mastermind 25590 points
    User,

    To confirm - you're talking about the average DC output dropping as you increase load, not the ripple.

    The LM73606 (and all other bucks for that matter) will regulate the voltage by adjusting the voltage at the sense point (the place where R65 the top resistor in the feedback divider is connected). If all is working properly, this voltage should remain constant over operating conditions.

    The voltage on the output, after L26 and L34, will drop when load increases due to the DC resistance of those components (and the traces on the PCB). You can take the sense point after these components but that will complicate the loop stability equations which you'll have to recalculate and retest.

    Let me know if I am not correct in my assessment and we'll keep working on this.

    Thanks!
    -Sam
  • 0 in reply to Samuel Jaffe
    Prodigy 200 points
    Hello Sam,

    many thanks for your response. The voltage drop takes place directly at the converter. Between L22 and ST8 I have a voltage drop of 41mV. The Voltage breaks already at L22 around 188mV together. I took the EVAL board again and tested it with a fixed resistance ratio of 560k to 40k. With these values, I also have an output voltage of 15.6V. However, the voltage breaks down again at 188mV. Next I reduced the resistances to 56k and 3.9k. However, nothing has changed in the nature of the circuit. I have increased the coil to 18μH also without success. I'm a bit clueless what I can taste.

    Apart from that, the switching regulator and the coil in my design will be 100 ° C hot. I think that this is a bit too warm. I'm still unsure what I can change there.

    Best regards Philipp
  • 0 in reply to user507251644
    TI__Mastermind 25590 points
    Philipp,

    What do you mean by ST8? I don't see that on the schematic.

    Okay so we've ruled out the inductor and the feedback resistors. How about the layout? Can you share the layout as well?

    And the hot inductor - is this the inductor from the schematic? Make sure the RMS current rating is acceptable, check the saturation current, and make sure the core is acceptable for switching regulators.

    -Sam
  • 0 in reply to Samuel Jaffe
    Prodigy 200 points

    Hello Sam,
    ST8 is the plug at the output of the circuit. Behind the test points TP14 and TP15.
    The power inductance is a Coilcraft XAL1010-682MED.
    DCR 8.9mR, SFR 14MHz typical, Isat 21.8A, Irms 14A @ 20 ° C rise. I think the inductance should fit especially with the low DC resistance. Sure, I'm adding the layout, but I have the same problems with the circuit on the EVAL board. I have omitted the label in the components so that it is a little clearer. The components are arranged on the top layer. directly below is the GND layer. Did I hide in the picture. There are no lines there are only the ground connection. The vias do not have a thermal pad. On the bottom layer, the surfaces are copied from above to release the heat via a heat conducting foil to an aluminum plate.

    Best regards Philipp

  • 0 in reply to user507251644
    TI__Mastermind 25590 points
    Philipp,

    I would definitely expect to see some drop from L22 to ST8 but as I understand it you're seeing droop at L22, measured at the trace where R65 is connected. This is probably due to the layout. I see SW goes through L22 to the output caps. This connects to a plane which makes its way from the right-side of the picture to the left side (by way of a connection below where the pic is cut off) where R65 senses VOUT. This sense point should be taken directly at the output capacitors, close to L22. The way it is now means you'll get a voltage drop from VOUT to R65 with whatever current is following that same path.

    I also see that the FB trace is routed directly under the SW node plane. This noisy SW node could be injecting into the FB pin which may cause poor regulation. You can try cutting this trace and connecting a wire from FB to IC15-B to see if this is the case.

    You say you also see this on the EVAL board. How are you taking these measurements and what are you modifying?

    -Sam
  • 0 in reply to Samuel Jaffe
    Prodigy 200 points

    Hello Sam,
    yes on the trace between coil and plug I have a voltage drop of 41mV. The output voltage goes down under load by 230mV down. Of these, only the 41mV are due to the wires and filters between plug and coil.


    For the cables running under the SW surface there is still a closed mass surface in between. Therefore, I did not think that it gives problems. But I can change the routing in the next revision.

    In the middle of the picture is a LMV358 and an MCP4661 to adjust the voltages of the two regulators. The third voltage regulator is not visible on the picture I think this would have become too confusing.

    in the

    I made some changes on the EVAL board. I replaced the output capacitors with the same ones that I use on my printed circuit board. The inductance I exchanged for the coil of Coilcraft. The FB network I have adjusted so that I get 15.5V output voltage.
    For comparison, I have ordered a new EVAL board again. These values ​​are measured with a digital multimeter. With the oscilloscope, I measured a ripple voltage of 30mV at the output.

    Best regards

    Philipp

  • 0 in reply to user507251644
    Prodigy 200 points

    Hello Sam,

    I have tested again with the EVAL board. I deactivated the Auto Mode and operated the DC DC converter with FPWM and an RT of 100k. Then the output voltage remains constant and I only have the voltage drop over the tracks and filters.

    I would prefer to operate the DC / DC converter in auto mode to have a higher efficiency over the wide voltage and current range.
    Maybe this helps you further to limit my problem with the controller.

    Best regards

    Philipp

  • 0 in reply to user507251644
    TI__Mastermind 25590 points

    Philipp,

    So it sounds like the EVM does not have the same issue you're seeing on your board, correct?

    So in normal operation you're seeing a droop in output voltage proportional to output current. This has to be caused by something between the VOUT caps and the FB pin of the IC. Please confirm the following:

    • The FB pin of the IC is 1V during all loads
    • The voltage at the VOUT pad of R65 has the droop issue
    • Pin 5 of IC15-B also has the droop issue

    -Sam

  • 0 in reply to Samuel Jaffe
    Prodigy 200 points

    Hello Sam,
    I think the voltage drop on both boards is the same error. I have also removed my board the whole FB regulation and replaced by two resistors. Unfortunately, I can not easily change the mode on my PCB. For this I would have to unsolder the LM73606. I do not understand why there is a difference between the automatic and the FPWM mode on the EVAL board. The data sheet shows on page 13:

    Mode setting only Affects operational at light loads.There is no difference if load current is above the DCM and PFM boundary conditions Discussed above.

    So should the output voltage remain the same in both cases?

    Best regards

    Philipp

  • 0 in reply to user507251644
    TI__Mastermind 25590 points

    Philipp,

    There is no difference in VOUT regulation when the part is operating outside of DCM/PFM (higher output current) so the output should remain the same in both cases. The only difference is at light load where the output is regulated differently between PFM and FPWM but both should still regulate properly.

    Please confirm the following:

    • The FB pin of the IC is 1V during all loads
    • The voltage at the VOUT pad of R65 has the droop issue
    • Pin 5 of IC15-B also has the droop issue

    -Sam

  • 0 in reply to Samuel Jaffe
    Prodigy 200 points

    Hello Sam,

    I do not test with my printed circuit board. I'm testing now with the EVAL board.
    www.ti.com/.../snvu579.pdf

    There I have the feedback about R1 R2 and R4 on page 6 in PDF.

    I have now measured in automatic and in FPWM mode:


    The resistance values for the FB network:

    R1 = 10R
    R2 = 56k
    R4 = 3.9k

    6x 10μF 50V X7R capacitors in the output
    L XAL1010-682ME

    So I can confirm that I the voltage drop caused by the DC DC controller.
    I can not investigate the behavior on my own printed circuit board but if the problem also occurs on the EVAL board it can not be due to the switching on my printed circuit board.

    To answer your questions:

    The FB pin is not always at 1V
    The voltage dip is also present at R1 (R65).
    The voltage dip is also present on R4 (pin 5 IC15-B).

    Best regards

    Philipp

  • 0 in reply to user507251644
    TI__Mastermind 25590 points
    Philipp,

    Those numbers are within the spec of the part but the datasheet shows that the FB voltage should not drift that much with temp. Are you measuring this with a twisted pair and a multimeter very close to the FB pin and AGND pin of the IC?

    -Sam
  • 0 in reply to Samuel Jaffe
    Prodigy 200 points

    Hello Sam,

    yes, I measure the voltages on the EVAL board directly at the resistor which goes directly to AGND and FB. On the 2mm can not be a significant voltage drop. What else can I do to get the problem under control?

    Best regards
    Philipp

  • 0 in reply to user507251644
    TI__Mastermind 25590 points

    User,

    Are you referring to a discrepancy between no load and full load? I misunderstood, I thought you were comparing some load to full load.

    In that case this is completely normal. The output is not drooping. Auto mode / PFM just has a higher average output than FPWM when there is no load. PFM will charge the output voltage up, then turn off to save power until the output voltage reaches a threshold. This ripple has been designed to be small enough to not disrupt circuit operation but large enough to take advantage of the sleep off-time efficiency gains.

    See Figures 45 and 46 in the datasheet.

    -Sam