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.

LM3150: LM3150 - very low output current limit

Part Number: LM3150

Hello, I have made a power supply using LM3150, it is a redesign of an existing unit that works fine.

The problem with the new design is that it seems that the output current is severely limited, the design was made for 7V and 8A  and the limit of the unit is actually around 1A.

I have used Webench with the parameters below (the nominal input voltage is actually 12V.  The only changes compared to the Webench BOM were:

- the inductor - I have used IHLP6767GZER100M01  instead of the XAL1010-103MEB (that was the initial choice in Webench) - but the IHLP6767GZER100M01  was also recommended as alternative by Webenc

- the output capacitor is 180uF instead of 120uF (16SVP180M) - but it is the same series and with similar ESR

The output looks clean at 7V for a low current but if I test it with a simple power resistor to get a mere 1.25A output current the voltage drops severely and the output looks like a sawtooth between 0V and 7V with a frequency around 200Hz.  If I use two resitors (2.5A) the frequency gets to 400Hz.

I am assuming the chip goes into current limit protection ?

The old unit was designed for 5.5V output at 10A and it worked fine, but it used other MOSFETS and inductor and the values of the components were slightly different.

I would be grateful for any help to get this solved, I have checked the components and values and assembly and it all seems OK.  And as mentioned the previous version is running OK.

Thanks,

Mihai

  • A few more notes:
    - at the input there is additionally an electrolytic capacitor 220uF (as recommended in the datasheet)
    - I have measured the output voltage on both power supplies ( the old one and the new one) and the ripple on the new one is about half compared to the old one, could this be the reason of the problems ? I know the LM3150 relies on the output ripple for working correctly; moreover the capacitor that I have used has a slightly lower ESR than the one recommended by Webench so maybe this contributes to the lower ripple (?)
  • Hi Mihai,

    I hope that the input power supply is not current limited .

    Here are the few things you can start debugging with :

    1. Is the load connected at the output when you turn on the input . Can you turn on the input with disabled load and then add the load once the output is high . If it solves the issue then you would need to increase the Soft start capacitor .You can increase the SS to 100nF .

    2. Further there are few recommendations for improving the design . Add another 180uF cap in parallel with the existing output cap . Further Add ripple injection circuitry across the inductor and couple it to FB . You can do the same with the help of Webench itself .Just choose Low ripple option under o/p ripple .Also Add 100nF bypass cap on Vin to GND of LM3150 .

    Let me know how your debug goes .
    ---Ambreesh
  • Hello Ambreesh,

    Thanks for your help.

    The input PSU is not current limited.

    The load is not connected at startup, I have connected it afterwards but the result is the same,  so it is not related to the soft start.

    I will test with the solutions that you suggested, thanks.

    Best regards,

    Mihai

  • I have made a few more tests (I did not add for now the ripple injection circuit as it is a bit difficult since it is not provided on the current layout).

    What seems to work is removing the Cff capacitor, the output current now looks OK, I have tested up to 7A and it works OK

    The value of Cff from Webench seems not correct anyway, based on the calculation in the datasheet it should be around 170pF while Webench gets 620pF

    However removing Cff has made the output very noisy, it seems there are spikes of around 2-300mV at each switching of the mosfets.

    Is it OK to run the circuit without Cff ?

    I have tested also with a smaller Cff (100pF) but it does not work (same behavior as before).

    Thanks

    Mihai

  • I have made some more measurements without Cff and I found some strange results.
    When I test it with power load resistors I get a lot of noise as described in the previous message.
    But when I test with the real load circuit the output looks very good (no noise).
    I was thinking it may be due to the additional capacitors on the real circuit but if I add the power resistor together with the real circuit the noise appears again ... It is true that the load resistors draw more power (around 2.5A) than the real circuit
    But anyway with the real circuit it looks quite OK.
    Which leads back to the previous question - Is it OK to run the circuit without Cff ?
    Thanks,
    Mihai
  • Hi Mihai,

    I beleive what you are seeing without CFFis unstable switching and hence the high ripple at the output . LM3150 is a COT controlller . Typical COT hysteretic controllers need a significant amount of output capacitor ESR (or ripple injection circuit) to maintain a minimum amount of ripple at the FB pin in order to switch properly and maintain efficient regulation. Try a simpler solution : Add a 50milliohm resistance in series with the Output cap Cout to GND which will replicate the higher ESR . Try to capture the switch node .THe switch node should be stable . You can also acheive stable switching through the ripple injection circuit too.

    ---Ambreesh
  • Hello Ambreesh,

    I will try to add the 50 milliohm and I will test also the ripple injection circuit.

    Regarding the unstable switching - the SW node is not very stable indeed - attached below a screen picture at the SW node. It does not miss any pulse but the duty cycle is variable. The first picture is with continuous acquisition, the second one is a snapshot.

    Should I expect a better waveform on the SW node (something more or less like a clock waveform) ?

    I have attached below also the ripple at the output voltage. All these pictures are without Cff.

    Thanks,

    Mihai

  • Hello again Ambreesh,

    I would like to understand better the logic behind this behavior.
    It is clear to me that the device needs a certain ripple at the output in order to work correctly.
    However as I understand Cff is used to add more ripple from the output to the feedback node so with Cff removed the ripple should be actually lower isn't it? So in this case why do I see a better behavior without Cff ?
    Regarding the 50 milliohm that you suggested in series with the output capacitor, this means that an output capacitor with higher ESR may be also a solution, right ? (the cost being a higher ripple at the output - but this may be acceptable).

    Thanks,
    Mihai
  • Hi Mihai ,

    Here is the link to an app note about Ripple injection , cause and effects .

    www.ti.com/.../snva166a.pdf

    I would suggest two main approach to solve the issue :

    1. Add the ripple injection circuitry . Also you may just add resistance in series with output cap . Just start with 200mohm and if you get the stable output keep decreasing the value .
    2. Verify the Layout .Make sure that the SGND and PGND is isolated and SGND is connected to PGND at on one point only . Please refer to datasheet and EVM User guide for more details . May be in your design the Switch noise is coupling to the SGND and causing unstable this behavior .

    Further if after trying the above mentioned steps , you are not able to fix this , i would suggest you to buy one EVM of LM3150 , make changes as per your requirement and evaluate the design .Once evaluation is done you can replicate the EVM layout in your design .

    ---Ambreesh
  • Hello Ambreesh,

    I have made several further tests:

    - I have changed the output capacitor to one with a higher ESR (50m) but the behavior is unchanged

    - I have added the ripple injection circuit and the SW node looks better (stable) but the behavior is still the same (the circuit cuts the output at very low load current)

    Please see also a picture with the PCB layout. The PCB has 4 layers and there is a full GND inner layer.

    Thanks,


    Mihai

  • Hello Ambreesh, do you have some further suggestion about this problem ?

    We should go to production but we have to fix this problem.

    As mentioned with the ripple injection circuit I get a nice stable output at the SW node but still the problem occurs. What could be the reason for this ?

    And why does it work OK at low current but not at higher currents ?

    How can I determine if the circuit goes into current limit protection mode or short circuit protection mode ?

    Best regards,

    Mihai

  • Hi Mihai ,

    LM3150 monitors current through the low-side switch and triggers current limit operation if the inductor valley current exceeds a user defined value that is set by RLIM and the Sense current.

    Is it possible for you to share the Switch node and inductor current waveform in both the condition ie when the device is switching properly and when the output is ramping up and down .

    Further please make sure that the Rilim value is correct on the board .  You can also  increase the current limit by increasing the Rilim resistor .(you may try using 4.7K or so) .

    On the layout side , below are my recommendations :

    ---Ambreesh

  • Hello Ambreesh,

    I do not have a current probe so for now I cannot take current measurements.

    About the RILIM - it seems to be correct from the Webench (2K7) and also from the calculations based on the datasheet, I have tested also with a slightly higher value (3K3) but there was no change. And since the circuit works at high current without Cff I would assume this means the current limiting circuit is OK.

    The main differences from the previous design (that worked) are the inductor (it was 3.3uH and now it is 10uH)  and the Mosfets. I wanted the circuit to be able to withstand  a higher input voltage so I set the parameters in webench such as higher voltage mosfets were selected (although the normal nominal operating input voltage is the same - 12V).

    Is it possible that there is a problem with the inductor and the mosfet choice ?

    Regarding the layout issues - do you believe the points that you mentioned above would justify this behavior of the converter ?

    Best regards,

    Mihai

  • Hi Mihai ,

    My suggestion would be get an EVM , replace the power components that are needed to be changed as per your design requirements and test out the performance .
    This way you would eliminate the issue of the layout and once the design is verified , you could follow the layout of the EVM in your design .
    ---Ambreesh
  • Hi Ambreesh, I just got the EVM, I will test it in the next few days.

    Best regards,

    Mihai