• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Subscribers 62 subscribers
  • Views 1966 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.

LM3488 SEPIC current problem

PPinto
Prodigy 190 points
Other Parts Discussed in Thread: LM3488

Hello All,

I am using a LM3488 to build a SEPIC converter with 5V @ 5A output. I have implemented the schematic from the WEBENCH Design Report, but I am having some problems. The converter works fine with 5V input and for very low load current. By low current I mean to power a LED and two other regulators (3.3V and 1.8V) without load.

I have a second board with a processor and other things. When I plug the processor board to the power board the output of the regulator drops to 2.5V. If I increase the input to 7.5V or above, the regulator works again. The processor board consumes approximately 100mA, that is not even close to the current that the regulator should supply. 

Any ideas of what is happening?

Attached are the two schematic sheets I used in this design (one from the WEBENCH and the one I implemented).

Some of the parts I used are different from the ones referred in the WEBENCH report. Those are listed bellow.

D1 - SS5P5

Cout - EMVL160ADA471MJA0G

M1 - SI4850EY

Lin - MSS1048-682NLC

Lout - MSS1210-153MEB

8233.Project.pdf

Thanks,

Pedro

webench_design_LM3488.pdf
  • 0
    TI__Intellectual 2930 points

    Hello Pedro,

    I will take a look at your schematic to see if anything jumps out that may cause this.

    In the mean time I have 2 requests that would be helpful:

         Could you make sure that your Power Supply is not current limited and can support the current needed.

         Do you have access to an oscilloscope? If so can you probe the Switch node (drain of the FET), input and output nodes when the part is not properly function and when it is?

    Best regards,
    Tommy Jewell

  • 0 in reply to Tommy Jewell - WEBENCH Design Center
    Prodigy 190 points

    Hello Tommy,

    Thanks for your answer.

    My power supply is limited to 5A, so I guess that is enough for this case, since I just need a few hundred mA. 

    Regarding the measurements that you asked for, they are attached in a zip file. All the results have the input voltage on channel one and a little description on file name.

    4380.Results.rar

    Thanks,

    Pedro

  • 0 in reply to PPinto
    TI__Intellectual 2930 points

    Hi Pedro,

    When it is not working are you seeing any oscillations on the output voltage or is it staying consistent as if it were regulating to 2.5V? (Pics only showed a couple of switching periods so I just wanted to confirm).

    Well the duty cycle is getting noticeably smaller so either there is an instability or something is limiting it. 

    My first check would be on the Isense pin. This triggers the current limit. If you are seeing around 160mV at this pin when it turns off it means its current limiting. Spurious signals can get into this pin and cause this to happen so your layout is very important here.

    Layout is also very important around the FB trace, make sure it is not routed close to any power traces because as the current increases they can begin to affect the signal. What is the voltage on the FB pin when its not working?

    Do you have any layout files or pictures?

  • 0 in reply to Tommy Jewell - WEBENCH Design Center
    Prodigy 190 points

    Hello Tommy,

    Regarding your questions/comments,

    1 - I can't see any oscillations on the output, just a little ripple which I think is normal.

    2 - I already have noticed that the dutty-cycle is smaller than it should, when the input voltage is 5V. For higher input voltages it seems to be fine.

    3 - Regarding to the layout, it could be a problem since the Isense and FB pin are just under the Csep capacitor, just where the power flows.

    I also realized that the LM3488 buzzes when the voltage is under 7V and above 17V.

    Anyway, I have probed input, output, Isen pin ans FB pin, when it is working (Situation1 folder) and not working (Situation2 folder). The results are attached.

    Also attached are the screenshots of the PCB.

    0121.Files.rar

    Thanks for your help.

    Pedro

  • 0 in reply to PPinto
    TI__Intellectual 2930 points

    Hi Pedro,

    Yeah, the SEPIC cap is is not only a power path, but also a high di/dt path, which means it will introduce EMI into anything right next to it (or under it if you don't have a ground plane in between the layers). You want to keep any sensitive analog traces away from this.

    I am also a little confused on the Sense resistor connection (R31), is there a Via going from the FET's source to this (its not depicted in the picture)?  This is also a high current path, I would recommend putting the resistor on the same layer as the FET if possible (as to not make the high current go through a via)  and then running the analog trace to r30 through the Via.

    From the waveforms you sent I don't see anything that points out as THE culprit, but a good layout will usually fix many problems.

    Attached is a presentation on SMPS design for EMI, it has some good notes on layout for EMI mitigation.

    Best regards,
    Tommy2465.Switching Power Supply Design_ EMI_Marc Davis-Marsh_03_02_11.ppt 

  • 0 in reply to Tommy Jewell - WEBENCH Design Center
    Prodigy 190 points

    Hello Tommy,

    Yes, there are two vias from the source FET to Sense resistor, I don't know why they don't appear.

    Thanks for the presentation, it has very good information and explanations about EMI. I just read half of it and found a lot of things I'm doing wrong.

    I´m going to redo the layout and lets see if it works, I hope it does.

    Best Regards,

    Pedro Pinto

  • 0 in reply to Tommy Jewell - WEBENCH Design Center
    Prodigy 190 points

    Hello Tommy,

    I have redesigned my PCB and, unfortunately, it is not working. 

    Now I have two strange behaviors. First, at start-up and without any load, the converter just work if my power supply is able to deliver about 2.5 amps. My question is, does the SEPIC converter need large amounts of current to start-up, even when there is no load?

    The second strange behavior is about the current that the converter can supply. From 5 to 5ms my system needs 2A for a period of 5 us (communication burst). When that happens the system just shutdown like if it can't supply that amount of current. I have measured the ISEN pin and this time I am almost 100% sure that the problem is there, since it reaches about 300mV which is much higher than the allowed 160 mV. I just can't understand why it happens because the converter was supposed to deliver 5A.

    Attached are two files, one is the schematic and the other one is the PCB layout. Can you please check if I've made anything that could compromise the good working of the converter?

    3884.Pictures.rar

    Another doubt I have is, how does the Ccomp, Ccomp2 and Rcomp values are calculated?

    Hope you can help me.

    Thanks,

    Pedro

  • 0 in reply to PPinto
    Prodigy 10 points

    Hello Pedro and others

    I have made a boost converter with LM3488. It is not working at all!

    First I tri to syncronize it. It did'nt work. Then I tried free run..  no response..

    It is not oscillating at all!

    I have seen faulty SMPS IC from National/TI before and I dont want to waist my time on a faulty IC.

    I have made boost converters before without any problems, so I dont think I made a mistake.

    TI should not try to blame you for your design when they have a problem with the IC.

    To TI: You have a quality problem! Fix it.

    Best regards

    Hans

  • 0 in reply to Hans Iversen
    Prodigy 190 points

    Hello Hans,

    Thank you for your comment. Since I'm not the most experienced engineer in DCDC design, in fact I have made wrong choices on my first design. The second fact is that I have corrected the design and it still doesn't work very well. I have made the calculations over and over again (according to TI datasheets), used different approaches to calculate compensation network, have design a load simulator board...but it doesn't work as I expect. I'll not blame TI because, as I said before, I'm not the most experienced DCDC designer, but I'm getting mad with this converter.

    Once again, thank you for your comment.

    Best regards,

    Pedro

  • 0 in reply to PPinto
    TI__Intellectual 2930 points

    Hi Pedro,

    It appears I completely missed your reply way back in September. I'm very sorry. Are you still having the same issues with this board?

    Best regards,
    Tommy