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 Inverting Buck / Boost circuit not working

Other Parts Discussed in Thread: LM25574, LM3478

Using Webench, for the design and subsequently building, I get very low O/P voltage.

I have +10.8V to +13.2V input and attempting to obtain -12V O/P upto 200mA.  Webench recommended the LM25574 part and this is the circuit I have built-up.  With a light load (<< 10mA) I have an O/P at approx -1.4V.  If I increase the load value to obtain ~100mA, the O/P drops to approx -1V.  Within normal prefered values I have followed the webench design.  I have 3 Prototype PCB's all exhibiting the same problem.  Not even sure where to start looking.  I have already downloaded several TI docs. on these modes of operation, but cannot see any particular errors.

This is the complete schematic where the LM3478 Buck / Boost works just fine but the following Invertor does not.

4606.3731.WebenchReportsServlet.pdf

Any help would be greatly appreciated !!!

Graham

 

  • Hello Graham, 

    I will forward this to our webench team. 

    Also, please provide the board layout you are using. 

    Regards,

    Denislav 

  • Hi Denislav

    Thanks for your response.

    The best I can do at the moment, is to provide a PDF of both the Top PCB layer with component idents and a PDF of the Schematic diagram.

    3678.7713.PCB Idents - BigPSU-1A.pdf8468.7142.Schematic - BigPSU-1A.pdf

    Many thanks in advanvce

    Graham

     

  • Hi All

    I wondered if there is any news regarding the above problem?

    I still don't understand why it wont work.

    Regards

    Graham

     

  • Hello Graham, 

    I will review the layout later today and provide some feedback.

    Regards, 
    Denislav

  • Hi Graham, 

    Here are a few layout/debugging suggestions.

    1. For debugging, please double check each node on the circuit and see if you have any shorts to GND on nodes that should not be connected to GND. The layout has this pour of ground copper with very small islands/copper necks of GND traces poured in between all pads. I would recommend limiting the minimum copper island/neck size so that you don't have the potential of exposed GND traces shorting out. I have circled some of these necks here to give you an idea of what I mean:

    The fact that you are getting the same results on several boards go against this random shorts to GND idea, but it is worth checking anyway.

    2. The bypassing of the VIN and VCC pins of the IC can be greatly improved if you run a thick -12V trace under the IC in the "north" direction. Then place the bypass caps C22 and C29 on the north side of the IC. Take a look at this for illustration:http://e2e.ti.com/support/power_management/simple_switcher/w/simple_switcher_wiki/2215.lm5574lm25574-pcb-layout.aspx

    The illustration is for Buck configuration but the idea is the same.

    3. Use thicker traces for the AGND pin (pin 9) and PGND (pin2) of the IC. As is, all the current has to go through this tiny trace:

    This can be fixed by the configuration described in point 2.

    Do you have any routing on another layer?

    Can you capture scope shots of the following IC pins: VIN, SW, Softstart, Ramp?

    Regards, 
    Denislav 

  • Hi

    Whilst I might agree that some of the tracking could be improved, there is clearly no unwanted connections between the circuit and the Ground-Plane.  I have more than 1 PCB buid exhibiting the problem.  Yes, the de-coupling could be closer with a layout change, but none of these things should actually stop it working.  The other circuits I have used - based around other TI I/C's:  a 10V to 30V  Variable I/P, 12V O/P (as a Sepic) and a 10V to 30V I/P, 0 to 18V @4A O/P all work perfectly well from the Webench designs.  There is clearly something wrong with the design in question.  At this time I am not in a position to grab waveforms but be aware that the VIN as mentioned above comes from the Sepic circuit just mentioned.  This also drives other circuitry which is all functioning correctly.

    Regards

    Graham