Actions
Forum Thread Details
  • Replies 4 replies
  • Subscribers 287 subscribers
  • Views 544 views
  • Users 0 members are here
Related

LTC3127 design won't work with voltages above 4.1V

DaveMcLaughlin

Hi All,

 

I have a buck/boost power supply design based around the Linear Tech LTC3127 and I can not get it to work with a DC input of 5V.

 

The design is battery powered with a Li/ion battery as the primary supply and a secondard DC supply for when charging the battery. I am using a power path charger from Microchip for the battery charging etc. This outputs 2.5 to 4.1 from the battery and when you plug in the 5V external DC supply, it provides this 5V at the output. This output is fed into the LTC3127. I have attached the schematic of the power supply design (the lower part) excluding the battery charger part.

 

With only the battery as the source of power, the LTC3127 outputs a nice clean stable 3.3V for the logic supply at each switch on. If I then plug in the 5V external power, the charger IC routes this to the output and the LTC3127 output suddenly drops to 2.6V and remains there until I unplug the 5V supply and go back to battery, where it recovers back to 3.3V. This output is nice and clean on the scope. A steady DC output of 2.6V.

 

If I connect a bench supply to the LTC3127 input, bypassing the charging circuit, and set this to 4.1V and switch it on, the LTC3127 outputs a nice clean 3.3V as expected. If I increase this up to 5V the output remains at 3.3V.

 

If I set the bench supply to 4.2V or higher and switch on, the LTC3127 initially goes to 3.3V and after about 1 second, it drops to 2.6V and remains there. If I drop the supply down to 4.1, it jumps back up to 3.3V.

 

If I run this exact same design through the LT Spice simulator it works with 5Vdc as the input.

 

I have tried increasing the output capacitance to 1000uF LOW ESR and no change in the behavour. I am only drawing approx 300mA from the 3.3V supply but as the design is capable of 1A this is well within the specification and the capacitance is calculated to be sufficient as the current is pretty steady state.

 

 

Can someone have a look at my schematic and see if I have done anything wrong and maybe have a look at the datasheet for this device? The design I have used is basically out of the datasheet.

 

Cheers,
Dave...

Attachments:
imageLTC3127design.pdf

  • As a follow up to this I have discovered today that touching the PROG input resistor with my finger with the 5Vdc input active, the device bursts into life and outputs 3.3V and remains stable and operational.

     

    Still no solution to the above problem though as it just won't go above 1V on powerup with the external 5Vdc supply. Battery has been working great.

     

    Need a working solution or I will have to consider an alternative supply via a small daughter board but that means new costs for the project!!! image

     

    Dave...

  • in reply to DaveMcLaughlin

    Hi Dave,

     

    It appears at first glance you might be having a problem with the compensation.  Have you tried using the compensation for the general purpose application on the bottom of page 17?

     

    -Mike

  • Hello Dave,

     

    I think your input capacitor may be the wrong type or that the input tracks may be too long - your schematic shows a polarised capacitor  - if it's a low esr tantalum you might get away with it but a normal aluminium electrolytic won't do.

     

    To quote the LTC data sheet:

     

    Multilayer ceramic capacitors are an excellent choice for
    input decoupling of the step-up converter as they have
    extremely low ESR and are available in small footprints.
    Input capacitors should be located as close as possible to
    the device. While a 10μF input capacitor is sufficient for
    most applications, larger values may be used to improve
    input decoupling without limitation. Consult the manufacturers
    directly for detailed information on their selection
    of ceramic capacitors. Although ceramic capacitors are
    recommended, low ESR tantalum capacitors may be used
    as well.

     

    Try a 10uF ceramic cap as close to the pins on the chip as you can get and see if it helps.

  • Hi Dave,

     

    Assuming input capacitors are close to the part as described in the datasheet the source of your headache is most likely your compensation.  When using proportional control (resistor to ground) the loop depends on the dominant pole to be the output capacitor, and to roll off the gain well before the phase drops off from the right half-plane zero.  The LTC3127 demoboard is configured with proportional control to compensate for the three large tantamount capacitors on the output for a pulsed GPRS application.  When using the minimum amout of capacitance (2x22uF) integral compensation has to be used (cap to ground).  This give a high DC gain and sets the dominant pole.  The application on the bottom of page 17 in the datasheet shows the exact application that you are looking for and were trying to accomplish.

     

    Another source of problems could be the 2-way header in series with vin.  These usually have a high impedance and can cause significant problems with switching converters.  I would recommend that you tie the VIN of both of your converters directly to the battery and use the header to pull up the SHDN_BAR pins.  This would require a pull down resistor on the LTC3127 pin, or the two converters could share R29 (LTC3127 pin is a logic pin with no current).

     

    I wouldn't give up on the LTC3127 when compensating it with integral control would fix your problem and not require more output capacitance.

     

    -Mike