LTC3103 Drop-Out Seems very high...


I am working on a design using the LTC3103 to provide a 3V supply at relatively low currents (50mA is a reasonable estimate).  We received our iniitial prototype design and have found that the part loses regulation, even with no load, somewhere around 3.4V on Vin.      

Given that this is a battery application, we'd like to get as much life as possible, and losing regulation at 3.4V is not going to be acceptable.  Our input range reasonably can be anywhere from 7.5V down to "nearly" 3V.       

LTSpice simulations show that with a 100mA load we should be able to easily get Vin below 3.2V, and approach 3.1V, before we start seeing any droop in the output supply.   For reference, we are pretty closely following the reference design, using a Taiyo Yuden NRS3015T100MNGHV 10uH inductor and 0.1uF 50V ceramic cap for the boost cap.  I have tried adjusting the boost cap value down to 0.022uF as provided in the datasheet with no considerable change.        

I'm not seeing anything obvious in the datasheet which would indicate an issue, and the behavior has been confirmed across multiple boards.  

Any thoughts on possible causes for this, or has anyone who used this part had similar experiences?


  • As a follow-up, there is a graph entitled "Minimum Input Voltage at Maximum Duty Cycle vs Load Current" which addresses duty cycle and the resulting "minimum" input voltage which seems to be consistent with the physical results we are seeing.  The current suspicion is that the LTSPICE model probably does not take the duty cycle into account, and probably reflects the performance of the part when approaching a 100% duty cycle.    

    Given this it appears we need a different part for our application, but if anyone has further insight we're definitely still interested in understanding if another solution is available.

  • 0
    •  Analog Employees 
    on Jan 6, 2020 4:28 PM in reply to andersoj68

    Hello Andersoj88

    The diagram you mentioned is correct and additionally you should check following parameter in the ds:

    As you have not much means to overcome your problem with a standard Buck regulator (n-channel based, DC always <100%) you might consider using a BuckBoost regulator like the LTC3129 for your application. Your guess with the model limitation might hold true, but this is beyond my capabilities to check right now.

    kind regards