LTC3115-2 Protections


We have a LTC3115-2 that has a nominal 28V input and a nominal 28V output. It should be noted that we are trying to power the converter into a 750uF output capacitance (the 9ms soft start time is nowhere near enough). For a while, we saw no issue with this. Here are some waveforms:

Here the green is the current on the output of the converter but before the 750uF capacitance. The blue is the output voltage. You can see that the voltage does reach the target 28V, but there is hiccupping on start up due to some protection.

Here the yellow is the output voltage (no current measured this time). You can see there are 3 hiccups to reach steady state this time. 

At first we thought the hiccups were coming from the overcurrent limit, and it was simply retrying, and would do so indefinitely, or (in our case) until the output voltage was reached. This was originally in the sub 50ms time frame. We noticed however, at lower voltages and colder temperatures, it would actually take much much longer (like hundreds of milliseconds). Why would this be the case? What protection are we actually seeing here?


Added a small clarification on how much "longer" was
[edited by: Nino at 10:06 PM (GMT -5) on 3 Dec 2020]
    •  Analog Employees 
    on Dec 9, 2020 12:33 PM 2 months ago


    Can you share the schematics and operating range for this application. It would be easier to comment on the startup waveforms if we know the complete spec. 


  • Please see below.

    The input voltage range is 26V to 33V and the output voltage is 28V and output current is 0A to 1A. During startup, the load is near 0A however, and this startup waveform is independent of input voltage, and seems to be more temperature dependent. It also varies from part to part.

  • Hello!

    After further testing, we have started to notice more parts are failing in different ways.

    Here is a waveform (with the same test conditions as the original post) where it hiccups over 20 times before reaching the set voltage. In this waveform, yellow is the voltage, and purple is the current at the output of the converter into the 750uF capacitance.

    Here is another waveform (with the same test conditions as the original post) where the output fails to ever reach the setpoint. In this waveform, yellow is the output voltage, and blue is the current on the output of the converter into the 750uF capacitance. In this total failure, the hiccupping duty cycle is too low and the small startup load current (the device with the 750uF capacitance) is greater than the current injected by the converter during its hiccups. The voltage decays too rapidly when the converter is not switching, and the final voltage is never reached. 

    •  Analog Employees 
    on Dec 11, 2020 4:38 PM 2 months ago in reply to Nino


    This behavior seems odd. I would have to check this in our lab. I have a few questions , what is the input source and I see an external RUN command in the schematics as well? Can you also provide waveforms of VIN, RUN , Inductor current  with LOAD with and without the 750uF cap along with during the startup.

    Have you taken thermal measurements as you mentioned there is temperature dependency. Can you please share what correlation you find with temperature and at what data points. 

    Have you checked the behavior without the large output cap 750uF or changed the load to verify if the design is good. Can you also add some large input caps and see if it helps.

    Please share your location and company details and means to contact you. You can direct message me if you want. I would ask our field engineer at your location for technical support as you are seeing failures.


  • Anisha,

    We are working on getting the data you requested. In the mean time we have a new waveform to share that we believe to be the culprit.

    In this waveform, yellow is the output voltage, purple is the current into the 750uF capacitance, and light blue is the VCC/PVCC pin voltage. This waveform was taken after we switched out the 4.7uF capacitor in the schematic with a 10uF capacitor. Again Vin is 28V (much higher than required to account for dropout on the LDO for PVCC).

    You can see that while switching, it appears as if VCC just continues to drop below the UVLO. It is as if the the LDO is not attempting to regulate VCC properly at all...What is the bandwidth of the LDO? 

    I will DM you our details.