Post Go back to editing

LTC3786 Delayed start / ITH pin stuck at 2.8V during startup

Category: Hardware
Product Number: LTC3786

We currently have an LTC3786 implemented as a 12 or 24V boost converter.  The input voltage for the device is either 3.7V (12Vout) or 11V(24Vout). This issue was noticed while testing the 12V configuration (the 24V config has not been tested yet).

The issue happens during start-up where we set the RUN pin from an MCU high to start the converter and the convert SS pin starts to rise linearly as expected, but as soon as the SS pin voltage rise above the FB pin voltage the ITH pin shoot up to around ~2.8V and stayers there with no switching output for around 200mS at which point ITH pin come back down until switching begins. At this point the SS is long over, and the start-up is extremally aggressive, drawing 7-8A from the supply. This behavior doesn't always happen, but it happens fairly frequently, and we can't seem to find the conditions that cause it. 

NOTE: There is no load on start-up other than the output bulk cap.

Above scope shot is an overview of the startup event, where the switching doesn't start until around 200mS after the run pin is set. 

Above scope shot is a zoom on the area where ITH comes back into a reasonable range and switching beings aggressively charging the output cap. 

Above scope shot is a zoom in on the beginning where ITH can be seen shooting right up to around 2.8V while the SS pin voltage continues to rise.  

A "warm start" where the output is already close to the regulation point doesn't seem to cause the same issue. As seen in the above scope shot.

Schematic of the design. Note that V5P0 is actually 5.6V used to bias the converted before the boost converter is started, VBIAS is typically around 5.4V before the converter is turned on. 

Any help on this issue would be great, without a full picture of the internals of the part it's hard to debug much further. 

Thanks!

  • After some more debug, I have some additional information. 

    1. This issue doesn't seem to occur at higher input voltages. The issue was originally noticed at 3.7V input. But at 4Vin this doesn't happen.

    2. I think the issue may be related to the BOOST voltage/charge pump. It's clear that at 3.7Vin it's not enabling and charging up the BOOST cap, where at 4V and higher it is. 

    3.7Vin 

    In this above image we can see the issue occur, where BOOST is simply tracking INTVCC. ITH can be seen railing at ~2.8V as with my prior scope shots.

    4Vin

    In this scope shot we can see a normal startup where after a short time after INTVCC turns on, the BOOST voltage increases as the charge pump starts. 

    Any help on this would be great, I have been trying to track this issue down for days. Thanks!

  • An additional update after more debugging. I swapped D3 from a BAT54 to a BAS140 and noticed that issue is improved at room temp (25C) and the supply will now start at 3.7V and low input voltages. However, the problem still persists at colder temps (Around 0C).

    I'm not sure why this change made a difference, there is not a lot of noticed change in the boost voltage or other external voltages. Not sure if this is due to the diode leakage or some of property of the diode. Could really use some guidance on what could be causing this or further debugging steps. 

    Thanks

  • Some more information. I had some suspicions on power supply sequencing, possible race condition. This is because in our application VBIAS comes from a different supply than the main input to the boost converter that supplies SENSE+.

    Again, colder temperatures make this happen consistently but power up sequency does have an effect.

    1. In normal operation when the power switch is turned on VBAT comes on and SENSE+ gets powered with ~3.7V. After our MCU boots it then enables the 5V supply for VBIAS and then sets the RUN pin high to turn on the supply. In this condition the supply will not start at all at cold temps or start late as shown prior posts.

    2. If however, SENSE+ is kept at 0V and VBIAS is then powered and RUN set high before SENSE+ gets power then convert will start properly every time, even at colder temps.

    I have no insight into the inner working of the chip other than what is in the datasheet, but is it possible that the switching logic is getting into an odd state at colder temps with my specific power sequence? 

    UPDATE 5-21-24

    Further exploring sequencing. I found that if I reverse the order of VBIAS, RUN (i.e. set RUN high, then enable VBIAS) it will work every time at low temperatures. This really strange as we would expect RUN to work as an enable pin for the supply.