LTC3643 Parallel Vcap

Hi,

I plan to use the LTC3643 in a design. I have three supplies, each one with its own LTC3643. Is it possible to tie the three Vcap pins of each supply together and to put only one storage capacitor for the three supplies ? Have you already try this configuration ? When charging (or discharging) the cap in current mode, I think it won't be a problem but when it enters voltage mode, it can be critical. As I understand, when the cap reaches its programmed level, it works as an hysteresis mode so maybe, when the cap is fully charged, the circuit with the highest programmed threshold will run whereas the others will be in sleep mode.

Regards,

Adrien

  • 0
    •  Analog Employees 
    on Oct 26, 2020 1:55 PM 29 days ago

    Interesting application. This is something that I have not tried and have more questions than answers. Do the three supplies have the same input? I expect the three supplies have different output voltages, loads, and maybe even backup time requirements. Is a single storage cap going to provided enough energy for the three supplies? I do have a few concerns about how the PFI might occur at different points, brown out conditions that might cause one supply to reach backup an the others might not. If one PFI is used to control the other regulators might help with that. In CC mode the parts should charge up the CAP no problem. The three chargers can enter CV mode and enter burst mode at different times but I would not expect any misbehavior with the charges but this has not be tested. This should be evaluated thoroughly before tried. 

    If the loads are light enough could two regulators be fed from the output of a single LTC3643? This might be a simpler solution and only need one input FET.

  • Here is a simplified synoptic of the second application:

    First application: each supply works independently with its own storage cap

    Second application: all 3V3 are paralleled to deliver more power. As the tolerance on the caps (+ internal reference + feedback resistors...) is important, I wanted to tie the Vcaps together such that the holding time will be the same for all supplies. But it raises a lot of question. Each supply doesn't have the same input voltage (outputs of auxiliary flyback), so they won't fall at the very same time and triggers for the back-up will be a bit different. It might not be a problem but I have to test it. What happens if the three Vcaps are in CV mode (I have to test it as well if you never test it)? Or maybe as you mentioned, it will be easier if only one LTC3643 (let's call it "the master supply") hold the intermediate bus and to share this bus between the 3 supplies (it raises other questions because I can't share this bus in normal mode as it's like paralleling 3 voltage source...). One detail, I don't plan to use the LTC3643 to limit the load, in other words, the load is located before the shunt of the LTC3643.

  • 0
    •  Analog Employees 
    on Oct 27, 2020 12:59 PM 28 days ago in reply to Adrien29

    Interesting application. How different are the three aux supplies and their loads?

    How long is the backup time needed?

    Is there a common PGOOD signal that can be used between the three supplies to set them all to backup at the same time?

  • The three aux supplies are 12V with a tolerance of +-2%. The min load is 4W for the internal circuitry. And the load on one 3V3 is about 12W. The backup time depends on the user application (the idea is that he put whatever he wants, maybe I'll have to specify a maximum volume of capacitors) so I have to adapt the compensation network to be stable with a large range of capacitors.

    I have one demoboard and I tested the following configuration to emulate two LTC3643 in parallel:

    I didn't notice any misbehavior. When the supply is below the regulation threshold of Vcap, the LTC3643 enters burst mode operation. When the supply is above, the LTC3643 stops switching. I also tried to start the LTC3643 (using the RUN pin) when a voltage is already present on Vcap, and nothing wrong.

    It might not be critical if the three LTC3643 don't enter backup mode at the very same time. I will try this with one second demoboard.