Problem Statement

I have two power rails: 3V and 12V. Both of them have variable loads (e.g., microcontrollers, motors) and power sources (e.g., solar cell, battery) on them. The typical max current is ~1A on both rails. I want to couple them, so that if there is more load on the 12V side than the 12V sources can supply, but the power sources on the 3V rail can provide power, power shall be converted from 3V to 12V and vice versa. Basically, what a transformer would do for AC power.

A block diagram of the setting might look as follows:

block diagram of the scenario

As requested from the comments, here are a few more details on the scenario:

  • Any kind of consumer/producer combination must be possible. Obviously, the system needs one producer on one of the rails.
  • Max total current on the 3V does not exceed 2A.
  • Max total current on the 12V does not exceed 1A.
  • If there is higher load than supply, any kind of degradation is acceptable. A voltage drop would be the best scenario, but any kind of failure is ok, as long as the hardware does not take damage and there are no excessive overvoltage spikes.

The potential solutions I found so far are:

  • A dual-active bridge: Here I only found components and controllers that are designed for several kW of power, which is obviously too much.
  • Bidirectional converters like the Richtek RT6190GQW. However, if I understand the datasheet correctly, these need external switching between Buck and Boost mode. (I imagine one would do this with a microcontroller and a simple control like if v_a > 3V then boost_enable, else buck_enable, but I have the feeling that this is prone to oscillations. Maybe a comparator circuit is the solution here?)


So the question basically is: Is there is a simple IC or circuit that could achieve this behavior for a low-power scenario. Simplicity and few parts would be a design goal. (I'm rather inexperienced with DC-DC converters, so perhaps I'm missing something fundamental here.)

Related questions:

  • 1
    \$\begingroup\$ Please draw a block diagram. It's not clear why you need to reverse any power flow. Just boost your solar panel to the 12 V rail and buck down to 3.3 V? \$\endgroup\$
    – winny
    Oct 10, 2023 at 11:44
  • \$\begingroup\$ @winny The solar panel might be detached, and then are only consumers on the 12V rail remain (and vice versa). Or the consumers are detached and only producers are on one side of the rail. \$\endgroup\$
    – Jounathaen
    Oct 10, 2023 at 13:28
  • \$\begingroup\$ Please show with a block diagram. Perhaps a synchronous buck converter with fixed duty cycle would solve the problem for you. It would happily boost backwards. \$\endgroup\$
    – winny
    Oct 10, 2023 at 13:47
  • \$\begingroup\$ I added a block diagram. Does this help in understanding the desired scenario? \$\endgroup\$
    – Jounathaen
    Oct 10, 2023 at 14:04
  • 1
    \$\begingroup\$ Can you clarify all your scenarios? In the question you only mention one rail siphoning some power from the other, but in the comments you say that the 12 V generator (solar panel?) can be disconnected... We need to know all producers capabilities, max total current of the consumers (for both rails), and all possible combinations of producers being present or not. It would be helpful to also understand what you want to optimise, is it power efficiency? Moreover, what happens if demand on one rail is too high and the other rail cannot compensate? \$\endgroup\$ Oct 10, 2023 at 14:17

1 Answer 1


Like winny said, a synchronous buck converter with fixed duty cycle can help here.

The duty cycle defines the relation between the two voltages, however there are problems remaining:

  • If the 3 V producer is offline and the 12 v producer is not stable, all deviations are forwarded to the 3 V side and vice versa. This is dangerous for the 3 V side if the solar panel delivers e.g. 16 V.
  • If both producers are online, very small variations in the duty cycle will have a large impact on the energy flow direction. If both were rechargable batteries, one would drain the other until the voltages match the duty cycle.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ I never had access to a real computer to add the schematic as a real question. Do you use voltage controlled switches instead of MOSFETs to aid some simulation? You covered most things already, but what I wanted to add is that for a fixed duty cycle, it would depend on the rail needing current to drop voltage for any power transfer to take place. \$\endgroup\$
    – winny
    Oct 11, 2023 at 6:58
  • \$\begingroup\$ I think, this is exactly the behavior I'm looking for. Thank you! \$\endgroup\$
    – Jounathaen
    Oct 11, 2023 at 8:29
  • \$\begingroup\$ @winny Yes, I finally understood how this PWSREPEAT syntax works and wanted to play with it. Up to now, I didn't find another simple way to create a PWM driven half leg including FET dead times in CircuitLab. \$\endgroup\$
    – Jens
    Oct 11, 2023 at 12:59
  • 1
    \$\begingroup\$ @Jounathaen I use this topology for active balancing of series connected LiFePO4 cells. I set 50 % duty cycle and the current arrives close to zero if both cells have equal voltage. BQ500101 or CSD95375 from TI or SiC 521 from Vishay work great here. \$\endgroup\$
    – Jens
    Oct 11, 2023 at 13:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.