0
\$\begingroup\$

enter image description hereI'm designing a module that acquires power from two different sources: USB (5V) or I2C (3.3V.) The USB power routine is converted to two different 3.3V outputs (say, 3.3V1 and 3.3V2) by two different chips (a buck converter and an LDO regulator.) In order to let the device could also work with I2C power supply only, I need to combine the two 3.3V outputs to the same I2C power node. Are there any possible problems in doing so? I've tested on a board with a similar setup and it seems to be OK, but not sure for long-term work load.

I also uploaded the datasheets of the two chips (EA3036 and MEC6211 series) here.

ME6211 EA3036

\$\endgroup\$
6
  • 1
    \$\begingroup\$ The devil is in the detail. Show how you implemented your test board with a schematic please. \$\endgroup\$
    – Andy aka
    Aug 27, 2021 at 8:26
  • \$\begingroup\$ Attached. Please ignore the RC values and the rest two output channels of EA3036. The two 3.3V are directly shorted at the I2C port. I guess there should be some kind of isolation between 3.3V and 3.3VE, but not sure whether it's compulsory. \$\endgroup\$
    – Keyan
    Aug 27, 2021 at 10:11
  • \$\begingroup\$ Add data sheet links to the chips please. \$\endgroup\$
    – Andy aka
    Aug 27, 2021 at 10:14
  • \$\begingroup\$ Please ask a specific question \$\endgroup\$
    – Voltage Spike
    Aug 27, 2021 at 13:40
  • \$\begingroup\$ The EA3036 is a buck converter, which converts the 5V USB power to a lower voltage based on the value of the resistor divider R15 and R16 (but I haven't set the values yet). The MEC6211 is an LDO that outputs 3.3V from the same 5V USB power as well. Naturally, there may be some tiny differences between the two 3.3V output (if not shorted). The links are: \$\endgroup\$
    – Keyan
    Aug 28, 2021 at 0:39

3 Answers 3

2
\$\begingroup\$

You get no guarantees regarding the current distribution between two DC sources in parallel. Including not getting a guarantee that one device will source a humongous current and the other will sink it (if one device has an inherent ripple, this may alternate quickly with little apparent effect on the output voltage while burning large amounts of power).

So at a minimum you will want to make sure that your devices are not capable of sinking current and will not complain about overvoltage at its output. Also each device on its own must be able to bear the full load.

\$\endgroup\$
1
\$\begingroup\$

I could not find any spec sheets for the two ICs in the circuit diagram so all of this is based on assumptions. Also, in the schematic, are the 5V rails powering both regulators the same? If so, it would mean both regulators would always be on, as opposed to what you imply about having redundancy. It also matters because having both on always is a different story than having one on and other off. For example, some linear regulators have a diode from their output to their input in which case, if the linear regulator is off, it will act like a diode between the power rail and ground.

So in general, it is not a good idea to "parallel" voltage regulators, at least not without diodes (which may affect the output, depending on the circuit) but it never hurts to look into the details and also experiment and see what is going on.

In your specific case, having an LDO and the buck regulator with external feedback resistors mays allow this work. The typical buck regulator topology means there is a MOSFET inside so theoretically, the current could back feed into the buck regulator but only if the output grows much higher than the 3.3V. And again, this is the concern if the buck has no power, the LDO would then just send current into the buck regulator's mosfet through the body diode. If both regulators are both always on, the LDO effectively biases the feedback resistor network (R15/R16) and so the buck converter will simply adjust the on/off interval for the internal switch to maintain the output voltage.

A few things to watch out for:

  • If the output voltages of the two regulators are not identical (and they will not due to tolerances), there will be adverse effects as one regulator will do more work than the other
  • The internal control loops of either regulator could be affected by the forcing of voltage on the output and moved to a region of instability, which may come and go depending on the load.
  • Given the typical tolerance of the regulator output voltage, you may get two parts that work but another pair of parts that don't. This to me is the biggest reason to not utilize such a circuit.

Long story short, this is not the optimal approach to doing things. It may work in a lot of cases but there are all kinds of situations in which you can get in trouble.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Thanks for the detailed explanation! I guess I'll keep the current design until some explosion occurs, since for most of the application, the module will be supplied by I2C 3.3V. \$\endgroup\$
    – Keyan
    Aug 31, 2021 at 8:51
1
\$\begingroup\$

The LDO will not share power if another supply is slightly higher in parallel.

The DCDC converter may be underpowered if using feedback from the LDO so the dynamics of open loop LDO and buck with shared feedback may cause complications with dynamic loads with buck overshoot on hi to lo load LDO is always higher slightly.

\$\endgroup\$

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.