1
\$\begingroup\$

I am trying to design a circuit that u would like to be as reliable as possible. One feature I would like to add to my circuit is an additional voltage regulator for redundancy purpose. The problem is that the voltage contains a 1.2 V rail that powers the MCU. I have considered using Schottky diodes and ideal diodes, but each of them have their own problems.

My biggest worry with the ideal diode is that the voltage-drop over the diode will vary depending on the MCUs current consumption. Those small voltage variations represent a large percentage of the whole, and MCU should have a very low ripple. In the added picture, the circuit is using a ORing design. But I’ve already decided to change it.

The Schottky diode has a much more stable voltage/current relationship, but the threshold voltage can very a lot depending on the temperature. Are there any other options that I am overlooking?

Voltage Regulator Circuit

\$\endgroup\$
9
  • 1
    \$\begingroup\$ MOSFET switch. They're used all the time to switch power sources to a load. \$\endgroup\$
    – SteveSh
    Sep 1, 2021 at 15:45
  • 1
    \$\begingroup\$ "My biggest worry with the Ideal Diode is that the voltage-drop over the diode will vary depending on the MCUs current consumption." Don't worry. "Ideal diode" is not an actual diode, but a controlled FET, thus ideal. Read the datasheet. \$\endgroup\$
    – jay
    Sep 1, 2021 at 16:10
  • \$\begingroup\$ FYI, electronics.stackexchange.com/questions/472127/… \$\endgroup\$
    – SteveSh
    Sep 1, 2021 at 16:11
  • \$\begingroup\$ @jay I have read the datasheet. The problem that mosfets have an ON resistance, and when operated with low voltages, that resistance can large enough (~100mR) create a reletively large voltage drop. ti.com/store/ti/en/p/product/?p=LM66100DCKT \$\endgroup\$
    – user127515
    Sep 1, 2021 at 17:00
  • \$\begingroup\$ Consider using an ideal diode controller with an external mosfet. That will allow you to select a device with a significantly lower Rds-on. There are also Buck converters that can be directly or'ed at the output, so you switch the two sources using EN on the buck converter and while there is still a mosfet in the switching path, it is inside the control loop for the buck converter so you don't have any voltage drop at the output. \$\endgroup\$ Sep 1, 2021 at 17:02

2 Answers 2

1
\$\begingroup\$

Ideal diodes are the way to go in this case.

You need a controller whose power is independent from the current path to make the oring. For example LTC4352 "0V to 18V Supply ORing or Holdup" enter image description here Use 5V (You have 5V available before the DC/DC) to supply the Vcc input and then Vin will work reliably with 1.2 V
This IC needs external mosfet, so you are free to pick a mosfet with the best Rds ON that you can afford. The mosfet will be turned ON with a Vgs bigger than 5V, which is enough for most mosfets. enter image description here
When one thinks about reliability, latent faults are to be avoided, so you need to find a way to diagnose whether or not the circuit is working properly, the outputs \status and \fault will give you valuable information.

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

EDIT: I just noticed the other linked article, which basically does what I explain here. I'll leave this here, though, maybe it's still useful.

An approach I've seen used to parallel DC-DC converters is by taking feedback not from the converter's own output, but rather from after the diode-steering arrangement:

schematic

simulate this circuit – Schematic created using CircuitLab

This causes the converter to compensate for the voltage drop across the steering diode, so you still get your 1.2V overall output.

However, we choose the fraction of feedback in each converter to be slightly different, so that one of the converters is presented with a slight "over-voltage", causing it to cease switching while the other converter functions normally.

In the above diagram, BUCK1 will operate normally, but BUCK2 will effectively be disabled, because its feedback signal exceeds its Vref. If BUCK1 fails, though, the output will drop slightly, enough to bring feedback for BUCK2 below its own Vref, at which point it starts up, taking over operations.

I have no idea if this is acceptable practice for any old switcher though, so I present this only as something to consider.

\$\endgroup\$
1
  • \$\begingroup\$ The possible problem with this approach is that you could have the control loops for the two converters fighting each other. This could lead to unstable voltage at the OR'd output. Unless you know the loop characteristics of each power supply and can analyze the combined stability of the system, or unless the power supplies are specified to operate in parallel like this, you may be opening up a can of worms, \$\endgroup\$
    – SteveSh
    Sep 2, 2021 at 13:18

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.