0
\$\begingroup\$

Electronics novice here... I hope this is a reasonable thing to ask here. I haven't "tried" anything because I'm not really having a problem. I'm just looking for guidance before I start.

I am cobbling together a device that will need 5, 3.3, 2.5 and 1.8 Volt power sources. The heart of this thing is going to be a Raspberry Pi Compute Module.

In the documentation for the Pi Compute, they give a fairly strong warning that the power supplies must come up (in time) from highest voltage to lowest to avoid forward-biasing internal diodes between sources to avoid latch-up.

I am planning to use devices like the LM2576 because they seem simple enough to use and have lots of online examples and application notes.

I have two multi-part questions...

First, is it reasonable to simply build discrete power supplies for each required voltage and cascade them? (12V input -> 5V -> 3.3 -> 2.5 -> 1.8). Or is there a better practice to follow when you need multiple voltages?

Second, does merely cascading the supplies ensure that they come up in order? Or is there something I should do (perhaps with larger in-parallel capacitors between them?) to create a short delay between power-ups?

Thanks!

\$\endgroup\$
  • 2
    \$\begingroup\$ you could cascade them, or start at 12V for each one or start at 5V for each of the lower ones or start at 5V and use a booster to get 12V. it all depends on the current required. \$\endgroup\$ – jsotola Feb 7 '18 at 20:25
  • \$\begingroup\$ The 12 volt input is a given (a design constraint). If I were to use the output from the 5V supply for each of the others, how could I ensure they come up in the right order? \$\endgroup\$ – Matt Redmond Feb 7 '18 at 20:34
  • \$\begingroup\$ can you post a link to the doc? \$\endgroup\$ – jsotola Feb 7 '18 at 20:59
  • 1
    \$\begingroup\$ Cascading works, although you can also get special power sequencing ICs for this. I suspect if you look at the normal Pi board you'll see they've cascaded them. \$\endgroup\$ – pjc50 Feb 7 '18 at 21:04
  • 1
    \$\begingroup\$ e.g. linear.com/product/LTC2924 \$\endgroup\$ – pjc50 Feb 7 '18 at 21:05
1
\$\begingroup\$

Cascading supplies does guarantee that they come up in order but you can't do it in your specific case.

The minimum input voltage for the LM2576 is between 3V and 4V depending on temperature. https://www.onsemi.com/pub/Collateral/LM2576-D.PDF

(See figure 11 on page 6 of datasheet)

Therefore you can go from 12V to 5V to 3.3V.

The 2.5V supply may or may not be able to run from 3.3V.

The 1.8V supply probably won't be able to run off of 2.5V.

If you want to sequence the supplies by running one off of the other then you need to pick a supply that has a minimum input voltage that goes all the way down to 2.5V.

Also it should be noted that the overall efficiency goes way down if you cascade so many supplies. For example if each of your supplies was 90% efficient then...
You get 12V at 90% efficiency.
You get 5V at 90% of 90% = 81% efficiency.
You get 3.3V at 90% of 81% = 73% efficiency.
You get 2.5V at 90% of 73% = 65% efficiency.
You get 1.8V at 90% of 65% = 59% efficiency.
And so on...

A better way to sequence the supplies and obtain high efficiency is to run them all off of 12V, but control the ON/OFF pin of each supply from the output of the other supplies.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
0
\$\begingroup\$

I'm going to answer/close this because it seems pretty clear that cascading will work and if it's not optimal at least it's not idiotic, which is what I wanted to avoid.

I'll watch current requirements and make sure each supply can handle its own devices as well as all of the downstream power supplies and their devices.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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