0
\$\begingroup\$

Forgive my obvious naivety below, as I am a software engineer and not electronics.

I have an application that draws a fair bit of power (~50A@48V), and I would like to support the use of multiple external power supplies of different source types. For example, a 12VDC combined with a 120VAC supply. The number of supplies should be scalable and not limited to 2, and I need to be able to distribute the load proportionally across them (e.g. 30% of the power from the 12VDC supply, 70% from the 120VAC supply).

I can develop circuits that convert any of these intended sources to a common voltage (e.g. 48VDC), that's not a problem.

I understand that it is particularly difficult to balance load across multiple supplies in parallel, as the supply with the slightly higher voltage (always the reality that they will never be perfectly identical) will end up taking the entire load. Even if you can match the output voltages perfectly, they will split the load evenly and you'll have no control over how much power each of them supplies.

So, I've been thinking about using multiple power supplies in series. Say my target voltage is 48V, and I want to assign the 30% to the 12VDC supply and 70% to the 120VAC power supply. If I develop both power supplies with adjustable DC/DC boost-buck converters, I could set the 12VDC supply to an output of 14.4VDC and the 120VAC supply to 33.6VDC. In series, this gives me 48VDC and the supplies should be taking 30% and 70% of the load, respectively.

The problems are that (a) I'm not an electrical engineer so I don't know if there's any obvious flaw in this design, and (b) it doesn't scale up to a large number of supplies as the maximum number would be the target voltage (e.g. 48VDC) divided by the lower bound on the DC/DC boost-buck converter (e.g. 6VDC) for a max number of supplies (e.g. 48/6 = 8 supplies).

So my questions are:

  1. Does the in-series multiple power supply scheme described above seem feasible, or are there flaws with the design?
  2. If the design makes sense, does this mean that each independent power supply must be rated for the max amperage of the system? In my example of a load of 50A@48VDC, would each supply need to be able to handle 50A continuous?
  3. Is there an alternative, industry-standard approach for this problem that I'm not aware of? I imagine there must be a way of balancing a large load across multiple power sources, e.g. powering a building off of multiple back-up generators or powering a large load across many solar panel systems.

I've reviewed this related post, but it only discusses even load sharing and not intentional asymmetric load dist

\$\endgroup\$
2
  • \$\begingroup\$ I would look more closely at load sharing across multiple supplies. \$\endgroup\$
    – SteveSh
    Mar 18 at 20:52
  • 1
    \$\begingroup\$ Series is an unusual design to say the least. You can make any sharing algorithm you like if all the controllers of all the buck/boost converters are linked together. Any switching converter works by moving discrete 'chunks of energy' from the input to the output; usually the output is regulated by adjusting how often it moves chunks. A more advanced controller could be made to control multiple 'chunk movers' at once, moving energy from several inputs to the same output. \$\endgroup\$ Mar 18 at 22:03

1 Answer 1

0
\$\begingroup\$

In-series is possible, but more complicated, all architectures are based on having a ground, and you are trying to have them separated. It's doable, but you need a really good reason for that. Instead, have all of them in parallel, especially if you can - have them in boost topology, as it naturally ensures current flow in output direction. And then you need one to set the voltage, and others to control their current according to current on the master.

Rating - prepare for the full power. If one of the supplies goes down, even during debug you don't want the rest to burn.

Not sure about standards, but I am sure I saw something similar with Victron.

\$\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.