I'm trying to arrive at a solution for lower cost ATX power supply which will supply about 250W, has two separate inputs each capable of supplying the full load in case of failure/disconnection but are normally balanced when both are on. (Single ATX power supplies are quite cheap - as low as about $20 USD, but dual power supplies much more expensive and typically run $500-800 USD).
I've considered putting multiple ATX power supplies in parallel but realize that won't work. It seems like a solution involving using multiple LM317 ICs would work (or maybe something with higher current and an enable pin could simplify even more e.g. https://www.microchip.com/en-us/product/MIC29502, but the basic pattern would be the same).
My question is in two parts:
Question 1: Are there any holes one can poke in this circuit:
It's the same pattern repeated with different output voltages and input lines.
I know the resistor values have to be figured out, but it seems this pattern above could be repeated however many times it would be needed to give me regulated output (12V and 5V shown but 3.3V is the same thing just different resistor values) from any input the LM317 supports, and the ATX "PS ON" signal of pulling that pin low to activate the system would work. It needs a timing capacitor for the "Power Good" line and it doesn't support the -12V rail. Let's assume those things are okay for my setup. And I know I'll need to cool this thing probably with one or more fans.
And I would then just repeat the above circuit pattern for 3.3V and for however much output power I wanted to support in order to get my expected outputs and current (which so far as I can tell is somewhere around 20A on the 3.3V and and 5V outputs and 15A on the 12V output). The LM317 datasheet shows an example of parallel supplies (page 16, figure 22) and from what I can tell as long as I have the outputs tuned correctly so each of the LM317s for a given voltage matches exactly at their ADJ pin, and I have this 0.2 Ohm resistor the datasheet shows - seems like I could get the required current.
Does this seem sensible? It looks like this would work but I've just never designed something to carry this much current, and I'm looking for a sanity check and hoping if I'm doing something crazy here I can find out before I learn the "hard way" with smoke and sparks and fire extinguishers.
Question 2: Rather than buying "off-the-shelf" 120VAC->24VDC converters, can someone suggest a transformer(s) I might use for this?
I've spent some time looking through Digikey but it seems like the vast majority of transformers will not deliver the expected output current. Most are rated for something like 50VA, whereas (if I understand the math correctly) if I want to deliver 240 watts of power, I need a 240VA transformer which would be capable of converting 120VAC to 24V and delivering 10A (and taking into account some power loss from the transformer, and then the rectifier diodes and whatever else I'm forgetting, to get actual usable input for the design shown above). Most of what I can find here https://www.digikey.com/en/products/filter/power-transformers/164 is rated for more like 1 or 2A.
Would it make sense to use several transformers in parallel? Or am I just not finding the right part?
I'm also just trying to decide if it's worth building out this AC transformer part of the circuit, because 120VAC->24V DC power "bricks" are cheap and very commodity and when I look at the prices it seems like I would be adding complexity and risk trying to build the transformer part into my circuit and not saving much money.
