AT Power Supply Schematic

http://www.pavouk.org/hw/en_atxps.html

More...

HiTekRedNek, You'll see that this supply schematic shows a bridge rectifier feeding high-voltage electrolytic capacitors. These are called "bulk" capacitors and have 300 to 350 volts across them, so that's the DC voltage you'd need.

One interesting thing: You don't necessarily need to go inside the circuit to connect an external DC supply. If you set the switch to the 230V position, you can simply power it from DC connected to the AC-line connectors!

Modern computer power supplies work differently. They no longer simply rectify the AC line into the bulk capacitors. That's because such rectification and capacitor charging occurs as high current spikes at the top of every AC-line cycle, which is very bad! Instead they use PFC, or Power-Factor-Correction circuitry. Yes, they usually have an input bridge rectifier, but it goes to a rather small storage capacitor (like 0.5uF), and immediately to a boost converter, to step-up and charge a larger bulk capacitor, which holds power throughout the AC cycle. The PFC is setup to draw a current from the AC line that's proportional to the instantaneous line voltage, as if the load was a resistor.

Depending on how the PFC circuitry is setup, you may be able to use the connect-DC-power-to-the-AC-line-plug trick as with the old designs. If you are willing to experiment, go for it, but be careful! Do it out on your driveway, there may be a fire!

One interesting consideration, you may be able to successfully use much lower DC voltages, perhaps as low as 100 to 120 volts. It'll also be happy working over a range of voltages, perhaps as low as 80 to as high as 350 volts. That's thanks to the boost converter aspect of PFC.

I see you now saying your available DC voltage is too low to apply to the bulk-capacitor node. But maybe it's high enough for the PFC input? If not, let me suggest that a simple boost converter from your available lowish DC voltage to say 120 volts DC out, is much more simple for you to make

(or even buy on eBay, see pic) than a full-fledged inverter, as some have suggested. BTW, this cheap eBay item is set for *)V max output, but it's easy to modify for higher voltages: change the output caps and feedback resistor. email me for a drawing.

I meant to write, 80V max. The PCB comes with 100-volt output caps installed, and the switching transistor, Q1, a SUP85N10, is rated at 100V.

This boost circuit was discussed in an earlier CR$ thread, Power-Supply-DC-DC-Step-Up-Converter-Problem, where, having bought one these eBay supplies, I reverse-engineered to get a drawing, and discussed its operation in several posts in the thread. You should start by reading those posts.

I modified one of these supplies for much higher voltages, using a single 100uF 250V output cap, replacing the MOSFET with a common IRF740 type, rated at 400V, and changing R10 from 3k to 2k, or lower. These simple changes worked fine, and allow you to boost to higher output voltages.

It's useful to understand the basic principles of boost-type switching converters, and pardon me for suggesting that Chapter 9 in our new book, The Art of Electronics, 3rd-edition, is a great place to learn about it. But you don't have to plunk down $93 to read it, because our publisher has offered it as a free sample chapter. Here's a DropBox link to Chapter 9, which you may freely share. Boost converters are discussed in 9.6.6, page 647 of the book, or page 71 of the file.

Anyway, we'd like to evaluate the power capability of this low-cost $12 eBay circuit, We can do this by calculating Iout. To do that we first need the ON duty cycle of the switch Q1, (E'qn 9.4d), which is D = 1 - Vin/Vout. For example, if our programmed loop-controlled output was 120 volts, and our input happened to be only 16 volts, then D = 1 - 16/120 = 0.867, or 87%. Looking at (E'qn 9.4a) we see Iout = Iin / (1-D), or Iout = 13% Iin, for our selected operating voltages.

I'm pretty sure the designer's intention was for you set to the (input) current limit to 10A. Under this condition, with Vin = 60V, the supply would work up to 600 watts, as advertised. But with only 16 volts in, we can only boost 160 watts max into our 120-volt output.

This means that even tho you hook our boost-converter's 120Vdc output to an ATX power supply, it'll only work properly up to about 150 watts of total load.

To go to higher power levels, you'd need to have higher input voltages, or need to run the MOSFET to higher maximum currents (choose a different type, like an FDP2710, or see Table 3.4b, page 189 in the book), and probably change the hand-wound toroidal inductor, L2, so it doesn't saturate.