I'd like to use a standard ATX PSU as the power supply for my 3D printer: they're safer (no exposed mains terminals), quieter, and can generate a lot more power for the money. I'd like this to be modular: I want to be able to use standard PSU connectors so that I can unplug and swap out power supplies if needed - I don't propose to start modifying the PSU in any way. Because that's dodgy as hell. My issue is current ratings. My current printer power supply outputs 240W at 12V (20A), which is adequate for my printer but doesn't leave a lot of headroom (IIRC my printer actually uses 16A-18A). Currently it uses a pair of 14 gauge wires to connect the PSU to the main board. The ATX24 connector only has 2x 12V lines, and those pins are only rated for 6A each - clearly that's not going to be enough. Which begs the question as to how Ooznest can get away with claiming that this ATX24 converter board is suitable to run a 3D printer... The 4-pin power connector however uses pins that are rated for 8A each. In theory that's 16A from one 4-pin CPU power connector, so if I use two 4-pin connectors that's a theoretical total of 32A across all four 12V lines. Therefore, if I tie those four 12V pins together that should be more than enough to run my printer. The current is going to be distributed across all four pins, so in theory I should have more than enough headroom. I will end up using the ATX24 connector, since that has +5VSB which will run a Raspberry Pi print server, and the 12V power pins on that connector are more than enough for fans. Have I got this right, or am I literally heading for a nasty shock?
Personally I won't try and use a computer supply. They are specialized supplies that are expecting a very limited range of operating conditions. If concerned with exposed terminals and noise then I'd build a shield to cover the terminal and stick the PSU in a muffler box (which could also be your shield). If you are replacing supplies on a regular biases this is a signal something is wrong with the printer. If you really want to go the ATX PSU route then: You'll want to watch out, some 8-pin EPS connector pins are only rate to 7 A, but as you are using less the 20A that should be fine. Your theory is correct however, you should consider a real world issue, line resistance. Each wire can have a different resistance and current will favour the path of least resistance. Thus current will not divide perfectly evenly, so you could end up with one line carrying more than the rated current. I not sure if computers have circuits to prevent this or rely on the power supplies manufactures ensuring minimal variation in resistance. If using the Ooznest board do not exceed 12A at 12V DC.
Thanks for the replies. Safety & noise aren't the only concerns. Noise is not an issue in the room the printer lives in - the printer itself is far louder than any PSU I'd want to use - and safety can be easily fixed. I haven't made a long-term decision on the power supply; if I stick with the style of PSU that I have now I can print a cover for it and wire up a fused mains socket. The main reason I want to use an ATX supply is convenience: an ATX PSU can supply more than enough power on the 12v rail to run the heated bed, hotend, and motors; I have additional 12v connectors for fans, separate to the "main" supply; the 5v rails or 5vsb are more than enough power to run a Raspberry Pi print server, without needing additional voltage regulators (in theory); I can use PS_ON signal to easily switch the power; and they all use standardised Molex interconnects. One of the project ideas I had was to design a latching power switch to control the whole system: a momentary switch activates a circuit which pulls PS_ON low and keeps it there, allowing the PSU to start both the Pi server and the printer; when the Pi shuts down I can use a GPIO signal to release the latch on PS_ON when the Pi halts. One switch boots both the Pi and the printer and the power is automatically switched off when the Pi cleanly shuts down. I originally planned to do this using discrete components, but now I've given it some more thought it really makes sense to use a microcontroller powered by the 5vsb line. I don't intend on replacing power supplies regularly (if my printer kills a PSU then there is indeed something very very wrong), but using ATX connectors makes it far simpler to disassemble the printer. Right now all the power cables are attached to the control board using screw terminals; if all I have to do to remove the power supply is simply disconnect a few connectors then it makes maintenance a hell of a lot easier. TBH I don't think line resistance is going to be a major issue. It'll likely have an effect, but it'll probably be very minor. There's no reason why 4 wires, when cut to the same length, should have major differences in the current being drawn by each wire (assuming they're all tied in to the same common load). Worst case scenario I can test the resistance on each cable and calculate the variance. I'm basing my info on what I've read here. The terminal might handle 9A, but the heat generated may be enough to melt the plastic; hence the 6A rating. I've seen the 6A per pin rating mentioned elsewhere, too.
Just to be clear I am not opposed to your plan to upgrade your printer's power, just want to make sure you haven't overlooked a simple solution. Good plan and probably the best argument for using the ATX supply. But since you are thinking of adding a micro controller, a second small 5V DC supply and set of relays could achieve the same one switch control. For easy of connectivity then I would suggest adding a plug and connector inline of your existing supply, plus you can get a connector that would be rate for the full current. Molex MLX™ Power Connectors are rated up to 20A. Alternatively a quick solution would be to use an IEC C13/C14 cable, cut in half, wire the C13 half to the power supply and the C14 to the printer (I think I have that the right way around). Four cut wire alone should be very similar, but more significant variations can come from the cable assemble, i.e. the quality of the crimp connection, and dirty or worn/damage pins in a connector. Just something to consider.
