ATX PSU, Heated Bed, and Wiring

After building a new PCB heater, I ran into a bit of a problem. The PCB Heater is in 4 sections, each with a resistance of about 0.85R. I know that's very low, and puts a tonne of power across it, but I wasn't getting near enough heat out of the last few heater solutions I was using. Long story short, I fried my 21A power supply pretty quickly. So, I invested in a new 700W ATX PSU, rated for 58A at the +12V rail. This has 2 problems, however. Firstly, as with all ATX PSUs, I don't need the 23million cables coming out of it. I need 6. 3 x +12v, and 3 x GND. Secondly, the cable is 18AWG, which I don't think will hold up to the ~16A needed for the print bed. So, I'm trying to work out a suitable solution.

So what are my options here? Can I run 2 or 3 of the 12V wires from the ATX to the HPB contact on the board, like this:



Also, I have heavier gauge wire, but I'm not sure of the best way to wire it up. I'm reluctant do desolder the 12V wires from the ATX's board to replace them with heavier wire. It's a messy job, and I'd rather a different solution if there's an easier one...

Your options are to either use all the +12v (yellow cables) and all the Ground (Black cables in paralel) or simply cut them very short and solder them as a bunch on a much thicker cable to feed the heated bed.

However even if your ATX psu is puting out 58amps it may only do so for short periods of time, and only when the other rails have some load... 5v rail comes to mind.

I had a similar problem with my old ATX psu, it was suposed to put out ~22amps, which would have been enough but the voltage drooped on the 12v rail down to 10.5-11v even with load on the 5v rail via large wire wound resistor. At this point I decided to ditch the ATX psu, and finding the chinese standalone PSU's not trust worthy, I opted for a 12v only server psu manufactured by delta. It is not the quietest thing on the planet, however even with the heated bed on, and a 150w auxiliary heater on it never really droops below 12v. I used heavy gauge wire soldered into its terminals and have seen stable power out of it for the last few months. Voltage droop is caused by thin wires heating up, or even a weak PSU which is not exactly living up to standard.

My particular model is rated at 53.3A with 110v input and 106.5A at 220v..... Here is a link to the work necessary to get my particular model to work. PSU If your counting the psu is puting out 650w at 110v and 1300w at 220v. those are the continuous ratings for it. it can put out much more, the guys at RC forums are getting nearly 1500w out of these with 220 and nearly 800w with 110v. In many respects the server PSU was much easier to modify than the ATX psu I used before. I simply soldered cables to it, and used a 2 pin jumper cable wired to a switch to turn it on or off. The down side to this psu is that it does not have a 5v rail if you need it you need to run a converter. I used a cigarette lighter plug with usb ports on it to power the wifi module for my particular setup.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

You can run multiple wires to share the current. Presuming the supply and ground are the same gauge, you'll need just as many ground wires as supply. A single larger gauge wire may be easier to route and provide more flexibility than a bundle of smaller wires.

If you run all 4 traces in parallel, and each trace is .85 ohms, then your overall resistance is 1/4 that, or .2125 ohms. To keep the resistance at .85 ohms, you'll need to wire it so that pairs are ran in series and then each pair is ran in parallel.

If you're worried about the aesthetics of having all the wires, you can open the supply (unplugged of course) and clip the unneeded wires inside the case. Just wrap them in a bit of electrical tape to make sure they don't accidentally contact the case or each other and they are out of the way.

Quote
cdru
You can run multiple wires to share the current. Presuming the supply and ground are the same gauge, you'll need just as many ground wires as supply. A single larger gauge wire may be easier to route and provide more flexibility than a bundle of smaller wires.

If you run all 4 traces in parallel, and each trace is .85 ohms, then your overall resistance is 1/4 that, or .2125 ohms. To keep the resistance at .85 ohms, you'll need to wire it so that pairs are ran in series and then each pair is ran in parallel.

If you're worried about the aesthetics of having all the wires, you can open the supply (unplugged of course) and clip the unneeded wires inside the case. Just wrap them in a bit of electrical tape to make sure they don't accidentally contact the case or each other and they are out of the way.

Ah! This may be the solution I'm looking for! I'm not picky about the aesthetics, and where I have to run multiple wires between, for example, the heatbed power connection on the board to the PSU, I've some nice black heatshrink to keep the multiple wires together.

So from what I can tell, the plan is to take 3 x 12V wires and 3 x GND wires from the ATX and "group" them so that they all connect to the HB + and GND respectively.
Then take a single +12V and a single GND from the ATX and go to the board power (for all the other stuff)
and then one last +12v and GND for the fans.

With the other cables in the ATX, I've the ATX case open, and they will all be cut short, and heatshrink put on the ends of them so that they'll be neatly bundled off INSIDE the ATX case.

That all seems ok, right?

Quote
twicx
So from what I can tell, the plan is to take 3 x 12V wires and 3 x GND wires from the ATX and "group" them so that they all connect to the HB + and GND respectively.
Then take a single +12V and a single GND from the ATX and go to the board power (for all the other stuff)
and then one last +12v and GND for the fans.

If you do that, your heated bed and fans will always be on when the power supply is on. You'll have no control over it. You don't mention what electronics you use, but typically ALL the power wires go to the electronics (RAMPS, Melzi, Gen7, etc) and then you have wires that go to the heated bed, fans, hot end, etc. Alternatively, some setups will hook the heated bed more directly to the power supply but use a a solid state relay to do the switching of power instead of the board electronics.

Quote

With the other cables in the ATX, I've the ATX case open, and they will all be cut short, and heatshrink put on the ends of them so that they'll be neatly bundled off INSIDE the ATX case.

You'll likely also need the green PS_ON wire to turn on and off the power supply via software. Some power supplies won't put out sufficient power on the +12V rail unless you have a load (typically a 4.7 ohm 10w resistor) on the 5v rail too so before you go snipping wires, double check that. You won't know until you try if you need the dummy load. You may also want to keep a +5v line around unclipped as well. You might need that down the road if you want to add a servo and the arduino can't supply enough of it's own +5V juice. The rest you can snip.

Quote
cdru

Quote
twicx
So from what I can tell, the plan is to take 3 x 12V wires and 3 x GND wires from the ATX and "group" them so that they all connect to the HB + and GND respectively.
Then take a single +12V and a single GND from the ATX and go to the board power (for all the other stuff)
and then one last +12v and GND for the fans.

If you do that, your heated bed and fans will always be on when the power supply is on. You'll have no control over it. You don't mention what electronics you use, but typically ALL the power wires go to the electronics (RAMPS, Melzi, Gen7, etc) and then you have wires that go to the heated bed, fans, hot end, etc. Alternatively, some setups will hook the heated bed more directly to the power supply but use a a solid state relay to do the switching of power instead of the board electronics.

Quote

With the other cables in the ATX, I've the ATX case open, and they will all be cut short, and heatshrink put on the ends of them so that they'll be neatly bundled off INSIDE the ATX case.

You'll likely also need the green PS_ON wire to turn on and off the power supply via software. Some power supplies won't put out sufficient power on the +12V rail unless you have a load (typically a 4.7 ohm 10w resistor) on the 5v rail too so before you go snipping wires, double check that. You won't know until you try if you need the dummy load. You may also want to keep a +5v line around unclipped as well. You might need that down the road if you want to add a servo and the arduino can't supply enough of it's own +5V juice. The rest you can snip.

Ah ok. I'm using a RUMBA board, currently with 1 extruder, but soon to go to two.

I'm going to have two fans wired to a 12V supply; one for the electronics board case to keep some air circulating over the stepper driver heatsinks (this seems like good practice, although I've seen some printers that don't do this) and the other always-on fan is at the hot ends, to keep them cool. There will be a switchable fan, but that's connected to the board itself.

I guess the best way for me to check if I need the PS_ON wire would be to leave it out for the moment and see if it will power the board and two fans without it. If not, I wonder could I use that 5V rail for the electronics board fan. The resistance across the fans I have is 240k though, but, the original setup for my printer (Felix 1.5, RAMPS and an 8A ATX) used the 5V rail to power the electronics board fans.

A few notes:

1. If you connect your heated bed sections in parallel, they will take 14A each at 12V (perhaps a little less after voltage drop in the wiring). That's 56A total, close to the limit of your ATX PSU, and you haven't allowed for the current draw of stepper motors or fans yet. Unless your heated bed is huge, you should get enough power if you apply 12v to two sections connected in series, and the same for the other 2 sections. That brings the current down to 14A total and the bed power to 168W. That should be sufficient to heat a bed up to about 250mm square to about 120C.

2. Some ATX PSUs have a single 12V supply internally, others have 2 or more. The label on the PSU should say. If your PSU has 2 or more separate 12V supplies internally, you must not connect them in parallel.

3. If you really want to drive all 4 sections in parallel, then you would need a 56A switch to control them. Your RUMBA board won't handle anything like 56A, so you will need to use a 56A external switch. That in turn would require several mosfets wired in parallel. So instead, I suggest you use 4 external switches, each rated at 14A, all controlled from the same RUMBA output. Then you can connect each heater section to a separate output from your ATX PSU, solving tour wiring problem.

4. If you do run all 4 sections of your heated bed from 12V, that's 674W of power in total. That represents a serious fire risk, for example if the firmware freezes and leaves the heated bed at full power.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
A few notes:

1. If you connect your heated bed sections in parallel, they will take 14A each at 12V (perhaps a little less after voltage drop in the wiring). That's 56A total, close to the limit of your ATX PSU, and you haven't allowed for the current draw of stepper motors or fans yet. Unless your heated bed is huge, you should get enough power if you apply 12v to two sections connected in series, and the same for the other 2 sections. That brings the current down to 14A total and the bed power to 168W. That should be sufficient to heat a bed up to about 250mm square to about 120C.

2. Some ATX PSUs have a single 12V supply internally, others have 2 or more. The label on the PSU should say. If your PSU has 2 or more separate 12V supplies internally, you must not connect them in parallel.

3. If you really want to drive all 4 sections in parallel, then you would need a 56A switch to control them. Your RUMBA board won't handle anything like 56A, so you will need to use a 56A external switch. That in turn would require several mosfets wired in parallel. So instead, I suggest you use 4 external switches, each rated at 14A, all controlled from the same RUMBA output. Then you can connect each heater section to a separate output from your ATX PSU, solving tour wiring problem.

4. If you do run all 4 sections of your heated bed from 12V, that's 674W of power in total. That represents a serious fire risk, for example if the firmware freezes and leaves the heated bed at full power.

By the way, this is the new PSU and a size comparison to the RUMBA board. Next task, as I'm sure you can imagine, is to get through the rats nest...


1. I should have been clearer on this, apologies, the whole heated bed takes 14A. Each section is much less than that. On a side note to this, I ran out of Kapton tape for my heater, so I'm a bit screwed on it till I get more (and it's impossible to find in Ireland) so I'm now having a look at what power resistors I have. The original setup was 4 x 4.7R power resistors, but the temperature maxed out at around 70C. Its fine (I guess) for PLA, but I'd rather more heat. I have some 3.3R and some 2.2R, power resistors, but I'll have to hunt down their data sheets to see what their max operating temperatures are. they're just the usual 15W Aluclad type ones. Pain in the ass though.

2. This one has 2, but from each +12V contact point on the board, there's about 6 x 18AWG wires coming from it. So I'm assuming that if I'm grouping 3 x +12V and 3 x GND for the heat bed, they should all come from the same +12V and GND contact point on the board. I've tried to take a photo here, but it's quite tight in the case! You can see that at both of the 12V contacts, there's a number of wires coming out of it. The ground wires are just about visible at the side of the photo too. I figure as long as I keep them grouped from the same contact points, I'm alright?


3 + 4. If i did have a printer using 56A and 674W, I'd be fairly worried, and I wouldn't have it in my bedroom!

Oh, and a huge thanks to everyone's help with this. Isn't the old saying measure twice, cut once?

Quote
twicx
2. This one has 2, but from each +12V contact point on the board, there's about 6 x 18AWG wires coming from it. So I'm assuming that if I'm grouping 3 x +12V and 3 x GND for the heat bed, they should all come from the same +12V and GND contact point on the board. I've tried to take a photo here, but it's quite tight in the case! You can see that at both of the 12V contacts, there's a number of wires coming out of it. The ground wires are just about visible at the side of the photo too. I figure as long as I keep them grouped from the same contact points, I'm alright?

According to this page which looks to be your power supply, there's only one 12V power rail despite what the circuit board seems to indicate. Examination of the traces may show if they are connected or not or just a continuity test between the leads. I had a similar power supply that seemed to indicate that there were multiple rails, but they were connected via a soldered in jumper. I could only guess that they use the same base circuit board for multiple models.

If you're really worried about it, yes, just use wires from the same grouping. All ground points should be connected so there likely isn't similar groupings for them. Use one set for the HB connection, use the other set for powering the electronics and hot end.

Quote
cdru

Quote
twicx
2. This one has 2, but from each +12V contact point on the board, there's about 6 x 18AWG wires coming from it. So I'm assuming that if I'm grouping 3 x +12V and 3 x GND for the heat bed, they should all come from the same +12V and GND contact point on the board. I've tried to take a photo here, but it's quite tight in the case! You can see that at both of the 12V contacts, there's a number of wires coming out of it. The ground wires are just about visible at the side of the photo too. I figure as long as I keep them grouped from the same contact points, I'm alright?

According to this page which looks to be your power supply, there's only one 12V power rail despite what the circuit board seems to indicate. Examination of the traces may show if they are connected or not or just a continuity test between the leads. I had a similar power supply that seemed to indicate that there were multiple rails, but they were connected via a soldered in jumper. I could only guess that they use the same base circuit board for multiple models.

If you're really worried about it, yes, just use wires from the same grouping. All ground points should be connected so there likely isn't similar groupings for them. Use one set for the HB connection, use the other set for powering the electronics and hot end.

Perfect. Thanks again.

Turns out I've 4 x 1.5R 15W Power resistors, these kind: [www.arcolresistors.com]

The old 4.7R ones were in parallel and the heatbed maxed out at about 60 degrees C, so I'm going to have to work out if I can get away with the 4 x 1.5R ones in parallel without them going over their max ratings. If not, I'm sunk till I can get some kapton tape.

14A for the entire bed is much more reasonable. Unless you want the 200mm wide stuff, Amazon should be able to supply you with Kapton tape in a day or two.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].