So how much load can I safely put on the wire

[velis]

For lack of a better forum:

I use my PSU to power some devices external to the PC, such as modems, routers and even some LED lighting.

This is done through an eSATA bracket which also has a molex extension for power.

If an 18 AWG wire is used (and by the look of it it is), the tables say that I could place 80A load on it and still have some (minimal) safety margin. This is way more than the PSU is even capable of, so the maximum sustained 30A load should be even more OK.

But just one look at those molex connector extensions gives me the creeps. Those pins are .2mm (visual estimate, could be even thinner ) thick metal curved around the wire which means the actual area current flows through is much smaller than what the wire provides. All those connectors are designed to handle loads of a HDD or two, not for serious loads.

I'd hate to fry something and I can't seem to find a proper connector that would put some metal to the job, so to speak, so how much load can I really safely place on one wire?

 

[PottedMeat]

looks like the old 4 pin molex connectors are limited to 13A per circuit - http://www.te.com/catalog/bin/TE.Connect?C=12881&F=0&M=CINF&GIID=1568&BML=10576,10782&PID=0&LG=1&I=13&RQS=C~12881^M~FEAT^BML~10576,10782^G~G

2 ground wires + 1 5V and 1 12V line

where are you getting 18AWG is good for 80A? this says 16A chassis/ 2.3A power transmission: http://www.powerstream.com/Wire_Size.htm

 

[_Rick_]

http://www.molex.com/pdm_docs/ps/PS-8981-4M4P.pdf

Molex themselves limit it to 10A with > 16 AWG, and 8A with 18 AWG.

 

[Paul98]

For lack of a better forum:

I use my PSU to power some devices external to the PC, such as modems, routers and even some LED lighting.

This is done through an eSATA bracket which also has a molex extension for power.

If an 18 AWG wire is used (and by the look of it it is), the tables say that I could place 80A load on it and still have some (minimal) safety margin. This is way more than the PSU is even capable of, so the maximum sustained 30A load should be even more OK.

But just one look at those molex connector extensions gives me the creeps. Those pins are .2mm (visual estimate, could be even thinner ) thick metal curved around the wire which means the actual area current flows through is much smaller than what the wire provides. All those connectors are designed to handle loads of a HDD or two, not for serious loads.

I'd hate to fry something and I can't seem to find a proper connector that would put some metal to the job, so to speak, so how much load can I really safely place on one wire?

You can add more wires if you want, or if you aren't using them all clip the end off of parts that you aren't using and wire them with the other wires of same voltage.

 

[velis]

looks like the old 4 pin molex connectors are limited to 13A per circuit - http://www.te.com/catalog/bin/TE.Connect?C=12881&F=0&M=CINF&GIID=1568&BML=10576,10782&PID=0&LG=1&I=13&RQS=C~12881^M~FEAT^BML~10576,10782^G~G

2 ground wires + 1 5V and 1 12V line

where are you getting 18AWG is good for 80A? this says 16A chassis/ 2.3A power transmission: http://www.powerstream.com/Wire_Size.htm

Thanks for suggestions guys.
Found out I've been reading the wrong column (http://en.wikipedia.org/wiki/Awg). Turns out the wire itself can only handle about 8A.

Time to cut down on ambitious plans about powering entire LED array (200W) through one molex

 

[lord_emperor]

Thanks for suggestions guys.
Found out I've been reading the wrong column (http://en.wikipedia.org/wiki/Awg). Turns out the wire itself can only handle about 8A.

Time to cut down on ambitious plans about powering entire LED array (200W) through one molex

Are you building some kind of death ray?

 

[velis]

Of course, I'm Doctor Evil, BTW

 

[SecurityTheatre]

Are you building some kind of death ray?

Good point. 200W of high-output LEDs would shine like a stadium light.

That's not exactly an indoor installation. :-D

 

[_Rick_]

Thanks for suggestions guys.
Found out I've been reading the wrong column (http://en.wikipedia.org/wiki/Awg). Turns out the wire itself can only handle about 8A.

Time to cut down on ambitious plans about powering entire LED array (200W) through one molex

A "6+2 PCIe"-plug should be able to give you something closer to that ballpark.
4x8Ax12V should easily be possible, giving you a theoretical 300+W.

Unless you're running SLI/XFire already, any power supply that has 200W to spare over your current configuration should have one of those dangling about. Extension leads should be available, or you could custom crimp it.

Also: 200W LED is enough to taken an eye out. Don't forget not to play without your goggles

 

[velis]

Why do you guys see 200W LED as something insane?
I just installed a total of 6 metres of LED lighting in the kitchen for 15W/m and it's not exactly something to write home about.
Sure, the light is plentiful, but not nearly too bright.

Alternatively, I have a 5 + 4 m 5W / m lighting in the staircase also giving me just a nice level of lighting though not nearly as bright as the two 100w classic bulbs I also have mounted on the ceiling.

Together that makes up for a bit less than 150W of lighting in two adjacent rooms. Not exactly rocket science nor too bright.

Edit: I'm looking to route this through PSU for efficiency reasons. The PSU is gold while those 220AC/12DC converters are not even bronze, not to mention they aren't optimally loaded.

Thanks for the 6+2 suggestion though. Seems like those things can carry 150W each though in 4 separate wires. I'm not sure what would happen if I just joined the cables, I suppose I'd just get the aggregate amperage. Would require some custom soldering, but that's not an issue.

 

[sm625]

Take a PCI Power Bracket and rewire it so that the 5V pin is shorted to the 12V pin.

You might also need to hack up a few of these:

When you are done you will have two +12V pins and two GND on each molex. That will give you 200W per molex. With 3 wired in parallel you'd have a max of 600W (300W sustained). Just make sure you think it through and wire it correctly if you are going to hack a 5V line into a 12V line.

Are you sure you are really using 200W worth of LEDs? Have you measured the power to be that high? I wouldnt believe that until I actually saw a kill-a-watt meter reading.

 

[SecurityTheatre]

Why do you guys see 200W LED as something insane?

I guess it depends on the spread.

If you're using a traditional bulb with a diffuser and reflector, especially one that's warm color balanced (which eats a bit of the light) and especially if it's an AC circuit (which eats more power), I guess...

I built a 50W LED panel for a project in school years ago and it was burning people's eyes... like retinal afterimage for 1-2 hours sort of intensity - the kind that can cause permanent damage.

I think these LEDs were around 100 lumens/watt or so, so 200W of those would be 20,000 lumens.

If it's directional to a circle of about a foot in diameter, I would wager that is sufficient to set paper on fire or cause skin burns with prolonged exposure...

20,000 lumens is what flashlight makers might call.. somewhere on the order of 1-5 million "candle power" in a tight beam, or roughly the power of a 700W metal halide stadium light if it's a bit more diffuse.

Of course, if it's not high output, highly directional light and you're just doing a massive board like a jumbotron, that's a different thing entirely.

 

[velis]

I'm using el cheapo LED strips
4,8W/m == 360lm/m (http://www.inled.si/led_trakovi/led_trakovi_standard/led_trak_48w_nevtralna_bela)
14,4W/m == 1060lm/m (http://www.inled.si/led_trakovi/led_trakovi_standard/led_trak_144w_nevtralna_bela)

These aren't the most precise of figures. Actual power consumption is lower, as is the delivered brightness, but I can't provide exact figures because I can't tell how much of consumption is due to changed PSU efficiency and how much because of actual load. I'm too lazy to bother with actually measuring it with a multimeter.

Anyway, combined staircase lighting is thus theoretically 3240lm (in reality the two 100W classic bulbs over-light it when turned on)
Combined kitchen lighting is 6300lm.

I am loath to believe these figures. Nowhere near eye-searing brightness. I know quality LEDs will output a lot: there's a company producing street-lighting lamps I know. Their lamps are 23W each and they are insanely bright compared to these strips I use. Maybe I should find me a supplier with better quality strips

 

[mindless1]

http://www.molex.com/pdm_docs/ps/PS-8981-4M4P.pdf

Molex themselves limit it to 10A with > 16 AWG, and 8A with 18 AWG.

Yes but remember, Molex is just not rating a "type" of connector, they are rating their specific brand or model of connector of this type to have the metal properties and manufacturing tolerances needed.

Some random generic connector we call molex which is only compatible, can't be assumed capable of the same current. Although I don't have a source, a vague recollection I have is that no matter the wire gauge, it should be limited to 6A per pin. In other words a 2 lead wire with positive and ground pins is limited to 6A.

 

[mindless1]

Edit: I'm looking to route this through PSU for efficiency reasons. The PSU is gold while those 220AC/12DC converters are not even bronze, not to mention they aren't optimally loaded.

It can't be assumed to be as efficient if powering only a 12V rail load. It may not even maintain 12.0V regulation (or at least not very near that), and could fluctuate the 3.3V and 5V rails enough to shut down with some PSU designs.

Thanks for the 6+2 suggestion though. Seems like those things can carry 150W each though in 4 separate wires. I'm not sure what would happen if I just joined the cables, I suppose I'd just get the aggregate amperage. Would require some custom soldering, but that's not an issue.

IMO, you're better off leaving the PSU for computer use and getting a proper 300W rated 12VDC power supply. It will be smaller, should cost less than the (new) value of the PSU you are considering if bought as surplus, and be simpler to repair if/when the day comes that it fails. It may not even be less efficient in this application if a decent quality product with full range (110V-220V) APFC input.

 

[mindless1]

Are you sure you are really using 200W worth of LEDs? Have you measured the power to be that high? I wouldnt believe that until I actually saw a kill-a-watt meter reading.

Kill-A-Watt meter will only tell the wall power, not the PSU efficiency so if not graphed for this particular output state, it will not give a dependable way to estimate actual output power.

A little closer would be to measure whether it maintains a 12.0V rail voltage at the LED strip, or whatever voltage it is, multiply by PSU output current except that many meters can't measure over 10A so basically in such a situation one would be left trusting the LED manufacturer that if it is a 12.0V input, it's operating at 200W.

The ultimate solution for this particular strips would be to get a PSU with adjustable output voltage. Measure the LED die temperature. If low enough, increase PSU voltage while continuing to measure temperature. When it gets just prior to what you deem unacceptably high temperature level for the lifespan you want, it's time to stop increasing drive voltage. For all we know they could already be running too hot for good lifespan with a high amount of use.

 

[velis]

It can't be assumed to be as efficient if powering only a 12V rail load. It may not even maintain 12.0V regulation (or at least not very near that), and could fluctuate the 3.3V and 5V rails enough to shut down with some PSU designs.

Why do you assume this kind of usage is any different for the PSU than starting a 3D game where the gfx card will suddenly hugely increase the load on the PSU?

IMO, you're better off leaving the PSU for computer use and getting a proper 300W rated 12VDC power supply. It may not even be less efficient in this application if a decent quality product with full range (110V-220V) APFC input.

I'd normally agree with the first part. Heck, I was pretty afraid when I first turned the LEDs on. However, this scenario is not something a PSU has issues dealing with. Usage spikes are pretty common during power-on and games. That's what the capacitors are there for.
As for the second part, I can't get an ATX PSU under 400W. My HTPC / server uses <50W even when playing TV / movies so I view this as just taking advantage of capacity that would otherwise be wasted.

A little closer would be to measure whether it maintains a 12.0V rail voltage at the LED strip, or whatever voltage it is, multiply by PSU output current except that many meters can't measure over 10A so basically in such a situation one would be left trusting the LED manufacturer that if it is a 12.0V input, it's operating at 200W.

This is most definitely not an issue as these strips are limited to 5m length. At 14W / m that places them below 6A per one single strip. I could actually measure the actual consumption if I really wanted to.

 

[PottedMeat]

Why do you assume this kind of usage is any different for the PSU than starting a 3D game where the gfx card will suddenly hugely increase the load on the PSU?

an atx supply is designed to supply multiple voltages, it will feed the gfx spike as well as the background 3.3v/5v. mindless1 is saying loads across voltages you don't need will matter because cheap atx supplies are group regulated to save cost/components.

http://www.jonnyguru.com/forums/showpost.php?p=57556&postcount=7


an original xbox360 power supply will hit 16A on 12V, dunno what else commonly available will give more