ATX PC Power Supply to a Lab Power Supply conversion doubt

[Locobox]

Hi everyone!

I just finished my conversion of a regular ATX PC Power Supply to a Lab Power Supply.
Really enjoyed the whole process.

I followed this tutorial:


Don´t get me wrong, that tutorial is very well explained, I just have some doubts.

For instance, the guy on the video explains that in order to get the PSU started and get always a stable voltage is important to add a 20 Ohm sandbar resistor between the neutral and the +5v or +12v posts (That is depending on the type of ATX PSU you are using) mine was an old PSU so I put my 20 Ohm resistor between the neutral and the +5v line ... However I want to know if the resistor value is dependable on the load I attach to the PSU?

Will a higher Ohm rated resistor will generate less heat at high loads?
Maybe the resistor is there just to start the PSU but I want to be sure... better safe than sorry.

Thanks in advance for any help on this matter.

[rob77]

the resistor is needed to load the supply - when unloaded the regulation might not work correctly.

[Richard Crowley]

20 ohms across 5V is only 1/4 amp or 1.25 watts.  You should be using at least a 5 watt resistor there.
However,  20 ohms across 12V is 600mA or 7.2W, so at least a 10W resistor would be in order. But 20 ohms seems excessive to me.

Note, however that putting fixed load resistors across computer power supplies is a debated practice.
If you power up your supply without a load, does it still hold regulation?

[Locobox]

Many thanks for your responses, however it is not all clear to me if the 20W sandbar resistor will hold...

Lets say I put the PSU under a high load +5v @20Amps will the PSU perform OK or the sandbar resistor will release the magic smoke at high working loads?

I'm concerned over the safety of the whole PSU.

Again many thanks Richard and rob77 for your input.

[HackedFridgeMagnet]

is important to add a 20 Ohm sandbar resistor between the neutral and the +5v or +12v posts

sandbar is it really called that?

almost certainly not the neutral but the common on the secondary.
As to 5v or 12v, I guess you would put it on the rail that provides feedback.

[mariush]

You don't have to put 20 ohms like he says in the video.

In his video, he uses an older model of an atx power supply. That design needs to have the computer use some power on 5v in order to  output a good 12v. Without some load, that power supply may or may not oscillate or output less or more than 12v.
Those kind of power supplies already have a resistor on 5v and 12v, which usuall waste about 0.5-1w of power. That load, plus a bit of load on 5v from the computer makes the power supply output the other voltages within good ranges.

Your power supply may not need such a big resistor, it depends on the design of your power supply, if and what resistors it already uses.

Measure the 12v output, see if it's steady,  put some device on that 12v and see how the voltage changes.  If the voltage goes down quite a lot, you may need to put a resistor on 5v to trick the power supply.

The basic idea with the resistor is to make the power supply produce some power on 5v all the time.   You have two formulas : V = I x R  and P = IxIxR  where V is voltage, I is current, R is resistance. 
So for example, if you have a 10 ohm resistor, you have  5v = I x 10 so I = 5/10 = 0.5 A or 500mA  and the power wasted is  I x I x R = 0.5 x 0.5 x 10 = 2.5w  which means a plain 5-7w resistor will do.

If connected on the 12v line, the same 10ohm resistor will be  12/10 = 1.2A and power is 1.2 x 1.2 x 10 = 14.4 watts... which is a lot.

This resistor behaves like anything you would normally connect to the 5v output, so if you connect something that pulls 20A from 5v, it doesn't matter that the resistor uses 2.5w, it's a separate thing.

Now of course, how much your power supply can actually produce is important.  Look at the label on your power supply.. it may say 5v @ 25A , 3.3v @ 30A or something like that, but most power supplies have some limitations, look on the label and you'll probably see there something like this " 3.3v + 5v maximum output = 100w" , which means no matter how much you connect to 3.3v or 5v, in total the power supply can only output 100 watts  ( 5v x 20A = 100w)
Even if it says that, power supplies designed like that usually don't like it to output so much power on just 5v and nothing on 12v, or the other way around. So don't assume that it can output 5v @ 20A just because the label says so.  It may do 5v if you also use a few A on 12v, otherwise it may not output a stable 5v.

And it's not neutral .. it's the ground (the black wires on the secondary side)

[What_NZ]

Actually not quite sure what the OP means by "regular ATX PC Power Supply" but as already mentioned the age of the PSU is important.

The best advice mentioned is just to try it with and without the resistor, under load.

If you want to open up the PSU trace where the voltage feedback optocoupler is connected. If it goes to the 5v rail then you may have to provide a load, when you are drawing current from 12volts.

[rr100]

I've never seen any trouble at all from powering stuff from only one (+5 or  +12V) rail on ATX PSes but I don't doubt there might be problems for some setups.

That would make for a very interesting blog (video or not) or at least a post here, especially if something really wild is observed. How would one go about this? I assume load on both rails is something expected so it makes no sense to play with various sizable loads on both. So, what to do if one wants to test this? Just test the output (with whatever you have best, voltmeter, scope) with no load and various loads like 1-2-5A (each time on only one rail) ?

Apart from this nowadays many desktop computers can do a lot of stuff when stopped (like charge devices over USB, power and detect mouse movement, WOL, etc). Over which rail is the power coming from (if there is any standard for this)?

[tautech]

I've never seen any trouble at all from powering stuff from only one (+5 or  +12V) rail on ATX PSes but I don't doubt there might be problems for some setups.

That would make for a very interesting blog (video or not) or at least a post here, especially if something really wild is observed. How would one go about this? I assume load on both rails is something expected so it makes no sense to play with various sizable loads on both. So, what to do if one wants to test this? Just test the output (with whatever you have best, voltmeter, scope) with no load and various loads like 1-2-5A (each time on only one rail) ?

Apart from this nowadays many desktop computers can do a lot of stuff when stopped (like charge devices over USB, power and detect mouse movement, WOL, etc). Over which rail is the power coming from (if there is any standard for this)?

+5V standby

[miguelvp]

On most ATX supplies you have to drive the Power On (green cable) is pulled up to 5 volts and you have to drive it to low (ground it) in order to turn the supply on.

[janoc]

I would also check whether the PSU doesn't already have a load resistor built-in. Many newer ones have one on the 5V rail to ground, to ensure reliable start and regulation even when otherwise unloaded.

[rob77]

I've never seen any trouble at all from powering stuff from only one (+5 or  +12V) rail on ATX PSes but I don't doubt there might be problems for some setups.

That would make for a very interesting blog (video or not) or at least a post here, especially if something really wild is observed. How would one go about this? I assume load on both rails is something expected so it makes no sense to play with various sizable loads on both. So, what to do if one wants to test this? Just test the output (with whatever you have best, voltmeter, scope) with no load and various loads like 1-2-5A (each time on only one rail) ?

Apart from this nowadays many desktop computers can do a lot of stuff when stopped (like charge devices over USB, power and detect mouse movement, WOL, etc). Over which rail is the power coming from (if there is any standard for this)?

the 5V VSB is a separate supply withing the ATX power supply - no issue with that one - it's supposed to be within the +/- 5% spec all the time.
the other outputs got a common single regulation loop - therefore some of them might be out of spec during no-load or strange (from atx spec point of view) uneven loads.
if the voltage is out of spec a bit - it doesn't necessarily mean you'll run into problems , your circuit's will most probably work - no issue at all.
bu you simply can't call that LAB SUPPLY. every lab supply MUSTprovide voltages within specs - no matter what.

[macboy]

...you simply can't call that LAB SUPPLY. every lab supply MUSTprovide voltages within specs - no matter what.

100% agree. You could call a converted ATX a "Bench" supply, but definitely not a "Lab" supply.

The 12V output would be suitable to power things normally powered by automotive DC. An example might be a charger for LiPo batteries commonly used for R/C models.

[What_NZ]

...you simply can't call that LAB SUPPLY. every lab supply MUSTprovide voltages within specs - no matter what.

100% agree. You could call a converted ATX a "Bench" supply, but definitely not a "Lab" supply.

The 12V output would be suitable to power things normally powered by automotive DC. An example might be a charger for LiPo batteries commonly used for R/C models.

I don't think the OP was using the dictionary definition of "Laboratory"

Quote "The 12V output would be suitable to power things normally powered by automotive DC. An example might be a charger for LiPo batteries commonly used for R/C models."
Really?..... remember these power supplies are good enough to power a computer!

If he checks the regulation as has been suggested, why shouldn't he use it for general electronics?  Even at bad regulation it will be under voltage, not over-voltage that may cause an issue. You should always be aware of the limitations of any power source.

[ctz]

A lab supply with 5v/12v at 20-30A with no current limiting is a bananas idea. Unless you like melting and welding things when you make a mistake!

[Locobox]

@HackedFridgeMagnet
Sure that's what they call it, see the back of the resistor, it resembles a sandbar.

@mariush
I forgot to mention it, my PSU is an old model (Most amps are on the +5V rail).
Without the 20W resistor in place voltages on most rails were to low or unstable to properly power my devices.
With the resistor on place all voltages on all rails are stable and good.

I think I undestand it all with what you wrotte:
"This resistor behaves like anything you would normally connect to the 5v output, so if you connect something that pulls 20A from 5v, it doesn't matter that the resistor uses 2.5w, it's a separate thing".

@What_NZ
it's an old PC PSU
I have attached a pic of the label.

@janoc
This PSU has no dummy load built-in.

@What_NZ
Lab or bench I used the word Lab 'cause that's what the guy on the video calls it, good to know the proper way to call it thou.

I'm using this PSU to power low current devices, PC speakers, Testing LED arrays, Toys ETC, But I'm also concerned about the safety of the PSU when powering high current loads lets say 10, 15 or 20 amps (Given that the max. amps available on the 5v rail are 25).

---

So far so good with this conversion (learned a lot in the process) I understand now that the 20W resistor is more to stabilize the output voltage than a safety measure.

Many thanks to everyone who helped me understand this matter.

[AmmoJammo]

I would, at the very least, have fused outputs.
Even have 2 or 3 separate fused outputs on each rail, with different sized fuses. 1amp, 5amp, 15amp for example.

Small electronics will only need 1amp, if you blow that fuse, either the circuit is wrong, or it needs more current, hence stepping up to the 5amp

[janoc]

So far so good with this conversion (learned a lot in the process) I understand now that the 20W resistor is more to stabilize the output voltage than a safety measure.

It is actually both - without that resistor the feedback loop is not working and there are some old AT/ATX power supplies that will send all kinds of high voltage spikes to the output in that situation. Some PSUs were even known to self-destruct due to the voltage rising too high if not (sufficiently) loaded.

Generally speaking, I am avoiding dealing with really old AT/ATX PSUs if I can - saving few bucks by reusing an old crusty supply is not worth the potential fire or electrocution hazard should anything go haywire in them.

A decent simple 30V/3A lab supply with a proper current limiting and voltage regulation costs around 100EUR, it is not really worth bothering with ATX supplies unless you need the high current capability.

[Locobox]

Many thanks for your responses.

@AmmoJammo
Have not thought of that, many thanks for the suggestion I'll put fuses in all rails.

@janoc
I have done this conversion just to learn more about electronics I'm not actually planning to use it for anything but powering a set of cheap PC speakers, a small new use for a device that otherwise would rest buried in a landfill.

Thanks to everyone.