HELP NEEDED WITH FUSE RATINGS FOR ATX PSU MOD TO LAB PSU

[cobra149]

Hi, please could some kind soul help me with correct fuse ratings to protect the atx psu
when connected to electical items. Picture is attached.

I:E if any attached items, go bad , I need the fuse to blow instead of the atx psu continuing to
give loads of current.


Thank you.

View attachment 41492

 

[Hero999]

You shouldn't need any fuses as long as the cable and terminal posts are rated to carry the full current. The PC PSU will be internally protected so short circuits shouldn't be a problem.

I have three main concerns:

1) The power supply may need a minimum load current to give the correct output voltage.

2) The 0V rail of most PC power supplies is bonded to earth ground so the ground terminal of the 0 to 22V LM317 supply won't really be 0V but -12V which would cause a short circuit when you connect it to an oscilloscope and the LM317 will burn out, if the output voltage is set to < 0V its output is connected to real earth/ground.

3) Although you don't need any fuses to protect the power supply you might need a fuse to protect the circuit connected to it. It depends on what you're using it for, if it's just to power small MCU/logic circuits, it's prudent to have an output protected with a 1A fuse.

You should also be aware that the LM317 can't provide 1.5A from 1.2V to 22V, you'll only be guaranteed the full 1.5A when the voltage is set to between 9V and 21V. Above 21V, the voltage might drop below the set value when the load is connected and below 9V the LM317 will reduce its output current to protect itself from overheating. This is all assuming the LM317 is on a really good heat sink.

 

[cobra149]

Hi,
thanks for the reply, the psu is showing correct voltages on all post terminals.As I have fitted an lcd volt meter
to the face off the psu cabiinet. I think the lcd meter could be the small load required thats making the Atx give
the correct voltages.

The -12v gnd from variable pcb is showing 0v. And the lcd meter - + connected to the variable terminal
output posts.

The 317 has an overkill heatsink fitted ,and the 317 is external to the pcb using 3x 16 guage wires.
So the reg is really cool after hours of usage.

I will use a 1A fuse to protect any items connected to the variable output as you suggest.

Not exactly sure what you mean about an osciliscope would short the -12v varaible terminals and
cause the 317 to overheat ?.

Thanks again.

 

[Hero999]

The -12v gnd from variable pcb is showing 0v

. And the lcd meter - + connected to the variable terminal
output posts.

To measure the voltage between the ground terminal and protective earth, you should connect the meter between the ground terminal and the PSU's case or the ground/earth pin on the plug and make sure there's good continuity between the meter and the

Not exactly sure what you mean about an osciliscope would short the -12v varaible terminals and
cause the 317 to overheat ?.

If the PC PSU's 0V terminal is connected to protective earth, it means the -12V rail will be -12V relative from protective earth. A typical oscilloscope has its chssis bonded to earth so if the -12V rail is connected to earth there will be a short circuit.

If the PSU's 0V terminal is connected to earth, the LM317's output voltage will vary between -10.75V and 10V relative to earth. The output of the LM317 is only designed to source current. If the output is below 0V and is connected to 0V current will try to flow from the 0V rail into the regulator which will damage it.

Remember that voltage measurements are relative, not absolute values.

 

[cobra149]

Hi,
Thanks for that. In view of your knowledge on this matter I have decided to do away with the -12v and +12v idea to get variable voltage from 22v down. I wasnt aware of the theory of how it would behave when used.

Would I now be ok to use my variable pcb to vary the +12v and black 0v gnd wire.
I realize that I will only get about 10v down to 2v variable adjustment with +12v input on the lm317
But I would rather go with safety if using the +12v and 0v is safer than the +12v and -12v method to .

Thanks

 

[Hero999]

Yes, that will work, you can also use the LM337 to get a negative 1.25V to 10V.

Here's a schematic to illustrate the point I made earlier. If the -12V pin is connected to 0V there will be a short circuit. An oscilloscope's metal case will be connected to the 0V pin.

View attachment 41497

 

[cobra149]

Great, thanks for the schematic I can visually see the problem now.
The thought of getting 22v variable sort of took over when I spotted this pc Atx mod on other sites.
But when I think of it, I probably wouldnt have needed that voltage much in the real world.

The -12v , & in fact all other - rails have now been snipped off the psu, to be gone for ever.
So I am now happy with 1.2v to 10v with a real 0v. I will go with +12v and + 5v on the other binding posts.

It was so kind of you to put some time into this, to avoid me having any further problems down the road.


Best Regards.

 

[Hero999]

The -12v , & in fact all other - rails have now been snipped off the psu, to be gone for ever.

It looks like you've made your choice but don't worry if you do change your mind later, it should be easy to undo the screws and reconnect the wires.