combining power supplies

[sn3rd]

But connecting in parallel might be an option. As I understand, the floating grounds could definitely be an issue. I don't think the rails need to be identical...

 

[Freakazo v2.1]

Back again!

So me on this topic again, this time after doing some research.
But sadly with that research came more questions...

I now have a much better idea of power supplies and how they work.

My question is this, what is the "weakest link" in a power supply?
It seems to me that it most likely is the rectifiers?
Would simply putting two rectifiers in parallel do the job?
Would I maybe require to improve the the final filtering stage?

 

[bdt]

Guys,If anything the two psu grounds need to be a reference to both psu`s. You cannot have the two separate and floating.You need a common ground or earth.

Indeed, and here's what we have:
- two power supplies, also known as metal boxes, that...
- happen to have the earth pin of the mains cable grounded, it's not called 'earth' for nothing you know!
- the black cable on the DC side == earth pin/PSU casing on the AC side
- these two PSUs bolted to the same chassis (is it a metal box being made?)
- ...a common 0Vdc reference.

Two supplies rated at two different current(amps) outputs do not matter.

Yep: connect each of 'em to what they need to do, paying attention not to overload anything.

The voltage is the sticky one here..A supply will deliver ONLY what the device drawing current from it needs.So there is no "un balance" regarding the current in two supplies..

1. there's no problem with voltage here: you can happily connect the HDDs, optical drive(s) and motherboard 12V (that 4pin plug) to one PSU and then the big motherboard and VGA to the other PSU, but..
2. current is on-demand: you can have a 1kW PSU feeding a 12cm fan if you like but you're not going to run that 9800GTX so well of some weedy PSU (that's being pushed into overload on account of being made to deliver more than its rated output)

If the current draw exceeds the first psu output delivery the second psu can be relied upon to make up for the shortage though...this is after all the reason that we went with 2 psu`s in the first place.

Yep: but all you need do is add up the loads and distribute accordingly.

The two most important criteria here are the volts that must be same and the common earth or ground as reference to both psu`s.

As long as volts off the respective PSUs are within the spec expected by the devices to which they're connected, there should be no problem. In fact, about the only way I can forsee one is by using a twin-molex to 6-pin PCIe power that gets fed from each of the PSUs and their respective 12V or 5V rails are substantively different.

Another reference to this.
http://www.overclock.net/power-supplies/14455-multiple-power-supply-guide.html

...also known as 'wake up the 2nd PSU by grounding the 'power-on' signal'

 

[bdt]

My question is this, what is the "weakest link" in a power supply?

Don't take this the wrong way, but ...the n00b with the mad-scientist gleam in his eye who's overthinking what's really a very simple thing

It seems to me that it most likely is the rectifiers?
Would simply putting two rectifiers in parallel do the job?
Would I maybe require to improve the the final filtering stage?

Dude, are you planning on opening those things up and stuffing around inside of fancy-pants POWER electronics that has potentially (heh, geddit?) lethal internal voltages when you patently don't have the training or experience for it??! (or, that's what it looks like to me - feel free to correct/chastise me if I misread you).

 

[sn3rd]

It depends what you mean by "weakest".
As for the filter, BE CAREFUL! ! Remember that a bigger capacitor in the filter stage will improve the ripple voltage, but will increase the surge current, and that's BAD! The rectifiers in parallel? That doesn't sound right... A large current draw on one side will cause instability on the other side. The weakest link, in my experience, is in the regulation stage.

 

[sn3rd]

That said, improving the filter stage will help the regulators (the heat dissipated and stability depends on the DIFFERENCE between the input and output voltage).

But I agree with bdt: you shouldn't be messing around with this unless you've had prior experience. Things can turn bad VERY quickly... Speaking from experience. Oh, and if you decide to go ahead despite the warnings, over engineer it... by a LOT

 

[Freakazo v2.1]

Indeed, and here's what we have:
- two power supplies, also known as metal boxes, that...
- *happen to have the earth pin of the mains cable grounded, it's not called 'earth' for nothing you know!
- *the black cable on the DC side == earth pin/PSU casing on the AC side
- these two PSUs bolted to the same chassis (is it a metal box being made?)
- ...*a common 0Vdc reference.
Yep: connect each of 'em to what they need to do, paying attention not to overload anything.
1. there's no problem with voltage here: you can happily connect the HDDs, optical drive(s) and motherboard 12V (that 4pin plug) to one PSU and then the big motherboard and VGA to the other PSU, but..
2. current is on-demand: you can have a 1kW PSU feeding a 12cm fan if you like but you're not going to run that 9800GTX so well of some weedy PSU (that's being pushed into overload on account of being made to deliver more than its rated output)
Yep: but all you need do is add up the loads and distribute accordingly.
As long as volts off the respective PSUs are within the spec expected by the devices to which they're connected, there should be no problem. In fact, about the only way I can forsee one is by using a twin-molex to 6-pin PCIe *power that gets fed from each of the PSUs and their respective 12V or 5V rails are substantively different.
...also known as 'wake up the 2nd PSU by grounding the 'power-on' signal'

Over thinking things will ensure nothing get's left out, I want to ensure nothing goes boom boom.

*There is a difference between ground and earth, (depends on the speaker) ground is often referred to as the 0V

*That's exactly our problem, there's no reference ground. Ground is just floating around (ground can be say for example 20V higher than ground(0V))

*almost all power supplies are a bit out from their rated voltage (5V can be 5V1) so that's why I'm going from combining two power supplies to combining two power supplies components (rectifiers)

Don't take this the wrong way, but ...the n00b with the mad-scientist gleam in his eye who's overthinking what's really a very simple thing
Dude, are you planning on opening those things up and stuffing around inside of fancy-pants POWER electronics that has potentially (heh, geddit?) lethal internal voltages when you patently don't have the training or experience for it??! (or, that's what it looks like to me - feel free to correct/chastise me if I misread you).

I know exactly how switch mode power supplies work and why they are more efficient at a higher frequency. The only things that makes switch mode power supplies more efficient is using PWM to get a higher frequency (and at lower loads change the pulse widths)

It depends what you mean by "weakest".
As for the filter, BE CAREFUL! ! Remember that a bigger capacitor in the filter stage will improve the ripple voltage, but will increase the surge current, and that's BAD! The rectifiers in parallel? That doesn't sound right... A large current draw on one side will cause instability on the other side. The weakest link, in my experience, is in the regulation stage.

I'm not sure exactly how the regulation stage works, the only reference to the workings of it is Wikipedia and it states that it adjusts the time of which the mosfets are on, if it refers to the mosfets in the primary circuit then it wouldn't be to much of a problem, if it is somewhere else I'm screwed

That said, improving the filter stage will help the regulators (the heat dissipated and stability depends on the DIFFERENCE between the input and output voltage).

But I agree with bdt: you shouldn't be messing around with this unless you've had prior experience. Things can turn bad VERY quickly... Speaking from experience. Oh, and if you decide to go ahead despite the warnings, over engineer it... by a LOT

I had prior experience in making things not work, will that help
Don't worry I will use every safety rule!


I also read that placing 2 rectifiers in parallel is in fact commonplace in higher end power supplies.

 

[sn3rd]

Hmmm at the rectifiers... sounds like you've got a diagram? Not sure how it'll go in parallel... Will you connect the ONE input to each of the rectifiers? Or will each rectifier have it's own input and you join the outputs? Either way, a phase difference might cause problems. Not saying you're wrong (never attempted a project like this), just helping you gather info

 

[sn3rd]

What you need to know about the regulation stage is that the regulators take a signal that is possibly at the wrong level or that fluctuates, and makes it a nice clean signal (in theory). The further off the input is from your desired output, the more your regulators will strain. Adjustable regulators provide better tolerances, but their operation can be iffy. For a fixed 5V regulator, check out the 7805. Can't remember 12V part number

 

[Freakazo v2.1]

I looked at my diagram again and it looks like it is a simple comparator which checks if the voltage output is the same as the reference voltage, if it is off it simply adjusts the time "something" (this part i'm still confused about)

[edit] While writing I just found out that it alters the primary transistor in charge of the switching before the transformer (for the transformer). So the regulators (which is just a PWM'er) won't have any extra strain of what they already have.

PSU's complexity is sometimes overrated, yes there is a whole bunch of math that is complex (even for me) but basic principles is over rated.

Now, is there something I'm missing? some unseen hazard?


What I'm going to do is basically just place a rectifier above the corresponding rectifier and using thick wires I'll carefully solder the wires directly down to the pins of the lower rectifier (maybe some heatshrink tubing to minimize chances of contact with the heat sink etc.)

Do some research on the final filtration stage and see what I can do to UPgrade it.

Something tells me S(uper?)nerd has some experience in electronics cause you would be 100% correct if it used another method of voltage regulation!
Do you know something about filtration? (smooth voltage output)

[edit]Oh yeah since how much current does the normal mobo use on the 3.3V? (since it uses the 5V output and changes it to 3.3V) would I be safe to asume most of the power for the cpu comes from the 12V rail?