Might help.

Curious if it backfeeds the 5v or only supplies it to the three aux connectors.

You could also make one pretty easily by buying a 12v to 5v converter and installing a molex on it. Pulls 12v and backfeeds 5v back to the 5v rail (multiple 5v sources in parallel, which is fine, we do this a lot in solar with higher votlage)

Sphere478 wrote on 2023-10-28, 15:59:

Might help.

Curious if it backfeeds the 5v or only supplies it to the three aux connectors.

You could also make one pretty easily by buying a 12v to 5v converter and installing a molex on it. Pulls 12v and backfeeds 5v back to the 5v rail (multiple 5v sources in parallel, which is fine, we do this a lot in solar with higher votlage)

Gut feeling was that if it back-fed it could cause issues with PSUs not designed for it, and they’d reduce costs by not bothering with the outputs.

Na, that’s not how dc works. You can parallel multiple DC sources of identical voltage.

Sometimes this can cause more (or less) line noise. But I don’t expect you will have problems relating to this. The condition of your mobo caps would be where I looked if you did.

Regardless, the input is EPS12V, so no way to backfeed from the input.

paradigital wrote on 2023-10-28, 18:33:

Regardless, the input is EPS12V, so no way to backfeed from the input.

Ah, misread thought input was molex or sata. So it isn’t paralleling then. 👍

paradigital wrote on 2023-10-28, 16:58:

Sphere478 wrote on 2023-10-28, 15:59:

Might help.

Curious if it backfeeds the 5v or only supplies it to the three aux connectors.

You could also make one pretty easily by buying a 12v to 5v converter and installing a molex on it. Pulls 12v and backfeeds 5v back to the 5v rail (multiple 5v sources in parallel, which is fine, we do this a lot in solar with higher votlage)

Gut feeling was that if it back-fed it could cause issues with PSUs not designed for it, and they’d reduce costs by not bothering with the outputs.

We could get into floating and bonded negatives and issues relating to that, but again, I don’t expect you to have an issue here either, especially with the psu being the source of negative. It really should work just fine and be simple as that. 👍

I do like the corsair’s unit’s purpose built design. And 3.5” mounting. That is nice.

Like I said, I’ve paralleled DC sources all my life, in computers, (dual psu, combining multiple rails, backfeeding a usb charger into a hub, etc) and in solar all the time. It’s possible to have issues, but those are rare and avoidable. They are the exception from poor equipment or poor equipment matching, not the usual rule/experience.

There are Chinese 10-20A $7-20 Buck converters on ebay doing same thing, $30 for brand name in nice case and with all appropriate sockets doesnt sound terrible.

rasz_pl wrote on 2023-10-28, 18:46:

There are Chinese 10-20A $7-20 Buck converters on ebay doing same thing, $30 for brand name in nice case and with all appropriate sockets doesnt sound terrible.

Indeed.

Sphere478 wrote on 2023-10-28, 18:30:

Na, that’s not how dc works. You can parallel multiple DC sources of identical voltage.

Sometimes this can cause more (or less) line noise. But I don’t expect you will have problems relating to this. The condition of your mobo caps would be where I looked if you did.

Differrences in MOSFET Rds will cause one MOSFET to do most of the work. You can't just slap them in parallel. You need matched pairs, use balancing resistors, etc.

kingcake wrote on 2023-10-28, 20:00:

Sphere478 wrote on 2023-10-28, 18:30:

Na, that’s not how dc works. You can parallel multiple DC sources of identical voltage.

Sometimes this can cause more (or less) line noise. But I don’t expect you will have problems relating to this. The condition of your mobo caps would be where I looked if you did.

Differrences in MOSFET Rds will cause one MOSFET to do most of the work. You can't just slap them in parallel. You need matched pairs, use balancing resistors, etc.

I mostly agree. However, with currents in the magnitude we're talking about for a powerful +5V fed cpu, 8 inches of even high area cable almost resembles a balancing resistor. Ideally one would have (ideal) diodes in series between to supplies, but practically I guess that voltage drop across worrying as well as across mosfets balances it out. Sphere478 even mentions that this is done in his profession - makes it seem pretty plausible IMO.

kingcake wrote on 2023-10-28, 20:00:

Differrences in MOSFET Rds will cause one MOSFET to do most of the work. You can't just slap them in parallel. You need matched pairs, use balancing resistors, etc.

Connecting another buck converter in parallel works just fine, mosfets are deep down the chain in front of coil/diode/filtering network. One caveat is Voltage level, but it works out in the end.
Julian Ilett Parallel Boost Converters https://www.youtube.com/watch?v=bIN-WEKI8BE

As ATX12VO gets established will it ever make sense to solve the strong +5V (and also -5V while you're at it) problem by making a new converter board design, or would such a converter be just as expensive as finding a strong +5V PSU?

rasz_pl wrote on 2023-10-28, 21:21:

kingcake wrote on 2023-10-28, 20:00:

Differrences in MOSFET Rds will cause one MOSFET to do most of the work. You can't just slap them in parallel. You need matched pairs, use balancing resistors, etc.

Connecting another buck converter in parallel works just fine, mosfets are deep down the chain in front of coil/diode/filtering network. One caveat is Voltage level, but it works out in the end.
Julian Ilett Parallel Boost Converters https://www.youtube.com/watch?v=bIN-WEKI8BE

The one with the higher output voltage will take most of the load. This is why you calibrate them to the same voltage under simulated load individually if you want to get this far into the weeds. But this is mostly over thinking. Loading one down will also pick up the other if they are close. Nothing bad is going to happen either way. If they are close. If one is 4v and one is 7v that is a problem. But 4.98 vs 5.56 won’t hurt anything other than balance

Indeed we deal with voltage drop over the wire as well which will occur in the small ga wire in a pc. (Resistor) when dealing with DC voltages this low, the voltage drop is nutty, the wires may as well be literal resistors. This is why all of our off grid solar installs are 48v we try to avoid 24v and 12v like the plague. It can work, but it is very limiting for the scale we are working with. (Running entire houses)

You can indeed just slap multiple dc sources together without harm. So long as those sources are the same voltage as I said before.

If you find one has more current than you would like, adjust its voltage trim pot. And the amperage will shift.

Most people don’t bother and it works just fine but when I do an install I balance the voltage offsets on controllers when we do installs. It’s basically the equivalent of geordie laforge trying to get that extra 0.05% out of the warp injectors to compete with the other ship’s engineer.

By the way, another way you could approach this is to use a dual psu atx splitter and a pico psu powered by one of the 12v rails

Sphere478 wrote on 2023-10-28, 15:59:

Might help.

Curious if it backfeeds the 5v or only supplies it to the three aux connectors.

You could also make one pretty easily by buying a 12v to 5v converter and installing a molex on it. Pulls 12v and backfeeds 5v back to the 5v rail (multiple 5v sources in parallel, which is fine, we do this a lot in solar with higher votlage)

rasz_pl wrote on 2023-10-28, 18:46:

There are Chinese 10-20A $7-20 Buck converters on ebay doing same thing, $30 for brand name in nice case and with all appropriate sockets doesnt sound terrible.

And it works for pre-P4 ATX builds?
I have some experience with electrical stuff, but is there kinda a schematic or a video that guides on how to this with an ATX PSU, I'd rather not do this blind?

jakethompson1 wrote on 2023-10-28, 21:40:

As ATX12VO gets established will it ever make sense to solve the strong +5V (and also -5V while you're at it) problem by making a new converter board design, or would such a converter be just as expensive as finding a strong +5V PSU?

I mean, buying old stuff, even "brand new unused", means 90s PSU quality that's already 20+ years old. Feels like chasing unicorns.
Getting a modern one with strong +5v rail feels like chasing different colored unicorns.
But yeah...

Sphere478 wrote on 2023-10-28, 22:05:

By the way, another way you could approach this is to use a dual psu atx splitter and a pico psu powered by one of the 12v rails

Isn't PicoPSU low-wattage?
And this would be basically using PicoPSU as a 12>5v converter, is PSU's +5v also used in tandem?
Would that PicoPSU+other 12v rail connected to mobo when used combined get overloaded?

Is this maybe necessary and/or another viable alternative (it has 5A per rail/7A combined max/tap Molex for more): https://github.com/dekuNukem/ATX4VC

This is *NOT* a load balancer.

Load balancer means a device have many outputs that is regulated individually to each load. Remember current is different on each load regardless of voltage. Good example: Charging battery per cell to keep them balanced.

What this you are looking at is, this device is a DC to DC converter to supply more 5V on one bus for extra current (0ut) using from 12V input.

Cheers,

rasz_pl wrote on 2023-10-28, 21:21:

kingcake wrote on 2023-10-28, 20:00:

Differrences in MOSFET Rds will cause one MOSFET to do most of the work. You can't just slap them in parallel. You need matched pairs, use balancing resistors, etc.

Connecting another buck converter in parallel works just fine, mosfets are deep down the chain in front of coil/diode/filtering network. One caveat is Voltage level, but it works out in the end.
Julian Ilett Parallel Boost Converters https://www.youtube.com/watch?v=bIN-WEKI8BE

Are you really posting a retired sign repairmen that teaches people how to build lithium death traps on youtube as a primary source? Did you even watch the video? He has to fiddle and fiddle to tune it. The SECOND the load changes you will have to start over.

There's a reason motherboard manufacturers don't put a bunch of cheap, dumb buck converters in parallel. Because you can't. Period. One converter will always be sourcing and one will always be sinking.

They use synchronous buck converters with multiple phases.

Here's a brief summary paper from TI about this: https://www.ti.com/lit/an/slyt748/slyt748.pdf

But if you want to keep believing randos on Youtube, be my guest.

2mg wrote on 2023-10-28, 22:50:

And it works for pre-P4 ATX builds? I have some experience with electrical stuff, but is there kinda a schematic or a video that […]
Show full quote

Sphere478 wrote on 2023-10-28, 15:59:

Might help.

Curious if it backfeeds the 5v or only supplies it to the three aux connectors.

You could also make one pretty easily by buying a 12v to 5v converter and installing a molex on it. Pulls 12v and backfeeds 5v back to the 5v rail (multiple 5v sources in parallel, which is fine, we do this a lot in solar with higher votlage)

rasz_pl wrote on 2023-10-28, 18:46:

There are Chinese 10-20A $7-20 Buck converters on ebay doing same thing, $30 for brand name in nice case and with all appropriate sockets doesnt sound terrible.

And it works for pre-P4 ATX builds?
I have some experience with electrical stuff, but is there kinda a schematic or a video that guides on how to this with an ATX PSU, I'd rather not do this blind?

jakethompson1 wrote on 2023-10-28, 21:40:

As ATX12VO gets established will it ever make sense to solve the strong +5V (and also -5V while you're at it) problem by making a new converter board design, or would such a converter be just as expensive as finding a strong +5V PSU?

I mean, buying old stuff, even "brand new unused", means 90s PSU quality that's already 20+ years old. Feels like chasing unicorns.
Getting a modern one with strong +5v rail feels like chasing different colored unicorns.
But yeah...

Sphere478 wrote on 2023-10-28, 22:05:

By the way, another way you could approach this is to use a dual psu atx splitter and a pico psu powered by one of the 12v rails

Isn't PicoPSU low-wattage?
And this would be basically using PicoPSU as a 12>5v converter, is PSU's +5v also used in tandem?
Would that PicoPSU+other 12v rail connected to mobo when used combined get overloaded?

Is this maybe necessary and/or another viable alternative (it has 5A per rail/7A combined max/tap Molex for more): https://github.com/dekuNukem/ATX4VC

Basically you would use the pico to power a few drives to take the load off. It would power up when the mobo does and do exactly what the corsair unit seems to be doing. Less elegant. Not saying it is better, just mentioning it. You could also use a full blown psu

pentiumspeed wrote on 2023-10-28, 23:30:

This is *NOT* a load balancer. […]
Show full quote

This is *NOT* a load balancer.

Load balancer means a device have many outputs that is regulated individually to each load. Remember current is different on each load regardless of voltage. Good example: Charging battery per cell to keep them balanced.

What this you are looking at is, this device is a DC to DC converter to supply more 5V on one bus for extra current (0ut) using from 12V input.

Cheers,

All this thing does is to offload (spare) power from 12V rail to power some external accessories.

Useful if your PSU's 5V rail is weak while you have a lot of external devices using 5V, like HDD/SSD, ARGB lights, etc. It won't be of any help for onboard stuffs so if Athlon boards still require much more power on 5V than your PSU could offer this will not make any difference.

By the way, it seems some PSU manufacturers once again increased capacity of +5V a bit. I'm seeing EVGA offering 24A on +5V.

kingcake wrote on 2023-10-29, 01:09:

Are you really posting a retired sign repairmen that teaches people how to build lithium death traps on youtube

:]

kingcake wrote on 2023-10-29, 01:09:

He has to fiddle and fiddle to tune it. The SECOND the load changes you will have to start over.

fiddling is for demonstration purposes, they are self balancing. If there is offset one will take the brunt of the load until it cant keep up and sag to the level of the other. https://www.digikey.com/en/articles/paralleli … er-output-power
If you worry this much add blocking Schottky diodes.