Thank's for inspiration .... At least finally I am doing something for the community that I love most ...
Dream's buddy .... dream's that made us fly with "Right brother's" .... ?
Why there is not much hype on active cooled ac-dc brick ? doesn't they make cases more smaller ?
- Thread starter graphichasan
- Start date
-
- Tags
- 300w 400w 500w ac-dc adapter power brick
i also didn't have enough heart to tell OP, that guryhwa has been making exactly what he's taking about, for several years now.
i have a 500W external AC brick, which has 2 fans, that i got from guryhwa.
works perfectly.
Personally I really admire his work .... he is the one who brought 800W pico-psu concept to life at affordable price .... Did mention his work on I think every post of mine .... My idea is not about just making a product that he is already doing .... but to improve & make it standard with as much safety feature as possible .... like accidental water spill is a vary bad situation for an actively cooled external brick .... & a new one like strip-down rail version for internal chassis mounting system .... why just limit a 800W powerful pico-psu to only 500W .... let the hunt began to make tinny cases overclocking champ .... ?
Project Thundercreck
400W active cooled ac-dc brick || Mini 6L Awesome Workstation
Why there is not much hype on active cooled ac-dc brick ? doesn't they make cases more smaller ?
Last edited:
With a fan and opening for air flow, it won't be water resistant. You'd need at least IPx4 level of resistance.
Not to sound hard, but good luck with that. Even HDPlex is not standard and quite a niche market. If it is not part of the ATX norm, then it is not standard, period.
Yeah, "making it standard" would require getting major industry players on board, not to mention a standards body such as IEEE or ISO. That is no small undertaking. Intel would also need to be involved, as they control the ATX standard. The best one could hope for is an unofficial extension to the ATX standard, but IMO we already have that. That there are multiple options in terms of DC input voltage is mostly a good thing, even if it can be confusing. 19V is arguably better for external bricks as they can be made noticeably smaller than 12V bricks and are far more available - but of course they have the drawback of needing internal conversion of the 12V rail, requiring bigger PCBs and heatsinks on the DC-ATX units.
The biggest possible gain from some sort of standard IMO would be for AC-DC units to adhere to ATX requirements for power quality, hold up time, etc (which not all do), as well as the possibility of passing the PS_ON signal from the DC-ATX board to the AC-DC unit (removing the need for load switches for other 12V lines). However both of these would make using currently available off-the-shelf AC-DC units impossible, making this concept kind of moot.
On a side note, designing an AC-DC brick like this at this point in time - with GaN units on the way in near-ish future from major OEMs, making them dramatically smaller - seems like poor timing. In a year or two we are likely to have far better options. GaN should also improve efficiency and output power quality.
Lastly, my main gripe with this design is that it seems too early and unfinished, and quite a few important parts are either omitted in the description, glossed over or overlooked both in the overall design and in the detailed descriptions of the parts. Load balancing for multiple AC-DC units is one. Ripple filtering for the higher output MeanWell units is another - IIRC the RPS-400 just barely clears the ATX spec for ripple and noise, and the RPS-500 is quite far out of spec. High quality ATX and SFX PSUs have DC output ripple an order of magnitude lower than this. While this isn't strictly necessary for operation of a PC (beyond being in spec - ripple outside of spec can and will kill components), it's important for stability when overclocking or otherwise pushing a lot of power through your PC. Then there's the lack of concrete plans for the distribution board (load switches? Ripple filtering? Load balancing?) and the seemingly complete lack of regard to how much space this will take and how difficult it will be to fit in a small case. The complete lack of discussion of load switches for the 12V rails that circumvent the DC-ATX board scares me a bit, as omitting that would leave not only your GPU but the CPU VRM receiving 12V even when the PC is switched off. Then there's the question of how to filter the DC inputs - just adding filter caps can lead to issues with different AC-DC units as smaller ones might not be able to handle the capacitive load required to filter the bigger ones sufficiently, which then creates a need for a large and potentially expensive filter circuit. And so on.
As someone mentioned above the current design will also struggle with airflow - remember that airflow drops as pressure drops, and pressure drops by roughly half for every 90° turn in the airflow path, so you would need a very high static pressure fan for your current design to provide the necessary airflow for the MeanWell units. For proper ingress proofing you'd also likely need some sort of membrane, which would again impede airflow dramatically.
Don't get me wrong, there is definitely merit to a concept like this, and making 12V DC-ATX more accessible is a very noble goal. But this is obviously in very, very early stages of development. Keep at it, though. If this pans out, it could be really good.
The biggest possible gain from some sort of standard IMO would be for AC-DC units to adhere to ATX requirements for power quality, hold up time, etc (which not all do), as well as the possibility of passing the PS_ON signal from the DC-ATX board to the AC-DC unit (removing the need for load switches for other 12V lines). However both of these would make using currently available off-the-shelf AC-DC units impossible, making this concept kind of moot.
On a side note, designing an AC-DC brick like this at this point in time - with GaN units on the way in near-ish future from major OEMs, making them dramatically smaller - seems like poor timing. In a year or two we are likely to have far better options. GaN should also improve efficiency and output power quality.
Lastly, my main gripe with this design is that it seems too early and unfinished, and quite a few important parts are either omitted in the description, glossed over or overlooked both in the overall design and in the detailed descriptions of the parts. Load balancing for multiple AC-DC units is one. Ripple filtering for the higher output MeanWell units is another - IIRC the RPS-400 just barely clears the ATX spec for ripple and noise, and the RPS-500 is quite far out of spec. High quality ATX and SFX PSUs have DC output ripple an order of magnitude lower than this. While this isn't strictly necessary for operation of a PC (beyond being in spec - ripple outside of spec can and will kill components), it's important for stability when overclocking or otherwise pushing a lot of power through your PC. Then there's the lack of concrete plans for the distribution board (load switches? Ripple filtering? Load balancing?) and the seemingly complete lack of regard to how much space this will take and how difficult it will be to fit in a small case. The complete lack of discussion of load switches for the 12V rails that circumvent the DC-ATX board scares me a bit, as omitting that would leave not only your GPU but the CPU VRM receiving 12V even when the PC is switched off. Then there's the question of how to filter the DC inputs - just adding filter caps can lead to issues with different AC-DC units as smaller ones might not be able to handle the capacitive load required to filter the bigger ones sufficiently, which then creates a need for a large and potentially expensive filter circuit. And so on.
As someone mentioned above the current design will also struggle with airflow - remember that airflow drops as pressure drops, and pressure drops by roughly half for every 90° turn in the airflow path, so you would need a very high static pressure fan for your current design to provide the necessary airflow for the MeanWell units. For proper ingress proofing you'd also likely need some sort of membrane, which would again impede airflow dramatically.
Don't get me wrong, there is definitely merit to a concept like this, and making 12V DC-ATX more accessible is a very noble goal. But this is obviously in very, very early stages of development. Keep at it, though. If this pans out, it could be really good.
Last edited:
Thank you for the time & thought you put through my concept .... by making Standard I mean by availability & sff builds like pico-psu 160W .... not like the OEM/Intel way ... English is my second lang so .... anyway concept design was to get what I am missing .... and your guide my friend is all I ever needed to go on .... working v0.2 .... hoping to be at 0.7L (200*85*40) with 2*40mm fan .... space for a pcb .... once done I will edit my post to proper goal & functionality .... see you there then .... ?
Last edited:
With a fan and opening for air flow, it won't be water resistant. You'd need at least IPx4 level of resistance.
I was kind of thinking about to go for IPX3 ....
No problem at all, I'm just glad to be able to provide some feedback
That being said, probably 95% of what I know about this stuff I have learnt from other far more knowledgeable users on these forums, so you should have plenty of opportunity to discuss and improve this over time. @Thehack's amazing thread A guide to 12V PSUs is how I first started learning about this stuff, and given that they also make and sell DC-ATX units (bundled with MW RPS-200-12s) there is definitely plenty of knowledge there. There are quite a few other knowledgeable users around here as well, so I would recommend reading through some of the more elaborate brickless build logs here - there are plenty of interesting 12VDC setups out there. My understanding of this is mostly on the conceptual side (as in: I mostly understand what functionality is needed and what pitfalls to watch out for), but I'm no electrical engineer by any means - just an enthusiastic DIYer who enjoys taking on projects a step or two beyond my level of competency and learning how not to blow things up while I go.
It seems that it´s difficult to make a 12v power supply big in power and low in ripple and noise. Here it is a 550w one measuring 127x76x38 but 240 mVp-p :/
what is your thought about this unit MeanWell EPP-400-12 datasheet at 120mv ripple .... just need a bit more ripple control ...
I don't think it's necessarily difficult as such - modern high efficiency ATX/SFX PSUs are mostly 12V PSUs with internal DC-DC converters for the minor rails, after all - but it might be more expensive, require more effort/complex designs, or just not be a requested spec for the industries these PSUs normally serve.
As such, one of the "easier" ways of making a high output external 12V brick would be to start with an SFX PSU, make a new enclosure for it, rewire the input and output(s), and change the fan placement to something more suitable for external use. Doing something like this ought to be comparable in volume (if not smaller) than any MW-based setup once you exceed 4-500W.
Hmm ... that can be done ... but will need a lot for cable, housing, connector's & will have to consider voltage drop over 4-6 feet .... I hate cable mess ... also case may be quite big at 125*125*65 1.15L .... let's design both ... then compare ...
While the datasheet says max 120mv ripple, if you take a look at the test report it suggests EPP-400 will be safely under that
66mVp-p in that specific test(Also I have been using EPP-400-12 to directly power my PC for the past year)
So Ripple Filtering, Over Voltage, Over Current, Short Circuit is checked for 400W Unit I need to weary about Load balancing, load switches & Airflow to combine two of them at 800W... am I missing something ...
Here I am searching ac DC converters. I haven't found anything special apart from this outsandingly expensive vicor unit. It makes 400w at 24 (there isn't a 12v one) but it measures just 35.54 X 124.77 X 9.3mm and seems very efficient... Only for 339$ if I buy 100 units.
If you're up to modding, ex server psu can be had for peanuts. In my country 600-800W goes for 12 - 20$
That's a good path ... I think Guryhwa is already doing that ... but it is vary unreliable ... you see part of why MeanWell is their safety & 3 years of warranty ... imagine you provide bricks to let say 20 people, only to realized 14 of them failed with in 6 month ... that's a vary bad reputation ... I want to supply only quality component's as in the case of overclock you must not want to power with a creapy supply to blow things up...
Sorry I should have stated that I was only replying to the previous post by @paulesko .
Of course, refurbishing server psu into a brick would be a bad business idea.
Of course, refurbishing server psu into a brick would be a bad business idea.
Similar threads
