I see. Better start asking now if chances are that it takes a while to find someone
Johnny has tested some super-niche PSUs, even proprietary form factors from shuttle PCs, so that could make a fair bit of sense.
-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Power Supply 500W FlexATX PSU
- Thread starter iFreilicht
- Start date
I see. Better start asking now if chances are that it takes a while to find someone
Johnny has tested some super-niche PSUs, even proprietary form factors from shuttle PCs, so that could make a fair bit of sense.
Yes and yes. If it had a -12V rail, I would've already plugged it in
I'll test it out with the -12V regulator on a breadboard first. Though, with the simplicity of the daughterboard, it should even be possible to mod that into the PSU directly. If we can't find enough people to buy a full unit, just selling an "upgrade" kit would be an alternative idea. It would just consist of a new daughterboard, maybe even just an add-on to the ATX24pin connector.
Let me know if you need any help with that. I have experience with designing PCBs and selecting cost effective components.
Oh, I'd like to see the review of that one! Link?
While typing up the PM to you (incoming in a moment), I noticed that by accident, I actually connected my mainboards -12V pin to GND, not to the converters output, and it still started up fine. WTF
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=281
Edit: Johny Guru seems to have reviewed the TFX one. However, hardware secrets have a review here:
http://www.hardwaresecrets.com/athena-power-ap-mfatx40p8-400-w-power-supply-review/
While typing up the PM to you (incoming in a moment), I noticed that by accident, I actually connected my mainboards -12V pin to GND, not to the converters output, and it still started up fine. WTF
That's interesting. I don't know how this particular motherboard detects -12V, but somehow the GND is fooling it to think that there is a negative voltage present at this pin. I don't even know why motherboards detect this voltage for POST as it is an obsolete voltage standard. -12V was originally intended to power serial port, PCI (non-express) and Opamps in sound cards ( +12 , GND, -12 for Opamps) .
Last edited:
When you tested previously, did you have the '-12V sense' pin grounded, or left disconnected (floating)?While typing up the PM to you (incoming in a moment), I noticed that by accident, I actually connected my mainboards -12V pin to GND, not to the converters output, and it still started up fine. WTF
That doesn't look too bad, though build quality is subpar. I'd really like to see a review of the 500W Flex unit, maybe there's a different reason why it failed for @Hahutzy?
I do know that it doesn't work with nothing connected, so that's something.
Are you talking about the "FB" pin? That's connected to "REF", just like the sheet says. I didn't change anything else when swapping the connection from GND to -12V and I confirmed with a multimeter that the potential between those two is indeed -12V. GND is also connected to the SATA GND directly, so it really is at ground potential.
That HWsecrets article is the one I consulted to instill some confidence before ordering the 500W unit.
I tweeted at them shortly after reading it, hoping they could test the 500W. Never heard a response.
Quick note on setups I tried on it:
1) i5-4440S + R9 280X @ 100% power limit - fails to finish 3DMark (shuts off)
2) i5-4440S + R9 280X @ 80% power limit - fails to finish 3DMark (shuts off)
3) i5-4440S + GTX 980 Ti @ 100% power limit - fails to finish 3DMark (shuts off)
4) i5-4440S + GTX 980 Ti @ 60% power limit - successfully finishes 3DMark
5) i5-4440S + GTX 980 @ 100% power limit - fails to finish 3DMark (shuts off)
6) i5-4440S + GTX 980 @ 60% power limit - successfully finishes 3DMark
I used a Molex to PEG to make sure I was balancing the load on both rails.
I don't exactly know what the reason is for all these tests to fail - none of them should if everything works as per specifications. I do hope someone can sort out the AP 500 though. It'd be a great option if it can live up to its paper spec.
Last edited:
So, does that mean the GPU was connected to both rails at the same time? Have you tried different configurations of this?
Yea I tried all the different configs I could.
At the beginning I had both on the same rail (with the stock PEGs).
Then when it shut down on me, I thought that was the problem.
So I went with MOLEX to PEG for one of them to get 150W from one rail, and the other 75W from the other rail.
Didn't help still, but at least I know the problem is more complicated than "overloading a single 12V rail"
Short Update:
I know there were no updates for a while, and I'm sorry about that, so here's a short roundup of the situation:
Freilite e.K. is now incorporated and has a proper website and email address, so I'm now legally allowed to sell stuff.
A better solution for the -12V problem was found, using an isolated power converter like this. While this increases reliability, it also will increase the cost of the PSU by something between 15 and 5 bucks. Not really a fan of that
Most importantly, though, FSP allegedly released some info about 600W and 750W FlexATX PSUs at electronica, with -12V rails. I don't know too much yet, but I'll pester them as long as needed to get some information about that. I'll only change the title of the thread once I can confirm that those will actually exist at 150mm length. While 600W are of course overkill for systems that profit from FlexATX, the additional power overhead would make it easier to integrate two 8pin connectors and reduce noise further.
I'll do my best to pick up the pace.
I know there were no updates for a while, and I'm sorry about that, so here's a short roundup of the situation:
Freilite e.K. is now incorporated and has a proper website and email address, so I'm now legally allowed to sell stuff.
A better solution for the -12V problem was found, using an isolated power converter like this. While this increases reliability, it also will increase the cost of the PSU by something between 15 and 5 bucks. Not really a fan of that
Most importantly, though, FSP allegedly released some info about 600W and 750W FlexATX PSUs at electronica, with -12V rails. I don't know too much yet, but I'll pester them as long as needed to get some information about that. I'll only change the title of the thread once I can confirm that those will actually exist at 150mm length. While 600W are of course overkill for systems that profit from FlexATX, the additional power overhead would make it easier to integrate two 8pin connectors and reduce noise further.
I'll do my best to pick up the pace.
It does, but until I actually get confirmation, I don't believe that unit is just 150mm long. That would be a crazy power density. On the other hand, they already have this redundant 700W supply, where both units can supply 700W each, and that's just 80Plus Gold. They also have 80Plus Titanium ATX PSUs already. Dreams might come true.
About that new, better PSU:
So I received the list (which apparently isn't confidential, so take a look here), and unfortunately the new Flex PSUs with 650, 750 and 850 Watts, while including a -12V rail, are only going to be available at 220mm length, which makes them completely unsuitable for most cases the FSP500-50FSPT could be used in.
Interestingly, there seems to be a different version of the FSP500-50FSPT, the FSP500-50FGGBN, which is almost exactly the same except that some rails don't have a minimal load specified and that it is only 80Plus Gold rated instead of Platinum. It seems that this could be an alternative that is less expensive but also louder than the Platinum version. Personally, I don't like the idea of bringing back the exact reason why Flex is so stigmatised as a form factor, but I'll try and get quotes for either version to see what the cost difference will be.
So I received the list (which apparently isn't confidential, so take a look here), and unfortunately the new Flex PSUs with 650, 750 and 850 Watts, while including a -12V rail, are only going to be available at 220mm length, which makes them completely unsuitable for most cases the FSP500-50FSPT could be used in.
Interestingly, there seems to be a different version of the FSP500-50FSPT, the FSP500-50FGGBN, which is almost exactly the same except that some rails don't have a minimal load specified and that it is only 80Plus Gold rated instead of Platinum. It seems that this could be an alternative that is less expensive but also louder than the Platinum version. Personally, I don't like the idea of bringing back the exact reason why Flex is so stigmatised as a form factor, but I'll try and get quotes for either version to see what the cost difference will be.
That thing can only be used with hearing protection, though. The 1U 700W Platinum FSP700-80UEPB that @LukeD used for his build is 20mm longer, 18.5mm wider, has two fans in parallel, 150W less to push and is more efficient. It is allegedly and presumably very quiet, though.
At 220mm, 1U is the better form factor anyway. Flex-L won't become a thing if I have a say in it.
At 220mm, 1U is the better form factor anyway. Flex-L won't become a thing if I have a say in it.
But it will work .... "oh so well" in my case
Oh I'm so excited now iFreilicht.I've already got 2 different FlexATX psu's at home and about another 4-5 IPC-1U's ... but this new FlexATX line .... looks soooooooo good
If only they made them modular
If only they made them modular
At 220mm you'd think there'd be enough space for that. Apparently not
Or it's just cheaper or safer for them to get them made to customer specs each time.
At 220mm you'd think there'd be enough space for that. Apparently not
Or it's just cheaper or safer for them to get them made to customer specs each time.
Probably the latter, but .....
I recently got a nice Seasonic which is the same dimensions and is modular. Its only 500Watt but GOLD
Similar threads
