-
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
See for yourself, jonny reviewed it a while back.
Way better than the SSP300, it has two connectors for the ATX24pin, one for 12V only connections and one for peripherals with 12V, 5V and 3.3V. That's good enough in my book.
See for yourself, jonny reviewed it a while back.
Way better than the SSP300, it has two connectors for the ATX24pin, one for 12V only connections and one for peripherals with 12V, 5V and 3.3V. That's good enough in my book.
So they effectively gave buyers even less reason to get the TFX 350 they tried so hard to market :s
Here's a fantastic case for flex
https://www.aliexpress.com/store/pr...D-2-PCI-slots-aluminum/326009_1581749885.html
https://www.aliexpress.com/store/pr...D-2-PCI-slots-aluminum/326009_1581749885.html
Nice find, though I like the GEEEK A10 better. I think there's even a thread about it on here somewhere.
-12V Progress!
This might be the most boring schematic I've ever done.
And the PCB isn't much exciting, either:
It's a very simple solution with an isolated DC/DC supply and a connector. The PCB is 27.4mm*13.7mm and is only using SMD components, so it can be glued to the side of the ATX24pin connector with double sided tape.
The DC/DC supply can be one of three possible units from XP Power, Recom Power or CUI Inc which are pin-compatible. Which one will be used in the end will depend on availability and assembly cost. The connector is a 2mm pitch JST PH, which is also available from TE, so mating plugs are easy to source for everyone who wants to customise their cable harness but still needs 12V.
As you can see, the PCB has a somewhat odd layout for the pins with GND in the middle. This has two reasons. First, it allowed to make the PCB single-sided, which reduces cost a little bit. And second, as we found out, some boards - like the Gigabyte B85N-Phoenix - that "check" the -12V rail on boot can be fooled by just shorting the -12V pin to ground. So if someone wants to try that without modifying the PSU, all they need is to get a matching connector and bridge the GND and -12V pins with some solder. So with this pin layout, it is very easy to see whether Vin is shorted to ground in the process.
At this point I have to mention two things. First, what I just described is of course not officially supported and you might risk damaging your mainboard. Second, this -12V unit can only serve as a voltage reference, it's output current is about 80mA, which is below the current recommended by the ATX standard. This means that on mainboards with PCI (which still exist), PCI cards (which also still exist) that require -12V will not work and could potentially damage this unit by drawing too much current. I don't know whether there'll be any issue with RS-232 devices, but I think that shouldn't be a problem. Good, we got that covered.
Now the next step for this thing is to panelise it, generate the gerber files and then ask my manufacturer (the company that makes the PCB) and assembler (that company that puts the components on the PCB) for a quote.
Luckily, I already did part of that:
The rest I'll do tomorrow. Thanks for reading!
This might be the most boring schematic I've ever done.
And the PCB isn't much exciting, either:
It's a very simple solution with an isolated DC/DC supply and a connector. The PCB is 27.4mm*13.7mm and is only using SMD components, so it can be glued to the side of the ATX24pin connector with double sided tape.
The DC/DC supply can be one of three possible units from XP Power, Recom Power or CUI Inc which are pin-compatible. Which one will be used in the end will depend on availability and assembly cost. The connector is a 2mm pitch JST PH, which is also available from TE, so mating plugs are easy to source for everyone who wants to customise their cable harness but still needs 12V.
As you can see, the PCB has a somewhat odd layout for the pins with GND in the middle. This has two reasons. First, it allowed to make the PCB single-sided, which reduces cost a little bit. And second, as we found out, some boards - like the Gigabyte B85N-Phoenix - that "check" the -12V rail on boot can be fooled by just shorting the -12V pin to ground. So if someone wants to try that without modifying the PSU, all they need is to get a matching connector and bridge the GND and -12V pins with some solder. So with this pin layout, it is very easy to see whether Vin is shorted to ground in the process.
At this point I have to mention two things. First, what I just described is of course not officially supported and you might risk damaging your mainboard. Second, this -12V unit can only serve as a voltage reference, it's output current is about 80mA, which is below the current recommended by the ATX standard. This means that on mainboards with PCI (which still exist), PCI cards (which also still exist) that require -12V will not work and could potentially damage this unit by drawing too much current. I don't know whether there'll be any issue with RS-232 devices, but I think that shouldn't be a problem. Good, we got that covered.
Now the next step for this thing is to panelise it, generate the gerber files and then ask my manufacturer (the company that makes the PCB) and assembler (that company that puts the components on the PCB) for a quote.
Luckily, I already did part of that:
The rest I'll do tomorrow. Thanks for reading!
Last edited:
The unit does have PMBus and I know you can check the fans status through that, but I don't think you can control it. The problem with semi-fanless functionality in a PSU like this is that there are no vents on the top or bottom, so there is no airflow through convection in most orientations, and that's what passive and semi-passive SFX and ATX PSUs rely on.
In theory it might be possible to reprogram the MC, but that's some serious and dangerous hacking that I wouldn't want to attempt. Maybe a later revision will add this, for now I just want to get it out the door. I've first seen this unit over a year ago, it's about time I deliver on it.
In theory it might be possible to reprogram the MC, but that's some serious and dangerous hacking that I wouldn't want to attempt. Maybe a later revision will add this, for now I just want to get it out the door. I've first seen this unit over a year ago, it's about time I deliver on it.
It's a little mystifying to me why FSP left off the -12v rail on this PSU. Is it because it's aimed at server boards and they don't require it, or something? if so, why make it flexATX, and not 1U...
Well, whatever the case, w360 has been in contact with them about this and a couple of their other flex PSUs, since we've got a project or two that will potentially be using the form factor.
Well, whatever the case, w360 has been in contact with them about this and a couple of their other flex PSUs, since we've got a project or two that will potentially be using the form factor.
Well looking at the pictures, it is chock full with components, I don't really know where they'd put a -12V rail. And yeah, they probably left it out because OEMs can just make boards that don't need -12V if they want to use this PSU.
Um, that's cool and all, but what does that mean for me? You've got a much stronger position to deal with them than me, do you want to take this over? Maybe PM me if the answer is confidential or something like that.
As it turns out, there may be something even better on the horizon: w360 also contacted Seasonic, and they say they're working on a Platinum-rated, 650W flexATX PSU, scheduled for sampling in Q3. I see flexATX as the next logical step beyond SFX in the SFF space, so this is very exciting news to me.
Is it 150mm long, tho? 190mm long Flex PSUs can eff right off. Do we know whether they want to release to retail? Or will that be OEM only, too?
Last edited:
So they claim. Believe me, I wouldn't be interested if it were 190mm.
That I don't know.
650W would allow plenty of builds and quiet operation in most use-cases, hopefully this will make it to market !
By the time the SFF case brands have adapted to SFX, they can't start redesigning around FlexATX
By the time the SFF case brands have adapted to SFX, they can't start redesigning around FlexATX
Should we really? To be perfectly honest, a safer approach to me seems to be using adapter plates for now. Many cases would profit from using an offset Flex PSU in the position they already have their SFX/ATX PSU in.
First version of drawing complete!
Today I've finally completed a first version of the drawing! Link to PDF.
If anyone finds a mistake or has objections, please tell me! The division of connectors to the 12V rails is not final, everything else seems pretty much fine to me.
I'll add this to the first post so people can find it easily.
Today I've finally completed a first version of the drawing! Link to PDF.
If anyone finds a mistake or has objections, please tell me! The division of connectors to the 12V rails is not final, everything else seems pretty much fine to me.
I'll add this to the first post so people can find it easily.
Similar threads
