Concept Flex-ITX motherboard form factor concept

[Thehack]

The only two suggestions that involve a new form factor is a 6x6 and the AIO (partially new since asrock already has one PIO).

All other suggestions, which are plenty extends current itx into a better quality of life for everywhere builder.

Concerning the marginal cost increase, I'm specifically targeting the net cost. Which is definitely not marginal, considering the R&D and low run. Would you pay more for a board with essentially no benefit in the ncase m1? I wouldn't.

Think of the cost of a custom board + pcb vs a riser + off-the-shelf. I cannot imagine the custom board + pcb accessory costing less, all to save 2mm in a sandwich case.

My opinion is 12v only compatible with atx, NOT Intel atx12vo is better move forward.

My comments may be intruding and possibly off topic, but you guys are industry studs and I want bring some feedback to what I think are more innovative features than just switching to a pcb riser while costing more. These can be parallel to this super flexible itx board but I feel like some of the focus is misplaced.

 

[pwnedundies]

Yes, that would be trivial to implement. However, the GPU fans would be facing the back of the motherboard, so you'd want to have a good reason for doing it vs. a back-to-back layout.

I meant it as leaving the layout alone with the gpu fans facing out. It would be a custom pcb or flip the pcie connector to have it oriented the correct direction. I totally get what you are saying though.

 

[Necere]

The only two suggestions that involve a new form factor is a 6x6 and the AIO (partially new since asrock already has one PIO).

All other suggestions, which are plenty extends current itx into a better quality of life for everywhere builder.

Anything that breaks compatibility with mini-ITX is out of scope for this concept and this thread. The point is to retain compatibility with the ATX ecosystem, while extending the utility.

Concerning the marginal cost increase, I'm specifically targeting the net cost. Which is definitely not marginal, considering the R&D and low run. Would you pay more for a board with essentially no benefit in the ncase m1? I wouldn't.

Think of the cost of a custom board + pcb vs a riser + off-the-shelf. I cannot imagine the custom board + pcb accessory costing less, all to save 2mm in a sandwich case.

I think you're still in the mindset that this is just for one board. It's not. This is for an extension to the mini-ITX form-factor. If it catches on with manufacturers, it could completely replace mini-ITX as the standard SFF form-factor.

A PCB riser is inherently more reliable and less expensive than a flexible riser, so the benefit is tremendous for any layout that currently requires one.

 

[Thehack]

Anything that breaks compatibility with mini-ITX is out of scope for this concept and this thread. The point is to retain compatibility with the ATX ecosystem, while extending the utility.


I think you're still in the mindset that this is just for one board. It's not. This is for an extension to the mini-ITX form-factor. If it catches on with manufacturers, it could completely replace mini-ITX as the standard SFF form-factor.

A PCB riser is inherently more reliable and less expensive than a flexible riser, so the benefit is tremendous for any layout that currently requires one.

I understand the lens of your approach now but I am still worried about the industry take up on it.

No big makers have stepped into the sub 10L game and the motherboard industry is hyper competitive even at these huge quantities. This spec extension would really only be useful for sub 10L.

At some point we will also at least need a proof of concept beyond a physical spec sheet so these comments are still considerations for that at least.

 

[Necere]

I understand the lens of your approach now but I am still worried about the industry take up on it.

No big makers have stepped into the sub 10L game and the motherboard industry is hyper competitive even at these huge quantities. This spec extension would really only be useful for sub 10L, which like asrock pointed out, sales are loss leaders.

That's part of the impetus for this: moving to a PCB riser for sandwich layout cases reduces cost and therefore makes it more accessible to more people. It effectively lowers the barrier to entry to sub 10L SFF.

 

[Poblopuablo]

Although mounting points are same and basic form factor is the same. Some cases with a very specific layout) might be exclusive to flex itx and not compatible with the standard itx.(similar to how some cases don't support dtx, but support itx (


Since the pcie riser no long comes out perpendicular to the motherboard it no longer requires ~15-20mm for the riser height, allowing the gpu(or other components) to sit directly above where the pcie slot would normally fit.


One cool idea would be a different t slot varriant.

 

[race2c]

Nice to see this endeavor taking place. I will be following this thread to see what happens.

 

[Necere]

Updated the first post with a more organized product brief. Hopefully it better explains the what and why.

 

[ermac318]

Updated the first post with a more organized product brief. Hopefully it better explains the what and why.

I understand what you're going for much better after that update. Now it makes perfect sense, switching to the Edge connector and bundling a 90 degree connector in the box essentially keeps you 100% compatible with Mini ITX, but now gives you more options for sandwich.

Would a PCB riser for sandwich cases in the back have any issues with SMD clearance to avoid shorts? Otherwise you'd need to put some non-conductive surface on the face (or hope the motherboard has a backplate that's sufficiently thin to still accommodate the riser).

 

[Necere]

Would a PCB riser for sandwich cases in the back have any issues with SMD clearance to avoid shorts?

No. It should go without saying that you'd want to design the riser such that it respects the motherboard keepout zone. The ATX specs call for a 6.35mm (1/4") standoff height, so that effectively sets the minimum distance the riser should be from the back of the motherboard.

 

[race2c]

I may have missed it, but has there been any feedback recieved from the motherboard manufacturers and whether this is a feasible endeavor?

 

[arthasdklol]

OK, I understand this is more of an evolution than revolution type of a project.
Still, I would pay the same amount of money to get fewer slots/features (dunno, call it superleggera and charge more for less like Porsche/Lambo...) and extra to integrate something @Thehack built with respect to the ATX power plug. For compatibility reasons it could have a space allocated and some guiding plastic element to allow people to plug in a full 24pin plug, just not use the irrelevant pins...

 

[dondan]

Good idea but keep in mind also long PCB traces will require retimer or redriver for PCIe future speeds.

 

[Necere]

Good idea but keep in mind also long PCB traces will require retimer or redriver for PCIe future speeds.

Yes, I listed it under advantages of PCB risers after @ExplodingWaffle101 mentioned it.

 

[Kilrah]

Pretty meh in my opinion. If you're going to do something that requires changing the mainboard then go with a solution like Intel presented, just with the "processing" board populated on the correct side.
For ITX form factor you put the PCIe edge connector on the "top" so that a sandwich case just needs a small backplane style "U" which can also include the power connections that go the the Flex PSU right under it, in console style it's a short I connector... Instead of having a stupid long PCB running the whole length for both form factors.

 

[Necere]

Pretty meh in my opinion. If you're going to do something that requires changing the mainboard then go with a solution like Intel presented, just with the "processing" board populated on the correct side.
For ITX form factor you put the PCIe edge connector on the "top" so that a sandwich case just needs a small backplane style "U" which can also include the power connections that go the the Flex PSU right under it, in console style it's a short I connector... Instead of having a stupid long PCB running the whole length for both form factors.

You're missing the point. This is not an entirely new form factor - it's a small change to an existing form factor.

SFF is a niche, and sandwich-style cases are a niche within a niche. You want specialized motherboards that only work with certain layouts? It's expensive to develop a new motherboard. The sales need to be there to justify the cost of development. Motherboard manufacturers aren't going to invest in a new form factor that only works with a small number of cases when they can just develop a mini-ITX (or ATX) motherboard and sell 50 (or 500) times as many. The idea here is to make a small, low-cost change to mini-ITX that would offer signficant benefits to niche SFF cases. We're not going to get anywhere by demanding the moon.

 

[scatterforce]

I think this is a great idea. With the NUC 9, sandwich style cases are getting some more attention. Now is the time to shine a spotlight on this and this is a very minor change that could bring about some major benefits.

 

[prava]

I understand the lens of your approach now but I am still worried about the industry take up on it.

No big makers have stepped into the sub 10L game and the motherboard industry is hyper competitive even at these huge quantities. This spec extension would really only be useful for sub 10L.

At some point we will also at least need a proof of concept beyond a physical spec sheet so these comments are still considerations for that at least.

The thing is that this implementation doesn't mean much in terms of cost, since mITX motherboards are already very expensive to begin with: in a $50 motherboard this would be problematic... but have you checked mITX prices latetly for high-end skus?

BUT, at the same time, it offers a huge functionality bonus since it streamlines the usage of different layouts that change the traditional orientation of gpu in relation to the motherboard without requiring shielded risers that are delicate, expensive, and a general pita in many regards (let alone space-consuming).

All in all this is a very very good idea, that I hope gets some traction.

 

[theGryphon]

What would go really well with this concept is a 90-degree hard PCIE riser, to enable some interesting case layouts that are deemed too fringe now.

 

[Kilrah]

SFF is a niche, and sandwich-style cases are a niche within a niche.

My suggestion doesn't cater for just sandwich, I used both same examples you did and it does both better...

You're missing the point. This is not an entirely new form factor - it's a small change to an existing form factor.

In terms of engineering it means a new board layout in both cases, so pretty similar.

Both solutions would command a premium on the pricing, so I don't think that "you can use it in a standard mini-ITX config" argument has much weight. The only people willing to pay for the premium will be those who actually want the feature, so then why not make it ideal for them increasing your chances that people find it interesting instead of making an intermediate no one else will use and won't drive sales numbers up.