Completed Salvo Studios S402: ~9L Steel Console Style Case

[robbee]

You also joined the conversation a bit late so the link you provided has already been discussed at length. The reply you quoted I had already mentioned it.

Oops, nevermind me then. Consider it to be fashionably late

 

[DSmithBYU]

Yo grsy!

A few questions:

1- Is it possible that the S401's spacers can be compatible for this case?
2- Like its predecessor, can it still support 3.5" HDDs so long as the GPU's length is measured below 200mm? It's fine if it's just one place that the HDD can be in, considering the usage of the 3-slot GPUs like EVGA's single-fan cards.

Best of luck with the project.

 

[grsychckn]

Hey @grsychckn, the Discord Server link (https://discord.gg/QRnBu8) isn't working, just as @jakemagee said. Are you sure it's not private? What's the name or address so we can search and join?

Have you tried it again? I checked and everything seemed configured correctly. I will try to be on later tonight as I'm still settling back into my routine after the long trip. If you can't join, I can send you an invite if you PM me your username.

 

[grsychckn]

Is 60mm CPU height possible? That will include Zalman CNPS8900, which is one of the best performers at that height.

I'll have to check the final tolerances. Right now, the maximum CPU cooler height would be around 59mm. That also doesn't eliminate the possibility of my making more panel spacers for the S402 like I did with the S401. This would increase the cooler height to around 64mm.

 

[grsychckn]

Yo grsy!

A few questions:

1- Is it possible that the S401's spacers can be compatible for this case?
2- Like its predecessor, can it still support 3.5" HDDs so long as the GPU's length is measured below 200mm? It's fine if it's just one place that the HDD can be in, considering the usage of the 3-slot GPUs like EVGA's single-fan cards.

Best of luck with the project.

1. Not currently. The addition of 60/70 x 25mm fan support meant I had to increase the height of the case (when vertical) by 10mm. The panel mounting points would be very similar to the S401 on the vented side, but on the solid side, the panel mounts have moved for clearance. I was hoping to keep the overall dimensions the same for the purpose of customers reusing their custom panels but I don't think this is possible while still supporting the thicker fans. Even with the fan support removed, the PSU inlet clearance has changed for the 3-slot height and thus had to be relocated so it is not just the new fan support that breaks backwards panel compatibility.
2. I'm trying to include this support, yes. The problem I'm facing now is that the PSU requires a new, larger bracket be made so that I can add an additional 2.5" drive mount underneath it. There is one 3.5" mount right now underneath the GPU, but that would require a single-slot GPU for clearance. I'm still massaging the design so I'll update here when I get to the next plateau of completion.

 

[ermac318]

There are about 4x the amount of options at 60mm than 70mm - 70 seems to be a very odd fan size. The usual suspects (Noctua, etc) don't make 70mm fans, just something to consider. 60mm support probably more important than 70mm.

Another big plus, which has been a boon to Sliger's recent SM560 3-slot case is Bifurcated Riser support. Even if you were on the 2-slot GPU bandwagon, you could use that extra slot for 10GbE, more M.2 cards, etc. So it just opens up more options, which I think is a good thing. You might ask @KSliger but it sounds like the SM560 is a much more popular case than the SM550. If that's true, it sounds like there's a lot of demand for 3-slot GPU cases.

How would you provide AIO support? Would it be the usual 120mm AIO bracket when using a short-length GPU? How would you get it to 240mm long as described at the start?

 

[KSliger]

100% you should keep the 3-slot just for the number of sales you will get.

 

[grsychckn]

There are about 4x the amount of options at 60mm than 70mm - 70 seems to be a very odd fan size. The usual suspects (Noctua, etc) don't make 70mm fans, just something to consider. 60mm support probably more important than 70mm.

Another big plus, which has been a boon to Sliger's recent SM560 3-slot case is Bifurcated Riser support. Even if you were on the 2-slot GPU bandwagon, you could use that extra slot for 10GbE, more M.2 cards, etc. So it just opens up more options, which I think is a good thing. You might ask @KSliger but it sounds like the SM560 is a much more popular case than the SM550. If that's true, it sounds like there's a lot of demand for 3-slot GPU cases.

How would you provide AIO support? Would it be the usual 120mm AIO bracket when using a short-length GPU? How would you get it to 240mm long as described at the start?

The case will support both 60 and 70mm fans. Also, all 3 fan locations (either 60 or 70mm) support up to 25mm depth.

Agreed on the 3-slot popularity. I'm thinking though that soon there will be new, more expensive PCIe 4.X riser/extenders that are going to be more expensive. I'm going to reach out to Li-Heat and see if they have plans already to update their product line. Either way, 4.X won't be mainstream for a while (at least in terms of devices that can effectively use the bandwidth).

The AIO support now is a 120mm AIO mount that would require a short GPU. 240mm radiator support will require an expert to cram in all the gear. Essentially, it would have to sit right behind a single-slot GPU and probably push the hot air onto the rear cover plate. Not ideal, but also not a major stretch for me to provide the mounts. This larger radiator will probably not be supported, but available for those that want to try it.

 

[grsychckn]

100% you should keep the 3-slot just for the number of sales you will get.

I'm definitely planning on keeping the 3-slot support. I might have to slightly alter the dimensions after I get the first prototype in though to make sure there is ample clearance for the rear cover plate and the fan shroud. I don't want anything to accidentally ground out so an extra 1mm here or there is not going to cause me to change the design.

 

[ermac318]

I'm thinking though that soon there will be new, more expensive PCIe 4.X riser/extenders that are going to be more expensive. I'm going to reach out to Li-Heat and see if they have plans already to update their product line. Either way, 4.X won't be mainstream for a while (at least in terms of devices that can effectively use the bandwidth).

I'll mention what I said in Sliger's thread when someone started asking about PCIe 4.0 risers:

What do you need a PCIe 4.0 riser for? I asked this same question and quickly realized it was irrelevant. PCIe bandwidth on GPUs is complete overkill at the moment. You don't need even 8 lanes of PCIe gen 3 to run any single GPU at full speed. All PCIe4 is good for at the moment is IO like SSDs that can actually saturate their links, and the only reason this becomes relevant to SSDs is because the M.2 form factor only has enough traces for 4 lanes, so M.2 slots are doubling in speed. If they were PCIe-attached in 8x or 16x slots, the jump to PCIe4 would be irrelevant.

Also, the PCIe Bifurcation case makes all this even more messy. The Bifurcation risers I believe need two cards running at the same PCIe speed to function properly, so I don't think you can even do one 8x slot at PCIe gen4 and one 8x slot at PCIe gen 3, but who knows since the whole "standard" is poorly documented.
So don't get wrapped around the axles on PCIe4 risers. The only place it seems to matter right now are on M.2 SSDs, and generally speaking these don't need risers.

 

[Thehack]

I'll mention what I said in Sliger's thread when someone started asking about PCIe 4.0 risers:

Is PCIe 4.0 risers actually possible? All the talks of shielding and retimers seems to indicate it is not possible.

 

[ermac318]

Is PCIe 4.0 risers actually possible? All the talks of shielding and retimers seems to indicate it is not possible.

In time, maybe, but I'm not holding my breath. It seems like motherboard manufacturers are having to add retiming chips or go to 10 layer PCBs just to get PCIe 4.0 to the 5th slot on an ATX board. I can't imagine getting that to work on a riser.

That said, I would've thought it was insane to daisy chain PCIe3.0 risers, but Linus managed to go half way around a room with them before hitting issues in FurMark.

 

[grsychckn]

In time, maybe, but I'm not holding my breath. It seems like motherboard manufacturers are having to add retiming chips or go to 10 layer PCBs just to get PCIe 4.0 to the 5th slot on an ATX board. I can't imagine getting that to work on a riser.

That said, I would've thought it was insane to daisy chain PCIe3.0 risers, but Linus managed to go half way around a room with them before hitting issues in FurMark.

I agree BTW on the need for PCIe 4.0 support for now. In an ITX case, the only thing that will even come close to using it is NVMe drives which most users will probably just use the on-board m.2 connectors. I do know that everyone wants some future-proofing so I checked with Li-Heat to see if they are going to release a 4.0 spec product. It seems they have had the spec for sometime and are simply waiting on the X570 boards to be released so they can start testing their products using PCIe 4.0. I think they will have a product available soon but just knowing they are already working on it makes me feel better. I have already followed up with them to check if the mounting points will be the same so users can use the S401/S402 with the newer tech when it is mainstream.

 

[annasoh323]

I agree BTW on the need for PCIe 4.0 support for now. In an ITX case, the only thing that will even come close to using it is NVMe drives which most users will probably just use the on-board m.2 connectors. I do know that everyone wants some future-proofing so I checked with Li-Heat to see if they are going to release a 4.0 spec product. It seems they have had the spec for sometime and are simply waiting on the X570 boards to be released so they can start testing their products using PCIe 4.0. I think they will have a product available soon but just knowing they are already working on it makes me feel better. I have already followed up with them to check if the mounting points will be the same so users can use the S401/S402 with the newer tech when it is mainstream.

It's pretty impressive Li-Heat has already been working on solutions for these. Good on them for doing that! The next best thing after that would be finding out you could use an NVMe drive riser card with bifurcation for a video card. That would be sick.

 

[grsychckn]

It's pretty impressive Li-Heat has already been working on solutions for these. Good on them for doing that! The next best thing after that would be finding out you could use an NVMe drive riser card with bifurcation for a video card. That would be sick.

The question is: would a PCIe 4.0 bifurcation riser be able to split into two 3.0 x 16 connections. If so, this would be the way to go until 4.0 becomes more mainstream.

 

[Piewalker]

Have you tried it again? I checked and everything seemed configured correctly. I will try to be on later tonight as I'm still settling back into my routine after the long trip. If you can't join, I can send you an invite if you PM me your username.

I've tried several times. Each time it shows invite invalid.

 

[ermac318]

The question is: would a PCIe 4.0 bifurcation riser be able to split into two 3.0 x 16 connections. If so, this would be the way to go until 4.0 becomes more mainstream.

This wouldn't be possible with a bifurcation riser, you'd need a PCIe switch board for this to be possible at all. By definition, Bifurcation boards take the existing lanes and split them between multiple cards, so the number of lanes can't change - it has to add up to 16x (if you're bifurcating a 16x slot). If you had a proper PLX switch, then this would work. You see PLX switches on cards like the older dual GPU cards (Titan Z and the like), or newer cards like the QLogic 10GbE + 2xM.2 card.

 

[grsychckn]

This wouldn't be possible with a bifurcation riser, you'd need a PCIe switch board for this to be possible at all. By definition, Bifurcation boards take the existing lanes and split them between multiple cards, so the number of lanes can't change - it has to add up to 16x (if you're bifurcating a 16x slot). If you had a proper PLX switch, then this would work. You see PLX switches on cards like the older dual GPU cards (Titan Z and the like), or newer cards like the QLogic 10GbE + 2xM.2 card.

Gotcha. I wonder if a PLX switch could sit directly on the riser then. But that would surely make it quite expensive.

 

[grsychckn]

Since Discord is down, I'm going to try and quickly post a message here regarding my current front panel USB options for the S402. Some feedback would be great, but for now I'm heavily leaning towards the standard 3.0/3.1 dual type A connectors (option 1).

#1 - Dual 3.0/3.1 type A
Pros
1. Less expensive.
2. Easier to source and less likely to be discontinued.
3. I can get replacements in days rather than weeks.
4. Smaller front panel footprint (takes up less space).

Cons
1. No type C without an adapter.
2. Mounting hardware will be visible from the front face.
3. 20-pin connector.

#2 Dual type A and dual type C PCB
Note: I would remove the PCB from the steel 3.5" drive caddy and then install into the S402. Ignore the steel caddy for my purposes.

Pros
1. 4 front panel USB interfaces (2xType C, 2xType A).
2. Simple type A cable output (doesn't require standard 20-pin connector).
3. SATA power connector for additional devices.
4. PCB mounts would not be visible from the front (cleaner front panel).

Cons
1. Almost 3x the cost of option 1.
2. Larger footprint inside the case.
3. Getting mount points correctly could be very costly to produce multiple prototypes.

 

[loader963]

4 front usb headers, whether a or c, is a bit much for an atx case, much less an itx one isn’t it? Personally I’d just go dual a.