• Save 20% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!

S4MAX: Brickless S4M w/ 3090 FE and R9 5950x - 800W, 5l, water cooled

BaK

King of Cable Management
Bronze Supporter
May 17, 2016
967
958
* Speaking of cooling: This gets me back to the clash I cannot resolve and have mentioned above.
My design still has a radiator inlet fitting clashing with the GPU water block (my other clash got sorted by getting the ultra low profile ram)...

...and that means my copper radiator will need an additional inlet port here:
In addition to the crossflow rads @dondan just proposed, there are some rads with extra ports.
Their thickness might be an issue though...

HWLabs SR2 140 (Width: 153mm Height: 173mm Thickness: 60mm)



Alphacool NexXxoS XT45 (Width: 144mm Height: 189mm Thickness: 45mm)



And, as this is requiring rather profound metalwork skills and special equipment (I don't possess either - and leaks aren't an option!), what that really means is that I need to find someone who can do this, ideally within the UK.

Any volunteers coming forward/ recommendations/ introductions are very much appreciated!
@aquelito from France had added two ports to one of his rad by soldering two fittings to it:



Finally, as someone asked earlier whether board and PSU would fit into the case:
They do!
I guess you don't plan on using any SATA connectors as the right angles ones of your ASRock x570 mITX/TB3 motherboard come into the PSU.
 
  • Like
Reactions: petricor

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
In addition to the crossflow rads @dondan just proposed, there are some rads with extra ports.
Their thickness might be an issue though...

HWLabs SR2 140 (Width: 153mm Height: 173mm Thickness: 60mm)



Alphacool NexXxoS XT45 (Width: 144mm Height: 189mm Thickness: 45mm)




@aquelito from France had added two ports to one of his rad by soldering two fittings to it:




I guess you don't plan on using any SATA connectors as the right angles ones of your ASRock x570 mITX/TB3 motherboard come into the PSU.
The extra ports on the Hardwarelabs radiator look like what I need - unfortunately they appear not to be an option on their 30mm thick rads - what @aquelito has soldered together looks pretty amazing, I might have to go down that route (used one of this load switches in my previous build!)... and yes, SATA is definitely ruled out!
 

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
Time for a hardware teardown!

First in line is the board with some pretty chunky metal attached...



...that should all come off fairly easily removing the screws on the underside.
The metal cross bar at the bottom belogs to the rear i/o shield, that is not only height adjustable (this will come in handy when shortening the motherboard standoffs), but also separate from the massive decoration block over the i/o ports:

With everything removed, the board appears much more SFF-able - some work will need to go into the cooling of chipset and VRMs, the latter probably getting custom heat sinks fitted I plan to get CNC’d, the former in all likelihood going to join the water loop - that said, if anyone has successfully passively air cooled an X570, let me know, the single rad already has a lot of load at hand...

Clearances under the shield are more or less within expectations, a predictable pinch point will be the audio module standing 36mm tall above the mobo PCB. Above this I‘ll still have to squeeze the GPU with water block - the added 3mm I can gain by shortening the mobo standoffs will make a key difference here:



The VRMs project 8mm vertically so in theory they could be covered by a large server CPU water block (along with the CPU) - I‘m currently between the beefy EK Annihilator Pro Square ILM (show in the visualisations) with a large contact surface that would allow for covering additional components, and the much smaller original EK Annihilator designed for LGA1151 CPUs. With limited radiator capacity in mind I‘m currently swinging towards convection cooling for the VRMs.



A closer look at the incredibly solid and heavy i/o shield shows that it’s not really designed to do any cooling whatsoever but more of a cosmetic item. It basically works as an adapter taking heat from the VRMs to the heatpipe connected to the chipset cooler - everything above the heat pipe may perhaps be useful to stop bullets, but not for the purpose of this build.

The GPU however, from all that we know, is quite reliant on its beefy cooler dressing to not turn into a rather expensive smoke generator. Despite its massive volume, the ASUS Turbo it is one of the smallest naturally aspirated RTX 2080ti around- and its about to become smaller:


Removing the connector plate shows the single line layout of the connectors, a rather rare configuration sacrificing one Display Port when compared to the reference design in favour of a low connector profile making it an excellent candidate for a low volume solution...



And indeed, with the cooler gone, we all of a sudden look at something significantly more palatable with perspective to stuffing it into a 5l case!



Interestingly though, the actual fin stack doing the work is only consuming 50% of the cooler’s volume - there is quite a void aroud the radial fan. The fin stack alone almost looks like something that could live in a small case - an airflow of sorts provided...


Comparing the volume of the parts that actually do the work to shielding, cooling and decoration tells an interesting story about upcoming sff callenges- and probably also explains why more than a year into 2080tis there is still no mini version in sight...

With everything stripped, its time for a first mock up of the GPU’s position in the case:


Looks like planned - The fan is propped up by 30mm with a space reservation for the radiator underneath (see last post - the radiator remains a challenge), with the space between mobo and GPU reserved for GPU/CPU blocks and cooling loop.

Up next: Cooling cooling cooling and some more design work
 

paulesko

Master of Cramming
Jul 31, 2019
415
322
I love What you Are doing. Really.

From my experience with x570 I don't think it is that difficult to cool. In my pc I have an ncase m1 the fan in the motherboard is always off and the two fans on the side blows to the mb at 650 - 900 rpm. The chipset normally doesn't go over 62-65 degrees with its fan unplugged and it doesn't go over 60 when I plug the chipset fan. Given the fact that the chipset heat sink on my aorus 570 is a piece of crap I think that if you use some proper copper heat sink you will be able to cool it passively. You can also use those 40mm 10mm thick noctua fans on it also over a smaller chipset heat sink.

Same for the vrm, they are so overized and are so efficient that they will not get very hot. Those are meant to go as high as 105 degrees and I don't see mine over 47 with the configuration decribes previously. Of course you will have worse ventilation than me, but I would be optimistic if I were you; I trust more the vrm than the chipset to be able to cool easily, I would just put some of those alphacool copper vrm heat sink, and call it a day. The chipset... Maybe a small hetsink with that small noctua could do it.

AlphaCool 17426 GPU RAM Copper Heatsinks 10x10mm - 10pcs Refrigeración por Aire Disipadores Pasivos
These for the vrm could work although maybe they are too tall? They have plenty of surface to cool though... Maybe if you cut them when necessary.. Don't know.
 
  • Like
Reactions: petricor

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
I love What you Are doing. Really.

From my experience with x570 I don't think it is that difficult to cool. In my pc I have an ncase m1 the fan in the motherboard is always off and the two fans on the side blows to the mb at 650 - 900 rpm. The chipset normally doesn't go over 62-65 degrees with its fan unplugged and it doesn't go over 60 when I plug the chipset fan. Given the fact that the chipset heat sink on my aorus 570 is a piece of crap I think that if you use some proper copper heat sink you will be able to cool it passively. You can also use those 40mm 10mm thick noctua fans on it also over a smaller chipset heat sink.

Same for the vrm, they are so overized and are so efficient that they will not get very hot. Those are meant to go as high as 105 degrees and I don't see mine over 47 with the configuration decribes previously. Of course you will have worse ventilation than me, but I would be optimistic if I were you; I trust more the vrm than the chipset to be able to cool easily, I would just put some of those alphacool copper vrm heat sink, and call it a day. The chipset... Maybe a small hetsink with that small noctua could do it.

AlphaCool 17426 GPU RAM Copper Heatsinks 10x10mm - 10pcs Refrigeración por Aire Disipadores Pasivos
These for the vrm could work although maybe they are too tall? They have plenty of surface to cool though... Maybe if you cut them when necessary.. Don't know.
Cheers- will give that a try!
 

BaK

King of Cable Management
Bronze Supporter
May 17, 2016
967
958
Some thoughts about the rad issue, relevant or not you to tell...

- 140mm rad with ports on the opposite side (PSU side)
- 140mm rad with ports on the opposite side (PSU side) / 120mm slim fan to allow routes for the tubing above the rad if needed


I know you are going to ask a lot of cooling to the rad already, but maybe a 120mm radiator will not loose so much on heat dissipation vs a 140mm one. Lots of people are saying the fan is more important...

- 120mm rad to allow enough offset of the blocked port
- 120mm XSPC thin rad (20mm) to gain 10mm if that helps with the port
- 120mm XSPC thin rad / Noctua fan NF-A12x25 for better airflow/pressure (if there is room for the extra 2mm, 20+25mm vs 30+13mm)
- 120mm rad with ports on the opposite side (PSU side)
- 120mm rad with ports on the opposite side to the motherboard, if enough room
 
  • Like
Reactions: petricor

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
So, finally I got my hands onto this:


...an EK-FC water block for the 2080ti, a key building block required to check out my hypothesis on squeezing the GPU into a case it clearly wasn't made for.


Inside the box we find a bag of screws and tools, some thermal pads and paste....


...and below the actual water block, being of pretty impressive weight:



Looking at all I have received there is an obvious thing missing: A manual! Should be pretty self explanatory though, still a bit odd given the rather hefty price tag...
On the upside, the block is gorgeous - the nickel plating is so even and shiny, it's almost impossible to photograph!

Time to marry board and block - lucky enought there is only one type of screws going into pretty obvious places, so this should be straightforward even without instructions. As you can see, the acrylic top is much larger than the actual water block, so there is quite some potential for freeing up space in case of clashes by trimming back the acrylic - might come in handy later.


And indeed, the screws pretty much only have one place to go, so very quickly I get to this neat sandwich here, leaving thermal paste and pads out for the moment as this is just to confirm dimensions:


Measuring it from screw to front edge I get to 17.2mm, within what I expected, however just about - will get tight...


Time to see how it fits into the case:


Stacking things on top confirms initial assumptions on dimensions...



When sliding the card in place, however, it becomes apparent that it rests on top of the Wifi- and audio modules, both projecting 36mm off the mother board - that will be helpful to carry the weight...




...but in the current alignment of the motherboard it leaves it to project out of the case.


Taking a reading on how much, I get to 2.8mm - luckily that's within tolerance: As established earlier, I plan on shortening the motherboard standoffs by 3mm to establish contact between the m.2 drive on the rear of the motherboard and the case's bottom panel for cooling, so i'm left with a generous .2 mm of hypothetical air gap between GPU and upper enclosure!
Thats plenty ;)

Plan B would have been ditching the waterblock's RGB (not a big fan anyway) and milling pockets for the Wifi- and audio module into the acrylic to gain a few mm - however, it looks like this would not be necessary. I will, however, have to insulate the inside of the case with a thin film of thermal tape to ensure I'm not shorting the GPU as contact between case and card will be certain.
A nice side effect will be that I can activate the case's upper panel for additional cooling of the GPU's PCB.


Now, placing the card over the case in it's desired position looks like I might just be able to get away with using full height RAM whilst using the original tubing adapter block coming with the water block (that's the black acrylick block with the two G1/4 plugs at the bottom of the card in the image above).

After slotting-in a corsair stick instead of the ULP RAM and attaching a riser as space reservation for the PCIe connector however, the waterblock adapter clashes with the RAM: Looks like I'll have to stick to ULP memory until I have built a smaller adapter.


Good news is that I should be able to fit low profile elbow adapters under the tubing adapter coming with the water block as there is ample space underneath with hardly anything projecting from the board. This means that I can start to use the GPU it whilst waiting for custom machined parts to ship - usually it takes 3 weeks for anything to arrive from China.
That said, delivery time for the CPU might even exceed this - I hope I won't get to regretting my platform choice and that AMD cranks up its output anytime soon!


Checking dimensions of the tubing adapter coming with the EK FC 2080ti, it appears to be identical to the ones used for their 1080ti series. As I have a precise CAD model of the 1080 adapter, it should be pretty straightforward to design a smaller version of it, swapping 1/4" fittings for 1/8" and allowing to save precious space that would enable the use of full-height memory.




Next up: Some revised cad models on basis of the real parts - and as I am under no pressure in absence of a CPU, I have some time for thorough planning...
 
Last edited:

reallamer

Chassis Packer
Jan 17, 2019
15
11
I dont understand how it will be working but i think this is really cool and amazing project. Thanks for your work!
 

nathanramos

Cable Smoosher
Jun 26, 2018
10
7
After a little setback (more below) and little time at hand, I’m back with the next Episode of my rather ambitious project: Proof-of-Concept for the power solution.

With the PSU squeezed to measure, it’s now time to verify my hypothesis on powering the build:
Idea is to use a 160w pico PSU to drive the 24pin ATX connector of the main board, but to have the heavy loads, including the GPU and CPU power feeds, driven directly from the PSU’s 600W 12v rail to exploit its capacity without the Pico-plug-in-board becoming a bottle neck.
Also, I’d want to avoid any load switch in the middle - the PSU is platinum rated and has a standby mode, and assuming that the PS_ON is designed to ATX specs, it should be possible to control it in parallel to the pico PSU, using the main board’s PS_ON signal.
This requires, however, to hijack the +5V standby power coming from the pico, and to feed the board directly from the PSU’s standby power feed instead.


As this is somewhat uncharted territory, I have equipped myself with an ATX PSU tester and ATX cable that I can place between pico PSU and board, allowing me to splice in the standby feed and to hijack the PS_ON signal from the board - and to test whether it all adds up to an ATX compliant output, before risking frying a precious board and CPU.


Plugging in the pico for an initial test with the PS_ON connected to ground directly at the PSU, I get a compliant reading- however, just for about 3 seconds, then the PSU switches off the 12v feed!


As it’s impossible to distinguish whether it’s just going to standby or reporting a fault without reading out the PMBus, I’m back to experimentation.
Something clearly doesn’t work as expected.
Either I have indeed missed a second pin on the PSU (PSKill?), which would surprise me as I tested it left, right and centre, or it is somehow related to the Noctua replacement fan I use.
After some quite time consuming testing it indeed turns out to be the Noctua apparently not spinning fast enough, and triggering a fault alert and PSU shutdown.


Searching the web, I have found similar cases reported by users replacing fans on Supermicro boards with slower spinning ones triggering fault alarms.
As there is now way to override this behaviour of the PSU, it appears that for the time being I’m stuck with the original fan, coming with notable noise and most critically: too much length.


Looking at the fan it appears that, similar to the Noctua, I can recess the fittings a bit, and taking a slightly more aggressive approach I aim to gain 4mm which should just about fit - I still have about 6mm left between the psu and radiator to play with.



After cutting back the fittings for the screws at the bottom, the fan looks like this...


...and fits nicely in-front of the standoff for the PCB.



On the top side, where I do not require screws to go in, I simply cut in a 1.5mm slot to interlock with the flanges in the case.


Fits!


Its all a bit tighter now, but it still looks like should be able to just about fit in a 140mm radiator next to it.


To not waste space on projecting screws, I cut sunk holes in to the fan...



...allowing for a flush fit of the screws.


With a new wire soldered to the fan and the PSU re-assembled i get to a rather tight package...



...that finally behaves as desired and runs continuously when switched on! Unfortunaltely this also comes along with a well-audible fan hum - not terribly annoying but miles off what you'd want to have when having been spoiled by noctuas in the past.


And whilst at it: here is the confirmed-to-work pin-out for the Supermicro PWS-606-p i'm using:



...and this image shows the rear side of the connector with the +12V and PS_ON wires soldered on. Forgot to post this earlier...



With fit resolved, there is still the noise issue which I will have to deal with at another point in time- initial research shows a very limited choice of low noise 40mm pwm fans, and it would be trial-and-error as I have no specification for the minimum rpm the Supermicro controller requires.

So, next step is finally verifying my power hypothesis:
Having tested and confirmed the connectivity between the ground input of the pico and the ground output on the atx side in “off” state, it should suffice to route a single wire from the PSU’s +5v Standby output to the matching pin on the boards atx Input to get the board into standby mode.

Using the ATX extension cable, I cut the +5v sb lead (violet), and hook the board side up to the PSU.


Second mod is to connect the PS_ON wire (dark green) between board and pico to the PS_ON Pin of the PSU: If the board successfully powers to standby using the PSUs standby output, it should be able to pull both PS_ON signals on pico and PSU low, 1) powering up the PSU, now feeding 12v to the pico, and 2) directly turning on the pico to feed the board.

I also connect the auxiliary 12v cpu feed Input of the tester directly to the PSU’s 12v rail to test whether it all stacks up.


And- it works! Both PSU and Pico come to life, with a compliant reading on the tester.

That gives me the confidence required for the ultimate test: Using my i7 8700k S4M as a testbed and driving it off my new power unit.


I disconnect both PSUs of the S4M and remove the HDPlex direct plug...


...replacing it with my pico bypass cable, and connecting the auxiliary 12v CPU connector to the supermicro’s 12v output.


With this setup, the board AND the PSU should go to standby upon connecting the PSU to mains...


Success...

...and now power on both PSU and board when pressing the button:



And it spins!


Looks like I now have a reasonably compact, working and a load switch free solution for my build, using the Supermicro PSUs standby capability, and allowing me to feed up to 600 watts directly to CPU and GPU from a single source. Major box ticked.

Next: Modding the pico to do away with the ATX extension and feeding it off four wires, directly from the PSU - and finding a better fan to replace that screamer...
I also had this issue with my delta flex psu. When I soldered a 40x20mm PWM Noctua the psu would start for about 5 seconds then shut down. I was reading online about other flex atx psu’s and saw that some had their polarity reversed. So I went ahead and soldered the Noctuas PWM wire to the psus sense wire and soldered the Noctuas rpm speed to the psus control wire and it worked when it booted up, no shut down! I’ve been using it for about a month now with no issues and I can hear the fan speed up like it’s suppose to when stress testing.
 

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
After a bit of research on fittings, and, most importantly, Alphacool coming up with their Eisstation 40 DC-LT, a compact reservoir-pump-combo with SFF written all over it, I have revised my cooling layout and believe to be looking at something workable now:


The Eisstation 40 is installed pump-down to use the reservoir for filling and air bleeding. Together with the spare front port on the radiator, I now have a"proper" loop with reservoir, fill port and drain port.
Aligning it to the bottom of the case, I have clear access to all connectors of the graphics card which has been challenging in my previous configuration.


Looking under the GPU, another change becomes apparent: I have swapped out the EK Annihilator Pro for an original EK Annihilator in LGA 1152 form factor (I will require fastening screws shorter than those coming with it). Reason for the change is that my original plan to connect the chipset and VRMs to the large cold plate of the Annihilator Pro using milled copper parts will add to the already significant thermal pressure onto the 140mm rad - with my recent spec upgrade to a 3950x this appeared to be a bad idea, so I'll separately cool chipset and VRMs (probably heat pipes connecting them to the case...). This means I can save significant volume by choosing a smaller block. Sourcing isn't easy as they are out of production since 2015 but I have a good lead on one...


Further I have skimmed through the inventory of more or less every fitting manufacturer to come up with a spatially permissible arrangement that can actually connect, using a mixture of Koolance, Alphacool and Bitspower fittings.

The soft tubing system is based-upon FESTO industrial pneumatic fittings and tubes, offering a wide range of options with G1/8 threads and 10mm OD tubing. The NCQH fitting range selected for this build has a fluid and high pressure rating and should be well able to cope with warm water (I guess we'll see!).

Only remaining clash is between GPU Waterblock Top and a 90 degree fitting. I have for now (and rather optimistically) assumed that I can screw in a non-rotary (and hence very low profile) 90 degree fitting straight into the radiator and get it to point in the right direction - should this work out, I should be able to mill a notch into the acrylic top of the water block to make things fit. In this scenario, the clash is now outside of the o-ring area and should be resolvable without creating leaks.
This means I might get away without modding the radiator which would take a major concern and source of leaks out of the equation.

 
Last edited:

riba2233

Shrink Ray Wielder
SFF Time
Jan 2, 2019
1,741
2,286
www.sfftime.com
Despite the progressive 7nm chipsets and the heat dissipating lid soldered to the crystals, the Ryzen 9 3950X needs more efficient cooling than the older AMD Wraith PRISM cooler can offer. Therefore, AMD decided to shift the issue of selecting a cooling system to the user's shoulders, giving only a vague recommendation that the Ryzen 9 3950X requires liquid cooling with a radiator size of 280 mm or more. In fairness, it should be noted that AMD itself for its marketing materials tested the Ryzen 9 3950X with the Noctua NH-D15S air cooler

And this is only a processor without a video card, your system will gain 90 degrees in 3 minutes in any game. Believe me, I am a professional who has assembled more than one system in a super compact package. Your option with this lineup is 2080 and 9900t. I wanted to save your money and time

3950x has same power as 3900x because of better bin, and you can enable eco mode with minimal performance loss (configuring it to 65w tdp.). Also cpu and gpu wont be utilized 100% at the same time. 2080ti can also be undervolted and slightly underclocked without much performance loss.


BTW petricor you are maddd!!! This is insane, congratulations :) ?
 

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
I also had this issue with my delta flex psu. When I soldered a 40x20mm PWM Noctua the psu would start for about 5 seconds then shut down. I was reading online about other flex atx psu’s and saw that some had their polarity reversed. So I went ahead and soldered the Noctuas PWM wire to the psus sense wire and soldered the Noctuas rpm speed to the psus control wire and it worked when it booted up, no shut down! I’ve been using it for about a month now with no issues and I can hear the fan speed up like it’s suppose to when stress testing.

no luck, unfortunately...


Also tried a 40x20mm Delta fan I have sourced as an alternative, promising higher RPMs at less noise than the original jet engine - to no avail, the low load rpm reading appears too low for the PSU to stay on...

By sheer co-incidence I then came across a vialble solution for the Noctua fan: Disconnecting the PWM signal has the fan spinning up to full speed (5000 rpm - still barely audible), and that reading seems to satisfy the PSU controller!
Also, indeed the polarity is somewhat reversed - the blue wire from the Supermicro PSU's connector needs to be hooked up to the Noctua's green tacho wire, the Supermicro's yellow wire is the PWM signal (disconnected below to get the fan spinning up).

No more jet!
I still have a TP354 signal generator in the post that i'll attempt to use for spoofing the tacho signal, it may even allow me to have the already pretty silent Noctua fan at non-audible low RPMs when not under load.
 
Last edited:

nathanramos

Cable Smoosher
Jun 26, 2018
10
7

no luck, unfortunately...


Also tried a 40x20mm Delta fan I have sourced as an alternative, promising higher RPMs at less noise than the original jet engine - to no avail, the low load rpm reading appears to low for the PSU to stay on...

By sheer co-incidence I then came across a vialble solution for the Noctua fan: Disconnecting the PWM signal has the fan spinning up to full speed (5000 rpm - still barely audible), and that reading seems to satisfy the PSU controller!
Also, indeed the polarity is somewhat reversed - the blue wire from the Supermicro PSU's connector needs to be hooked up to the Noctua's green tacho wire, the Supermicro's yellow wire is the PWM signal (disconnected below to get the fan spinning up).

No more jet!
I still have a TP354 signal generator in the post that i'll attempt to use for spoofing the tacho signal, it may even allow me to have the already pretty silent Noctua fan at non-audible low RPMs when not under load.
Unfortunate but awesome that you still find a way!
 
  • Like
Reactions: petricor

riba2233

Shrink Ray Wielder
SFF Time
Jan 2, 2019
1,741
2,286
www.sfftime.com
Can I ask a stupid question, why don't you use alphacool pump-block combo, so save even more space?

 
  • Like
Reactions: petricor

petricor

Airflow Optimizer
Original poster
Bronze Supporter
May 12, 2018
346
1,845
Valid question, and I have indeed been looking at this one- too tall though (~25mm if I remember correctly) and clashing with the GPU above
Key feature of the server blocks I’m looking at are G1/8 fittings (as compared to the G1/4 used in the Eisbär) allowing for an ultra low profile