-
Save 15% 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
- Thread starter petricor
- Start date
Holy cow - just looked it up - i'm sure it can also invert gravity
definitely bullet proof...
I'm sorry @petricor, but I'll have to classify your thread for 18+. Way to much porn for our members to handle 
.
.
Would this be of any help to bypass the RPM timer and allow for the use of a slim fan?
hardforum.com
Generate Fan RPM Signal W/O Fan
The title pretty much says it all. I am looking to generate a "fake" RPM signal for a switch that I modded to fit a 120mm fan. Basically, I have removed the 40m 8000RPM fan which was noisy, cut a hole in the top of the case and affixed a 120mm fan. My main issue is that the IOS now thinks there...
hardforum.com
Wow, glad I've found this thread, congrats on the job done so far!!!
Really enjoying all the information you are sharing along with nice pics, especially the electrical part on the PSU!
Everything is indeed so well explained that my questions are not related to this topic...
I am wondering what software you are using for the grayed 3D designs of the first post?
And also, how do you 'free hand' annotate such images?
Really enjoying all the information you are sharing along with nice pics, especially the electrical part on the PSU!
Everything is indeed so well explained that my questions are not related to this topic...
I am wondering what software you are using for the grayed 3D designs of the first post?
And also, how do you 'free hand' annotate such images?
Yeah this looks like it could do the trick. On the list!Would this be of any help to bypass the RPM timer and allow for the use of a slim fan?
Generate Fan RPM Signal W/O Fan
The title pretty much says it all. I am looking to generate a "fake" RPM signal for a switch that I modded to fit a 120mm fan. Basically, I have removed the 40m 8000RPM fan which was noisy, cut a hole in the top of the case and affixed a 120mm fan. My main issue is that the IOS now thinks there...
Glad you like it! In all brevity:Wow, glad I've found this thread, congrats on the job done so far!!!
Really enjoying all the information you are sharing along with nice pics, especially the electrical part on the PSU!
Everything is indeed so well explained that my questions are not related to this topic...
I am wondering what software you are using for the grayed 3D designs of the first post?
And also, how do you 'free hand' annotate such images?
3D - Rhino 6 (“Arctic” shading)
Markups/ Annotations - iPad pro with Autodesk Sketchbook Pro
@petricor Have you considered using the alphacool eisblock solo lt? Integrated dc-lt, ultra low profile and am4 native - only issue I see is the lack of extra inlet/outlets
www.aquatuning.us
Alphacool Eisbaer LT (Solo) - black
The popular Eisbaer LT AiO liquid cooler is now also available in a solo version. This gives you the opportunity to design your own AiO solution according to your wishes. All connections use the G1/4" thread which we have all become to...
www.aquatuning.us
OK that was a somewhat lengthy downtime - tl;dr: The project is alive, and I got to do some shopping in the meanwhile!
First major conclusion I have arrived at in the past six weeks is that I have one irresolvable clash and that the only way around this will be tinkering with the radiator - more below* -, requiring the help of someone who can weld or solder copper fittings into place (and that's not me).
Assuming that this will be difficult but doable, I have decided to move on and equipped myself with some of the more juicy ingredients required to make the build work:
There are obvious bits missing:
As I need to strip my brand new toys of all cooling and decoration in order to photometrically scan them, voiding the warranty in the process, I want to test everything for functionality first. In absence of a CPU this requires a slightly sketchy setup, also giving me an opportunity to entertain my modded PSU with some more load attached:
The legend has it that some x570 boards would boot to bios without a CPU installed, and with a screen connected to the GPU this should give me sufficient reassurance that everything arrived in working condition:
...this one, however, does not appear to be one of them. I get the Mobo's and GPUs RGB firing up, the chipset fan spinning, but that's about it.
To make sure the stuff really works (and because I'm curious to see it in action!), I slot the GPU and RAM into my "old" z370 S4M build - NB that I power everything from the new PSU, as the old dual PSU setup does not deliver the wattage required by the RTX:
So far, so good:
An RTX 2080ti connected, 32 gig of Ram showing- and running benchmarks with no signs of flames or smoke! 10,788 points on Timespy of the cuff
Replacing the 2666 ULP memory with 3600mzh CL18 gets me up to 11,862 - indicating that I might run into a memory speed bottleneck further down the line. But this test here is on an intel platform, and team red's memory controllers are entirely different beasts...
Key outcome here is that most of the stuff I bought is verified to work and can now be happily stripped of it's warranty - and that my PSU does the trick, at least with a 2080ti at stock speed and an i7 8700k overclocked to 103%. Looking at R3000 TDPs I do not expect any more load on my final build.
But boy, that PSU gets loud when pushing it - we are comfortably within jet engine territory here. It spins up immediately when stressing the system indicating that the fan is controlled by load and not temperature. Which is a good sign with perspective to silencing it.
But most amazingly, the PSU is actually cool to the touch after three consecutive Timespy benchmarking passes! That tells me that there is a lot of scope for silencing once I have overcome the RPM feedback loop issue. Also, I have another fan in the post that might deliver the RPM reading required by the controller at significantly lower noise level...
Finally, as someone asked earlier whether board and PSU would fit into the case:
They do!
Next up: Stripping GPU and Mobo down to the bare PCB for a scan to verify clearances and packing hypothesis in 3D - and making some key decisions re cooling loop...
-----
* 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:
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!
First major conclusion I have arrived at in the past six weeks is that I have one irresolvable clash and that the only way around this will be tinkering with the radiator - more below* -, requiring the help of someone who can weld or solder copper fittings into place (and that's not me).
Assuming that this will be difficult but doable, I have decided to move on and equipped myself with some of the more juicy ingredients required to make the build work:
- an ASRock x570 mITX/TB3 (it packs the most complications with the biggest heat sinks, and as I have to come up with my own design for chipset cooling regardless of manufacturer, I might as well get the beefy one with Thunderbolt...)
A look at the board's business end shows the TB3 connector and the DP in port user to re-route the GPUs video signal into the TB controller:
- a 1 TB Sabrent Rocket PCIe 4.0 M.2 SSD, note the absence of a heat sink - that's intentional...
...as the AsRock has only one M.2 slot fitted to the rear of the board. Luckily enough the S4M is one giant aluminum heat sink, and the plan is to cool the SSD directly by creating contact with the case:
This requires shortening the mobo offstands by 3mm, also adding valuable headroom to the case which I'll desperately need to make the CPU-GPU sandwich fit.
A quick visual check confirms that the SSD is the highest part on the rear side, so this should be possible without other clashes, or let alone short circuiting vital parts.
- 32 gig of ultra low profile Kingston 2666 CL19 RAM, picked up during a rather exciting visit to the Shenzhen Huaqiangbei electronics market...
... next to some full height Corsair Vengeance for size comparison:
Using ULP memory is actually Plan B, resolving one of my design clashes: Plan A remains to fit full height RAM; an R3000 CPU clearly wants a higher RAM clock swimming in sync with the Infinity Fabric, but there appear to be severe limitations on ULP memory specs. The 2666 CL19 were the fastest I could find.
If anyone comes across higher frequencies or tighter timings for 16GB ULP sticks: Please let me know!
- and the ASUS RTX 2080ti Turbo...
...chosen for its rather rare line-up of all rear ports in a single line, allowing it to be stripped down to a true single slot package when water cooled - and it's nominal OC headroom for some added fun!
There are obvious bits missing:
- the CPU - not available yet
- and the cooling, which will need to be verified against the stripped components before purchase as I don't have accurate 3D PCB models of motherboard or GPU.
As I need to strip my brand new toys of all cooling and decoration in order to photometrically scan them, voiding the warranty in the process, I want to test everything for functionality first. In absence of a CPU this requires a slightly sketchy setup, also giving me an opportunity to entertain my modded PSU with some more load attached:
The legend has it that some x570 boards would boot to bios without a CPU installed, and with a screen connected to the GPU this should give me sufficient reassurance that everything arrived in working condition:
...this one, however, does not appear to be one of them. I get the Mobo's and GPUs RGB firing up, the chipset fan spinning, but that's about it.
To make sure the stuff really works (and because I'm curious to see it in action!), I slot the GPU and RAM into my "old" z370 S4M build - NB that I power everything from the new PSU, as the old dual PSU setup does not deliver the wattage required by the RTX:
So far, so good:
An RTX 2080ti connected, 32 gig of Ram showing- and running benchmarks with no signs of flames or smoke! 10,788 points on Timespy of the cuff
Replacing the 2666 ULP memory with 3600mzh CL18 gets me up to 11,862 - indicating that I might run into a memory speed bottleneck further down the line. But this test here is on an intel platform, and team red's memory controllers are entirely different beasts...
Key outcome here is that most of the stuff I bought is verified to work and can now be happily stripped of it's warranty - and that my PSU does the trick, at least with a 2080ti at stock speed and an i7 8700k overclocked to 103%. Looking at R3000 TDPs I do not expect any more load on my final build.
But boy, that PSU gets loud when pushing it - we are comfortably within jet engine territory here. It spins up immediately when stressing the system indicating that the fan is controlled by load and not temperature. Which is a good sign with perspective to silencing it.
But most amazingly, the PSU is actually cool to the touch after three consecutive Timespy benchmarking passes! That tells me that there is a lot of scope for silencing once I have overcome the RPM feedback loop issue. Also, I have another fan in the post that might deliver the RPM reading required by the controller at significantly lower noise level...
Finally, as someone asked earlier whether board and PSU would fit into the case:
They do!
Next up: Stripping GPU and Mobo down to the bare PCB for a scan to verify clearances and packing hypothesis in 3D - and making some key decisions re cooling loop...
-----
* 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:
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!
YESSHHH!!
I can sleep now.
Ohhh btw I sent someone on NFC's discord your build log, inspired him to do the same, and he managed to spoof the rpm feedback.
I can sleep now.
Ohhh btw I sent someone on NFC's discord your build log, inspired him to do the same, and he managed to spoof the rpm feedback.
Amazing: Is that solution/ build documented anywhere? And how do I get an invite to the NFC discord?
How did you cut and splice the wires to get the 2 8 pin connectors working? I’m trying the do the same with my flex psu. Thanks!OK that was a somewhat lengthy downtime - tl;dr: The project is alive, and I got to do some shopping in the meanwhile!
First major conclusion I have arrived at in the past six weeks is that I have one irresolvable clash and that the only way around this will be tinkering with the radiator - more below* -, requiring the help of someone who can weld or solder copper fittings into place (and that's not me).
Assuming that this will be difficult but doable, I have decided to move on and equipped myself with some of the more juicy ingredients required to make the build work:
- an ASRock x570 mITX/TB3 (it packs the most complications with the biggest heat sinks, and as I have to come up with my own design for chipset cooling regardless of manufacturer, I might as well get the beefy one with Thunderbolt...)
A look at the board's business end shows the TB3 connector and the DP in port user to re-route the GPUs video signal into the TB controller:
- a 1 TB Sabrent Rocket PCIe 4.0 M.2 SSD, note the absence of a heat sink - that's intentional...
...as the AsRock has only one M.2 slot fitted to the rear of the board. Luckily enough the S4M is one giant aluminum heat sink, and the plan is to cool the SSD directly by creating contact with the case:
This requires shortening the mobo offstands by 3mm, also adding valuable headroom to the case which I'll desperately need to make the CPU-GPU sandwich fit.
A quick visual check confirms that the SSD is the highest part on the rear side, so this should be possible without other clashes, or let alone short circuiting vital parts.
- 32 gig of ultra low profile Kingston 2666 CL19 RAM, picked up during a rather exciting visit to the Shenzhen Huaqiangbei electronics market...
... next to some full height Corsair Vengeance for size comparison:
Using ULP memory is actually Plan B, resolving one of my design clashes: Plan A remains to fit full height RAM; an R3000 CPU clearly wants a higher RAM clock swimming in sync with the Infinity Fabric, but there appear to be severe limitations on ULP memory specs. The 2666 CL19 were the fastest I could find.
If anyone comes across higher frequencies or tighter timings for 16GB ULP sticks: Please let me know!
- and the ASUS RTX 2080ti Turbo...
...chosen for its rather rare line-up of all rear ports in a single line, allowing it to be stripped down to a true single slot package when water cooled - and it's nominal OC headroom for some added fun!
There are obvious bits missing:
- the CPU - not available yet
- and the cooling, which will need to be verified against the stripped components before purchase as I don't have accurate 3D PCB models of motherboard or GPU.
As I need to strip my brand new toys of all cooling and decoration in order to photometrically scan them, voiding the warranty in the process, I want to test everything for functionality first. In absence of a CPU this requires a slightly sketchy setup, also giving me an opportunity to entertain my modded PSU with some more load attached:
The legend has it that some x570 boards would boot to bios without a CPU installed, and with a screen connected to the GPU this should give me sufficient reassurance that everything arrived in working condition:
...this one, however, does not appear to be one of them. I get the Mobo's and GPUs RGB firing up, the chipset fan spinning, but that's about it.
To make sure the stuff really works (and because I'm curious to see it in action!), I slot the GPU and RAM into my "old" z370 S4M build - NB that I power everything from the new PSU, as the old dual PSU setup does not deliver the wattage required by the RTX:
So far, so good:
An RTX 2080ti connected, 32 gig of Ram showing- and running benchmarks with no signs of flames or smoke! 10,788 points on Timespy of the cuff
Replacing the 2666 ULP memory with 3600mzh CL18 gets me up to 11,862 - indicating that I might run into a memory speed bottleneck further down the line. But this test here is on an intel platform, and team red's memory controllers are entirely different beasts...
Key outcome here is that most of the stuff I bought is verified to work and can now be happily stripped of it's warranty - and that my PSU does the trick, at least with a 2080ti at stock speed and an i7 8700k overclocked to 103%. Looking at R3000 TDPs I do not expect any more load on my final build.
But boy, that PSU gets loud when pushing it - we are comfortably within jet engine territory here. It spins up immediately when stressing the system indicating that the fan is controlled by load and not temperature. Which is a good sign with perspective to silencing it.
But most amazingly, the PSU is actually cool to the touch after three consecutive Timespy benchmarking passes! That tells me that there is a lot of scope for silencing once I have overcome the RPM feedback loop issue. Also, I have another fan in the post that might deliver the RPM reading required by the controller at significantly lower noise level...
Finally, as someone asked earlier whether board and PSU would fit into the case:
They do!
Next up: Stripping GPU and Mobo down to the bare PCB for a scan to verify clearances and packing hypothesis in 3D - and making some key decisions re cooling loop...
-----
* 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:
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!
They are soldered together and end in 14AWG wires connected to the Supermicro's single 50A 12V output. You might exceed permissible loads if you try to squeeze the 220W (~20A) a non-oc'd 2080ti pulls through a single ATX power connection... a single ATX pin is rated for 5A max
OK that was a somewhat lengthy downtime - tl;dr: The project is alive, and I got to do some shopping in the meanwhile!
First major conclusion I have arrived at in the past six weeks is that I have one irresolvable clash and that the only way around this will be tinkering with the radiator - more below* -, requiring the help of someone who can weld or solder copper fittings into place (and that's not me).
Assuming that this will be difficult but doable, I have decided to move on and equipped myself with some of the more juicy ingredients required to make the build work:
- an ASRock x570 mITX/TB3 (it packs the most complications with the biggest heat sinks, and as I have to come up with my own design for chipset cooling regardless of manufacturer, I might as well get the beefy one with Thunderbolt...)
A look at the board's business end shows the TB3 connector and the DP in port user to re-route the GPUs video signal into the TB controller:
- a 1 TB Sabrent Rocket PCIe 4.0 M.2 SSD, note the absence of a heat sink - that's intentional...
...as the AsRock has only one M.2 slot fitted to the rear of the board. Luckily enough the S4M is one giant aluminum heat sink, and the plan is to cool the SSD directly by creating contact with the case:
This requires shortening the mobo offstands by 3mm, also adding valuable headroom to the case which I'll desperately need to make the CPU-GPU sandwich fit.
A quick visual check confirms that the SSD is the highest part on the rear side, so this should be possible without other clashes, or let alone short circuiting vital parts.
- 32 gig of ultra low profile Kingston 2666 CL19 RAM, picked up during a rather exciting visit to the Shenzhen Huaqiangbei electronics market...
... next to some full height Corsair Vengeance for size comparison:
Using ULP memory is actually Plan B, resolving one of my design clashes: Plan A remains to fit full height RAM; an R3000 CPU clearly wants a higher RAM clock swimming in sync with the Infinity Fabric, but there appear to be severe limitations on ULP memory specs. The 2666 CL19 were the fastest I could find.
If anyone comes across higher frequencies or tighter timings for 16GB ULP sticks: Please let me know!
- and the ASUS RTX 2080ti Turbo...
...chosen for its rather rare line-up of all rear ports in a single line, allowing it to be stripped down to a true single slot package when water cooled - and it's nominal OC headroom for some added fun!
There are obvious bits missing:
- the CPU - not available yet
- and the cooling, which will need to be verified against the stripped components before purchase as I don't have accurate 3D PCB models of motherboard or GPU.
As I need to strip my brand new toys of all cooling and decoration in order to photometrically scan them, voiding the warranty in the process, I want to test everything for functionality first. In absence of a CPU this requires a slightly sketchy setup, also giving me an opportunity to entertain my modded PSU with some more load attached:
The legend has it that some x570 boards would boot to bios without a CPU installed, and with a screen connected to the GPU this should give me sufficient reassurance that everything arrived in working condition:
...this one, however, does not appear to be one of them. I get the Mobo's and GPUs RGB firing up, the chipset fan spinning, but that's about it.
To make sure the stuff really works (and because I'm curious to see it in action!), I slot the GPU and RAM into my "old" z370 S4M build - NB that I power everything from the new PSU, as the old dual PSU setup does not deliver the wattage required by the RTX:
So far, so good:
An RTX 2080ti connected, 32 gig of Ram showing- and running benchmarks with no signs of flames or smoke! 10,788 points on Timespy of the cuff
Replacing the 2666 ULP memory with 3600mzh CL18 gets me up to 11,862 - indicating that I might run into a memory speed bottleneck further down the line. But this test here is on an intel platform, and team red's memory controllers are entirely different beasts...
Key outcome here is that most of the stuff I bought is verified to work and can now be happily stripped of it's warranty - and that my PSU does the trick, at least with a 2080ti at stock speed and an i7 8700k overclocked to 103%. Looking at R3000 TDPs I do not expect any more load on my final build.
But boy, that PSU gets loud when pushing it - we are comfortably within jet engine territory here. It spins up immediately when stressing the system indicating that the fan is controlled by load and not temperature. Which is a good sign with perspective to silencing it.
But most amazingly, the PSU is actually cool to the touch after three consecutive Timespy benchmarking passes! That tells me that there is a lot of scope for silencing once I have overcome the RPM feedback loop issue. Also, I have another fan in the post that might deliver the RPM reading required by the controller at significantly lower noise level...
Finally, as someone asked earlier whether board and PSU would fit into the case:
They do!
Next up: Stripping GPU and Mobo down to the bare PCB for a scan to verify clearances and packing hypothesis in 3D - and making some key decisions re cooling loop...
-----
* 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:
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!
Epic update!
As I understand it, you need a custom fitting?
I have access to a cnc mill that belongs to a friend (still haven't got mine working). I could probably do it.
Corn crackin crazy son of a gun...
I need a cold shower after viewing this thread.
???
I need a cold shower after viewing this thread.
???
Epic, can't wait to help!
Yeah, he's doing some insane work!
I'm glad you saw it, josh!
Maybe this radiator will work: Alphacool NexXxoS ST30 https://www.alphacool.com/detail/index/sArticle/21146
or this
Hardware Labs Black Ice Nemesis 140GTS XFLOW
or this
Hardware Labs Black Ice Nemesis 140GTS XFLOW
Similar threads
