Production SENTRY 2.0: Evolution of console-sized gaming PC case

[Idle2824]

The best my phone can do really.

The 1st one is the drilled hole, the 2nd is the current stripped screw.

Ouch. Stripping a security torx screw is quite an achievement, honestly. The cam-out torque on these screws should be far higher than what you should be tightening them to on the Sentry. Maybe check the head of the screwdriver you are using, it could be chipped or broken in some way that's causing it to slip more easily than it should.

Otherwise it's probably just that you've overtightened them or crossthreaded them when you last reassembled it. If they were fine when you got them, it's probably not a bad tap during manufacturing. Depending on how much material is left around the newly drilled hole, you might be able to tap it yourself and fit a bigger screw to fix it. For the one you haven't drilled out yet, I would keep trying to unscrew it as long as your screwdriver is not camming out. Sometimes a tap on the back of the screwdriver with a hammer can help, but that risks damaging your tools and/or the case, so it may not be good advice...

 

[NeroFX]

Ouch. Stripping a security torx screw is quite an achievement, honestly. The cam-out torque on these screws should be far higher than what you should be tightening them to on the Sentry. Maybe check the head of the screwdriver you are using, it could be chipped or broken in some way that's causing it to slip more easily than it should.

Otherwise it's probably just that you've overtightened them or crossthreaded them when you last reassembled it. If they were fine when you got them, it's probably not a bad tap during manufacturing. Depending on how much material is left around the newly drilled hole, you might be able to tap it yourself and fit a bigger screw to fix it. For the one you haven't drilled out yet, I would keep trying to unscrew it as long as your screwdriver is not camming out. Sometimes a tap on the back of the screwdriver with a hammer can help, but that risks damaging your tools and/or the case, so it may not be good advice...

Thanks for the advice. I've tried the screw driver and hammer trick, I also have other Torx drivers of various sizes, it won't budge.

Going to look into a screw removal kit.

My temps are amazing with the black ridge now. I removed the chipset fan for access. It hasn't hit 80c, I don't think it will either. My 3060 ti also stays at 70c, so all is good, other than the screw issue.

I'm also using liquid metal which I think is a lot of the reason the temps are good.

My CPU hasn't hit 80c, not the chipset

 

[pulp_party]

It fits:
EVGA 3080 xc3 ultra (only model on stock).
No testing so far.
Paired with a i5 8400, have to see if the bottleneck is bad of its ok playing 4k on the lg CX. If im not happy with performance/temps i will transfer to a bigger case. An keeping the 1070 on the sentry

As an update, im currently playing Cyberpunk 2077, 4k, rtx ultra, dlss performance. Im getting 60 fps with 40 fps downs on some moments.
The temps on the GPU are peaking at 83 C. I did a quick undervolt.
I was thinking on 3d print the fan duct for the cpu and installing 2 tiny noctua fans on the top of gpu side, where the hot air accumulates.

The tiny fans has already been discused i think, can improve thermals or are not worth it?

 

[joker.mn]

@pulp_party - can you share the RPMs & Temps of CPU/GPU to get a feeling for noise?

 

[timmeh]

Long time lurker here. I've loved my Sentry 1.1 (white ) for some years now but until recently have mostly been plug and play. Based on some of the recommendations here and elsewhere, I settled on a 5600x for it's price, thermals, and performance for games and productivity. I was lucky enough to snag one recently and was able to improve performance and go from ~84C to ~69C with a simple undervolt in Ryzen master, ~64C by swapping out my L9i for the L9a, and finally a cool 60C by adding the J-Hacks 92mm fan shroud (10mm height) under full load in Cinebench r23.

Satisfied with the CPU, and with stocks we'll start to see picking up, my attention turns to the GPU. The more traditional EVGA XC3 series seems to be most recommended for the Sentry, but those 3xxx FE cards are just so dang purty... And with airflow to the back, it seems in theory we could see some very nice results. I've seen @AlphaDave and some others mention modifying the panel to accommodate the new design. @ZombiPL rightly seems to be taking a 'wait and see' approach to how well they are received and whether we see a long term movement to this style.

I am wondering whether anyone has tested this at all yet, specifically with the 3080 FE. I don't see myself taking a drill to my case, but am considering having something custom made (because fun), perhaps with a removable air pocket insert that would play nice with different style designs in the future, and printing shrouds for both fans. Would love feedback on this if anyone has, cheers!

 

[phinix]

I'm wondering how the RTX3080 EVGA XC3 or those newly introduced blower versions from Asus and Gigabyte will work in Sentry.

 

[SaperPL]

I'm wondering how the RTX3080 EVGA XC3 or those newly introduced blower versions from Asus and Gigabyte will work in Sentry.

The XC3 should fit, but I couldn't get my hands on it. Up until the news that there's a new card coming, we were ready to get one right away for testing once the notification from the EVGA store came, but it never happend.

In other news, HardwareCanucks made a video about the risers:


Watch till the end. Note riser vendors going back on their 4.0 compliance statements and recommending to switch back to 3.0

Also check out this tweet:


And maybe like/share/comment if you want to encourage HardwareCanucks to take a look at the issue of single pci-e switch for both x16 slot and nvme m.2 slot on ITX boards.

I've got the B550 from asrock, I'll be looking at it hopefully in the near future, but I've got a lot of stuff piled up that is urgent before the end of the year so I didn't even have time to update the bios.

Sorry for not going through all the comments, I took a glance and I've seen you've been supporting each other which is good. Thanks for that whoever helped.

By the way, since a lot of people were waiting and thinking that we are just sitting and doing nothing and making excuses to not make more cases, here's just one of the projects that the company is making:



If you look closely, you'll find our logo at 0:40. Now you know why Sentry is a side project for us, hobby-like in comparison to stuff that actually is our daily business.

And there are other big projects like that where Zombi and the rest of the team are occupied with, simply because a lot of industrial plants had to scale down production due to covid situation and a lot of them figured that it's a good occasion to have some improvements in the process, machines serviced and upgrades etc.

 

[volopasse]

I'm wondering how the RTX3080 EVGA XC3 or those newly introduced blower versions from Asus and Gigabyte will work in Sentry.

XC3 definitely fits. I have the XC3 Black (=no backplate) 3070 in my Sentry 1.1. 3080 has an identical cooler, and the backplate shouldn't be a problem since it mounts on top off the PCB

 

[volopasse]

Long time lurker here. I've loved my Sentry 1.1 (white ) for some years now but until recently have mostly been plug and play. Based on some of the recommendations here and elsewhere, I settled on a 5600x for it's price, thermals, and performance for games and productivity. I was lucky enough to snag one recently and was able to improve performance and go from ~84C to ~69C with a simple undervolt in Ryzen master, ~64C by swapping out my L9i for the L9a, and finally a cool 60C by adding the J-Hacks 92mm fan shroud (10mm height) under full load in Cinebench r23.

Satisfied with the CPU, and with stocks we'll start to see picking up, my attention turns to the GPU. The more traditional EVGA XC3 series seems to be most recommended for the Sentry, but those 3xxx FE cards are just so dang purty... And with airflow to the back, it seems in theory we could see some very nice results. I've seen @AlphaDave and some others mention modifying the panel to accommodate the new design. @ZombiPL rightly seems to be taking a 'wait and see' approach to how well they are received and whether we see a long term movement to this style.

I am wondering whether anyone has tested this at all yet, specifically with the 3080 FE. I don't see myself taking a drill to my case, but am considering having something custom made (because fun), perhaps with a removable air pocket insert that would play nice with different style designs in the future, and printing shrouds for both fans. Would love feedback on this if anyone has, cheers!

Those are some impressive temps on the CPU! What voltage/PPT are you running in Cinebench? I have the R5 3600 (which i know is less efficient than the 5600x), L9A and a fan shroud (this one and not the J-Hack one), and I had to go Eco-Mode, or manually drop PPT to 50W-ish, and still get temps in the high 70s-low 80s under stress loads

 

[timmeh]

Those are some impressive temps on the CPU! What voltage/PPT are you running in Cinebench? I have the R5 3600 (which i know is less efficient than the 5600x), L9A and a fan shroud (this one and not the J-Hack one), and I had to go Eco-Mode, or manually drop PPT to 50W-ish, and still get temps in the high 70s-low 80s under stress loads

I'm currently running at 1V, 4200MHz. I could possibly go lower, but I was satisfied and stopped at that point. At first I thought my results might be misleading - OptimumTech has a video on this here, which GN has covered as well. But running Cinebench I did indeed see my scores improve, and PPT ends up something like 50-55W at these settings.

I'm new to tweaking, so my methodology for this was pretty basic. I ran the tests stock and monitored the clocks, voltage, and PPT in HWiNFO and Ryzen Master to figure out a baseline. For me, this was something like 1.25V and 4100MHz - 4200MHz. I then set my clock speed to the higher end of that range (4200MHz), lowered peak voltage one or two ticks, and ran the test again, monitoring PPT and temperature. Then repeat Iterations are fast when doing this in Ryzen Master (versus BIOS). I also didn't run the full stress test each time - just a couple of minutes to get a sense of results, with every 2nd or 3rd iteration running the full test to get a sense of stability.

Working with set clock speed and peak voltage this way I found much more straightforward and consistent than playing with PBO and trying to manually set PPT, TDC, and EDC, which for me always felt like fighting the algorithm. PBO2 is another new thing for us Sentry / SFF folks to be testing, however it will be Ryzen 5000 series only :/

 

[phinix]

The XC3 should fit, but I couldn't get my hands on it. Up until the news that there's a new card coming, we were ready to get one right away for testing once the notification from the EVGA store came, but it never happend.

Size wise people mentioned it should be fine.
I wonder how 3080 - power wise would work?
320W card, I guess small undervolting would need to be done, but do you think 320W card could be ok in this case?

 

[timmeh]

@volopasse Note that I am also still running a 1060 - your GPU sounds much beefier (and hotter) than mine

 

[Runamok81]

Grabbed a screenshot from the HardwareCanucks PCI4.0 Riser cable review. Interesting to see which cables work for gen 4, and which fail.

 

[Dax911]

Grabbed a screenshot from the HardwareCanucks PCI4.0 Riser cable review. Interesting to see which cables work for gen 4, and which fail.

He mentioned at the end that some riser will fail on one setup and not on another one even if the build's parts are identic .

 

[SaperPL]

He mentioned at the end that some riser will fail on one setup and not on another one even if the build's parts are identic .

Like I said over and over again before. We're at bleeding edge right now with first generations of cards that support pci-e 4.0. Stuff like this will keep happening. This is a teething problem where we've got a spec, but before we got both the cards and the boards, vendors of both wouldn't be able to test things out properly to smooth things out on either side. And now we're at stage where cards can talk to boards and software is smoothed out, but comes a third part into the equation, that it's not that common. which is a riser and we'll have to wait some time for it to get to a point when the risers will get stable.

I'm not really sure about those long passive ribbon riser being able to properly work on 4.0 at all without some additional power delivered across the ribbon. We've got the boards that are squeezing out the last drops of the old pci-e slot design to handle 4.0 and now we are plugging in extension that tries to force the board to push more power - to handle higher frequency you need more clocks so the current raises (4.0 improvements), to handle bigger distance and thus higher resistance (the riser) you need more voltage to distinguish value differences.

Power delivery to the slot may vary from board to board of the same model and one person might get lucky because he got slightly better power delivery for the slot and that slight difference is just enough that his good quality riser works on 4.0 and he will state that this riser works with this board and this GPU and after that people will just go for it and it'll be hit or miss.

This is something that someone who chewed through the pci-e 4.0 electrical part would would need to test on the boards and would need to know where and how to clamp on the boards to test out the things how this actually behaves and how close we are to the boarder or stability on those risers.

Btw anyone of you tried running a 75W card that doesn't have a PEG connector through a ribbon riser? This is interesting part here, but remember we had those weird risers that had additional power plug to handle this, but it wasn't generally a problem with cards that have PEG. But depending on how the card power delivery is built, if you can't deliver those 75W from the slot, it might mean you're starving your card and it runs lower clocks because of this.

And here comes the fun part where we have brands promoting their designs with long ribbon risers showing off low temps without showing how much such system compares to how this would work in a chassis that doesn't use the riser. 75W is a significant difference even for a card that can pull 300W. Not necessarily all of this 75W will be lost, but let's say we get 45W instead of 75W, so it's a 10% drop from 300W and for those 75W cards it might mean crippling performance significantly if they'll run at all.

 

[phinix]

As an update, im currently playing Cyberpunk 2077, 4k, rtx ultra, dlss performance. Im getting 60 fps with 40 fps downs on some moments.
The temps on the GPU are peaking at 83 C. I did a quick undervolt.
I was thinking on 3d print the fan duct for the cpu and installing 2 tiny noctua fans on the top of gpu side, where the hot air accumulates.

The tiny fans has already been discused i think, can improve thermals or are not worth it?

So how is this 3080 in Sentry? Did you use it for longer period now? How are the temps? Is it throttling?
How much undervolt you had to do?
Would you say that 320W card like this 3080 is ok for Sentry?

 

[Runamok81]

Like I said over and over again before. We're at bleeding edge right now with first generations of cards that support pci-e 4.0. Stuff like this will keep happening. This is a teething problem where we've got a spec, but before we got both the cards and the boards, vendors of both wouldn't be able to test things out properly to smooth things out on either side. And now we're at stage where cards can talk to boards and software is smoothed out, but comes a third part into the equation, that it's not that common. which is a riser and we'll have to wait some time for it to get to a point when the risers will get stable.

I'm not really sure about those long passive ribbon riser being able to properly work on 4.0 at all without some additional power delivered across the ribbon. We've got the boards that are squeezing out the last drops of the old pci-e slot design to handle 4.0 and now we are plugging in extension that tries to force the board to push more power - to handle higher frequency you need more clocks so the current raises (4.0 improvements), to handle bigger distance and thus higher resistance (the riser) you need more voltage to distinguish value differences.

Power delivery to the slot may vary from board to board of the same model and one person might get lucky because he got slightly better power delivery for the slot and that slight difference is just enough that his good quality riser works on 4.0 and he will state that this riser works with this board and this GPU and after that people will just go for it and it'll be hit or miss.

This is something that someone who chewed through the pci-e 4.0 electrical part would would need to test on the boards and would need to know where and how to clamp on the boards to test out the things how this actually behaves and how close we are to the boarder or stability on those risers.

Btw anyone of you tried running a 75W card that doesn't have a PEG connector through a ribbon riser? This is interesting part here, but remember we had those weird risers that had additional power plug to handle this, but it wasn't generally a problem with cards that have PEG. But depending on how the card power delivery is built, if you can't deliver those 75W from the slot, it might mean you're starving your card and it runs lower clocks because of this.

And here comes the fun part where we have brands promoting their designs with long ribbon risers showing off low temps without showing how much such system compares to how this would work in a chassis that doesn't use the riser. 75W is a significant difference even for a card that can pull 300W. Not necessarily all of this 75W will be lost, but let's say we get 45W instead of 75W, so it's a 10% drop from 300W and for those 75W cards it might mean crippling performance significantly if they'll run at all.

@SaperPL - No worries. I get it. I also have the luxury of not having a wildly successful case with an adoring fan base wanting to know if you can solve the PCIe Gen4 problem for them.

In other news, I saw that @yawacool (LiHeat) has got another concept on the forum and he mentioned his team might be working on Gen4 PCIe riser. LiHeat has supplied risers to a few of the other concepts on SFF, so It would be great if they could get an inexpensive (less than 3M) riser knocked out for some of our cases. I asked him over in his thread. He posted an interesting PCI Sig article discussing the relevancy of PCIe "redriver" and "retimer" chips being required to help attenuate and "push" the newer higher frequency PCI gen 4.0 signals further. I believe images from the same article were referenced on this forum earlier this year. It sounds like integrating a "retimer" chip may be required? Retimers - signal conditioners - have been a thing for a long time, but I think the PCI Sig team formalized the retimer spec for Gen 4.0. So, they are less of a crapshoot now as they are "part of the spec." We might be seeing more of them? An interesting counterpoint is that I don't think the 3M or LinkUp cables have a retimer do they? I don't see one. Both of those manufacturers seem to cap the cable length to under 400mm and I suppose, they just use better shielding and more exotic materials (silver) and build quality to preserve the signal? In regards to the article, it's not abundantly clear to me WHERE one would integrate a retimer? Is this something we do add to the motherboard just before it exits the PCIe connector? Do any existing motherboards have them? Or, is this something we should expect to see in an active "powered" riser? A riser that has a bit of PCB to include a chip.

Not being an electrical engineer, this is all just a bit over my head. But, I sure I hope someone is successful figuring this out.
It's clear to me that the market has a need for an inexpensive and reliable Gen4 riser cable right now. First one to market wins!

 

[Dax911]

Like I said over and over again before. We're at bleeding edge right now with first generations of cards that support pci-e 4.0. Stuff like this will keep happening. This is a teething problem where we've got a spec, but before we got both the cards and the boards, vendors of both wouldn't be able to test things out properly to smooth things out on either side. And now we're at stage where cards can talk to boards and software is smoothed out, but comes a third part into the equation, that it's not that common. which is a riser and we'll have to wait some time for it to get to a point when the risers will get stable.

I'm not really sure about those long passive ribbon riser being able to properly work on 4.0 at all without some additional power delivered across the ribbon. We've got the boards that are squeezing out the last drops of the old pci-e slot design to handle 4.0 and now we are plugging in extension that tries to force the board to push more power - to handle higher frequency you need more clocks so the current raises (4.0 improvements), to handle bigger distance and thus higher resistance (the riser) you need more voltage to distinguish value differences.

Power delivery to the slot may vary from board to board of the same model and one person might get lucky because he got slightly better power delivery for the slot and that slight difference is just enough that his good quality riser works on 4.0 and he will state that this riser works with this board and this GPU and after that people will just go for it and it'll be hit or miss.

This is something that someone who chewed through the pci-e 4.0 electrical part would would need to test on the boards and would need to know where and how to clamp on the boards to test out the things how this actually behaves and how close we are to the boarder or stability on those risers.

Btw anyone of you tried running a 75W card that doesn't have a PEG connector through a ribbon riser? This is interesting part here, but remember we had those weird risers that had additional power plug to handle this, but it wasn't generally a problem with cards that have PEG. But depending on how the card power delivery is built, if you can't deliver those 75W from the slot, it might mean you're starving your card and it runs lower clocks because of this.

And here comes the fun part where we have brands promoting their designs with long ribbon risers showing off low temps without showing how much such system compares to how this would work in a chassis that doesn't use the riser. 75W is a significant difference even for a card that can pull 300W. Not necessarily all of this 75W will be lost, but let's say we get 45W instead of 75W, so it's a 10% drop from 300W and for those 75W cards it might mean crippling performance significantly if they'll run at all.

No worries Saper , I was just pointing this fact to Runamock for him to be careful if he intends to buy one.
I'm actually not waiting after a pci-e 4 riser since my mobo can split pci lanes to get the nvme on pci-e 4.0 .

 

[aaaadrian]

So how is this 3080 in Sentry? Did you use it for longer period now? How are the temps? Is it throttling?
How much undervolt you had to do?
Would you say that 320W card like this 3080 is ok for Sentry?

I've been enjoying my 3080 xc3 and 5600x for almost 2 months now, played CP2077 for 70+ hours. Everything works perfectly fine and temps are also reasonable. I undervolted my 3080 to run at 1905mhz on 850mv and set a more aggressive custom fan curve so my gpu temp never exceeds 80 degree Celsius.

 

[AlphaDave]

Long time lurker here. I've loved my Sentry 1.1 (white ) for some years now but until recently have mostly been plug and play. Based on some of the recommendations here and elsewhere, I settled on a 5600x for it's price, thermals, and performance for games and productivity. I was lucky enough to snag one recently and was able to improve performance and go from ~84C to ~69C with a simple undervolt in Ryzen master, ~64C by swapping out my L9i for the L9a, and finally a cool 60C by adding the J-Hacks 92mm fan shroud (10mm height) under full load in Cinebench r23.

Satisfied with the CPU, and with stocks we'll start to see picking up, my attention turns to the GPU. The more traditional EVGA XC3 series seems to be most recommended for the Sentry, but those 3xxx FE cards are just so dang purty... And with airflow to the back, it seems in theory we could see some very nice results. I've seen @AlphaDave and some others mention modifying the panel to accommodate the new design. @ZombiPL rightly seems to be taking a 'wait and see' approach to how well they are received and whether we see a long term movement to this style.

I am wondering whether anyone has tested this at all yet, specifically with the 3080 FE. I don't see myself taking a drill to my case, but am considering having something custom made (because fun), perhaps with a removable air pocket insert that would play nice with different style designs in the future, and printing shrouds for both fans. Would love feedback on this if anyone has, cheers!

Yeah I managed to snag a 3080 FE after several attempts, it is the one I wanted, plus it’s the cheapest variant. As for the case I cut out a section of the air pocket insertion bit and drilled out vent holes. I use weather stripping to seal the GPU to the side of the case and also the exhaust as best as I could.

Additionally I drilled all of the existing vent holes to the GPU compartment to 6.5mm diameter to help intake. Seemed to have little effect on the structural integrity of the case, though though going round each hole with metal files neatening it all up was an utter pain in the ass.

My specs are as follows:
i7 8700
LP53 + Noctua NF-A9x14
ASRock Fatal1ty Z370 Gaming-ITX
Corsair Vengeance LPX 16GB 3200mhz
Geforce RTX 3080
Corsair SF 750 Platinum

3080 is undervolted to 1875mhz at 850mV, anything less crashes Control despite passing other stress tests.

All stays relatively quiet, quieter than my old 1070 FE was under load anyway.