-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Production Modultra LOBO: Direct die kit gauging interest.
- Thread starter Alloy Craft
- Start date
Hey, would like to know what Alphacool fittings exactly are this? Can you link them please?hey guys!I received the first prototype of the LOBO. Thanks to @Alloy Craft for the oportunity to test this beautiful product. <3
It really is a masterpice of engineering and precision. Just for everybody interested: my unit is still a prototype that will differ slightly from the final product
and was allready in use. Eventhough it looks just stunning and BOY IS THIS THING HEAVY!
I've took some pictures during my video shoot that should give some insights and a better view on the product.
Lets start with the assembled LOBO:
Hope you like the pictures and that you will tune in for the videos!
Kind regards
Victor from GucksTV
Yes we will be shipping to Europe.
Production is a bit delayed, I had some trouble getting material and some of the tooling is still not available yet. But all of the material is here now and just waiting on a few tools and production can begin.
People have been asking about pre orders and I will be accepting those soon. However I want to make samples of all the heatsink styles before I start taking pre orders, that way people will know exactly what is being offered. Hopefully I can have those done next week. After that pre orders will open up.
People have been asking about pre orders and I will be accepting those soon. However I want to make samples of all the heatsink styles before I start taking pre orders, that way people will know exactly what is being offered. Hopefully I can have those done next week. After that pre orders will open up.
I managed to salvage my HWinfo logs, hooray! It'll take me a day or two to get my results together, but there will be graphs and temperatures and photos, oh my! Have to wrap my head around how to present the data best (what's more interesting - straight CPU temps, CPU over air intake deltas, or CPU over water (over air intake) deltas?) and try to avoid spamming you with too much data, but at the very least there will be a thorough comparison of the Aquanaut and the Lobo at several pump speeds, fan speeds, and CPU power levels.
A couple of photos from the installation process:
A couple of photos from the installation process:
Will you be able to change mount between intel/amd or are they separate units?
Curiosity
Too busy figuring out if I can to think if I shoul
Platinum Supporter
Bronze Supporter
M...M...M...M...Multi-Tier...Subscriber...
- Apr 30, 2016
- 724
- 838
Super curious about this guy. I'm currently using a Asus Z490i where I had to remove the top heatsink in order for the Barrow pump/block/res to fit (which, don't use the "intelligent" version, it's garbage. It reverts to max pump speed 95% of the time after sleep or shutdown).
Will be keeping my eyeballs on this.
Will be keeping my eyeballs on this.
This will be perfect for my triple radiator setup in the formd t1, so i'll be buying one definitely!
If beta testers are needed for this cooler, i will gladly take part!
i'm a long time sff modder/enthusiast, and it has been long since i was this hyped about a product!
If beta testers are needed for this cooler, i will gladly take part!
i'm a long time sff modder/enthusiast, and it has been long since i was this hyped about a product!
All right, so it's test report time. This will be a bit lengthy, so I'll be posting it unfinished and updating as I go.
Edit: finished (though let me know if there's any other data that might be interesting - I have tons of logs). See conclusion at the bottom of the post.
A bit of a TL;DR for those of you wanting that: The Lobo is pretty good. It's relatively easy to install, build quality is great, and thermals are noticeably better than the original Aquanaut, particularly at low pump speeds. It also looks really good, though being just about the most hidden component in your system (beneath a pump and pump housing, fittings and tubing) that likely doesn't matter much. Sadly I haven't tested the Aquanaut Extreme, which seems like a compelling alternative with its non-reversed flow orientation, but this is clearly a superior product to the original.
For reference, this is the setup I started with: an Aquanaut (first run, with the bad mounting screws - current ones are AFAIK better). Intake from the 280mm rad in my Meshlicious and through the micro res/bubble catcher; outlet leads to the GPU. Pump and fans are all controlled with an Aquacomputer Quadro.
Initial impressions
Being a beta unit, this is how my LOBO arrived - well packed, but nothing fancy. I didn't actually order the pump housing, but apparently there was a shipping snafu, and I got one. Due to not having any spare thermal pads I sadly didn't get a chance to test it out, but it seems good - its walls are noticeably thicker than my Alphacool one, though judging by the weight that is copper/brass while this is aluminium.
Edit: sticking the rest of this in a spoiler tag, so that the thread doesn't completely bog down due to photo overload. Click to read:
The pump housing is by far the most 'beta' of these units, with the side slots being particularly rough. Not unexpected though, and likely easily figured out before any full manufacturing run.
It still fits my Laing DDC perfectly. It's really too bad I didn't have the thermal pads to test it.
At the very least, it is Fiona approved.
As is the main unit.
The quality and finish of the unit itself is superb.
While mine had a few blemishes in the cerakote (about 10 o'clock above the pump impeller cavity in the photo above), @Alloy Craft pointed out those beforehand and has reportedly moved to a different paint shop due to this. Even if not, and even considering the premium price this will have in the end, I wouldn't consider it a major issue, as they are hidden behind the pump and even if visible would be negligible (especially on the more matte finishes). The fit of the parts is very, very good, with everything slotting together perfectly. The included O-rings fit perfectly and were easy to install after disassembly, as they slotted nicely into their grooves. When screwed together everything felt extremely solidly mounted - with the exception of the mounting plate, which "floats" between a lip on the cold plate and the top housing. This is a great design, as it avoids bowing from the mounting plate transferring into the cooler, which can cause issues.
Paint was a tad thin on the underside of the mounting plate, but really nothing to be concerned about. Look at that fit, though!
The outside of the cold plate is milled well. Not a mirror polish, but more than smooth enough, with no notable grooves, dents or chips. I didn't check for flatness, but going by thermals, contact is good.
My cold plate's microfins had some very minor milling errors: if you look closely towards the sides of the fin stack, you can see that one groove on each side is wider than the rest, and has slightly bent its neighbouring fins. Definitely not enough to cause issues, but worth noting.
One thing worth noting here: all fasteners on the LOBO are (seemingly) in Freedom Units™, so non-US buyers/people with metric tools beware! Ideally the correctly sized hex keys/bits would be included though of course this would be an extra cost and might not be feasible. I made it work with a 2.5mm hex bit for the mounting screws (can't remember the size for the cold plate or the side plate), but that felt very loose and had me a bit worried. The last thing you want is a stripped head on a CPU cooler mounting screw, after all. I'll be ordering an imperial hex key set to use with this in the future, as I don't trust the screws' ability to hold up to this kind of torque on too-small tools for long.
Edit: the above might not be accurate - @Alloy Craft assures me that the screws are supposed to be metric, though I wonder if they might be metric threading with imperial heads. Either way, mine were workable with metric tools, though far from ideal. IIRC the screws on the side cover plate fit my metric bits perfectly, but both the cold plate screws and mounting screws had some slack, fitting a 2.5mm bit loosely but not fitting a 3mm bit. Either way, I'm sure this will be straightened out come release.
Conclusion:
So, taking everything into account, where does the LOBO land? As I said above, I like it. It seems like a very good product, with excellent build quality and good performance. In the latter aspect it manages to both cool better than the Aquanaut and deliver marginally higher clocks while doing so. It is especially impressive at lower pump speeds, where the Aquanaut falls behind by nearly five degrees.
This testing being done on a notoriously difficult-to-cool CPU makes this particularly impressive to me. As we can see, the thermal results are still rather hot - 70 degrees at max fan speed under water isn't cool. But on a less thermally dense CPU with its heat better centered under the cold plate, such as any Intel CPU? I would expect noticeably lower temperatures at the same power level.
A note on noise: I don't have any means of doing reliable noise testing, and there's too much noise coming from the window next to my desk for this to be reliable anyhow. Going by ear, there's no discernible difference between the LOBO and the Aquanaut with the same pump. If you're worried about a metallic pump mount vibrating or ringing, don't be. My motherboard and GPU VRMs' coil whine is far louder than the pump at 3400rpm, and neither are audible through open-backed headphones.
Now, my beta unit did have some minor QC issues. Hopefully the new paint shop will avoid blemishes like on mine, and there won't be any rusty springs included either. If the final units came with the appropriate hex keys included (or switched to metric head sizes), that would be fantastic. But other than that, this early production run is essentially ready for prime time.
Overall, the differences between the LOBO and the Aquanaut aren't massive, and the LOBO is more expensive. If you're on a budget, that's still a good option. But a strict budget and SFF custom loop building don't really go together all that well, which makes the LOBO look better - you're getting much better build quality, better performance, easier bleeding, a better mounting system and better aesthetics for your money. Would I recommend it? Yes. Easily.
Edit: added a note on the mounting screw situation after hearing back, see above.
Edit: finished (though let me know if there's any other data that might be interesting - I have tons of logs). See conclusion at the bottom of the post.
A bit of a TL;DR for those of you wanting that: The Lobo is pretty good. It's relatively easy to install, build quality is great, and thermals are noticeably better than the original Aquanaut, particularly at low pump speeds. It also looks really good, though being just about the most hidden component in your system (beneath a pump and pump housing, fittings and tubing) that likely doesn't matter much. Sadly I haven't tested the Aquanaut Extreme, which seems like a compelling alternative with its non-reversed flow orientation, but this is clearly a superior product to the original.
For reference, this is the setup I started with: an Aquanaut (first run, with the bad mounting screws - current ones are AFAIK better). Intake from the 280mm rad in my Meshlicious and through the micro res/bubble catcher; outlet leads to the GPU. Pump and fans are all controlled with an Aquacomputer Quadro.
Initial impressions
Being a beta unit, this is how my LOBO arrived - well packed, but nothing fancy. I didn't actually order the pump housing, but apparently there was a shipping snafu, and I got one. Due to not having any spare thermal pads I sadly didn't get a chance to test it out, but it seems good - its walls are noticeably thicker than my Alphacool one, though judging by the weight that is copper/brass while this is aluminium.
Edit: sticking the rest of this in a spoiler tag, so that the thread doesn't completely bog down due to photo overload. Click to read:
The pump housing is by far the most 'beta' of these units, with the side slots being particularly rough. Not unexpected though, and likely easily figured out before any full manufacturing run.
It still fits my Laing DDC perfectly. It's really too bad I didn't have the thermal pads to test it.
At the very least, it is Fiona approved.
As is the main unit.
The quality and finish of the unit itself is superb.
While mine had a few blemishes in the cerakote (about 10 o'clock above the pump impeller cavity in the photo above), @Alloy Craft pointed out those beforehand and has reportedly moved to a different paint shop due to this. Even if not, and even considering the premium price this will have in the end, I wouldn't consider it a major issue, as they are hidden behind the pump and even if visible would be negligible (especially on the more matte finishes). The fit of the parts is very, very good, with everything slotting together perfectly. The included O-rings fit perfectly and were easy to install after disassembly, as they slotted nicely into their grooves. When screwed together everything felt extremely solidly mounted - with the exception of the mounting plate, which "floats" between a lip on the cold plate and the top housing. This is a great design, as it avoids bowing from the mounting plate transferring into the cooler, which can cause issues.
Paint was a tad thin on the underside of the mounting plate, but really nothing to be concerned about. Look at that fit, though!
The outside of the cold plate is milled well. Not a mirror polish, but more than smooth enough, with no notable grooves, dents or chips. I didn't check for flatness, but going by thermals, contact is good.
My cold plate's microfins had some very minor milling errors: if you look closely towards the sides of the fin stack, you can see that one groove on each side is wider than the rest, and has slightly bent its neighbouring fins. Definitely not enough to cause issues, but worth noting.
One thing worth noting here: all fasteners on the LOBO are (seemingly) in Freedom Units™, so non-US buyers/people with metric tools beware! Ideally the correctly sized hex keys/bits would be included though of course this would be an extra cost and might not be feasible. I made it work with a 2.5mm hex bit for the mounting screws (can't remember the size for the cold plate or the side plate), but that felt very loose and had me a bit worried. The last thing you want is a stripped head on a CPU cooler mounting screw, after all. I'll be ordering an imperial hex key set to use with this in the future, as I don't trust the screws' ability to hold up to this kind of torque on too-small tools for long.
Edit: the above might not be accurate - @Alloy Craft assures me that the screws are supposed to be metric, though I wonder if they might be metric threading with imperial heads. Either way, mine were workable with metric tools, though far from ideal. IIRC the screws on the side cover plate fit my metric bits perfectly, but both the cold plate screws and mounting screws had some slack, fitting a 2.5mm bit loosely but not fitting a 3mm bit. Either way, I'm sure this will be straightened out come release.
Installation
Intalling the LOBO was reasonably simple. Sadly I don't have any photos of the process (I only have so many hands!). The mount is a plastic cap on top of a thick, sturdy spring, through which a hex head screw is threaded. The screw pushes the cap down which compresses the spring, and the length of the cap limits how much the screws can be tightened. For the AM4 version at least, the screw goes directly into the stock backplate, which is great - simple, reliable, and easy to install and remove. The plastic of the cap is thick and sturdy, and I'm not at all worried about it bending or flexing under the pressure of the mount, unlike the thin nylon washers used in the first revision of the Aquanaut.
(If you were wondering why there were pliers in the background of one of the photos above, that's wat happens when you have dual-threaded rods with a screw-on cap for mounting like the Aquanaut (photo link) - you'll most likely need pliers or a very small wrench to remove the rods from the motherboard/backplate after they've been fully tightened. The Lobo has no such issues.)
The drawback of this style of mounting (and the advantage of the Aquanaut's style) is that holding it in place while mounting, at least on AM4, is fiddly. You'll need to hold the backplate to the motherboard (or have your motherboard on a flat surface, which will make it wobbly with the backplate underneath), line up the cooler with the socket mounting holes (without dragging your thermal paste all over the place while adjusting it), and hold it in place while getting at least a few screw+cap+spring assemblies sufficiently threaded in place to hold it all. I'm making it sound harder than it is in practice here - likely because I installed mine on a motherboard already in a case, with the inherent difficulty spike of doing that - but it takes a bit of dexterity. That being said, I still prefer this mounting style to the Aquanaut's.
Another (hopefully) beta "issue" - a couple of my springs were had rust spots. It was fine after spraying with some 5-56 and rubbing down with a paper towel, but these should ideally come greased/rust proofed from the factory.
Screwing fittings into cerakote-covered G1/4 threads feels really weird - a strange combination of resistance and low friction - and one of my fittings took a bit of force for the first installation. That was a one-time thing, as it was much easier after unscrewing once, and the seal was good from the beginning. Despite this initial tightness, there were no signs of wear or cracking on the paint where the fittings threaded in, so the finish seems well suited to the task.
The insides of the unit were also flawlessly clean. If this is the level of finish for retail units, there's no need to flush these. I scrubbed my unit down with q-tips and alcohol, and didn't find a single speck of anything inside - no paint chips, no metal shavings, no debris of any kind.
The block with attached tubing+QDCs passed pressure testing with EK's leak tester without issues (as expected).
Lastly, thanks to the different port locations I had to rethink my loop layout just a tad - this won't do:
Nothing a couple of spare fittings couldn't fix though, and the bleed/fill port on top of the micro res is easier to use thanks to rotating the unit 90°.
Overall I'm very happy with how it came out looking in the end.
Intalling the LOBO was reasonably simple. Sadly I don't have any photos of the process (I only have so many hands!). The mount is a plastic cap on top of a thick, sturdy spring, through which a hex head screw is threaded. The screw pushes the cap down which compresses the spring, and the length of the cap limits how much the screws can be tightened. For the AM4 version at least, the screw goes directly into the stock backplate, which is great - simple, reliable, and easy to install and remove. The plastic of the cap is thick and sturdy, and I'm not at all worried about it bending or flexing under the pressure of the mount, unlike the thin nylon washers used in the first revision of the Aquanaut.
(If you were wondering why there were pliers in the background of one of the photos above, that's wat happens when you have dual-threaded rods with a screw-on cap for mounting like the Aquanaut (photo link) - you'll most likely need pliers or a very small wrench to remove the rods from the motherboard/backplate after they've been fully tightened. The Lobo has no such issues.)
The drawback of this style of mounting (and the advantage of the Aquanaut's style) is that holding it in place while mounting, at least on AM4, is fiddly. You'll need to hold the backplate to the motherboard (or have your motherboard on a flat surface, which will make it wobbly with the backplate underneath), line up the cooler with the socket mounting holes (without dragging your thermal paste all over the place while adjusting it), and hold it in place while getting at least a few screw+cap+spring assemblies sufficiently threaded in place to hold it all. I'm making it sound harder than it is in practice here - likely because I installed mine on a motherboard already in a case, with the inherent difficulty spike of doing that - but it takes a bit of dexterity. That being said, I still prefer this mounting style to the Aquanaut's.
Another (hopefully) beta "issue" - a couple of my springs were had rust spots. It was fine after spraying with some 5-56 and rubbing down with a paper towel, but these should ideally come greased/rust proofed from the factory.
Screwing fittings into cerakote-covered G1/4 threads feels really weird - a strange combination of resistance and low friction - and one of my fittings took a bit of force for the first installation. That was a one-time thing, as it was much easier after unscrewing once, and the seal was good from the beginning. Despite this initial tightness, there were no signs of wear or cracking on the paint where the fittings threaded in, so the finish seems well suited to the task.
The insides of the unit were also flawlessly clean. If this is the level of finish for retail units, there's no need to flush these. I scrubbed my unit down with q-tips and alcohol, and didn't find a single speck of anything inside - no paint chips, no metal shavings, no debris of any kind.
The block with attached tubing+QDCs passed pressure testing with EK's leak tester without issues (as expected).
Lastly, thanks to the different port locations I had to rethink my loop layout just a tad - this won't do:
Nothing a couple of spare fittings couldn't fix though, and the bleed/fill port on top of the micro res is easier to use thanks to rotating the unit 90°.
Overall I'm very happy with how it came out looking in the end.
Filling and bleeding
My impressions here aren't really representative of most users, as I have a spare pump and res that I use (with my QDCs to connect it all) that makes filling and bleeding much easier. QDCs also let me fill individual portions of the loop if I want to.
Still, the pump was obviously entirely empty at first. I first filled it+the directly attached tubes before installing it all, then topped it up after installation. Bleeding took maybe ten minutes of power cycling the pump, and I quickly switched to running the DDC, as it's more powerful (and the external pump can't really rotate it to get air out of the DDC impeller).
All in all, my impression is that this is easier to bleed than the Aquanaut - likely due to better machining with fewer sharp angles and better overall geometries for allowing water to flow easily. The last time I rebuilt my loop with the Aquanaut I had some issues with air bubbles being stuck in the pump and making a lot of noise. With the LOBO, the pump was clear within a few cycles, and it did an admirable job of not letting new air get stuck as bubbles passed through the system.
(I still very much recommend a micro res setup like mine though. The "res" is an EK flow indicator with the rotor removed, stuffed with some anti-bubble foam, which effectively prevents air bubbles from being recirculated through the pump.)
My impressions here aren't really representative of most users, as I have a spare pump and res that I use (with my QDCs to connect it all) that makes filling and bleeding much easier. QDCs also let me fill individual portions of the loop if I want to.
Still, the pump was obviously entirely empty at first. I first filled it+the directly attached tubes before installing it all, then topped it up after installation. Bleeding took maybe ten minutes of power cycling the pump, and I quickly switched to running the DDC, as it's more powerful (and the external pump can't really rotate it to get air out of the DDC impeller).
All in all, my impression is that this is easier to bleed than the Aquanaut - likely due to better machining with fewer sharp angles and better overall geometries for allowing water to flow easily. The last time I rebuilt my loop with the Aquanaut I had some issues with air bubbles being stuck in the pump and making a lot of noise. With the LOBO, the pump was clear within a few cycles, and it did an admirable job of not letting new air get stuck as bubbles passed through the system.
(I still very much recommend a micro res setup like mine though. The "res" is an EK flow indicator with the rotor removed, stuffed with some anti-bubble foam, which effectively prevents air bubbles from being recirculated through the pump.)
Aquanaut Comparison
The LOBO is noticeably smaller than the Aquanaut. The rounded mounting plate takes up less space; the unit itself is more compact in every dimension, and the angled pump mount makes it appear even smaller. The unit itself is ~3mm shorter than the Aquanaut, with the pump mounting slightly lower than this again.
I measured the LOBO with my pump mounted as follows:
49mm from the bottom of the cold plate to the tallest fin
44mm from the bottom of the cold plate to the top of the pump housing, below the fins
Seeing how the Alloy Craft pump housing is in-between these two in height, I would estimate a total mounted height of ~46mm using that.
All of this is of course not including fittings.
The LOBO is noticeably smaller than the Aquanaut. The rounded mounting plate takes up less space; the unit itself is more compact in every dimension, and the angled pump mount makes it appear even smaller. The unit itself is ~3mm shorter than the Aquanaut, with the pump mounting slightly lower than this again.
I measured the LOBO with my pump mounted as follows:
49mm from the bottom of the cold plate to the tallest fin
44mm from the bottom of the cold plate to the top of the pump housing, below the fins
Seeing how the Alloy Craft pump housing is in-between these two in height, I would estimate a total mounted height of ~46mm using that.
All of this is of course not including fittings.
Test setup
The thermal testing was done using OCCT 8.2.1, on its default CPU stress test settings, meaning Large data set, Normal mode, Variable load, Auto instruction set (=AVX2 for Zen3), Auto threads (=16 for my 5800X). Each test was run for 15 minutes, and data was logged using HWiNFO64.
System specs:
AMD Ryzen 7 5800X (config below)
ASRock B550 Phantom Gaming ITX/ax
2x16GB G.Skill FlareX 3200c14 @ 3800c15 @1.38V
PowerColor Radeon 6900 XT Liquid Devil Ultimate
Adata XPG SX8200 Pro 2TB
Corsair SF750 Platinum PSU
J-Hack M2426 PSU adapter
Corsair XR5 280mm radiator
Laing DDC 1T Plus/3.2(?) PWM
2x Arctic P14 fans
1x Noctua NF-A12x15 over the RAM and CPU block (B-die doesn't like heat )
Aquacomputer Quadro fan/pump controller
Aquanaut/LOBO water blocks
EK ZMT tubing, EK and Alphacool fittings, EK flow indicator, fluid thermal sensors after the GPU and radiator
Alphacool Industrial Nylon quick disconnect fittings
The system is run "as is", i.e. with my normal system config, no applications closed or other optimizations done for testing, save for a fresh reboot before each test cycle. That means various game launchers and the like running in the background - but the OCCT AVX2 load is sufficiently dynamic to maintain 100% load across all cores regardless of this. I do not lock my CPU to specific clocks or voltages when testing. According to the test logs, the testing kept the CPU pegged at within .01W of the power limit throughout testing.
I have run my CPU with a PPT/TDC/EDC limit of 120/95/120 rather than the stock 142/95/140, mainly due to thermals using the Aquanaut. Testing was done both at stock and with this profile applied. In addition to this, PBO is increased by +100MHz regardless of power limits, and Curve Optimizer is used to undervolt each core (negative offsets of 7, 11, 20, 20, 20, 20, 20, 20) in both power profiles.
This means a few things: the system will boost dependent on the logic inherent to the CPU's governor module, including power, voltage and thermal protections. Clocks (both reported and effective) have been logged during testing - differences will be noted below. In general, differences across runs are marginal.
Fans and pump speed are controlled through the Quadro, and testing is done in three profiles:
- 100% radiator fan speed (~1600rpm), 45% PWM pump speed (~3400rpm), 60% RAM fan speed (~1200rpm)
- 65% radiator fan speed (~1200rpm), 45% PWM pump speed (~3400rpm), 23% RAM fan speed (~550rpm)
- 65% radiator fan speed (~1200rpm), 20% PWM pump speed (~1200rpm), 23% RAM fan speed (~550rpm)
A couple of test notes: HWiNFO64 doesn't seem to like logging Quadro sensors, so depending on the test run some were missing, but they are sufficiently consistent that this is not an issue for the results - every run is consistent in these metrics. Also, the 45% PWM setting for the pump is not 45% of its maximum speed - it maxes out at around 60% and 4500rpm. I don't test at that setting as it's annoyingly whiny at that speed. 3400rpm is essentially inaudible. At 1600rpm the fans are audible but not loud, while at 1200rpm they are barely audible (there's typically more noise coming through the window next to the PC).
The thermal testing was done using OCCT 8.2.1, on its default CPU stress test settings, meaning Large data set, Normal mode, Variable load, Auto instruction set (=AVX2 for Zen3), Auto threads (=16 for my 5800X). Each test was run for 15 minutes, and data was logged using HWiNFO64.
System specs:
AMD Ryzen 7 5800X (config below)
ASRock B550 Phantom Gaming ITX/ax
2x16GB G.Skill FlareX 3200c14 @ 3800c15 @1.38V
PowerColor Radeon 6900 XT Liquid Devil Ultimate
Adata XPG SX8200 Pro 2TB
Corsair SF750 Platinum PSU
J-Hack M2426 PSU adapter
Corsair XR5 280mm radiator
Laing DDC 1T Plus/3.2(?) PWM
2x Arctic P14 fans
1x Noctua NF-A12x15 over the RAM and CPU block (B-die doesn't like heat
)Aquacomputer Quadro fan/pump controller
Aquanaut/LOBO water blocks
EK ZMT tubing, EK and Alphacool fittings, EK flow indicator, fluid thermal sensors after the GPU and radiator
Alphacool Industrial Nylon quick disconnect fittings
The system is run "as is", i.e. with my normal system config, no applications closed or other optimizations done for testing, save for a fresh reboot before each test cycle. That means various game launchers and the like running in the background - but the OCCT AVX2 load is sufficiently dynamic to maintain 100% load across all cores regardless of this. I do not lock my CPU to specific clocks or voltages when testing. According to the test logs, the testing kept the CPU pegged at within .01W of the power limit throughout testing.
I have run my CPU with a PPT/TDC/EDC limit of 120/95/120 rather than the stock 142/95/140, mainly due to thermals using the Aquanaut. Testing was done both at stock and with this profile applied. In addition to this, PBO is increased by +100MHz regardless of power limits, and Curve Optimizer is used to undervolt each core (negative offsets of 7, 11, 20, 20, 20, 20, 20, 20) in both power profiles.
This means a few things: the system will boost dependent on the logic inherent to the CPU's governor module, including power, voltage and thermal protections. Clocks (both reported and effective) have been logged during testing - differences will be noted below. In general, differences across runs are marginal.
Fans and pump speed are controlled through the Quadro, and testing is done in three profiles:
- 100% radiator fan speed (~1600rpm), 45% PWM pump speed (~3400rpm), 60% RAM fan speed (~1200rpm)
- 65% radiator fan speed (~1200rpm), 45% PWM pump speed (~3400rpm), 23% RAM fan speed (~550rpm)
- 65% radiator fan speed (~1200rpm), 20% PWM pump speed (~1200rpm), 23% RAM fan speed (~550rpm)
A couple of test notes: HWiNFO64 doesn't seem to like logging Quadro sensors, so depending on the test run some were missing, but they are sufficiently consistent that this is not an issue for the results - every run is consistent in these metrics. Also, the 45% PWM setting for the pump is not 45% of its maximum speed - it maxes out at around 60% and 4500rpm. I don't test at that setting as it's annoyingly whiny at that speed. 3400rpm is essentially inaudible. At 1600rpm the fans are audible but not loud, while at 1200rpm they are barely audible (there's typically more noise coming through the window next to the PC).
Thermal data
120W testing:
As you can clearly see, at the 120W load, the Lobo is clearly ahead. In fact, the Lobo at the lowest fan/pump setting is just barely warmer than the Aquanaut at the highest. That's a clear victory.
More in detail:
The LOBO has a 4.5°C advantage at the 20% setting, 2.7°C at 65% and 3.0°C at 100%. As there's no change in pump speed between 65% and 100% I can't for the life of me understand why the Lobo has a bigger advantage at 100% - but it's possible that it produces more flow and is as such better able to make use of the airflow from the faster fans. That's just speculation though. If I have time later I'll have a look at the fluid temperatures to see if those can give a clue to this.
Stock power testing:
As you can see this graph is much, much messier. The main reason for this is that a 5800X at stock, especially one with PBO activated, will boost quite aggressively, leading to significant single core thermal and power spikes (we start seeing single cores significantly exceeding 10W), which bring up Tctl/Tdie readings as well. This is what the Aquanaut essentially didn't handle, and why I made my 120W profile, as it limits this behaviour. This makes this workload inherently more variable, and the results thus somewhat less reliable. The LOBO still comes out ahead, though marginally. This is also why no test was run at this power level with the Aquanaut - OCCT was already throwing errors at most of these settings (though never crashing or erroring out - it just show that my undervolt isn't entirely AVX stable at these power levels)
Clock speed analysis
Due to the messy data above at stock power, I decided to take a look at the effective clocks for each run and compare them. This turned out to be quite enlightening. I'm looking at effective clocks rather than discrete clocks, as they more accurately represent the actual clock speed of the core while performing work (even if this is more of a concern at idle, there's still a minor difference).
Looking at the stock power testing, like in the thermal results the LOBO again comes out on top - though this time the LOBO at the lowest setting actually manages an extremely marginal victory over the Aquanaut in the 100% profile. We can also see that across both coolers, clocks increase ever so slightly when moving to the next higher profile. There's one exception to this: the LOBO at 65% has the lowest minimum effective clocks of the set, and the highest max effective clock. This is most likely an outlier, but at least it demonstrates how thermals affect boosting, and how boosting higher at a given thermal and power limit might also entail slowing down more to compensate - leading to less clock stability overall. Still, these differences are very tiny, and the averages still scale as expected with cooling.
At 120W, things are much more controlled than at stock power. The overall results are the same: the LOBO at 20% beats out the Aquanaut at 100%, and with the Aquanaut tested at 20% we can see that it's struggling a bit, falling significantly behind the other results. In this comparsion that is - we're still talking about the highest average clock speed being 0.013% higher than the lowest. This is extreme nit-picking, and might be within run-to-run variance (though given the consistency of the results across extended testing I'm inclined to believe the readings).
120W testing:
As you can clearly see, at the 120W load, the Lobo is clearly ahead. In fact, the Lobo at the lowest fan/pump setting is just barely warmer than the Aquanaut at the highest. That's a clear victory.
More in detail:
| 120W load, Tctl/Tdie | Aquanaut 20% | Aquanaut 65% | Aquanaut 100% | LOBO 20% | LOBO 65% | LOBO 100% |
| Average °C | 71.2 | 67.9 | 65.9 | 66.6 | 65.2 | 62.9 |
| Max °C | 74.4 | 70.9 | 68.8 | 69.9 | 68.5 | 66.0 |
| Min °C | 65.6 | 61.1 | 54.3 | 62.1 | 60.0 | 60.0 |
The LOBO has a 4.5°C advantage at the 20% setting, 2.7°C at 65% and 3.0°C at 100%. As there's no change in pump speed between 65% and 100% I can't for the life of me understand why the Lobo has a bigger advantage at 100% - but it's possible that it produces more flow and is as such better able to make use of the airflow from the faster fans. That's just speculation though. If I have time later I'll have a look at the fluid temperatures to see if those can give a clue to this.
Stock power testing:
As you can see this graph is much, much messier. The main reason for this is that a 5800X at stock, especially one with PBO activated, will boost quite aggressively, leading to significant single core thermal and power spikes (we start seeing single cores significantly exceeding 10W), which bring up Tctl/Tdie readings as well. This is what the Aquanaut essentially didn't handle, and why I made my 120W profile, as it limits this behaviour. This makes this workload inherently more variable, and the results thus somewhat less reliable. The LOBO still comes out ahead, though marginally. This is also why no test was run at this power level with the Aquanaut - OCCT was already throwing errors at most of these settings (though never crashing or erroring out - it just show that my undervolt isn't entirely AVX stable at these power levels)
| Stock power, Tctl/Tdie | Aquanaut 65% | Aquanaut 100% | LOBO 20% | LOBO 65% | LOBO 100% |
| Average °C | 72.9 | 71.9 | 72.8 | 71.3 | 70.1 |
| Max °C | 78.8 | 76.5 | 78.0 | 76.9 | 74.4 |
| Min °C | 67.3 | 67.8 | 66.9 | 62.6 | 66.1 |
Clock speed analysis
Due to the messy data above at stock power, I decided to take a look at the effective clocks for each run and compare them. This turned out to be quite enlightening. I'm looking at effective clocks rather than discrete clocks, as they more accurately represent the actual clock speed of the core while performing work (even if this is more of a concern at idle, there's still a minor difference).
| Stock power, clocks | AQN 65% | AQN 100% | LOBO 20% | LOBO 65% | LOBO 100% |
| Max effective clock, any core | 4727,0 | 4737,4 | 4742,3 | 4768,9 | 4761,3 |
| Min effective clock, any core | 4601,7 | 4630,5 | 4628,5 | 4574,5 | 4646,2 |
| Average effective clock, all cores | 4675,9 | 4686,8 | 4687,7 | 4702,3 | 4718,9 |
Looking at the stock power testing, like in the thermal results the LOBO again comes out on top - though this time the LOBO at the lowest setting actually manages an extremely marginal victory over the Aquanaut in the 100% profile. We can also see that across both coolers, clocks increase ever so slightly when moving to the next higher profile. There's one exception to this: the LOBO at 65% has the lowest minimum effective clocks of the set, and the highest max effective clock. This is most likely an outlier, but at least it demonstrates how thermals affect boosting, and how boosting higher at a given thermal and power limit might also entail slowing down more to compensate - leading to less clock stability overall. Still, these differences are very tiny, and the averages still scale as expected with cooling.
| 120W, clocks | AQN 20% | AQN 65% | AQN 100% | LOBO 20% | LOBO 65% | LOBO 100% |
| Max effective clock, any core | 4677,6 | 4732,3 | 4716,4 | 4728,1 | 4733,0 | 4719,3 |
| Min effective clock, any core | 4520,4 | 4120,8 | 4545,2 | 4562,7 | 4562,2 | 4564,3 |
| Average effective clock, all cores | 4598,2 | 4622,7 | 4640,8 | 4644,4 | 4652,7 | 4657,7 |
At 120W, things are much more controlled than at stock power. The overall results are the same: the LOBO at 20% beats out the Aquanaut at 100%, and with the Aquanaut tested at 20% we can see that it's struggling a bit, falling significantly behind the other results. In this comparsion that is - we're still talking about the highest average clock speed being 0.013% higher than the lowest. This is extreme nit-picking, and might be within run-to-run variance (though given the consistency of the results across extended testing I'm inclined to believe the readings).
Conclusion:
So, taking everything into account, where does the LOBO land? As I said above, I like it. It seems like a very good product, with excellent build quality and good performance. In the latter aspect it manages to both cool better than the Aquanaut and deliver marginally higher clocks while doing so. It is especially impressive at lower pump speeds, where the Aquanaut falls behind by nearly five degrees.
This testing being done on a notoriously difficult-to-cool CPU makes this particularly impressive to me. As we can see, the thermal results are still rather hot - 70 degrees at max fan speed under water isn't cool. But on a less thermally dense CPU with its heat better centered under the cold plate, such as any Intel CPU? I would expect noticeably lower temperatures at the same power level.
A note on noise: I don't have any means of doing reliable noise testing, and there's too much noise coming from the window next to my desk for this to be reliable anyhow. Going by ear, there's no discernible difference between the LOBO and the Aquanaut with the same pump. If you're worried about a metallic pump mount vibrating or ringing, don't be. My motherboard and GPU VRMs' coil whine is far louder than the pump at 3400rpm, and neither are audible through open-backed headphones.
Now, my beta unit did have some minor QC issues. Hopefully the new paint shop will avoid blemishes like on mine, and there won't be any rusty springs included either. If the final units came with the appropriate hex keys included (or switched to metric head sizes), that would be fantastic. But other than that, this early production run is essentially ready for prime time.
Overall, the differences between the LOBO and the Aquanaut aren't massive, and the LOBO is more expensive. If you're on a budget, that's still a good option. But a strict budget and SFF custom loop building don't really go together all that well, which makes the LOBO look better - you're getting much better build quality, better performance, easier bleeding, a better mounting system and better aesthetics for your money. Would I recommend it? Yes. Easily.
Edit: added a note on the mounting screw situation after hearing back, see above.
Last edited:
Yeah, i realised that if you put it in 3 slot mode, with the gpu in single slot, in the outer slot, there is 40mm free space, behind the gpu.
With some 3d printed parts, i am able to stack x2 alphacool rack raditors behind the gpu.
Im expecting my 2nd radiator to arrive tomorrow, and my tx240 later this week.
From my testing so far, i was able to keep an i7 8700 below 80, with the 6 40mm noctuas at only 60%
If you want, i can post pictures whenever i have something to show?
I have serious doubts that that would outperform a dual rad 240 x2 set up...seems like more work for less performance when 2x240 rads have been done before.
Yep, it's an EK flow indicator with the rotor taken out. Definitely not my idea (sadly I don't remember where I saw it first, so I can't give credit), but it is indeed genius - works really well, especially with that foam inside (otherwise the small volume makes it likely that air bubbles will be pulled back into the pump). Though I would probably recommend going with something other than this EK unit (it was the only one I could easily get ahold of): it has a really thick back wall - nearly half the unit's thickness - which takes away a lot of what could have been usable volume. I've considered drilling it out, but I'd need a drill press and even then I'd likely crack it (and it would require tearing down the loop again), so I'm leaving it as is. Something like this Alphacool unit makes much better use of its total volume. Anything with three ports and where the rotor can be taken out should work well.
I'm totally aware that a dual 240 setup is safe and tested, but i'm not into the thought of having a radiator dangling freely in there.
I'm also a lot into tinkering, and a "proof of concept" build like this was right down my alley, so i wanted to try it.
I don't know if it may have been me that inspired you, but this is the flow meter I've used, compact and cheap! @Curiosity
Similar threads
-
- Article
