The NRZ N2 Waterblock Beta Thread

[thenrz]

Looking forward to it!

Final pre-production coldplate slot/fin design cuts in the morning. 42% more slots/fins and 100% deeper cut, and will create a entire new order of magnitude worth of surface area.

This will be the one going to you. My personal block just leak tested for almost 3 days. McMaster is saying Thursday, but that's guaranteed. I hope the argb cables arrive by then, too. I have high hopes it doesn't suck...! But, I like to keep it in perspective. This is an idea conceived, drawn, revised, programmed, setup, cut, finished, assembled, tested, troubleshooted, revised, redrawn, redone... all by one person all over a couple of weeks. I might have access to CNC machines and the tools, but the rest has been pretty much hundreds of hours to figure it all out. You guys here and on a popular CNC forum helped me put some of the missing pieces together, and I'm super thankful for that.

If I can figure out how to compete with the polished brands and the cheap stuff from overseas, I will push push push into making this a business. I have a ton of ideas, and no real way or time to bring them to reality. If I had a free month, a workshop, two CNC machines, a CNC lathe, and my credit card...

 

[Blaylock]

...and this lamp!


Jokes aside I can't wait to see the Beta product. I'll do what I can to put it through the rigors and offer every bit of feedback I can. If the ARGB cables aren't ready it's not a huge deal. They are definitely cool and will aid in pushing sales, but in the end if this thing isn't thermally capable you will need to revisit the ol' drawing board.

 

[thenrz]

Finalized (for now...)

So the design and thermal characteristics of the coldplate are now complete and my full and official beta model design is all wrapped up.

Today started with scrapping nearly $20 worth of already cut copper coldplates. Some modifications to the CNC had been made without anyone telling me, and I couldn't for the life of me figure out why things kept going so wrong. Not a great start to the day at 5am. Went over every little thing I did and came up empty, and then I was told that there had been some modal changes made to some of the parameters and that was that. I didn't pay attention, so that's my fault. So, I got a couple of first sides cut, finished the flips and side twos after work, and then set in motion my toolpath of all toolpaths to complete the slot and fin operation.


Specifications:

Fin Width: 0.015" (0.381mm) [I think these qualify as "microfins"...?]
Slot Width: 0.025" (0.635mm)

Depth of cut: 0.100" (2.54mm) [Approximately 1" of full depth of cut across the entire lateral surface of the active area, and a {heavily} truncated 1/2" fillet up to the beginning and end of cut.]

# of fins: 32
# of slots: 34


Tomorrow the mounting equipment arrives, the ARGB cables, a new shipment of copper 145, and my new SSD and memory (woohoo). This means that this weekend I will have assembled the first full and complete beta block ready to ship. Already have the packaging materials and all of that, so now things get fun. Work has begun on permanent fixturing for low volume production. My runs will start out at 10 a piece and feel things out from there. Before this happens, the beta blocks need to mature in the proper hands so that any and all issues can be hammered out before I invest in production equipment, materials, and all the other accouterments of manufacturing fun things.

Enjoy the photos... I know I am! Now to do some relaxing after a 13 hour day. These kind of hours have become far too common lately. Averaging 9-10 on the clock, and 2-4 off the clock. Sheeeeesh...


View attachment 209271

View attachment 209272

View attachment 209273

View attachment 209274

 

[trents]

thenrz, how's it going? We have heard from you in a couple days. Did the mounting hardware come and have you had a chance to check thermal performance?

 

[thenrz]

thenrz, how's it going? We have heard from you in a couple days. Did the mounting hardware come and have you had a chance to check thermal performance?

Worked pretty much the whole weekend on top of overtime all week... 65 hours on the clock over the last six days. The mounting hardware did come, as well as the the backplate today and things for my own testbed. I haven't had time to do anything sans work and my other side gig of fixing computers and electronics. Everything hit all at once, unfortunately... Going to be a really great paycheck, and my side business is really kicking in with everyone so bored at home and getting their computers upgraded/fixed. (I'm sick of wiping down cases and laptops with lysol...)

I'm going to get everything installed into my new rig tonight. Can't boot it yet, as I'm still waiting on memory and a new nvme drive, but newegg is saying tomorrow for that. I'm going to plan for Wednesday since everything has been a day late, except for McMaster of course, due to the current covid situation. At least I will be able to play with the different springs, screw lengths, washer combos, and orientations this evening. I will get the new liquid loop all sorted out, too. I will be posting a craigslist/ebay build in the one thread sometime early next week once the new fans and controllers arrive. I was able to get an i7 7700k, z270 asus rog mobo, gtx 1070 founders edition, a brand new in box 750w corsair fully modular psu, and a brand new in box corsair obsidian 750D airflow case off of craigslist. I have just $295 into all of that hardware. I'm adding 16gb of ddr4 3600mhz memory (or 3200mhz if it won't xmp to 3600) and a 500gb pci-e 3 4x m.2 ssd from newegg. All together that's $430. I also have an entire high end loop's worth of parts from ebay/craigslist, excluding waterblocks, for which I paid just under $140. nexxxos monsta 240 (scratch and dent, brand new otherwise), alphacool 1500lph pump, 250mm tubular reservoir, and a brand new unopened primoflex flexible 1/2" ID tubing bundle. $570 for everything listed above with shipping. Oh, and ordered some RGB stuff just to see about developing an open source addressable controller at some point. We sorely need something unified without the proprietary nonsense. That's well after the waterblock business is off the ground. Could easily swap in a Ryzen CPU/mobo when the time comes. Can you tell I've been busy...

Will update with some photos this evening or tomorrow. Thanks for worrying about me

 

[trents]

Take care of yourself, dude.

 

[thenrz]

Hey all, been a wild week. Even though Ohio is under a stay at home order, the company I work for is considered to be essential so I am lucky enough to continue working. We have had several jobs come through that are for the medical industry that are obviously really important for the current situation... So needless to say, things have been quite crazy around the shop. I haven't been updating simply because I don't have time to rest and sleep and my project has had to take a temporary back seat.

But... The last parts for my testing computer, namely the m.2 ssd and the memory, show as delivered finally. The package got routed to Indiana when it was already within 30 miles of my house. It sat Indianapolis for 4 days until it finally started moving again. The block is mounted in my system, leak testing for 3 days now, and all the fun stuff is done to the case. I'm sort of side developing an open source argb controller and software, but that's on the second back burner.

I actually got out on time today, and will get the system running this evening... Or so is my goal anyway. I made a hardware revision to the coldplates, namely modified the gasket area. I was using the acrylic frame as the third side, but I honestly didn't trust it long term. There is now a full 3 sided gasket groove all in one piece. That means I need to cut two more to have the sets for blaylock and my other guy to test. Not ideal, but I'm feeling great about the design...but not so great about scrapping three pieces of copper.

Pictures on the way...

 

[Blaylock]

Glad to hear you all are staying busy. Next week I'm being forced to take vacation time as our plant is taking part in manufacturing medical grade face shields. Only employees tasked with this emergency project will be allowed to work, even from home. I digress.

I'm glad your are taking the time and effort to make potential improvements even at this stage. It seems you are committed to making the best product you can.

I'm looking forward to getting my hands on the beta version and seeing how it stack up to others.

Stay safe.

 

[thenrz]

Glad to hear you all are staying busy. Next week I'm being forced to take vacation time as our plant is taking part in manufacturing medical grade face shields. Only employees tasked with this emergency project will be allowed to work, even from home. I digress.

I'm glad your are taking the time and effort to make potential improvements even at this stage. It seems you are committed to making the best product you can.

I'm looking forward to getting my hands on the beta version and seeing how it stack up to others.

Stay safe.

I think a big part of things for me right now is not being able to adhere to my timeline. It's frustrating... Obviously there is a lot going on in the world, and I try to frame my personal situation as it is: a whole lot better than a majority of people. I'm still working, copious overtime in fact, have my health, and am still able to do what I want to do on a limited basis.

I just got back from microcenter, donning a mask and gloves, with a ryzen 3600x and a gigabyte x570 motherboard. The mobo was on sale, the ryzen chips have been through a price cut, and microcenter has a $20 savings when buying both the CPU and motherboard. Got the parts cheaper than anywhere else which is excellent. The 7700k setup has had issues and this is mostly due to it being used. I haven't bought a new setup in ages, as I simply haven't needed to do so. I will work on it here and there as it will make a solid server type setup in the future. My one big regret is getting a gen 3 SSD since I didn't think I would be going ryzen any time soon... Oh well, can upgrade later. Going to switch out the mount on my complete test unit, get the system together, leak test, and then report back. The convex surface on the coldplate was simply perfect for Intel and I hope it also applies to AMD. One could tell by the feel and the paste pattern that the level of contact was superb.

Anyway, stay safe everyone, stay healthy, and keep yourself and your family protected. I felt ridiculous in my mask today, but I also know that when I go to drop off groceries and stuff with my parents that I am not going to be endangering them. It's important for everyone to do their part in this.

 

[thenrz]

A quick update:

The only known beta version of the N2 block in existence is currently running on my brand new Ryzen 3600X system. (sounds awesomely cool phrased like that...) Current idle temps are at 26C and full load in prime 95 does not go past 51C. The current ambient temperature in this room is 22.2C. So far, so good. Really quite pleased with the performance. Hammering out some issues with the RGB, but I am thinking I have two of the wires flipped around and the Arduino setup was able to work regardless for whatever reason. It's really the only missing part of the package at this point. I kind of became obsessed with a mechanical keyboard they had on sale at microcenter, so I ended up picking up one of those as well. Boy oh boy is each and every keystroke satisfying now.

So a very good day, and I will update with pics tomorrow. I am tired, and I need to shut this bad boy off and go to bed. There's a good chance tomorrow will be a 10-12 hour day at work.

 

[EarthDog]

Cool!

Which version of p95 and what test? 51C is incredible, but something tells me that wasnt small fft or a version that is using fma/avx?

I havent been following too closely here... what radiator is this attached to?

 

[trents]

Yes, mechanical keyboards are awesome! Just got one for my wife and a new one for me this past week.

thenrz, tell us about the rest of your cooling setup to help us get some perspective on those performance numbers. What rad, fans and pump are you using?

 

[thenrz]

Alphacool 1500lph pump (I forget the specific model #) running at 12 o'clock which is mid power, Alphacool Nexxxos Monsta 240 in a push/pull (the pull has one fan up from dead center due to the case), random 200mm reservoir, and all 1/2" ID soft tubing. Fans are some Thermaltake 120mm RGBs that came in a kit for the push (CFM close to 60, not sure on static pressure) and the pulls are the Corsair joints that came with the 750D airflow case (again not sure on numbers, the entire system was put together this weekend). I don't have specific fan RPMs set yet.

Prime 95 was an old version I randomly had on one of my external drives that some of my "necessary" desktop aesthetic images and stuff reside. So, it is not a complete test by any stretch, but I knew it would get me some quick information. I want to ramp into this rather than plunge because the entire system is still completely fresh. Also, my 3600X is also not idling properly, the cores won't go below the rated GHz level even at complete Windows idle, so the thermal capacity of the liquid system is constantly being worked more than what I would consider normal. I want all of this to be normalized before I profess any sort of hard numbers. All I know is the block is working, has really solid idle numbers, has a pretty surprising load number, and I am just beginning here. The 51C number was the max number in HWMonitor by the way, I wasn't watching Prime run as I was cleaning up my gigantic mess from the build. I am trying to do this without getting the absolute ideal case for the block, so my numbers are more aligned with what would be a real world use case.

 

[trents]

It would seem to me that at some point you need to swap out your crafted water block with some other commonly used commercial blocks while keeping other components the same. That way, you can arrive at some hard data that says to people, "compared to block xyz my custom water block lowers temps by x amount."

 

[thenrz]

It would seem to me that at some point you need to swap out your crafted water block with some other commonly used commercial blocks while keeping other components the same. That way, you can arrive at some hard data that says to people, "compared to block xyz my custom water block lowers temps by x amount."

This is the exact reason why I have testers that will be receiving blocks as soon as I can get them out. With everything going on in the world and work, I haven't been able to stay on my schedule. I had hoped to have both my test kits out by last week... and I'm just now utilizing my own system. It is what it is, though. This period without machine time has given me the opportunity to start thinking ahead on other products around which to build this business. The N3 block is in the works and drawn, as well as a unique distro plate system, a fill/drain port system, and a new kind of acrylic plate tubing system. I won't be a one trick horse, but I also understand the importance of getting each and every product absolutely perfect to my ridiculously high standards. Given the fact that I don't have my own workshop/factory to support these ideas, I have to do what I can in the future hopes of being able to build a business.

 

[trents]

Yes, I just hope your testers have a consistent approach to remove as many variables as possible. I'm thinking you may do well to give them some guidelines. I'm sure you have seen how cooler and thermal paste reviews give results that are all over the map.

 

[thenrz]

Yes, I just hope your testers have a consistent approach to remove as many variables as possible. I'm thinking you may do well to give them some guidelines. I'm sure you have seen how cooler and thermal paste reviews give results that are all over the map.

You are absolutely correct here. This is why I have selected my testers very carefully. Blaylock and I have discussed things, and he's going to be my devil's advocate. I want to know the bad. I need to know how to improve. The fact is I am shooting for the best. It may end up the overall design needs modified. It may end up being pretty damn good... What I can say is that I have put a ridiculous amount of hours into making the N2 block high quality and high performance while making it manufacturable within my scope. That's the key for me right now. In a perfect world I can invest, or ascertain seed money, and be able to build a real manufacturing base.

But yeah, my testers are prepared to do some real work here which is incredibly appreciated.

 

[thenrz]

Here are some photos of the setup and block. Unfortunately, I am not sure what I did to the LEDs... they worked with the Arduino the last time I tried, and now I can't get them to turn on for the life of me either by the addressable motherboard headers or by the Arduino. I even made a neat little crossover cable. Oh well, I know it works, and I know it works with the regular addressable protocol, so I'm going to call that a headache of the beta stage.

View attachment 209445
View attachment 209444


It's weird how the temps shot up straight to 66C, and then sat at 51C for about 15 minutes. Now I am at 75C after about an hour and a half straight. Obviously something within the test changes after that initial timeline. I don't know much about Prime other than I have seen very robust systems get their butts kicked.
View attachment 209448


This is much more akin to what I expected to see from a hard prime95 run.
View attachment 209449


After stopping the test, I timed the temperature decrease back to baseline: 7 seconds back to 31C and holding there. Ambient in here is 76F by the way. This room gets really hot when my fiancee decided to keep closing the vents in the bedroom.

So, I've learned that my block doesn't blow up/melt/disintegrate/stage a coup/get snarky/otherwise conscientiously object to being a waterblock. I have also learned that throwing a stress test at the thing shows that my thermal design is actually pretty good all things considered, as it took Prime to get me over 51-52C. It seems that at some point, the ability of the system as a whole became a bit too heat saturated to continue to operate in homeostasis.

My current rad fans aren't great at all, and I want to improve those as soon as I can to remove the weakest part of the system. I also am only running the pump at mid power, so more flow would for sure help. I will continue to run tests and change a single variable as I go to see what happens. A basic "stress" test in CPUz wouldn't get me over 52, but a constant and thrashing assault of power draw over 102W and every core running at 4.391MHz for a long period of time put me where I needed to go to see just what the block would do.

I have spent too much on my computer these last few weeks, so I need to chill until the weekend so my fiancee (who is the coolest woman in the world and actually takes an interest in all of this...) doesn't have to give me the "you're spending too much on your nerdy stuff" eyes (even those eyes are irresistible, though).

Thanks for staying tuned everyone. I will do what I can to keep some tests going the next few days, and I really hope that I get some more machine time ASAP as these rush medical and transportation jobs at work start to finish up. This would allow some other numbers to be created and to truly begin the task of augmenting the design to achieve the best possible performance case. If it ends up that things cannot get much better and that the end of the design has been reached, then so be it. I will drop N2 and move to N3 and will have gained some seriously invaluable experience along the way, but I hope that some tweaks to the flow design and perhaps cutting another .025" or .050" (maybe wishful thinking for .050") deep on the slots would tend the performance in the right direction. What I don't want to give up is the very open flow design, but perhaps some restriction would benefit things here... hmm... this is when I am glad that the design is modular and it will be really quite easy to cut a pressure plate with a reverse of the slots/fins to a certain height.

I'm tired and rambling. Good night everyone

 

[EarthDog]

Ok... a couple of things...

1. What you saw in p95 is expected. It goes through different size ffts. Did you run small fft? What version of p95?? Use the latest version...
2. The loop needs to normalize/saturate, yes. For water loops, at least 30 mins to do so.
3. What is your water temp at the end of the test? The beginning?
4. Makes sense a stress test is needed to get much over the 50s on that cpu under water. I dont go over over 60C with a 4.5 ghz 16c/23t 7960x which is using a lot more power (3x120mm rad and kryos next block)
4a. Testing like this is how its done.. worst case
5. So long as you flow rate is above 1-1.5 gpm going running faster yields little change to temperature in most cases.
6. If you are testing the block, you can't really swap out parts and test.. the test system needs to stay the same for the datasets you are gathering.

Looking good so far... keep it up!

 

[thenrz]

Ok... a couple of things...

1. What you saw in p95 is expected. It goes through different size ffts. Did you run small fft?
2. The loop needs to normalize/saturate, yes. For water loops, at least 30 mins to do so.
3. What is your water temp at the end of the test? The beginning?
4. Makes sense a stress test is needed to get much over the 50s on that cpu under water. I dont go over over 60C with a 4.5 ghz 16c/23t 7960x which is using a lot more power (3x120mm rad and kryos next block)
5. So long as you flow rate is above 1-1.5 gpm going running faster yields little change to temperature in most cases.
6. If you are testing the block, you can't really swap out parts and test.. the test system needs to stay the same for the datasets you are gathering.

Looking good so far... keep it up!

1. I figured as much... Yes, ran small FFT, or whatever said the most heat and stress in the options. Wanted worst case.
2. I let the system sit at idle for around that kind of timeline before testing. When I say idle, I mean installing some software on the new Windows 10 Pro setup.
3. Right now, I don't have a way to read liquid temps. This needs to change. I'll see what I can find this weekend... I'm so far behind on all this new fancy liquid cooling stuff. I'm still somewhat stuck back when we were using 120V pumps and 7/16" OD tubing.
4. I am wondering if my arc angle on the coldplate is correct for Ryzen. I think the next copper I cut will be dead flat on the bottom. Something tells me that heat conduction from the IHS to the copper isn't as it should be. Right now I'm sitting at .005" of arc height from edge to center. From the information i've ascertained, I calculated this to be correct. That and I am still working on the mounting situation. This is what testing is for...
5. Again, need to measure this somehow.
6. I am going to change out the fans to something more applicable to performance and leave it at that. The current fans do not possess the static pressure to flow enough air over the larger depth of this radiator.

I am going to draw up a couple of different pressure plate options tonight. I have a feeling I need to modify my flow input to the coldplate. I am sweeping the water over the entirety of the surface area upon entrance right now where I think I need to focus things first. This is the beauty of my situation... I can make whatever changes I want with the only restriction being the size of my material.

Thanks for the input!