S A N D W A S P [ HEDT | Soft Tube | PC-O11D XL ]

[HM-2]

Yep, definitely a RAM issue it looks like. Whether one if the sticks is bad or the IMC is unhappy about dealing with 8 heavily overclocked sticks I don't know. Back down to 16GB for now and giving it a bit of stick to make sure it stays stable. Interestingly I seem to be getting slightly less heat from the same clocks and voltage. Previous AVX torture testing I would hit ~89°C on core 5 but now we seem to be stabilized around 85°C. So there should be some decent headroom in the chip on a proper cooking solution.

 

42x all core multiplier

36x ring ratio

DDR4-2400 at 3000MHz

1.265v VCORE

1.900v input

VDroop/LLC at 75%

Rest handled auto.

 

What's the collective noun for a group of RAM sticks? I like the idea of "a Family of RAM"

 

 

Resleeved 8-pin. 140mm rad, connector location and AIO cabling makes clearance tight so not the neatest job, I'll rejig it when the rad comes out.

 

Also tidied the 24-pin and 6+2s as best as I could. Might change routing here when the loop goes in. 

[HM-2]

More bits on order, this time a motherboard tray cover from Cold Zero

 

[HM-2]

More Watercool parts on the way.

 

 

Also take 2 of the IO cover and a 3D printed pump mount should turn up tomorrow. This time SLS, as I got a ridiculous price on them.

[HM-2]

Trident Z RGBs now in:

 

 

I've tested them at BLK100 3000MHz CL14 and BLK133 3250MHz CL15. Despite the latter having higher core clock speeds, memory speeds and cache multiplier, the latter runs hotter and performs worse in synthetic benchmarks. Couldn't get it to boot at 3400MHz.

 

Also received the two bits I had done in nylon via selective laser sintering.

 

> Rerun of the I/O cover

> D5 pump bracket

 

 

Both look pretty incredible, especially given they were less than £35 for both of them.

[HM-2]

More German engineering

 

 

Heatkiller RTX2080 GPU block and backplate. Not sure what I'm going to go with the RGB cabling.

 

Currently painting the IO cover up. But ran out of 100% Yellow Montana Gold.

[HM-2]

I/O cover fitted.

 

 

It looks...okay.  It's definitely too long and probably a bit too wide as well, might take a hacksaw to it and shorten it a bit if I can be bothered.

 

I'm far from the best painter in the world but the lines are reasonably crisp. I think if I was redoing it I would probably print the black top piece and yellow underside separately with dowels, then glue them instead to get a nice crisp division. Given I've got some GPU supports to print as well I might have a go at Mk3 done via SLS again, as the finish is incredible compared to the traditional 3D printing of the first iteration. But possibly with a slightly more judicious approach to measuring so that I can get a cleaner fitment. 

[HM-2]

After months of zero stock, Watercooling UK have finally got a shipment of HardwareLabs Black Ice Nemesis radiators in. 

So one GTX360 and one GTS360 ordered.

 

Everything is now either here or on order except:

• Pump
• Fittings
• Tubing
• Fixtures for distro plate assembly

No word from the machine shop who are doing the distro plate at the moment, will chase up at the end of the week. I know he'd been awaiting Acrylic stock which was due in last week.

[HM-2]

More shiny-shiny:

 

 

 

They're bloody heavy things!

[HM-2]

On 5/19/2020 at 9:58 AM, HM-2 said:

I'm not expecting to finish up until probably mid June to July. 

This is looking rather optimistic at the moment...

 

I've got some lovely pictures of the custom alu workholding piece my CNC machinist has made but he's fitting it around other tasks so it will be at least a couple more weeks.

 

 

Also doing a "Mk3" run of my IO plate, but properly measured up this time so I won't have to take a hacksaw to it. Will add a grab from Fusion 360 in a bit.

 

[HM-2]

Well that's some of the fittings arrived. Still need another 6x 45 degrees, 16 compressions, drain port and a few blanking plugs

 

 

8 45 degree fittings

2 90 degree fittings

Temperature sensor

 

All in matte black.

 

Currently considering whether or not to paint my Heatkiller bracket black too.

[HM-2]

Had some unfortunate news. The guy doing my distro had not succeeded in his attempts despite several goes and due to a backlog in existing work has had to admit defeat. I'm now moving onto a new supplier, recommended through the Bit-Tech forums, who has considerable experience in building these plates. Need to make a couple of minor changes to make toolpathing easier but hopefully that should be it.

[Hasle]

Hope the new place works out! Gonna look so clean with that motherboard cover in place - when are you fitting it?

 

(It's Az from GTAF, btw! )

[HM-2]

38 minutes ago, Hasle said:

(It's Az from GTAF, btw! )

Hey Az! To be fair it's kind of your fault I came here with your Juicebox build (and Trooper before it).

 

I did consider doing a partway swap where I moved some bits into the system ready for the loop but given I need to take a Dremel to various bits of the case to fit the distro plate I think it's safer to leave fitting the motherboard cover until final assembly. 

 

I'm hoping I can get everything sorted and ready before the end of July 

[HM-2]

More 3D printed parts from 3dpeople, again nylon SLS:

 

GPU support brackets, sized wide enough to enable me to use two screws to brace the GPUs and still provide spacing for two cards. Huge thanks to @kkpatel87 for the initial models from which these were printed.

 

Run 3 of the I/O cover, this time broken into two components for ease of painting (bottom yellow, top black). It's got a near-interface fit with the IO ports, just needs a ~2mm shim at the bottom.

[HM-2]

Finally nearing the end in terms of parts acquisition.

Big bag o' fittings:

 

Spent a bit of time prepping, painting and then matt lacquering some of the 3D printed parts. GPU supports and pump bracket shown below, generally happy with the finish on these (not perfect around the screw holes but very flat elsewhere). I/O plate needs a little more work but already a world away from the previous run.

 

Pump should arrive tomorrow along with leak testing bits. Awaiting completion and delivery of distro plate. Only things I need now are fluid and tubing.

 

Having given it some thought I'm going to reuse the existing Corsair fans for now until the Noctua NF-A12x25 Chromax are released. Don't see the sense in splashing ~£120 on fans now only to want to upgrade them in the future.

[HM-2]

Pump test fit in the support bracket:

Lovely. Even came with the right size O-ring for the distro plate.

Now I know it fits and works, I can void the warranty by drilling a bigger cable hole in the back and heatshrinkless sleeving the cables

 

Also got to a point where I'm happy with the I/O cover:

It's not a perfect colour match for the board, but it is for my fans and cables, so maybe I'll paint my heat sinks to match.

You can't really tell with the side panel on anyway.

[HM-2]

P U M P  |   S U R G E R Y

 

Well that was exciting scary.

 

 

I've done little bits of cable modding here and there but never on anything like a pump before. A learning experience all round.

 

Cables are heatshrunk on the inside, the power/ground cables will be done heatshrinkless and I did the PWM cable heatshrinked just because it's probably going to go into an extender. I couldn't finish the power/ground cables today but I'll come onto why later.

 

 

Heatshrink isn't great on the PWM cables because I used a piece too small initially and burned through my sleeve when heating it so had to redo it.

I got quite lucky ion that the soldering job in my pump is rather good; some of the ones I've seen pictures of have been terrible.

 

Stepped bit did an okay job of the cut, I cleaned it up with a scalpel and sanded the edges a bit.

 

All three cables together. VERY tight fit getting the pump housing back on, I really hope it's not putting pressure on any of the soldered joints but it seems okay.

 

I also cut down a couple of cable combs for the 12v/ground cables.

 

Now I have a bit of a quandry. Initially I was going to use affix another SATA connector so I retain a degree of flexibility but as far as I'm aware I'll need to solder the connectors for this. My soldering skills are nonexistent, so I'm now erring towards using ATX pins and wiring it straight into my PSU. That means I'll have to swap the connector if I change PSUs, but seems like the easiest approach. However I've mislaid my pack of female EPS pins so either way it'll hae to wait.

[HM-2]

So I ummed and ahhed about whether to graft on a PSU connector or another SATA one. In fact I initially tried to do a PSU one but my ATX/EPS pins are terrible eBay specials and I couldn't get good crimps even with my new crimper- the rear wings are too short and you don't get a proper heart around the insulation, just a mangled mess. So I ended up going with SATA which meant breaking out the soldering iron.

 

My SATA connector was a cheap one that was definitely not sized for sleeved cables. So I had to take a drill to it and tidy up with a file. I also filled the unused holes with plastic putty and then painted black with a paint pen. It's not the best job in the world but you'll never see once it's in the system.

 

 

I'm not going to subject anyone to the quality of my soldering. I've not done it for the best part of 20 years and never on anything I actually cared about. The 12v is terrible but at least reasonably sound. The ground was actually okay. I clipped the unused pins off the connector and filed down the nubs to limit the likelihood of my incompetence causing a short.

 

Threw on a little blue heatshrink to cover them over inside the connector body (ran out of my usual black and my set only seems to have it in stupid sizes) and then glued the connector together.

 

 

 

I was convinced my ham-fistedness would have killed it but no, it works absolutely fine on my test PSU.

[HM-2]

Distro plate milling begams today. Had to make a few adjustments so my machinist is happy with the design-

• Reduced some channel depths around 1.5mm in a couple of places to reduce likelihood of cracking when tapping M4 threads
• Moved a couple of screw positions for similar reasons (proximity to channels or other screw holes on the opposite side of the plate)
• Dropped from M5 to M4 for the pump bracket because of concerns around thread engagement depths in 10mm holes.

The latter one is a bit of a pain as it might require some re-engineering of the pump mount. I'm hoping I can use an M4 bolt with a larger 9.5mm head and use the countersinks to ensure alignment, or finding a stepped-down bolt with an M5 shaft and M4 threading, but if I have to reprint the pump bracket it's not the end of the world.

 

All other parts are here and ready now including some spare pump seals and EK-ZMT in 10/16 matte black.

[HM-2]

C U S T O M  |  F A N S

 

 

Initially my plan had been to wait for the release of the NF-A12x25s in Chromax Black but those have been pushed until 2021 now. I thought I could hack my existing SP120s but despite being "quiet editions" they're anything but and really grate.

 

Enter the NF-P12 Redux.

 

Picked up three of these for a little over a tenner each with the intent of experimenting on them. They're one of the few Noctuas on sale that actually still come apart in a non-destructive way (sort of, more on that later) and they're still pretty top drawer for radiator use. These are 1300RPM PWM models so nice and quiet.

 

I followed a couple of different guides for disassembly and ended up with a hybrid approach of popping the fan blades off without any kind of heating and extracting the motor/stator assemblies with a plastic spudger. However I didn't realise my cheapo one was damaged which left some marks on the PCB- thankfully despite their ugliness no damage to the traces and everything works as intended. New tool promptly ordered...

 

 

I used the rather excellent Plastikote matte black to spray the frames; despite being extremely careless, doing absolutely no prep or primer, and the first coat in the rain, they came out brilliant.

 

 

Unfortunately I forgot to put a drop of machine oil in one of them and when I went to pry it apart again, this happened:

 

 

Bugger.

 

 

I've tried gluing the blade back on and will test for balance, but strongly suspect 'tis fucked. Oh well, at a tenner I don't really care. Already ordered two more...

[HM-2]

That's three complete and working fans.

 

 

They all run at +/- 3% of their 1300RPM design speed so clearly nothing about the disassembly process has negatively affected their performance.

[HM-2]

Work continues on distro plate, now including an LED channel and some changes to screw layouts, drain port and pump recess.

 

 

 

Example of some of the previous work done by the same machinist:

 

[HM-2]

D I S T R O  |  P O R N

 

So...

 

 

Distro plate has been milled. To say I'm over the moon would be an understatement, after months of waiting it's happened very quickly.

I'll let the pictures do the talking.

 

[XGoodGuyFitz]

that looks amazing. way beyond my skill leevel lol

 

[HM-2]

7 hours ago, xXFitzXx said:

that looks amazing. way beyond my skill leevel lol

 

Thanks- a lot of this project has been trial and error and I've been really trying to push my comfort zone by doing lots of things I've never really done before- custom 3D printed and CNC parts, component modding etc. 

 

You never know what your skill level is until you try something, and you never improve without practice . You just have to be prepared to mess up a few things along the way...