This is an overview of already finished projects which took me the most effort and time. For the most part I still have all project files (paperwork and digital files) but not really an instruction on how I finished those 😄.
Building a Watercooled PC
Date: Late 2020
Probably anyone who is kind of into PC building of some sort wants to have a completely hard-line watercooled system. As for that I also attempted myself with this challenge, without having any experience whatsoever in watercooling. I had still my 6 year old setup at this time and thought that I could easily upgrade it with a complete hard-line build.
The Build
- Custom hard-line Watercooled
- Case: Corsair Obsidian 750D
- Mostly EK components (CPU block, radiators, pump-reservoir combo)
- GPU Block - Bykski, from AliExpress
As for the outcome I think its pretty decent for my first watercooled setup. I also included custom-sleeved cables and some modifications to the case to fit all the components. Because the case itself wasn’t really made for total custom watercooling, I had to modify some parts of it. For one I constructed a custom 3D-printed mounting plate for the EK FLT 360 Pump-Reservoir Combo. It mounts to the back where the HDDs used to live. As I didn’t wanted to also spent money on upgrading my age-old HDD mass storage, I needed to move them.
Because Corsair didn’t provided any PSU cover for this case I build one myself with some metal sheets and some 3D-printed parts. It covers the whole bottom where the PSU lives and the cables run through the back of the motherboard tray. At the end it came out pretty nice, because I could then also hide some of the cable mess 👍.
On the inside it also looked stunning with the fittings and hard-line tubes. The CPU block and GPU block were also quite nice and worked flawlessly. Later I just realized that my RAM run quite hot because my GPU was radiating a ton of heat and the airflow in the case was a bit restricted.
I configured it with two intake fans on the front with a 280mm radiatior and two on the top as exhaust with a radiator. The fan in the back also stayed there from before the watercooling. So in total the airflow wasn’t that great because the fans from the front were also a bit obstructed from the reservoir.
As for the CPU and GPU blocks they worked great. Especially the EK one was really nice (at least in 2021). The GPU block also worked really well. When I ordered it I checked the non-RGB variant but for some reason they sent me the RGB variant instead 🙃. The block was from Bykski which I didn’t really know until then but it seems that they are a pretty solid manufacturer in the Chinese market when it comes to watercooling. When you trust online communities they are also quite reliable and a good choice for price-sensitive builds. They were also the only ones still selling blocks for my (at this point) old Radeon Fury Nitro.
Conclusion
For my first watercooled PC build I had lots of fun and also learned a bunch of stuff. I’m also quite surprised that everything worked on the first try. No leaks, no broken components or something else.
But don’t be fooled! I prepared myself quite a bit before I tackled this journey. I researched a lot on possible configurations and possibilities how to layout the system. Because I noticed early that the case had some major drawbacks when it comes to watercooling, I modelled the whole thing with parts and cooling in Fusion 360. Only after that I was confident that everything would fit.
For the building process itself I looked into many different tutorials and build advices from experienced people (mostly on YouTube). Anyone interested in going down that route I would recommend to check out some of the following channels:
- Probably the best-in-class tutorials you get on watercooling
- Tons of experience
- Has a video on probably every edge-case, when it comes to watercooling
- Many tech-related content; Some also in respect to watercooling
- Good PC building tutorials
- You probably already know them
Alex Banks, formerly @ bit-tech
- Without a doubt, one of the best PC modder known to men
- Has also great tutorials on scratch-building parts (or even whole projects)
- If you want some inspiration on what is possible, check out the channel
- Also, the video quality is 👌
- As bit-tech shut down, he has now his own business going
As for my system, it ran really well. But as unproblematic as the build process went, some issues appeared after a bit of time. First of all, if you want to still use the Corsair Obsidian 750D, buy the Air-Flow Front. The solid one I had from the beginning has terrible airflow.
When it came to maintaining the system it worked quite well. I included a drain port which helped a lot but I also realized that you should plan that ahead in your design. Getting every last drop of water out was almost impossible and took some “shaking” the case too 😄.
Mentioning maintenance, I noticed later (after a year or so) that some pieces where floating around in my reservoir. Because the looked too sharp and crusty to be some growth-related residue I suspected some issue with some nickel plating.
Following every advice I used only nickel or brass parts inside the loop itself. The tubes were acrylic as the other plastic parts. I used an inhibitor from EK, one of their premixes for a pink water-color. After a quick look under the GPU I already saw that this thing was the culprit.
After the disassembly I also took a look inside the CPU and GPU block. The CPU block was pretty clean, it just had some erosion marks around the fins and the jet-plate. But the GPU block was probably, what you would consider, done. Some of the water was creeping under the acrylic lid and had left some color particle residue in those places. That discolored the nickel on the surface where the acrylic would contact the block. It looked a bit ugly but wasn’t a major problem.
The bigger problem was that the plating inside the block was eroding as well. It came of in the runways of the water and the fins. To my looks it seems not that it was the erosion of the water (or the coloring particles), it looked more like it just delaminated from the surface. I think that here the plating itself wasn’t that good of a quality and that the surface wasn’t prepared well enough before the plating process. So probably after some time and usage little bits of the nickel plating came lose. After that erosion took its way and lifted off even more and distributed the nickel pieces inside the loop.
Because of that I would probably not take another one of the blocks from Bykski. Maybe they have some good stuff as well, but in regards to nickel plating I wouldn’t recommend them.
Custom-Built Nixie Clock (with IN-14)
Date: May 2019 - June 2019
As I already tackled some smaller electronic projects I also wanted to have something more substantial that I could use every day. Having just finished watching Steins;Gate (btw, great Anime, would recommend 👌), I took inspiration from the Divergence Meter and found out about those number tubes. Back then I knew some tubes existed with some sort of indicators, like Magic Eye tubes, and also VFDs. But especially the Nixie tubes and their derivatives sparked particular creativity. And for that a clock would be most suitable.
After researching into the topic I found the IN-14 to be the most suitable (and affordable ones 😛) for my project. Getting these was also quite easy because they were used abundantly in the USSR and are also quite common nowadays in Russia. Ordering them on eBay was a bit of a hassle because good ol’ DHL screwed up the delivery 👍, but eventually I got my tubes. I think today I would probably not buy tubes from Russia. Besides the obvious reason these tubes seem to be used in many military applications as well. If they would get stuck in customs because of import regulations, I think the discussion with the officials there would not be that beneficial for myself 🙃.
The brain of the operations is a simple ATMega328P, basically used as Arduino (because that was the only µC I knew back then). The PCB was also fully custom designed and order via JLCPCB, which also included some design errors that I needed to fix afterwards. Driving the tubes was probably the most problematic part.
Figuring out the controlling with the ATMega and the tube drivers (K155ID1) was quite easy. I added a DS3231 RTC for time tracking along with two buttons to set the time. To save pins and not having to multiplex the tubes, I utilized some shift registers (HEF4015B) as decoders for the tube drivers. For driving the tubes I didn’t wanted to use a simple boost-converter circuit for the ~170 V. For stepping up voltages from quite low values (12 V in my case) to such high levels a better approach is to utilize flyback transformers.
Most of the flyback transformers out there a dimensioned to drive rather specific application, mostly CRT TVs and electron tubes but also PoE stuff. There wasn’t a prebuilt configuration for Nixie tubes but I found someone who probably had the same idea before me. tonyp7 engineered a pretty nice PSU for Nixies driving them with a flyback topology. I bought one of those on Etsy although they are know already sold out. Good thing is, all of the schematics and BOM are open source and available on the GitHub page.
Finishing off the design of the clock I used a bit more of a “fancy” approach. I chose a combination of ebony and brass for the base of the clock with the tubes open. The black of the wood when its finished with some linseed varnish and the look of the metallic brass look amazing in combination. It resembles a little bit antique furniture where ebony is used in combination with gold inlays. The glow of the tubes housed in the case looks stunning.
DIY-CNC
Date: Late 2019 - Summer 2020
This is, with no doubts, my most complex project yet. Already summarizing it here is too much. Some people learned a language or some instrument while the pandemic lasted but I chose to built myself a working CNC router. To make things even more ridiculous my fields of studies were more in electronics, not machinery. The basic mechanics that I knew helped a lot but I still needed to read through some more stuff to build this thing.
It actually started out with an idea to construct a router with a budget under a 1000€ but I quickly realized this wouldn’t suffice. I modelled everything in Fusion and came up with some calculations dimensioning the mechanics of the machine. After that I realized that I wanted something a bit more sturdy and capable which would unfortunately also raised the budget.
It ended up costing about 3500€ to built the machine with everything built from scratch completely custom. I constructed the machine myself using not really expensive tooling. Basically I just had an electric drill, a DIY drill press for that, drill bits and tons of screws. The most precise tools were my calipers, a feeler gage, a precision square edge and a granite surface plate. The electronics mostly used crimped cables and a little bit of soldering.
This thing works (for the price) really outstandingly well. Only the linear rails on the X and Z axis on the portal are a bit more wobbly than I like. Also for the invested materials it is shockingly precise with a reverse play of under 50 µm if adjusted. Cutting aluminium, plastics or wood is no problem. Also more tough materials like brass are easily doable. I didn’t try steel yet, but with the cooling system in place and more sturdy rails on the portal this should also be possible (at least for shallow cuts).
Building the CNC router and also operating it has taught me many things about machinery and also mechanics. For the future I will probably investigate more in that field but I don’t think I will build some that complex in the near future 😄.