Scratch Build – In Progress Logic - V 01 - The cook book!

Hi I'm back

Real life intervened, but I am back working on the case. I have just run over the plans and notes I made on this forum and am correcting some, upgrading other parts of the design. I will take a few more iterations to make sure, the plans are as ready as I am able to make them.

I will make the 1st prototype out of glass fiber so I can cut and fit any changes as I see fit before making the final version of carbon fiber. Cost and convenience are the reasons why.

The SAS4CM is still being developed more fun then expected.

The Reservoir: I have decided to make an acrylic likely CNC'ed version, also the pivot function will likely be CNC'ed into the surface then glued into the sides to avoid a shaft all the way through the reservoir (I have had my concerns/nightmares regarding the reservoir leaking at that point). Now the challenge is making that look good!

Since I am also working on a emergency cooling solution I am considering an emergency stop button at an accessible point of the case, no clue of where to put it.

Design I have decided to try and create a curved front below the mesh/fabric that works as a filter. I will elaborate later to why I am choosing this aproach.

The fabric is a challenge 1st because I'm a man, then because it has to function as a filter, but also has to be see through enough for the OLED's to be readable. There is an alternative but I'm not ready to go there.

 

1st sketch: Not ready yet
The case production method.

The case is planned from the inside and out. 1st a form shaped like the inside of the case is made 42 x 42 x 46 (actually 70cm long) including rounded corners, the corners with a radius of 2 cm. It's made longer than needed to get a mold with an extended area to mount the fabric, glass or carbon fiber fabric on.

Next a layer of wax is mounted, 3 mm deep (the thickness of the frame) it's wrapped around the form and is 50 cm long so a surplus of the frame can be cut off later.

A 2 part mold is created around the wax layer, the 2 parts are connected on 2 opposite corners. Thick layers of glass fiber are used to make the mold.

The molds are also worked over before they can be used, flattened sanded etc.

Then a frame is made with the mold. This initial frame is a stand in it has no holes or removed material. It will be flattened and sanded to a top shine to be optimal for the next part of the process.

On the newly produced frame we repeat the process now with 3 levels of wax a total 4 mm wax. The 4 mm layer of wax is divided into a 3 mm and a 1 mm layer. The 1 mm layer is covering the entire form and is a stand in for the space between the shell and the frame. The 3mm wax will become the shell.

For the shell a new 2 part mold is made over the wax, connected on 2 opposite corners on the form of the frame . Again a thick layers of glass fiber is used to make the mold.

Remember not to create a molds with parallel sides. Parallel sides requires a mold with several parts if you want to be able to remove the form from the mold.

This is a very simplified description of the process but it will do for now.

The next part is where the complexity hits, its not just about cut outs, but about creating spaces that are and look organic, spaces for components like panel mounted connectors, the reservoir, radiators and more. I am still updating and revising the design and finding mistakes made by me.

Mistakes:
Its stupid small stuff like placing the SD card reader slot below USB connection ports and having the wires when inserted being an annoyance while trying to access the SD card slot or worst case damaging a card.

I am also revising the SSD/HD docking bays for better thermals and usage and for a more intelligent wire handling, I might fall short in this area time will show.

Design related

On the left side of the frame the panel connectors USB, card reader etc, I would like to mount on a 45 degree angel or maybe less. Why to avoid the connectors or inserted components being damaged by accident (at least trying to reduce the risk). A pleasant advantage to this is, that they become easier to see and access. This is one of the reasons why I need some organic spaces.

Updating in a short while

 

I have without a doubt and purposely designed the case, the cooling system with specs that are far better then needed. One of my reasons to do this is having a lower noise burden ( I like it quiet) the other part is having alternative options and thus a longer lifespan of the system.
In the start I looked into what a system needs not just in real estate terms m³ or ft³ for component dimensions but also for the individual components power usage before i got sidetracked by real life.

We all know that the Components Power consumption went up with the latest GPU's, but the next generation of GPU seems to bring it back down, a good example is a Nvidia RTX 3080 with around 320 watt usage that will have a follow up of a RTX 4060 with around the same ability but only 200 watt usage. If any of you have seen a list with Components Power consumption I would appreciate a link.

Component Power required
The central processing unit (CPU) 55 – 150 Watts
Graphics card (GPU) 25 – 350 Watts
Hard disk drive 0.7 – 9 Watts
Solid-state drive 0.6 – 3 Watts
RAM 2 – 5.5 Watts
Case fans 0.6 – 3 Watts
Source

I am currently putting together the weight of some components and I imagine that the radiators might take the crown (with 2 in the system) if any of you have information of your radiator weight with and without liquid inside I would appreciate it : (On the commercial sites this data is not available)

Radiator name - size - with fittings (wf) or without fittings (wof) - without liquid - with liquid.
Alphacool NexXxoS XT45 Full Copper - 360mm - wof - without liquid 1110 g - with liquid xxxx an example.

Still refining the design before putting it all into CAD drawing.

 

Wiring is Bi@tch and the following shows partly why. I have placed the DATA wires ONLY for the Stand Alone System for Monitoring and Control (SAS4MC) and there are a lot! It's approx 1 quarter of the total wiring needed, the data wire will in general go the the SDA connection on Arduino Board then there is a SCL wire aka the Clock and the usual 2 wires for power this is true for break out boards and OLED's. So the design is being adapted.



My lack of experience when it comes to micro electronics has made me cautious. I have almost decided to cut this part of the build into 2 segments , meaning I will use a temporary short cut using a FAN RGB HUB to drive the FANS and Possibly the PUMPS since this transfers the control of these components to the motherboard. This gives me time to learn and understand the communication protocols for the Motherboard and takes the 12V part of the built into the near future. I would hate to **** up the onboard emergency shut down and fry the system.



The System functionality as a whole will not suffer from this since the control part is with the Motherboard until I get up to speed.

I will still get the DATA from the Motherboard to display and log on the SAS4M(C) now less control thus (C). This (for now) leaves the Monitoring and Display part on the SAS4M(C). This is the truth to 90% since the control of LED's, DATA, CASE STATUS and RESERVOIR WATER LEVELS are still placed with the SAS4M(C).

Updated SAS4M(C)


The Emergency Cooling System. Its positioned on the top of the cooling Rig. Still looking at better design options picture on left top corner is for inspiration / direction. The ECS can be mounted while shifting the SHELL to get access to the ports in the left side of the frame.

The access to the ECS can be found on the SHELL's right side above the Vents.

Front


The last picture above is from the back of the case.

Case:

 

Happy to help: (custom) radiator weight ~25kg, ~26.7kg with water.

 

Appreciated mate!

Impressive radiator. I have been considering building one myself and I didn't know how to before I watched your project so cheers mate!

 

I'm intrigued at the level of planning you've put in - is the plan still for carbon fibre, or aluminium? With radiators doing the geat dissipation I'd be surprised if the aluminium frame is getting anywhere near the water temps (and surely even more so if using carbon/epoxy) and so wonder if the concerns about thermal expansion are justified: One thing I found surprising was that even when putting a copper pipe over a gas hob to anneal it, and then heating and getting to perhaps 200 degrees C or more, for copper it was still cold enought to hold the pipe with bare hand maybe 10cm away or less even.

 

I am still going all carbon on the final stretch. There will be a stand in made of glass fiber (carbon is expensive) where I plan to fix problems and design issues, this is after all a labor of love.

I just found a big clue on how to handle the Fan controller part but I am still not sure if I can scale this up to fit my needs, exciting what would life be without challenges and new stuff to learn:
https://emc2arduino.wordpress.com/side-projects/automatic-4-channel-pwm-pc-automatic-fan-controller/
and
https://www.instructables.com/Use-Arduino-with-TIP120-transistor-to-control-moto/

I am as we speak updating the initial part with the Expectations, Purpose, Hardware and a design. The project has progressed and my world has gotten bigger so have the details that are part of the project. I do enjoy upgrading the entire setup and fixing challenges so essential tinkering.

The case meaning the Shell, Frame and Cooling Rig will now all be made mainly of carbon fiber. Some panels will be made of glass fiber because radio waves can pass through it, carbon fiber blocks radio waves, I might need WiFi, BT or NFC so better plan for it .

One of the ideas not yet implemented on the drawings is a Hull (frame)-mounted active row of antennas. Not yet on the drawings but maturing I might use laptop antennas that can be covered by the shell, the carbon shell effectively working as a hardware off button, but better looking IMHO.

How to make a radiator is for my next project.

I love how its all connected.

I have enlarged the SHELL and given the text a light color, details matter

 

Copper and heat have you checked heat pipes essentially copper tubes with steam inside (very little water) its has a high degree of efficiency, it has always been a mystery to me how they could be that good at transporting heat with that little mass.

It might be the mass of your tubes that create the effect, since copper is that good at transporting heat . More mass equal more area to spread the heat on. Did you try with a short tube?

 

Afaik heat pipes work by bulk transfer of heat energy - the water inside is at a sub-atmospheric pressure to lower boiling point so as to be useful (needing to get to 100c boiling point isn't very useful for cooling a processor!). The water vaporises at the heat source end of the heatpipe, vapour produced diffuses to the cool end of the heat pipe, and in condensing, transfers the latent heat energy (which is really high for H2O) to the pipe and contacting heatfins, condensing back to water to wick back to the hot end of the heatpipe. I'd guess its effectiveness comes down to the high specific heat capacity of water and the throughflow rate of the steps of the cycle of heating/vaporisation -> diffusion -> cooling/condensation -> wicking (like coolant flow rate in a watercooled pc). So eg restricting wicking by orientation of heat pipes becomes the limiting step in the cycle so the heatpipe becomes much less effective. It's like watercooling, in transferring heat energy over a larger distance, rather than being limited by thermal conduction (and the temperature gradient needed to drive that, meaning a hotter 'hot end'/hotter chip being cooled).

I wasn't clear - what I meant is that from my experience, even with holding and heating a single copper pipe over a gas hob, the distance between the copper at say 200c where it was heated, to being cold (room temp or within single digits C of it) was surprisingly small, and I guess that's where mass comes into it - a single thin walled pipe just doesn't have the mass of material to act as an effective heat conduit, whereas lots of kg of it does.

I could be completely wrong, but I'd imagine thermal expansion of materials probably isn't a major concern for PC construction unless tightly clamped and restricted materials (think freeze-thaw erosion of rocks from water) put under stress from compression/expansion (even in a Scandinavian winter the pc is still probably in a relatively unchangeable room temp rather than getting to -20C, and for most materials I'd be surprised if it's a concern, except for things like brittle epoxies, glass or glued wooden veneers that may be susceptible to differential thermal expansion I imagine.

I'll shut up now!

(/autist !)

 

I think that the point where a heat pipe works is above freezing 0. c at this point energy transfer starts, a heat pipe is a two-phase heat transfer device with a very high effective thermal conductivity. Link FYI: https://www.1-act.com/resources/heat-pipe-resources/

My point, regarding your experience with the heat transfer of the copper tube is, since copper is supposed to be a really good heat conductor the heat might spread out over a larger mass so fast that the experience you have with the pipe 10 cm from the heat source might be because the energy (heat) has spread the to rest of the tube at speed, not concentrating in the area around the source, I might be wrong! A short piece of tube with no mass to transfer heat to could likely prove or disprove my idea, If you test it let me know I am interested.

Thermal expansion I found a few sources regarding that subject and with the tolerance I was seeking vs the material size, thickness and shape it could be an issue since I planned to mix different types of material in this case carbon and aluminium. I seem to remember sharing them on this forum if not send me a msg. My behavior and location is part of the reason I can with the equipment walk outside in winter to -25 deg c to inside with 24 deg c. so a 50 deg c diff.

For whats its worth don't shut up! It's hard to find my own mistakes if nobody points them out!

 

This version of the Cabinet design with wiring is not the latest. but it will do for now .

It has required some relocation of parts inside the PC case and I have extended the case by 2 cm so far.

Placing micro switches for the status of the Cooling Rig and Shell for warnings when needed.

A light sensitive sensor to turn on LED's for the areas with connections USB, LAN, GPU etc.

Next are TOF sensors (Time of Flight) or Distance Range Sensors which are laser based, the idea is to KEEP THE VENTS CLEAR outside the case it would be silly to focus on air flow and forget the location of the case, these sensor should help remind me. Still working on placement and mounting points of sensors.

Large Area fabric based filters - Decreasing air resistance and thus increasing FAN efficiency. Never thought I would spend time ordering fabric samples and testing air flow just a part of the job it seems

Wiring and planning why do I use so much time on this part of the proces, the answer is in the material the carbon fiber, cut up and add parts to often, it looses appeal IMHO. So planning channels for wires that are "baked" into the material looks better, reduces gluing and alternatively cutting into the fibers. No reason to mention that cutting the boards/fibers decreases strength, not really a concern of mine on this project since I simply do not expect the case will meet that kind of strain.

Comments are as always welcome!

 

Hi

I have some parts of my present design that doesn't work for me, so I made a muck up of a new front panel. Why you might ask the short answer are the OLEDs I have incorporated a new OLED 1,5" 128 x 128 and I am still not sure that the text on the OLED will be read able below the fabric I want to use as a filter. So I looked for alternatives and a panel on the front is not ideal, especially not in this mock up, I think its way to vulnerable to be carried around but then again I might as well share this before trying to improve it.

Its not all bad a front panel made of glass fiber (why not carbon) can also hold antennas in a prime location and as mentioned before, glass fiber does not block signals like carbon fiber does.

On the bad side the panel does block the Large area filter from easy access and does not work with my modular idea. All parts should be easy to assemble or disassemble tool-less when possible. so back to the drawing board.

It might be worth mentioning that the weight of the frame is behind the front 11 cm of the frame so sliding in and out of the shell i not a risk/problem.

I have at this point in time not checked if the hardware will fit in panel, so also a to do.

 

I ran over the options that I find functional concerning the risk, of having the OLEDs outside the shell. I decided to make the panel flush with the front of the case and to create the panel front with a slight bend inwards to protect the OLEDs. I am not done with this part but on the right track. Keeping the panel gives me a lot of options I want to see where it takes me.

Rough estimates of the impact of a smaller filter area on the airflow are neglect-able and the fabric used will have the last word in this variable.

Some interesting options regarding fabrics and 3D printing popped up from MIT and people working with this, try to search on YouTube for - fabrics + 3D printing.

My Original idea was Nanofiber filter fabric, its see through. I just haven't found it yet in the sizes needed. So the option above with 3D print might be a need too.

When deciding to use a center panel I had to find a work around my specs with tool-less removable Large Area Filters. I created a sketch with 3D printable buttons working as locks, its essentially the same principle that will secure the Shell. The Filters has been divided into 2 parts in the front and are now removable when needed.

 

Updated

Why I am updating this post I might not have stressed the advantage of a ducted fan and the performance increase that can be had doing this. the GAIN can be 80% in efficiency, meaning same power use and and 80% performance gain.



When creating the front with the Air intake for the fans I used the calculator on this www.engineersedge.com this has relation to my SAS4MC. I haven't mentioned this because its a tat dry by its own, but for those of you interested in a REALLY GOOD explanation on why and how to use it - check YouTube channel Wyman's Workshop Fans Air flow and more



My input for those of us that like to know how the world works and like to optimize stuff.

 

I found a project on Smart PC fan and watercooling controller on https://www.hackster.io by seventz

Always appreciated when others have taken a big step in the right direction!

https://www.hackster.io/seventz/smart-pc-fan-and-watercooling-controller-8b260f

 

Today I was ecstatic and depressed at the same time.

Why hmm it comes with a story - When I started looking at products that I could use for this project a bit more then a year ago, I used a few hours researching (under fecking statement of the year)

Then within the last 24 hours and by pure luck the google algorithm pushed some options in my face after I searched for sensors and connections, one of these options was chosen and Voila Spring Loaded Connectors. I had been thinking about creating some myself but decided that this was outside my skill set.

Why is this a big deal imagine a time when we had working docking stations for our mobile phones (notice I didn't say smart phones) Mobile phones that could be mounted in the docking station without having to touch a wire. This is what I want for the modules in my project, without water damage being a potential problem.

Now why is this such a kick in the male parts because I based a lot of design decisions (and many hours) on this part not being available to me. Sighhhhhh. The front panel might now comeback with the Large are Filter in 1 peace. The Docking Bay has a better (more functional) design. With more options comes more responsibility!

It influences the entire wiring, all NODE POINTS - Connections with any modular peace I have or might choose for the project.

I hope these also give you guys some options, Enjoy!

NoX out for a few days.

 

This is 1 of 3 areas that I am working at, the areas are inter connected.

Last Updated (July 14th 2022)

Design decisions: (Yea well I need to put it to paper to focus)

I have used the last days to work through my expanding options created by the discovery of the spring loaded connectors (pogo pins).

What I find appealing about pogo pins is their use with modular devices making the modules easily exchangeable for future upgrades or repair (I like this part since some sensors and other parts could be places in exposed positions and easily replaced. All connections to the Cooling rig, Front Panel and Motherboard tray are now planned with pogo pins – power, serial and i2c. Water Prof Modules might also be a result of implementing the pogo pins, TBD.

It does take some time for the new wire routing to be decided on, this also applies to pogo pin mounting/connections. In progress!

* This is need to know for me, from Instructables.com - https://www.instructables.com/Complete-Guide-for-Tech-Beginners/

Project order for the Stand Alone System for Monitoring and Control, Sensors, data and more

Why oh Why do I do this; Simple, when we have a functioning SAS4MC that receives data and can print it to the Displays (OLEDs), the SAS4MC still needs to be programmed to act on specific changes, levels, or events. This is a 1st sketch with purpose and new ideas.

Turning on the SAS4MC (Checklist):

1. Display status is obvious 13

2. Arduino Due to Arduino Due serial connection is live and transferring data. 1

3. All the sensors are checked by the system aka alive and in contact.

4. Reading of data from all sensors, the data is used when we later (in 1/10th of a second) are calibrating the sensors.

5. Fans are online and working.

6. Pumps are online and working.

7. Calibration (when needed)

8. Checking (after the PC is turned on) that the USB to PC is working

Components purpose and specs:

· Arduino Due

Why an Arduino Due because it has an upgraded processor (CPU) faster by a factor of 10 compared to an Arduino, and it holds more connections for sensors. It’s also has a better performance with more displays = area.

· Displays

Displays to see what’s going on in the loop in as close to real time as possible within my budget. Why so many Displays, it’s easier to scale the amount of displays to what is needed, to show the information I want available.

Waveshare 128×128 General 1.5inch OLED display Module, Color: White - SPI or I2C interface - 3,3V or 5 V

· Sensors

Here starts the 5hit show, task and purpose. When I started I thought it would be easy and it could be, get the data and display the data. Let’s be realistic next level is minimum the use of the data to improve performance do yourself a favor don’t think about it. 5hit I thought about it, yeah well improve performance hmm how to….

If we had 1 sensor I would expect it to have the same accuracy / precision / tolerance every day if I wanted to compare data from day to day, expect hmm ignore who knows. With several sensors in a row I can’t have this varying accuracy so I will have to measure the data delivered from the sensors and calibrate the sensors to a common level or to what is possible.

Temperature sensor accuracy (Air and Water) can vary from ±0,2 to 1,5°C or °K from sensor to sensor and from manufacturer to manufacturer, this is normal. Sensor temperature Range: –60°C to +150°C.

The air temperature sensors are tested before the system is turned on. A sensor is placed in front of the air intake to the hardware chamber and another inside the hardware chamber on the top and shielded from direct air flow; this is done to avoid false data. Data from these sensors is used to secure the conditions for the water temperature sensors, to calculate the air mass (mol) and watch the climate inside the hardware chamber.

The sensors for the water temperature are tested when the SAS4MC is turned on, they are in a more complex environment with the possibility of residual heat, making calibration futile at that point in time. These sensors show what is going on in the system and are responsible for pushing the right buttons for fan and pump activity.

The water Flow sensors are tested and the calibrated before being mounted, this might have to be done with a regular interval. The sensors are mounted after the pumps to be a 1st line of defense to secure the water reaches the water blocks. The water flow sensor data can be compared to data from the water pumps speed (PWM) and reservoir level.

Liquid/Moisture sensor Because of the problems adding a connection/wires to a grid of parallel conductive lanes drawn in the bottom/floor of the Shell, a Shell that is meant to shifts position vs the frame where an Arduino is mounted in the top. I had to a certain degree dismissed this option. The pogo pins have restored this possibility, I hope.

The Liquid/Moisture sensor is the 2nd line of defense to secure the water doesn’t create damage in case of an accident. The sensor is used to shut down the entire system in a situation with leakage. The sensor lanes are placed on the inside bottom of the shell (with a thin layer of glass fiber below the lanes to insulate the current) with a female pogo pin.

A photovoltaic sensor is for lux levels in the room and to turn PC LEDs on or off. The sensor is placed on the top back end of the frame to avoid light from the displays interfering. Data retrieval is only necessary every 2-5 min (for a profile) and together with a RTC is used as a timer for the LEDs on the connection points. I will need easy access to control the level that fits my wishes (read button).

RTC is a Real Time Clock might be GPS controlled (likely not it’s a personal choice).

Energy Monitoring System (EMS); a Current (hall) Sensor is used for AM/Volt meter to Watt usage. I am not done considering how-to. The PSU is the challenge. Power use of the GPU and CPU is delivered by the components themselves (accuracy may vary) to the PC software and transferred to the SAS4MC by a USB connection. A total power use can in my view only be delivered by retrieving it from the PSU or before. Without being aware of my power consumption it’s hard to change my ways.

Capacitive touch sensors: 2 boards are used with 10 sensor points each. One board is used for the water level in the reservoir. The other sensor is meant to be used for touch buttons, the placement is for now a challenge. Each Capacitive touch sensor has 10 connections.

Micro switches; to get a position status from the Shell, Reservoir and the Cooling rig, with the implementation of pogo pins a necessity for the SAS4MC to work with this plug and play state. I still need to check if this is a viable option disconnecting parts while powered.

Laser Range Sensors to secure the air vents (intake and exhaust) are not blocked with a minimum distance to other objects.

A sound level sensor could also be mounted in and outside the case to look for abnormalities in the sound level if it is feasible, I am undecided.

Last Updated (July 17th 2022)

Humidity sensors (3); Are used one in each of the chambers and one outside the case, to compare the data from each of the sensors and to be an indication if there is a water leak on the way. The idea is that water will start to leak very slowly and will evaporate before we have a full blown leak, this might show up as a change of humidity. RH (Relative Humidity) is also used to calculate the Air mass.

Barometric pressure sensors (BMP 280) (2): Is also used to find the mass of the air being pushed though the fans. Sensor in the hardware chamber and in the back (outside) for the ambient pressure. The Pressure at sea level and on location are also used to calculate the Air mass.

Why the Air mass is important - The Airs ability to transport energy (heat) is depending on its mass. The mass is depending on the Temperature, Humidity and pressure. This part is for fun but also for comparing data by time of year and conditions in the environment.




More to come


Information overload; the initial architecture of the SAS4MC is based on 2 Arduino Due. I am well aware that the amount of sensors, displays and data might lead to information overload of the Arduino's and alternatives are being considered. The smart fix is to bring down the amount of information (not sensors) by only pulling e.g. temperature data once per second for every temperature sensor and not 100 or 1000 times per second, for other sensor like the Photovoltaic sensor retrieving data every minute is likely more than adequate. The hard fix would include more Arduino's dividing the load.



· Fans
10 Best Radiator Fans in 2022- more to be added.
https://pcmecca.com/best-radiator-fans/

· Pumps
Top 5 Best Water Cooling Pump Reviewed
https://10scopes.com/best-water-cooling-pump/
I would like to hear what your preferences are?

· Radiators
https://www.xtremerigs.net/2015/02/11/radiator-round-2015/7/
I doubt the radiators have changes to a significant degree since tested.

· Boards

· LED’s

Updated June 21, 2022 further clarification

Calibration:

1. How to

2. Precision

3. History

How To:

Assumptions: The water in the cooling circuit will in a neutral state have a temperature equal to the ambient air.

Fact: The Air Outside the enclosure is where the heat from all the components will end up. The Water is just transportation for energy (the heat) around the loop.

No need to mention: We want the air to be cooler than the water.

The collected sensor data when starting the SAS4MC is used to determine the status of the sensors, calibrating the sensors is done in the software. Sensor precision around 0.2 °C is my target. TBD.

Link to YouTube sensor calibration with Arduino!

Consideration: Comparing the data from the sensors inside the case, with data from sensors outside, we take into account the last time the PC was running and the residual heat.

Sensor Groups are as follow; the reservoir, the loops 1&2, HW Chamber, Coolant chamber and the ambient (outside the box). Then there is a sensor group for control. On the Cooling Rig (in the coolant chamber) are mounted the Reservoir and the Loops 1 & 2 with the radiators.

Reservoir;

1. H2O level. Absolute and Comparison

2. H2O temp

3. H2O Quality – 2 of 12-Key Capacitive - Touch Sensor - Breakout – MPR121

4. LED/RGB

5. 10 of 12-Key Capacitive - Touch Sensor - Breakout – MPR12 – Water Level


Loop 1;

1. Reservoir

2. Pump rpm lpm

3. Flow lpm

4. CPU

5. H2O temp

6. Radiator

7. Fan rpm

8. H2O temp

9. Reservoir


Loop 2;

1. Reservoir

2. Pump rpm lpm

3. Flow lpm

4. GPU

5. H2O temp

6. Radiator

7. Fan rpm

8. H2O temp

9. ECS

10. H2O temp

11. Reservoir


HW Chamber;

1. Humidity sensors

2. Barometric pressure sensors

3. Air temp Case C


Coolant chamber;

1. Humidity sensors


Ambient;

1. Humidity sensors

2. Barometric pressure sensors

3. Air temp C

4. Photovoltaic sensor


Control groups;

1. Humidity sensors

2. Liquid/Moisture sensor (Bottom of Shell)

3. Laser Range Sensors Front intake (2), Side vents (2) and Rear air intake (1).

4. Touch Buttons - 12-Key Capacitive - Touch Sensor - Breakout – MPR121

5. Micro switches

Important note: The Calibration of sensors is more important then I expected. By now I have seen error values of more then +/- 6 C from the temperature sensors. Humidity sensors have been up to 20% off from established values.

Calibration of the sensors:

For calibration of the sensors we need two points of reference to make a map in the software. The problem with only using one point of reference is obvious if we think about it. There are no guarantees that a difference of a +/- value at one point of the line will be the same at another point. This is why we need 2 points and a mapped difference.

For the temperature sensors and the 1st point of reference we use ice water, by inserting the sensors in a plastic bag (eventually mounted on a stick) and immersing the bag into the water for an extended amount of time. This gives us 0 deg C = 32 F and we note what the different sensors postulate and calculate the difference.

The secondary point of reference for the temperature sensors, we combine with the 1st point of reference for the humidity sensors and add a HG thermometer.

As the environment for the test we use a plastic box with a lit to isolate the sensors and the thermometer, we include a little bowl of salt (NaCl) saturated with water, this setup is now left for 3-6 H. The NaCl and water will produce a stable environment of 75% humidity inside the box.

Again we note the temperature sensor postulates and the thermometer value. The same is done for the humidity sensor 1st reference point. *This time the 75% humidity is noted with the humidity sensors values.

For a secondary reference point for the humidity sensors, we use another ingredient a humidity absorbers, in the same test setup this is done after removing the salt and water, this should give us an environment with 40% humidity. Now we note the postulates and the difference to the 40%.

The difference can be noted as +/- value per sensor in the software or as a map, in the software also per sensor. My personal preference is as map.

Barometric pressure same procedure comparing it to a mechanic device. The barometric pressure sensors have a very high level of precision.

Photovoltaic sensor we don’t need a comparison just a value that feels right for turning on/off the LED’s.

RTC compared with another watch.

Water level sensor are compared to the visual reading of the water level on the reservoir.

Laser Range Sensors tested with their distance and a measuring tape, comparing the measured to the sensor values receiving 2 reference points.

Pumps and flow sensors: a testing setup that is close on the case setup. The reservoir with 50 -100 L of water mounted 30 cm above the pumps, then a test of the flow speed from 0-100% in 10% intervals. This is done with both pumps in turn then adding the Flow sensors testing the influence of another component in the loops performance and the flow sensors.

Profiles:

· Nominal levels the components work within

· History and comparing with the present

Rules:

· Max temperature levels

· Maximum and Minimum performance levels of components

· The Sweet spot

 

Ducted Fan:

The ducted fan’s advantage, is the ducts ability to remove the turbulent air because the high and low pressure sides are clearly divided = less turbulence, higher efficiency and less noise.

When there is no divide between the high and low pressure side vortex’s are created wasting energy by creating chaos, noise and heat.

Turbulence: imagine spaghetti in a tube being the streams of air that make up the laminar flow, now imagine a few pieces of this spaghetti having been boiled and starting to bend around in this flow stopping other flows and putting pressure on tube side effectively hindering the flow. This is a short description of chaos, where we want laminate flow (order).

Insulated engine noise is an easy advantage of putting the fan in a duct.

Creating a lip around the front/top of the duct, creates a shape that essentially enlarges the fans reach by using the low pressure created suction and by air turning around the lip aka the Coanda effect (liquid or gas, to attach itself to a surface and flow along). This lip also secures a laminar flow (no vortexes) on the air intake. Lip size and shape is still in the wind and testing has to decide the final form unless I wander into knowledge of best practice.

From Wikipedia: essentially the setup needed to measure Fan and duct efficiency. This is also how water flow sensors work.

https://en.wikipedia.org/wiki/File:VenturiFlow.png



A flow of air through a venturi meter. The kinetic energy increases at the expense of the fluid pressure, as shown by the difference in height of the two columns of water.

Bernoulli's principle - https://en.wikipedia.org/wiki/Bernoulli's_principle in its latest edition a description of the loop IMHO. Depends our your point of view.

Centrifugal force: F=m*ω2*r is a part of my fan math

ω – https://en.wikipedia.org/wiki/Angular_velocity

I will elaborate on this later; I am still working on the math for parts of this. I am actually enjoying this to a minor degree a shame I didn’t feel this way when I was in my late teens or early twenties.


Fluid dynamics and the math behind it have taken some research. Choosing to optimize the water loop and Air flow since it has easy gains and should be a no brain er. I will try to implement the ideas where I’m capable. In general the rules that apply to gas (air) do also apply to liquid (water). What fluid dynamics is about is to a good degree explained by the following YouTube videos. I am not trying to create a detailed real time model of anything Fluid dynamics (I am not that smart) but a working approximation for fun and (my) purposes. I will try to optimize real life data and applications for the system to push the envelope. There is without a doubt considerable real life application to be had from a basic understanding of this subject.





Fan equation: https://youtu.be/T2v8660QcHo?list=PL87M7FRSLxj0dDSui-u5FWLGNlRtzwy2h

 

Expected heat generated in a given PC case by its components;

Of the energy used in the case, be it by the motherboard, the GPU, or the CPU 99.99% is converted into heat. The other .01% is the actual signal out of the GPU to your display. The power used by a 400 W GPU, will not be 400W the entire time it is in operation 400 W is the Max power use. Newer components especially GPU's can peak at levels of more then 3 x that level so 1200 W, that is a topic already covered by GN.net and other YouTube channels.

This means I will be looking at the maximum heat that can be generated and that a loop needs to be able to eject. I start with the radiator and look for an energy profile - fan speeds, water flow vs energy dissipation = performance. I am also aware that I would prefer a curve profile that looks at performance for fans from 800 - 1500 rpm because of the noise level.

1st I run into www.techpowerup.com that actually has a review database of RADIATOR PERFORMANCE @ https://www.techpowerup.com/reviewdb/look for radiator I choose a review of the Alphacool NexXxoS HPE-45 Full Copper 360 mm Radiator, this is still not what I am looking for but interesting.

Worth testing - I then used OuterVision® Power Supply Calculator this takes Overclocking into consideration when looking at a PC's total power consumption.

https://outervision.com/b/oA9Bai My here and now build if I was shopping today I'm not.

Expanding my search to radiator performance curves. A good e.g. is:
from https://www.ekwb.com/blog/radiators-part-2-performance/

Looking at those curves my hope of finding a radiator where I can expect the fans running at 800 - 1500 rpm will have to be revised I'll explain further down!



Updated July 13 2022

Taking another step back: For my imagined setup I chose a AMD Ryzen 9 5950X Maximum Operating Temperature which is 90C surpassing 90C the processor will automatically throttle (slow) down to keep the temperature below 90 C. CPU longevity suffers when close to that temp.

If the CPU stays around 70 C there will not be any difference in performance to the CPU at 50 C. It is a waste of energy to keep the CPU 20 C in temperature. Not to mention noise pollution from pumps and fans. The idea is to have a large amount of extra cooling capacity to avoid temperatures above 70 C when the system is running at a 100% or 120% when OC'ing. I would love to hear an arguments why to keep the temperature lower (not talking OC'ing).

Rule: the higher the Temperature difference (TΔ) of the water temperature in the radiator vs ambient temperature, the easier it is to deliver Watts of heat to the ambient air. A realistic ambient room temperature is 20C to 24C when estimating/calculating the needed radiator size. So I will be looking at Radiator curves that deliver in that area. This brings me back to having a CPU running at 70C should give a higher TΔ and thus a better efficiency of the cooling loops.

The cooling in the loop is determined by the amount of water that at any given time is in the radiator being cooled and in the water block of the CPU/GPU or both being heated. The easiest fix to better performing cooling is to a given extend a larger radiator with more fans, not having a faster water flow.

Next tubes.

A good guide can be seen at https://koolance.com/hx-360xc-radiator-3-fan-120mm-30-fpi-copper

Tube 6 mm pressure vs lpm - next tube ID 10 - 13 mm. The flow difference is significant! Finding the other components that are fitting the max water flow could define the system performance. For my loops I have chosen 10 mm ID tubes why should be obvious!

Doubling the Tube size and doubling the water flow speed at half the pressure equals an easy life for the pump. Does this double the performance of the system, likely not!

More to follow!




A decade ago I ran into Martins Liquid lab took me a few hours to dig it out of the archives it is worth looking at ! I will try to filter/reproduce the core of it here.

Depending on the Radiator, not all are created equal, doubling the flow rate will not double the cooling performance here we have to look at the watt the radiator delivers to the ambient air at a given water flow rate pared with the fan speed! This is where a lack of data makes this difficult. The fix is to look at the radiators and choose a cooling system that can handle 873 Watt when peaking.

The total watt usage by the PC is max with dual GPU's and capture cards = 1073 Watt of those are 873 Watt water cooled.

CPU: 1 x AMD Ryzen 9 5950X - 105 Watts
Video Card Set 1: 1 x NVIDIA GeForce RTX 3080 12GB - 384 Watt
Video Card Set 2: 1 x NVIDIA GeForce RTX 3080 12GB - 384 Watt

In daily use the loop performance is estimated to be 873 Watt x 60% = 524 Watt.

This is my target for the Radiators at 50 % of max load (a lower load is better)

If I chose EKWB's COOLTREAM XE 360 one radiator would do the job with fans running below 2000 rpm @ a 60% load of the components- To have a 50% backup capacity at 2000 rpm i am installing a total of 2 COOLTREAMS XE 360. This gives the system the needed capacity with the fans running below 1100 rpm with a 60% load. I still need info on the water flow speed to be sure. I expect most radiators in this performance range can deliver what is needed.

Motherboard: 80 watts and the rest of the components are in the hardware side and air cooled none of these components are relevant for the water cooling loops.

Back to the performance of the PC.
Since most systems are normally not running at more then 60% of their maximum performance level in daily use (when not running a synthetic test or demanding encoding or 3D application) I am looking for a radiator + fan + water flow performance combined with the CPU/GPU water blocks performance, that are running at ½ of their capacity securing a lot of performance for peak performance . As mentioned above.


FYI from https://www.xtremerigs.net/2015/02/11/radiator-round-2015/7/


and a total of radiator performance


I actually found the radiator estimator discussion with Martin from liquid Lab on https://www.overclock.net/threads/radiator-size-estimator.1457426/

Next water cooling blocks: