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

Hi again

Today my thoughts regarding some of the small but not unimportant stuff.

The case has some small troubled areas, one of these are the OLED cut Outs. Why you might ask, because all the curves are different for each OLED cut out (See 1st and 2nd Picture). But that is only half the challenge the front wall thickness will be 3-4 mm deep and the OLED screen is only 1 mm above the board its mounted on (See 3nd picture the yellow area is the screen area), meaning the OLED will be 2-3 mm deep into the front giving a lot om space for dirt to collect especially since air is traveling past the OLED 24/7.

In my view there are several ways to fix this;

1st could be to bring the wall thickness down to 1 mm in the area with the OLED cut outs when creating the front the problem with this is not having enough materiel to mount the OLED, so that's a no go.

2nd approach is to mount acrylic windows that are 2-3 mm deep into the cut outs thus removing the dirt factor. The acrylic would have to be heated, likely on top pf the front for the material to get the same curve as the front OLED cut out, so no cutting out OLEDs windows before shaping the acrylic. To be honest this still has some built in problems and I will likely have to return to this point.

The prototyping of a Carbon Fiber Shell and frame also have some inherent challenges because I need to create them in one piece. Imagine to have to create the cardboard tube (from the a toilet roll) by shaping it on the inside of a plastic tube with pieces of fabric to get the right surface, that's where I am starting.

Some time ago I did pick up a big piece of the solution on how to mount and form several layers of fabric on the inside of a form by using magnet in a pattern and clamps to keep the fabric in the right position and inside the corners while mounting the fabric and later infusing the fabric with epoxy, Inspiration was found in a video by Brian Oltrogge https://youtu.be/b0d8RftCfPA

I have been aware of the challenges before I stated this project and to be honest it makes me feel alive.

Uhh yea the pictures I almost forgot, must be getting old.




and the 3rd picture

 

An update on the Water Cooling Rig (WCR) i f0cked myself with the wrong dimensions on the Radiators. I used the specs on the Radiators I want to use and not the largest size 3 x 120 mm radiators on the market this is both in thickness and length of the radiators. MY plan was to use a max thickness of 60 mm and 415 mm in length. I didn't check the specs of all the Radiators that I had put on my list, shitty work from my side ... So now my specs are not being able to use a monster radiators of 80 mm thickness but I have updated the WCR - to 430 mm in length. This has consequences for the frame too that I will also have to redo! I might just redo the total size of the case its good practice in the 3D software and a happy weekend to me.

The problem with increasing the width of the case and getting added volume is that I will expect more value to be added.

Now updates with 20 pictures and several areas that I worked on and how I found out about the wrong specs.

On the WCR I added rubber anti vibration dampeners for the pumps, carbon if made right can vibrate as a bell so no way around this, the mounting plate will also under go some work to bring the vibrations down likely by cutting pieces out of it.

The WCR has been made longer by 15 mm






Fans are added to the model, the adding of data and power wires is to be done by the water tubing leading the way.





The Boards included in the Stand Alone System for Monitoring and Control (SAS4MC) on the water cooling side are all mounted on the inside space created by 2 acrylic plates that are mounted under the Rail where the Spring Connectors are to be added. This should minimize the chance of an accident with power and water inside this part of the case.

I will look into some hinges for the acrylic so its easier to update the boards.

Mounting position of he Arduino Due




Mounting position of the Mutiplex board I2C




Mounting area of the Capacity Touch Sensor for the Water level If I can get it to work.





Last but not least in the acrylic chamber an OLED display for a stand alone option of the SAS4MC in the WCR.




With water pumps nice to see that they fit.




A side view of the Water Tube guides V 0.2



Seen from the top




Why the tube guides are mounted in those positions.






Why this has to be adjusted checked by looking at a test of tube positioning





The space limitations of Radiators when the PSU is mounted.





Looking at the last 2 pictures with a Radiator width of 45 mm + a 25 mm depth Fan are optimal for the WCR adding a 60 mm Radiator still works but 80 mm is a no go. The PSU is 230 mm long and is AFAIK the longest available on the consumer market.

Coffee break time, might go and bake a cake before I start looking at options.

 

New hardware GPUs from Nvidia and CPUs from AMD how do you all feel about the increased power usage in general and relative to your projects cooling ability ? Not to mention size of the new GPU's.

 

When I learned about the power spikes and expected power draw of the new GPUs
I got a Corsair AX1600i , which will be mounted externally on the back of the case! Cooling wise it's more of a load than ideal, but have cooled similar total wattage before with a heavily overclocked and overvolted GTX480 and CPU before, so should be okay hopefully.

Size-wise it's bigger for those aircooling, smaller pcb for those watercooling - the blocks are only around 20cm long.

 

Power consumption is far too high for my preference. 100w for cpu and 200-250w for gpu is my max rationality. I'm not really interested in the new GPU sizes, two fan sized GPUs is what I prefer. The new Nvidia GPUs are SOOOO big,... I think I lost interest already.
By the way, German YouTube Igors Lab reported that the Nvidia GPUs were designed for 600w, but after the the switch to TSMC, the thermal load shall be at 450w, so coolers shall be way too big and could be really silent at full load.

 

Hi @miculago

I share you sentiment regarding the power consumption. I have a hard time understanding that the new GPU can't throttle down when not under pressure. We might have seen the same video by IgorsLab and what I took away from that video was that the new RTX 4080 has 300+ watt usage even when not under load, please correct me if I'm wrong. I do really have a hard time seeing the logic in using a+400 watt on a workstation when we are only reading mail or browsing especially when low cost mobile phones and tablets have that function. This essentially diminishes the value of a workstation for me to around ½ of what it should be, to save power I will have to use a laptop or tablet when sitting at the same desk its just stupid, saving power is not stupid but having equipment not able to is.

I really wanted the next generation components these are here but looking at efficiency I might just stick to a 5950 at half of the power consumption with only a 30% performance drop compared to the 79xx. Same goes for the new Nvidia 4K vs the 30xx generation again if the information available is correct.

This perspective may not be an option for everyone on a workstation, but for those who can send work to a server or just buy server time, it is less problematic.

For gaming or real time rendering I have a hard time to see if the cost benefit will make the new generation of hardware worth it.

What I can see and I am not a Apple - Mac fan boy is that the new ARM architecture or should I say, the prove that a low power architecture is possible and can perform on level or really close to what my target area is when talking performance, is pointing the way to some new hardware if companies are willing to produce it at a reasonable price.

 

I do not follow all of Igors videos, but I do not think that rtx 40 series will run 300w at idle, might be a misunderstanding. I skipped rtx 30 series and rx 6000 series, because of availability, mining prices and upcoming new generation, so have no valid experience, yet.
In my opinion AMD and Intel on the CPU side and Nvidia on the GPU side are overclocking their chips to the extreme at stock levels. If you reduce the clocks and voltages manually, you can reduce power consumption a lot.... and maybe, if you clock an ARM chip to 5GHz you end up with high power consumption, again.

 

I hope you are right but the video review I watched showed 300+ watt usage idle (in windows not doing other work) I am trying to find it again but now i am not sure it was IgorsLab. I will post it here when I do.

The Clock speed has gone up for most new products. I think this to a degree comes with the smaller nano meter lanes, I have no idea how close we are to the limit when it comes to this technology. I have seen new semiconductors with Optical CPU's that are now used by the financial sector, very fast, very efficient and very expensive! I have no idea when this will bleed into the consumer market, but I look forward to this happening.

 

Hi I Retract my Previous statement about the 4090 power consumption. It looks like the 4090 in 1440 resolution and up has a lower power usage then 3090 and to some degree the 3080 at a higher or the same frame rate. If Der Bauer is correct and a 70% watt load of the 4090 can be targeted a power saving might be within reach, with a 3-5 % performance loss, but a power use around 300 watt and not 420-450 watt.

I still haven't seen any data on this card in office use this would be interesting! But it throttles!

For me at home and typically using in the PC in the evening the price of 1 Kwh can reach 1$ this makes a system that is not using power throttling unacceptable, I presume this is a general problem in large areas of Europe atm.

 

But you'll use less gas as you won't need as much heating, gotta look at the small victories

 

Yea expensive heat

I did redo the Water Cooling Rig (WCR) not that I wanted to but it seems necessary for my long term goal of having a Case with longevity. When increased power consumption is to be expected, the only way to counter this seems to be to upscale the cooling ability.

So I did increase the size of the WCR with a few test cases before making a choice, the 1st picture is on of these discarded, not a choice for me since the increase in size of the WCR doesn't give the case more options.

So I did choose another track seen in the next pictures for the WCR this does also entail some changes in the Frame and a minor change for the Shell. The Frame Front and Back plate have been updated for added stability, with frame around the parts.

The Front was originally made curved for more air flow through the Air Ducts with the added bonus of the curved front giving more strength and stability to the part, with the added frame its perfect for transportation. This also goes for the back plate where an added frame is a plus with the mounted components. The frame on the front and rear panel makes the Mounting of the front and rear panels with screws to the skeleton easier just like upgrading parts or just repairs.

Discarded option.




Chosen option for WCR. ITs longer by one Fan MNT aalso making the water tube mounting easier.







WCR with cut outs for the Spring Connector and the Micro switches. The Spring Connectors should make wiring up this part some what easier and taking the WCR out off or inserting it as easy as disconnecting the QD3 Quick Disconnect No-Spill Couplings x 4.








Back Panel










Air Tunnel Front





Front Panel





Fan Ducts with Fan MNTs Still need to make a choice on what shape to choose, presently I prefer the top one. The second is Mickey and the low option has some naval association.



Front Panel





Font Panel with Sensor Panels




Front Panel with Fans and the Rear Panel of Air Tunnel.





WCR frame support/rail and for connectors.




Inside frame




Inside Frame Motherboard





Thats all folks!

 

Yesterday i wrote a few lines without posting so today I start from scratch.

When I was a teenager i watched a movie, the The Last Starfighter, the movie was made with the first use of integrated CGI for real world objects. The CGI if I remember correctly was made by a Cray 2 Super computer. Why is this worth mentioning? Because the Cray 2, the 1985 super computer was liquid or gas cooled. This Super Computer had when peaking a whooping 1.9 GFLOPS of performance.

Compared this to Mobile Phones made in 2022 that now can perform at 5 times that level. The multicore performance of a Google Pixel 3 phone is 19 GFLOPS. Impressive when considering that the Mobile Phone as a concept 1st time ,to my knowledge, was described by Alan Turing in a paper around 1930. To be honest it was described as a computer fitting in a pocket that could communicate around the world. Smart guy!

Oh well back to the basics, I have tried to improve the concept and continue to find things that need to be changed.

I started a few days ago, by playing with the hexagon Cut outs profiles as can be seen below:




Power and reset buttons with a lid to avoid pressing the wrong button by mistake. I need to add an extra button. I have the PWR on/off of the PC, reset 1 and 2 of the Arduino's and need a reset of the PC.



Added text on the frame and Spine for now. The text is 0.2 mm inside the material.



Front IO has acrylic around the access points for LED light to make it visible when opened. I will redo this.



I am playing with the idea to include Aero GPU Profiles that could include GPU Support. This is the 5h generation of this concept. They would be mounted to the Spine.




LED light was added under the rear fan for light to shine on the PCI brackets.



Last but not least Cut outs and PCI bracket Flanges.




As usual to be continued!

 

I couldn't sleep, so i used the last 3 hours to experiment, it's 0400 now. First I wanted to know how the Case could look if I kept the last experiments, its hard to choose when its about looks and not function. So here a short picture show.

Locked Case

Front



Front Right



Front Top Right



Front Top Left



Open Case

Still Front Top Left



Front Top Right



Rear Left Top



Front Button Left - next to last picture.



Last Picture Top



Feel free to comment and have a nice night.

 

Hi

I have been skipping updates the last month. Why you might ask, its simple the case is to large (read long) for the desk to be comfortable. Before being opened the case will be 65 cm (about 2 feet) long, opened will add 6 cm. So I have been looking at options to include the functionality and make it at least 10 cm shorter. I am not yet done fixing this challenge, the easiest fix is building a wider desk. Updates will follow when I have made a decision, after exploring as many options as possible.

 

Hi

I looked at options to make this enclosure smaller and a bigger table it is.

This mobile solution needs standard components in case something breaks, when I'm not in an area with access to a workshop, so custom components are not an option for this build. The enclosure is still custom scratch build but components like the radiators are not. The Stand Alone System for Monitoring and Control (SAS4MC) is also a custom project.
If the SAS4MC breaks down the motherboard can take over the majority of the workload controlling cooling some information of how the system works is lost and some detailed adjustment of the cooling system is also not an option. To make sure that is an option more planning of wire channels is needed.

I have updated my approach when considering how to build the case. For precision, 3D printing is still the method of choice and should bring each component in around 0.2 mm from target.

On this iteration I started with the frame of the Spine (aka as the frame of the motherboard tray) it will be 3D printer then a mold is made with Glass Fiber. In this mold the real Spine Frame is made of either Carbon or Glass fiber. Into the Spine Frame holes are created for Brass Knurled Inserts this is done before creating the Spine core of Glass Fiber and aramid honeycomb with a thickness around 5 mm for the aramid the Spine is 10 mm deep. The Spine frame will be fixed on a Mirror or on glass for a better finish, so are frames inserts like the REAR Panel to access the CPU Mount. Cable Channels are printed in mirror mode to also be mounted. Wire tunnels might be printed and inserted into the frame TBD.

The advantage of using a mirror is the extreme flatness since every uneven part will be visible.



Into the honeycombs around inserts extra epoxy will be added to gain higher stability (I hope), it seems logical.





The Spine Frame with Motherboard Mounting Points and next pictures spacers.





The top front of the Frame has room for a handle/belt to drag the Frame out of the Shell, earlier design had this belt on the bottom of the frame. This is not optimal having to push the buttons on top of the Shell to unlock the frame from the Shell while dragging the Shell and Frame apart at the bottom of the frame. So this has been corrected, almost.

I should explain this a bit more but I will leave this for the next update.




A picture with different Motherboard sizes the ones this enclosure might hold have been chosen and mounting inserts placed.




Motherboard with spacers, Wire channels and tunnels, Wire Pass Through Cut Outs.




To make this a bit easier for me to plan I created a motherboard with the main connections added to check if the plans are correct and after a bit of adjustment they are .The Motherboard used as a template is https://dlcdnets.asus.com/pub/ASUS/mb/Socket AM5/ProArt X670E-CREATOR WIFI/E21293_ProArt_X670E-CREATOR_WIFI_UM_V2_WEB.pdf?model=ProArt X670E-CREATOR WIFI









The Panel for the access to the REAR of the CPU can be seen here above and below.




The Base for the (3BL) 3 Buttons Lock can be seen below I haven't added the Buttons, Springs or bolts. the base is mounted on the Top Front on the left side. Small changes have been made with the positioning of a LED light mount into the base. A bolt at the button of the base should make it easy to adjust the height of the buttons so the Shell and buttons are at the same level.




Last but not least a new front is (might be) shown on the next two pictures the idea is to stream line the building process and avoid to smack to many small peaces together when its not needed.




The entire process of redoing the design is to get a higher precision since I started with seeing if I was in the ball park this round is for the building process.

 

Hi again

I got some back lock including this:

This is the next post to state the obvious!


On the picture above, the Exo-Frame of the Spine can be seen, the part to notice is the Mounting area for the Strap (Belt) Clip. Check the next pictures. The straps purpose is to be able to pull the frame with potentially 15 kg of weight out of the Shell. The Strap, Spine, Rear Panel and Front Panel are all mounted inside the frame that will carry the stress of most actions.




The Strap Clip will be made of Carbon Fiber or of Aluminium I am still undecided. The Clip/Collar is important to stabilize the track of the belt and to avoid the filter frame from being bent when using the strap.





Picture with the Strap. The strap is made of a Belt, check the 2nd picture.







The Filter Frame is as mentioned mounted with magnets the small strap on the left is to eject this part out of the frame when needed. The frame has a (likely glued) filter fabric on the front. The filter frame is designed to be pulled out and turned when in transport to protect the filter fabric by inserting it a cm into the frame.

The filter frame around the handle will be split, to be explained at a later date.








On the picture above the 3 sensor panels that are actually 4 sensor panels on the front left side, are made to be exchangeable with different sensors and easy accessible when needed.

On the right side the OLED Cut Outs can be seen. The Air Intake Tunnel is also reestablished (The light green color), below it can be seen from the rear.




Last but not least, On the picture below, can be seen purple areas that are stand ins for my idea of telescopic sensor panels. The reason being that I can't see any other working options for reliable sensor (laser) coverage of the area in front of the case. The original idea was to have small tunnels in the filter fabric for the laser to shine through. To be honest I was not a believer in the earlier design, but until a new idea pops up I try to keep the idea alive.

The disadvantage of this idea is the human factor aka me. Remembering to push the telescopic sensors in and out leaves a lot of problems on the table. In daily use the chance of the sensors being bent by anybody in the area is a real posibility.





Yes the 4th sensor panel can be seen above with USB slots inserted Its at the bottom center of the picture.

 

And its Friday Night again and I can't sleep so here I am. A shame that Lance Reddick is dead he is missed!

I looked at the front of the case and yea well the Reservoir is to small for me to Enjoy it, so I did what We do I changed its size. I have for some time known that this needed to be done to have enough water for the attachment of the Emergency Cooling System, here it is:






Also had to check that it still fits on the Water Cooling Rig (WCR)







Picture above with the 1st Arduino Due mounted still needs more planning of the remaining components and wire channels.

The Back Panel of the Reservoir also needed a work over for the position of inserted metal components likely brass to be able to register water levels and not to forget silicone rings to keep it all water tight. Reminder to self, get access to a CNC or save up and buy one.





Each insert in the Reservoir indicates 10% of water level until 100% the last is 105% indicating overflow is likely!




Since I had time for it I designed a 1 Button Locks (1BL) here they are, they are mounted on the inside of the frame (like the 3BL) to secure the WCR. Long overdue!






The Tube on "top" of the 1BL is for a 5mm LED to have the option of mounting some light that indicates the position of the Button.

I have been improving the WCR Rail inside the frame, I am not done yet but here are some updated pics with Cross Bracing:




Last but not least











This is just one of many **** happens situations experienced when I (also known as the noob) work with CAD programs



That's it for now






More adjustments and preparations for the end

 

The Reservoir next to last iteration.

I did look on one of my check lists for this Bit-tech video to get the details right and I found it so here it comes:

1st to mount the Panels and be a bit safer from water damage 40 mm was recommended between bolts and now each panel has 4 more bolts added.

Next the O-rings are now on the inside of the bolts else each bolt would have needed an O-ring, I was working in that direction but lazy is the way to go, so work smart not hard.

The channels for the O-rings are 1 mm deep on the Reservoir body and ½ a mm deep on the Front and Rear Panels. Reminder to self Chamfer the channels. O-rings have a Diameter of 1.5 mm.

The Bolts for the entire project (here not including the spacers for the motherboard) are going to be Allen Hexagon Socket Flat Head Countersunk Screw Bolts M3 (3 mm diameter) in different lengths. Note to self might be a good idea marking what holes need what length bolts. Why these Bolts, 1st of I can not have any bolt higher then the surface of the Reservoir or Frame this would kill the project. Parts of the head and the thread of the bolts can be removed where needed, and it will be needed The other reason for using 1 type of bolt is that 1 Allen wrench can be embedded into the frame surface to fix what might need adjustment.





Lets start with the 1st design failure and why it is a no go. The wire channels can be a problem when the reservoir needs to pivot so this one is out:




The reservoir front a basic version with markings indicating the water level.




Updated Front Panel with text






The water level metal parts are inserted into the Rear Panel to be closer to the electronics and be less of a problem when wiring and using the Pivot function of the Reservoir. The Wire Channels are cut into the Panel above the markings on the front so not to be a problem when reading the markings. I will add a picture at the end showing why.

Metal Parts




Rear Panel Basic with a hole in the Axis for wiring seems to be the easiest solution but it took me more time then I'd like to admit to get there!



Next the break through the wire channels are cut into the Panel 1 mm deep and lead to the Hole in the Axis. The Wires (0.8 mm in diameter) are continuously added from the top thus the Pyramid shape of the wire channel.





The Axis is mounted after the wire channels are cut with a "glue" that is used with large scale aquariums it doesn't glue it fuses the material, like a welded solution.

The wires are going below the Axis and coming out of the center Axis hole in a bundle.










Text added to the Reservoir Top Holes




Text added to the Reservoir Top Holes




Their is actually some logic to the madness try to imagine pivoting the reservoir with the water out going to the pumps. Then refilling the Reservoir while the system is running and the Fill port is out side the frame.

Wire Channel Disaster, rest where deleted no idea why :/




The last iteration to come should include wire channels or holders for the wires potentially added from the holes on Top and at the Bottom of the Reservoir.

That's all folks!

Almost had to add these 3 pictures of an all Acrylic Reservoir ->

 

Hi

This is something that has been lingering in the back of my mind for some time, CPU Cooler Heads and this is also true for the the GPU variant, so I want to putt this question out here in the case some of you have thought about this or even better have experience and knowledge in this area.

How to measure the efficiency of the CPU or GPU liquid Coolers to have a true comparison.

My take on this is (A) that I would start by measuring the amount of liquid that is inside the CPU Cooler head to the milliliter (aka 1/1000 of a liter or ml) (B) next measuring the amount of liquid passing through the head per minute and last but not least (C) measuring the temperature difference of water entering and exiting the part.

This should give a clear picture of energy removed from the CPU under load especially (D) when compared with the CPU temperature in the same interval.

Updated: Since we have point B. time as a factor is introduced by Liters Per Minute (lpm).

My questions are:

Do any of you in this forum have another take on this ?

and have some of you made the calculations?

My take on this I am not going to reinvent the wheel if others have done the job

Looking forward to get your input!'

I am not in the right mindset today to go over this but I found some links with information that might be useful one of those is:

https://theengineeringmindset.com/chiller-cooling-capacity-calculate/

and

https://www.overclockers.com/guide-deltat-water-cooling/

 

A design update on:

The telescope edition of the TOF (Time Of Flight) sensor these are an addition to the Stand Alone System for Monitoring and Control (SAS4MC). The sensors are meant to ensure that Airflow isn't blocked on the Air Intake Areas and potentially on the Air Vents (Air Out) areas too.

To be honest its a secondary security system since its not a challenge to monitor the cooling system in general, if there are temperatures above nominal then the system can shut down. So its to avoid a untimely shut downs and increase system uptime and hardware security.

This is a 1st draft, there are still unsolved problems like how to make it pop out after being transported. The Sensors are only supposed to be 1cm out of the Filter fabric (read front of case) the corners are rounded in the present edition.

TOF Specs:



Estimating an angel of 75 degrees for the a 10 cm area in front of the Air intake (front of case) to be free of blocking elements. Please comment if you have a better ideas.



My 1st virtual of the TOF




The test of a revolved TOF for the initial design to be able to se volume needed.



Design of side view not sexy yet but it gives a good idea of the volume needed for the final options.




Next Edition



And another version still very incomplete:





The advantage of this approach of rapid incomplete prototyping is fast getting close to the envelope/dimensions that the telescope sensor can be within. It helps me create ideas for the final form and access points into the object. Rounded corners is to make the part less vulnerable to essentially me!

To be honest I had to do something else, I have been sloppy in my approach to CAD with the fix points and Planes in the design of the entire case. So karma hit me, when it came back and kicked me in the private parts 2 days ago :/ Naturally when I tried to change a small detail it all came down like a house of cards, after 3 hours of looking for a fix I decided to start from scratch OK not really form scratch since all the detail can be transferred to a new design. So don't do as I did in this case

I'm looking forward to se more design from all of you in here!