Design Methods

From TheBestCaseScenario

[edit] How should one make sure his project will be a successful one? 3D!

To make things clear, I'm not talking about popularity, but about one's goal...

It is a bit hard for me to restrain myself each time I talk about graphics, but I’ll make an effort and try my best to help those who read this. I wish to elaborate on the planning stage of a build, mainly on design methods. And on cheaply designing something beautiful and never before seen (maybe) without actually doing any hands-on work… Well, excluding the fact that you operate the mouse and keyboard with your hands.

It is clear to me now that you can do a good job the first time you try, but in most cases you might not know what you exactly want… or you might not like what you thought you wanted… and maybe you don’t have that much material to work with. For instance, take me: I had a cheap no name case and a lot of ideas, but as I explored them the 3D way (and maybe applied them…), I figured out that they weren’t that good (I love criticism, especially when I apply it on myself). I have lots of .max files that on a second look seem just stupid and make me wonder what I was thinking. But what 3D does is take your mind to another level. Even though you don’t get experience in the manual labor field, you know this is right or wrong from the “looks of it”.

[edit] What are you building?

But, before delving further into the subject, think about what you are trying to achieve: aesthetics and/or functionality? Maybe it is very important for a particular piece of hardware to work 24/7 and maybe under high load and maybe overclocked… Well, then maybe you want to try something ingenious as a cooling assembly/power supply. Or maybe you have something that looks too common for your taste and you want it unique. Or why not go with both? When designing a new computer, one must take into account his tastes, needs and maybe abilities, but an iron will is quite a powerful asset. And if you have the patience to back everything up... You can go a long way :) Learning as you go is not bad at all. Only after repeated use of a technique you gain a skill, right? So everyone should have a rock-hard determination. For instance, some modders I’ve seen have taken perfectionism to a whole new level for me: they do it again and again, until they like it. RESPECT for that.

[edit] Then what?!

After you decide what you want it to be, consider drawing things down a couple of times. I learned that by sketching dimensions and filling in the details again and again helps things shape up. Some shapes disappear, some are refined. You actually might discover a better way or end up doing something else, that can be miles away from what you started with... but as long as you like it better, as long as it works better, it’s a win-win situation. It’s even better if you have the skills to do this 3D… I won’t recommend any specific program, as there are tons out there. Just find one that is most suited from your point of view: easy to work with so as not to make the planning more difficult. Consider it an extension of your mind :) I like numeric input of dimensions and position as it tells a lot and a good mouse movement implementation in a CAD app. Most important, the learning curve is essential: after a couple of years of steadily learning about 3D and computers, I found out that common sense is all it takes (besides patience). Whether you are tackling a complex shape with lots of tight tolerances or a sinuous design, made natural and pleasant looking by its curves, take it one step at a time and approach it in a tiny-step by tiny-step way. Although sometimes this might be hindered by your pc’s abilities, usually you can simulate any geometric shape that exists (or will exist). I have tried several CADs and all of them require some knowledge and getting used to, so anything is useful after you learn how to use it. (Although… My opinion is that a more professional design utility is much better for true design and if you should dedicate time to learning something, than it might be best to learn something a bit trickier… )

So, what to do when you come up with a great idea? Well, think about airflow, clearance and "do-ability".

[edit] Do-ability

I'll start with the last: can you build what you want? Even if you know you can't, maybe you know it's in your ability to learn how to do the job. All who have done a decent amount of work can say it's best to try your abilities on scrap first... if it's the first time you do it :) this way, you turn scrap into art, not the other way around ;) Also, think about tools, methods and the investment... and if there is another way to do it :) Read and watch what others have done before anything else, unless you already know... No, forget that; even if you do have some sort of experience you should still do research before doing the actual work. Even if one thing is of value in a work log, it was worth reading the whole thing. And if you know all that was said, than you can be reassured you are doing it right of course, once you’ve learned how to solder a resistor onto a LED, calculate the required value for it to safely work and installed it, maybe it would be faster to just do the work time after time, and forget about the reading part :)

Considering the complexity of the project beforehand is important because this way you won't run into (as many) problems when doing the actual work. And something else that helps is to "divide and conquer". Think about the pieces you will incorporate in your design and take it apart before it's completed. Analyze each component and do revisions as necessary. Here 3D helps because after refining pieces of virtual material you can test-fit them… if they don't fit, all you have lost is time and you can go back and refine some more... if they do fit, <excellent> It’s important not to give up at this point and not to be discouraged when something just doesn’t want to be made. Remember that you learn just once and after that you can do it again and again, better and better. And if you can do it in virtual reality, chances are high it will be as easy in real life, with the proper tools and materials, of course.

[edit] Clearance

After this stage, a nicely calibrated ruler and a rested eye are best suited for avoiding errors down the road. I always measure several times and do the math to see if everything is in order. For instance, take one DVD burner, one 80mm fan and a regular PSU. I measured, cut, test fitted and… oops, the case is not big enough for all of them… at least in 3D. In real life, I forgot to take into account 30mm of front panel space. Still, even with that slack, another PSU wouldn’t fit, because mine was smallish in depth (and quality).

Therefore, a thing you should consider when assembling the virtual model is clearance: does everything fit with slack? 'cause there will be wires and maybe tubes... And who knows what else you planned to put in there. So, either be very thorough and sketch everything or leave room to breathe :) And depending on your budget and design, this is a good time to change parts as well... A slim optical unit? a smaller cooling rig (at least as good as before ;) )? Maybe another method of laying the innards? Hmmm... You decide... but it's best to be open minded all the time, as there is always another way, maybe better and easier. Again, 3D helps because you can explore :) and you can never screw-up.

Looking around the web and on different sites, I’ve seen that not much importance is given to the actual variations of tolerance zones in real components. I admit I haven’t seen everything, but I’ve seen only one or two virtual copies of… motherboards (just popped into my mind) and dozens of boxes with a bitmap on top. I’m not saying it is wrong, but keep in mind that once you have a good copy you don’t have to remake it over and over again. You can just use a copy of that and rethink around it. Plus, I know some people might think better with the actual piece of hardware in their hands and that the picture only helps put order in thought. But, I think a small investment of time pays itself hundred-fold. Also, I’m not telling anyone to go obsessive and start measuring caps and ICs and take pictures of them and create a replica of the thing they work on, but try to take the most important dimensions into account, not just the volume. Anything that you might need to use or that might get in your way is a good suspect for “virtualizing”. Take the mating of a motherboard, CPU and HSF. Tons of problems can arise here even if no modding is involved.

A good advice is to keep a clear, calm head and again, never give up. One you master something, the satisfaction is huge. Trying to re-engineer (‘cause basically, that’s what we do) pieces of hardware to fit our purposes and tastes is hard, but by putting another step in the process, that can be as challenging as the build itself, just increases the overall satisfaction of having built something unique and personal.

[edit] Airflow

The last variable in the equation is airflow. This is what actually influenced me years back to do some fidgeting around… It all happened after buying my “second” PC (the first was a 486 Christmas gift I had no say in, called him Bronto, the second PC my parents bought being my first one to configure and order). It worked “nice” for almost a year, but after that period, I had one hell of a full time learning about parts and stuff. basically, I couldn’t put up with the guys in the shop acting all knowingly and trying to fix my PC by replacing the keyboard. After realizing that electronics need to be kept cool, that the thing they sold me was crap and that you could get more bang for the same buck, I changed my fans and heat sinks… at that time, I had no idea of what was actually available “out there” and there weren’t many possibilities either. But time changed, I got my hands on a nice CD with a nice 3D tool and much changed. Then I bought a modem and started learning. Then physics and mathematics came along and helped further. Cooling grew into an obsession and… sorry for ranting. This takes me back to airflow: if it’s air-cooling, water-cooling or something else, nothing helps lower the temperature better than more cool air (given the same heat sinks). There are 2 ways in my opinion to ruin a good rig: low efficiency heatsinks or the lack of airflow.

The precision with which manufacturers make their products, the calculations involved and their efforts (as well as ours), can be swept aside in an instant. Think of the old AT computers. As heat output grew, they needed another way of arranging the components: ATX.

Then, after some CPU producers decided that heat output and energy savings are not as important as a silly speed race, BTX appeared to further improve cooling efficiency.

Nowadays, the energy saving options and low voltage components cooled the situation down a little, but why not keep the same high efficiency to get noise levels lower? Or get that nice overclock?

This is where a careful positioning of things inside a PC using a 3D utility can help reduce noise and temperatures. Think BTX. I always found wind-tunnels a blessing because if air has nowhere else to go, it has to pass through the fins of the cooler, right? And you can keep cold and hot air in separate places and prevent them from mixing. Whether you are building a custom case or customizing an existing model, after you have the parts and tolerances in the virtual “battlefield”, you can find the best way to route air from the intake to the exhaust. The internet helps a lot here, because you don’t necessarily have to reinvent the wheel. There are some truly brilliant designs and lots of support (think forums, guides, and work logs), so only your imagination is the limit (hopefully). And although water and phase change are very nice for cool and quiet computing in any conditions, the problem is they alone just throw away all the benefits of ATX or BTX designs. Why? Because a computer is made of 2 things: things that are powered and things that power things. The digital parts don’t need cooling all the time (think of you NIC or non-hi-end sound card), just high frequency components, like processors and memory. But the voltage regulators, some caps, some ICs and lots of other black tiny pieces can suffer when you remove the gentle breeze of air that flows through a case… there is a lot of science behind this, but I hope you’ll trust me. Even if you have a full water loop cooling all the usual chips, you could let something fry because with no airflow, small parts with currents running through them tend to get hotter and hotter and by heating the air around them, they harm other small parts… and as you can see, even an ultra low noise fan will make the difference in the long run. (This reminds me! If you ever get a chance to work with a CNC machine, taking the time to measure the distances between holes and their diameter is a blessing ;) )

[edit] Wind-tunnels?

An important characteristic of cases is that they can act as wind-tunnels. By using small compartments with dedicated intakes and exhausts (with a fan at least at this stage) you essentially turn the case into several wind-tunnels and force air to move more efficiently. This also influences the choice of heatsinks. If you want to have a silent and also efficient system, low restriction radiators with tall pins, with wide spaces in-between positioned in the air path are a must. Some stores supply small radiators for all kinds of parts, so even if you have a water-cooled chipset, voltage regulators, GPU, CPU etc. you can still improve cooling if that gentle breeze exists.

[edit] Wrapping up

After carefully selecting parts, measuring them, copying them in 3D, assembling them and putting all stuff together, it is a good idea to analyze the created geometry and if all fits 100% perfectly, you can put on a coat of paint Here, 3D is as amazing as a good airbrush. Actually, I think absolute flatness can be achieved only in virtual reality. It is true that flames and intricate designs seem impossible to achieve, but with a matching 2D utility, some lights and patience, photo quality can be achieved and not only does the “finished” product look real, but if you were employed to do it, this would be like a prototype… taken to a professional photo-session.

[edit] Short exercise

I hope my enthusiasm 'till now convinced you that planning and building something virtually, using 3D utilities, is very helpful. Not to mention it makes you feel good at the end. Remember to measure twice and never cut ;) Of course, for those that don’t like reading that much, I have some pictures too. It took me a fortnight’s time, but I considered the exercise to be beneficial. So I started designing a complete cooling system, suited for what I modestly own now. I looked at my motherboard and thought what needed active cooling and what not... what needed to be reproduced in 3D and what left behind… and of course, the design changed constantly... If this would have been made in real life, I would be a lot poorer now and not much wiser.

[edit] Motherboard

I started with a high resolution photograph of the motherboard and used this to position the components I beforehand measured for clearance... What I considered important was the socket, the northbridge and southbridge chips, the voltage regulators, the slots and the plugs... plus all chips and capacitors that are visible and of a relatively medium to high size. Also, mounting holes are essential.

[edit] Video board

I searched for pics of the video board too, because I planned on watercooling it too… virtually. In real life, it wouldn’t actually be worth it, except for exercise purposes. A silencer is enough for this “beast”. Because there are RAM chips on both sides, I used 2 maps and lots of measurements... this was made easier by the fact that I have 2 video adapters to play with. In my opinion, planning for a one piece waterblock is trickier than using RAM-sinks and a GPU-block, so I took that way. Seeing that the video board plugs in the mother board, this was my first challenge.

[edit] Video block

I used the stock cooler to design the new one, plus some modifications at the RAM contact areas. The first design was simple, but as an experienced modder can see (in a higher resolution picture ), the turbulence was unwillingly diminished. Although not that bad, why not correct this problem? It didn’t actually happen, so it’s easy.

[edit] Video block rev2

This revision used a pin matrix for both GPU and RAM areas, as this is low restriction and turbulent enough for my modest budget (therefore a cheap pump could move enough water, and when possible a high-flow model would push the performance a notch or two).

[edit] Video block filled

Finally, I could mount the new block and see how it matches with the board… I even filled it with greenish coolant, without using tubes or fittings ;) It took about a day or two of refining, but I was happy with the way I made the block too. It is comprised of a copper plate, machined to create the pin matrix and the contact points and an acrylic or similar material top, machined to obtain the coolant path. Both parts are sealed with an O-ring and lots of bolts, counter-sunk in the base.

[edit] Video blocks assembled

After working with 3D models for a while, you learn that there are ways to ease the creation of derivates from the parent design. So I used the previous block to create the VRAM only water-block, for the remainder of the RAM chips on the board. Then I connected the two and well… picture says it... I liked the simple way it could be made, so I used this design further… copper, seal, acrylic, bolts.

[edit] The system

After having the motherboard and the video board made, I connected them and continued with the larger picture: The CPU-block and the chipset-block. Again, having measures and mounting point like in real life helps, so I made some primitives to work with (the more objects in a scene, the harder it is to work, so I sometimes go back and forth between scenes and find that I need an upgrade ).

[edit] CPU block

First version: The CPU block I had I mind was “conceived” during a quiet voyage, with just pencil and paper at hand… I was thinking of jet impingement and pin matrix blocks… I hate the idea of having a high-flow system if there is high pressure involved… basically, I am no master crafter and if the block would spring a leak, I would be toast. So I combined the 2 designs, with cooling in mind of course. The cool water enters in the middle and is pushed through tiny hole in the intake fitting. These high-pressure streams hit the base, but they enter small wells first, to maximize contact and turbulence. Then the water flows back and goes through a labyrinth wetting even more of the base surface. I think the design helps cool hot small cores (socket A) and big heat-spreaders (new CPUs), while still putting maximum coolness in the center

[edit] Northbridge block

I used a pin matrix for the northbridge-block, but again I was pondering what would the water do if it had the choice to flow through this… and put some more resistance in the middle part, so as to divert the flow and try to make it hit as hard as possible on all pins. 3D models help, but a good mathematics CAD utility makes the difference ;) At first, I thought that 3/4 “ fittings would be best, but I reconsidered afterwards, because of the complexity of the tubing… so I changed them to 3/8” for lots of blocks…

[edit] Voltage-Regulator block

In a normal system, a well cooled voltage regulator module is important for stability. Some manufacturers use aluminum or copper or heat-pipe assemblies to cool these (others leave the MOS-FETs bare…), but sometimes, it may not be enough. So a voltage-regulator-block is due. Same design, small inlets, pin matrix. But this one is also calculated, not just put there (there must be a balance between height, width and spacing of the pins for proper thermodynamics to occur).

[edit] Assembled system: take 1

The assembled system, which showed me the problem I had to solve. I needed a fitting to split the flow or a reduction… didn’t like the idea of having bottlenecks in the flow… and I knew there is a way to avoid Y’s or T’s in the loop… and I was thinking of running 2 parallel lines now… so I redesigned some blocks

[edit] CPU block rev2

This is the redesigned block for the CPU. Again, I must stress that if I had to use real materials and real time… the costs would have been… well, I can’t really tell, ‘cause all I will pay is the electricity bill now Same principles are applied with the difference that now I’ll use a 14mm ID input and two 10mm ID outs, and that the inner labyrinth changed to accommodate more water volume, but with higher speeds. I removed the pins and replaced them with a more intricate turn system that hopefully keeps luminary flow at a low enough level not to hinder performance. As nothing is really forged, anything can change (I have some ideas right now actually, but I’ll keep them for a later time).

[edit] Assembled system: take 2

New assembly, not completed, but getting closer… with this particular motherboard, I was forced to use a heat sink for the southbridge. This is why I dropped the idea of fully cooling my rig now. I wanted a low-pressure-drop, quiet radiator, a quiet pump and quiet fans cooling first the CPU, then the voltage regulators and the video board and on the parallel line the northbridge, southbridge and the audio processor (yes, why not, it gets a bit warm )... eventually. But then others ideas haunted me. And I gave up on the project for a while. I thought I may continue or redo it: no harm done anyway, nothing got wasted… only electrons were harmed.

[edit] Just a final thought

I hope that by sharing my thoughts and experience, I managed to improve the community and help out anyone considering using 3D CAD as a tool for modding.

-Dgrmkrp