Completed STX160.0 - The most powerful ATX unit, in the world!

Hahutzy

Airflow Optimizer
Sep 9, 2015
252
187

I like the fact that this orientation lines up the 4x PCIE to 16x PCIE really nicely!

Two thoughts on the HDPLEX pcie ribbon:

1) You mentioned that you're using the P4SM2, and it's got a 4x connector on it whereas the HDPLEX ribbon comes with a 16x to 16x gold finger board by default. Can they make a 4x to 16x board for this?

2) The HDPLEX pcie ribbon has angled female connectors on the same side of the cable. The way your layout is currently, the connectors need to be on different sides of the cable OR you need to fold one side 180 degrees. But looking at @iFreilicht's review of the cable, folding it like that should be ok.
 

CC Ricers

Shrink Ray Wielder
Bronze Supporter
Nov 1, 2015
2,233
2,556
How did I miss this thread yesterday.. it quite blew up since then.

Thumbs up on going STX and with the HDPLEX 160w and trying to get the most juice you can from it. Looks like figuring out a good way to route and connect the PCIe ribbon without interfering with the CPU cooler would take some time.

Don't have much else to add, but maybe you should print a custom label for the STX160.0 enclosure to go with a faux PSU look.
 
  • Like
Reactions: Soul_Est

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
I like the fact that this orientation lines up the 4x PCIE to 16x PCIE really nicely!

Yup, it's almost perfectly aligned.

1) You mentioned that you're using the P4SM2, and it's got a 4x connector on it whereas the HDPLEX ribbon comes with a 16x to 16x gold finger board by default. Can they make a 4x to 16x board for this?

The P4SM2 is open ended so using a x8 or x16 card or extender is not a problem for the adapter itself. In the next update or two we'll see that the HDPLEX extender has other fitment issues though.

folding it like that should be ok.

That's what I was planning, since there's enough cable length to bend it over to make the connectors line up.

Don't have much else to add, but maybe you should print a custom label for the STX160.0 enclosure to go with a faux PSU look.

That's what I'm thinking. Though I don't know if I'll get around to it since it won't be visible the way I'm planning to use the case.
 

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
I just meant to write a paragraph or two about my choice of CAD software before getting into the case design but got a bit carried away :p Feel free to skip this wall of text post if a brief introduction to CAD software doesn't interest you.

A brief introduction to CAD software

As I mentioned in the previous update, SketchUp is a really useful tool for roughing out a case layout but I don't find it very suited to actually designing a sheet metal case and producing the files needed for manufacturing.

One problem is that SketchUp is a surface modeler and not a solid modeler, Aidan Chopra, author of the SketchUp for Dummies book (and really nice guy) has a simple breakdown of what that means here. Basically SketchUp treats everything as being composed of vertexes, lines, and surfaces, they're "hollow". This can be worked around to a degree (in fact the Pro version of SketchUp has solid modeling tools that emulate solid modeling) but one major issue for designing computer cases is that it has to approximate all curves and circles as polygons.



It may not be noticeable at first since by default SketchUp creates arcs with a higher number of segments and hides the lines between segments, giving the illusion of a smooth curve. Here I've reduced the segment count and have unhidden the lines to demonstrate. I've also deleted the face nearest the camera to show how it's hollow on the inside.

If SketchUp is just used to create 2D drawings then these limitations don't matter as much. But if used to create 3D models then it's a problem since it'll make for a messy import, requiring the manufacturer to spend time either cleaning up the model or just redrawing it from scratch in their CAD program (both of which costs money). And that's ignoring the fact that it's typically difficult to get a SketchUp file into a file format such as IGES or STEP that manufacturers can work with.

---------------------------​

To be perfectly clear though, I'm not hating on SketchUp! In fact I just spent all of the last update going over how I used it for the beginning stage of this project. It's easy to use, there's a large selection of computer parts made in SketchUp available online (though of wildly varying degrees of accuracy), and it's free :D

But I just wanted to go over its limitations because I think anyone serious about designing computers cases for production should seriously consider a more full-featured CAD program. There are many different CAD programs out there, but by far the one most commonly used in small and medium-sized sheet metal shops is Dassault Systèmes SolidWorks. From what I've seen Autodesk Inventor and Siemens Solid Edge are also used but are not nearly as common as SW.

The main problem is cost though, using SolidWorks would make it easier to communicate with manufacturers since that's what most of them use but it's $5000 or more upfront plus annual maintenance fees!

There are a few free options however: Onshape, Fusion 360 (for students, hobbyists, and startups), and FreeCAD for example. Fusion 360 has sheet metal functionality but Onshape and FreeCAD currently do not. Onshape has it on their roadmap and hopefully the first stage of it will be out later this year though and there are a few plugins to add sheet metal functionality to FreeCAD.

Update: Onshape added basic sheet metal functionality in February 2017.

Dedicated sheet metal tools aren't strictly required though, they just make things easier:



Base sheet, two flanges, and two mitered flanges off those in just a few clicks! So nice :)

---------------------------​

I've mentioned surface vs solid modeling, but within solid modeling there are two main camps: parametric and direct modelers. Most CAD programs (at least those I've looked at for sheet metal) are parametric (also called feature-based or history-based). They rely on defining each new feature by referencing existing features.

Examples of parametric modeling include SolidWorks, Onshape, Inventor, and FreeCAD.



Here I'm using Onshape to demonstrate. This is a 2D sketch of a rectangle with a smaller rectangle inside it, except as far as the software is concerned it's really 8 lines with perpendicularity, coincident, parallel, horizontal, and dimension constraints applied. You don't typically have to specify each and every one of those constraints (the software adds many of them automatically depending on what you're modeling) but that's basically how they work. This can be extremely helpful if the constraints are setup in a logical manner since it means making a change in one place will correctly propagate through the model.

For example, in the above illustration I have the inside rectangle with a 10mm dimensional constraint from the top edge of the outside rectangle. If the intention is that the inside rectangle should always be 10mm from the top edge then this is great! I can change the size of the outside rectangle and the inside rectangle will always be 10mm from the top edge.

The problem with this though is if I had meant for the inside rectangle to be centered vertically within the outside rectangle. It happens to be right now, but if I change the height of the outside rectangle to 45mm then the inside rectangle will now be 10mm from the top edge and 15mm from the bottom since it's constrained to always be 10mm from the top edge.

Easily fixable in this contrived example but you can imagine working on a more complex design and accidentally boxing yourself into a corner with a poorly thought out feature tree.

---------------------------​

So on the other end is direct modeling. Like it sounds, this involves directly pushing and pulling on the geometry, free of the constraints of, well, constraints and feature histories and is well-suited for the conceptual stage or working with existing models where the feature tree is either poorly setup or nonexistent.

Examples of direct modeling include ANSYS SpaceClaim (shown in the above sheet metal animation) and PTC Creo Elements/Direct (at least I think it's that one, PTC's portfolio is confusing).

Direct modelers have the advantage of being easier to learn for users familiar with programs like SketchUp, 3DS Max, Blender, and other 3D modelers since they operate similarly.

The lack of a feature tree is direct modeling's strength but also its weakness. A thoughtfully constructed feature tree can make modifications to existing designs very easy. For example, a sheet metal manufacturer may have a standard model of rack mount chassis that different customers need in different widths.

With a direct modeler, changing the width would involve selecting and moving all the flanges, cutouts, and other features to match the new width.

With a parametric modeler with a feature tree setup with this exact change in mind, it could be as simple as changing a single dimension and everything else falls into place via the constraints.

---------------------------
There are also hybrid systems that use a combination of both parametric and direct modeling. Solid Edge is an example of this.

---------------------------​

Last year I had tried several different CAD packages and ended up settling on SpaceClaim, a powerful direct modeler with a robust sheet metal package.

I would actually encourage new designers to try a parametric modeler first because I think they're a more powerful tool for chassis design though. Personally I'm too used to SketchUp and SpaceClaim is often touted as "SketchUp on steroids" and I've found that to be true so it made sense for me since I was able to transition and become productive quicker.

Also, it's not as apparent from this build log so far since I'm currently writing these posts after the fact, but my workflow is very disorganized. I'll model something, and change and redo it several times before finding something that works the way I want. So I would be constantly running into the problem of poorly setup feature trees.

One feature unique to SpaceClaim that also appealed to me is that it has native import/export file support for SketchUp. SketchUp is a surface modeler so it generates surface-based files, so when imported into most CAD programs the model is imported as a bunch of surfaces.

SpaceClaim on the other hand will parse the SketchUp file, and as long as the objects are "watertight", meaning no holes or gaps, it will convert them to solids. If the objects aren't watertight, SC has a suite of tools for repairing imported models to fix things like missing faces, duplicate edges, etc. Not only that, but if a circle in the SketchUp model is a native circle, it will get converted to an actual circle and not the polygon approximation. I specify native circles though because while SketchUp circles are really just polygons, SketchUp does treat them differently from normal polygons (mostly from a metadata perspective though and not the actual geometry).

---------------------------​

Quick note about evaluating and buying CAD software, it's very much like purchasing a car. If you give your actual email address and phone number when signing up for trials of CAD software, fully expect to get an email or phone call from a sales rep. And annoying as that is, I'd recommend not giving fake addresses/numbers if you're seriously considering purchasing that CAD package because oftentimes you can't just buy the software online, you usually have to go through the sales rep.

Also, it's normal for prices to not be listed anywhere on the website, they do this because prices are negotiable. YMMV but one trick to save money is to wait until near the end of the year since they may offer special pricing to up their sales figures for the year.

---------------------------
Next update I'll start going over the actual case design process, promise! :)

Table of Contents

Next Update
 
Last edited:

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
Just a suggestion, maybe you've thought about it, but you could flip the motherboard so it could ventilate from the bottom, maximizing cooler size potential, like a Scythe Big Shuriken 2 rev. B.

Oops, forgot to reply to this earlier. I have thought about this quite a bit actually, you'll see why in a future update ;)
 
  • Like
Reactions: Phuncz

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
This needs to be it's own post in the resources section

I'll add a link to the Useful References guide for now. For a formal writeup I need to go back and re-evaluate some of the CAD software I mentioned. Fusion 360 didn't have sheet metal in the free version last time I was looking.

Very cool idea! I like the "Next update" link in each update, that's very helpful. Good luck!

I'm trying to keep this build log more organized. We'll see if I can keep it up once I catch up to the present :p
 

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Fusion 360 didn't have sheet metal in the free version last time I was looking.

Fusion 360 didn't have sheet metal tools at all in the beginning, it was mainly designed for quick prototyping and small products, so 3D print first, then mould. They have added that by now, though, at least from what I know.

I'm trying to keep this build log more organized. We'll see if I can keep it up once I catch up to the present :p

I think it's a good approach. Certainly easier than what I'm doing, copying every damn update into the first post with a spoiler.
 

CC Ricers

Shrink Ray Wielder
Bronze Supporter
Nov 1, 2015
2,233
2,556
I'm gonna try FreeCAD for Windows later today. I don't remember who else was showing technical drawings from it here, but I really liked the quality of them, they were on par with some of the professional paid software.
 

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
Certainly easier than what I'm doing, copying every damn update into the first post with a spoiler.

Yeah, it's time consuming enough writing these things no way I'm going to do that :p
 

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
I got a bit sidetracked last update with a overview of CAD software but for real this time I'm going to go over some of the details of designing a case for production!

Designing the case part 2: U's, brakes, and ghosts

As I said at the end of Part 1, I had messed around with the rough layout as much as I could in SketchUp and it was time to move to my weapon of choice for chassis design, SpaceClaim, to start the real work. Since SpaceClaim supports native import of SketchUp files I simply opened my SketchUp mockup and could get right to work.



BTW, if the screenshots from SpaceClaim look "odd" to you, it's because, like most CAD programs, its default display mode is parallel projection. That means parallel lines in the model are displayed as parallel lines on the screen, regardless of orientation or distance. This is useful because it's much easier to check it everything lines up but it can look unnatural/unreal if you're not used to it.

Next was deciding a rough design for the sheet metal. For this project it was a pretty easy decision, I wanted the chassis to be composed of just two pieces to keep costs down and I wanted it to resemble a typical ATX PSU. So that meant a simple U-shape enclosure, unimaginatively named after the U-profile of the two halves:



As a note though, this type of enclosure will not work if I wanted to make a future shorter, GPU-less variant that exposed both the front and rear IO of the Mini-STX board. Protocase has a good explanation of why here.

Another limitation of U-shaped enclosures is the proportions can't be too deep because otherwise there isn't room for the press brake tooling. To grasp this problem requires a basic understanding of press brakes though.

---------------------------​

Prototypes and very small runs may be done with a manual brake while mass-production is typically done with stamping but press brakes are the primary method of bending sheet metal for the small to medium-size production scales of interest to SFF cases so that's what I'll focus on.

Press brakes work by pressing the sheet metal between two vertically-oriented pieces of metal to force it to bend. The upper piece is the punch, which is held in a punch holder, and the bottom piece is the die. More detailed overview here.


Image source (flipped and size reduced)

Here's an animation showing a press brake in action. That red piece to the left is the backgauge, it's a computer controlled backstop that the machine operator pushes the work piece against so that the bend ends up in the correct position.

Side note: the backgauge is why it's helpful to make sure the opposite edge of the sheet from the bend is parallel to the bend. Not all manufacturers will have fancy press brakes with backgauges that can position themselves at an angle to the punch/die.

And here you can see those same elements in reality:


Image source

So now, let's look what happens when trying to make the second bend in a U-shape enclosure that's too deep relative to the width:



Uh oh! The work piece hits the punch holder, certainly damaging the sheet metal part, but also possibly damaging the tooling or even injuring the operator :(

Obviously that's no good so if you're uncertain you'll want to rotate the part 45° and mockup the punch vertically above the bends and check if the other flanges will interfere:



You're probably thinking it looks like it'll just barely fit right? But you'd be wrong!

This won't actually work because of what's called spring back. More of a problem with steel than with softer aluminum, but the issue is that the metal tries to return to it's original shape after being bent. So to compensate the press brake software will bend the sheet metal a few degrees past 90° (and thus would still hit the punch holder in the above illustration) so that when the metal springs back it'll end up at 90°. If we look at the press brake animation again, but watch closely you can see this happening.



Because of this it's hard to ensure that all bends are at exact right-angles so be careful not to design in such a way that the case needs perfectly exact bends to work otherwise you'll pay for it later with a high parts reject rate.

Or at least, that's true of air bending. If you look at the animation you'll see that the work piece only touches the die at the outside edges of the V cutout, the bend is created in the air, hence the name. The alternatives would be bottom bending and coining. Both of those methods are less commonly used so to be safe assume your part will be formed using air bending. It's also probably safe to assume the air bending won't be done by a short bald kid with a big arrow on his head :p

Talk of spring back and air bending aside, there are special punches and holders to allow for deeper designs but those are specialized tools so be sure to verify with your sheet metal shop if they have the necessary tooling before designing a case that relies on it.

BTW, if you find this kind of thing interesting and want to learn more, I'd recommend checking out this list of useful references. In particular I've found SheetMetal.Me to be a hugely valuable resource for learning about sheet metal fabrication.

A good homework topic would be fabrication tolerances. I mentioned how the bend angle can't be perfectly formed with air bending above, but the length of the uncontrolled dimension won't be perfect either since the metal is both stretched and compressed at the point of the bend, the backgauge won't be absolutely perfectly positioned each time, the operator may not have pressed the work piece firmly against the backgauge, etc.

---------------------------​



In the previous step (before I got sidetracked by talking about press brakes) there are some sharp looking corners and one thing I like to do is round off outside corners as I create them. This can be done at the end of the design process but I've done that and let me tell you, it's no fun spending an hour tracking down and rounding every corner in your case.

This is perhaps not strictly necessary but I swear, if you don't round off the corners in your case and I slip while working in it and slit my wrist and bleed to death, I'm going to come back and haunt you! :mad:

If you do decide to round off the corners (good for you!), then be sure to double-check the manufacturer's approval drawings because sometimes when they prep the model for manufacturing the rounds get stripped out by accident. And it's your responsibility to ensure the drawings are accurate to what you want so if the rounds are missing and you sign off on the drawings then it's still your fault and you're still getting haunted ;)

---------------------------​

Okay, I had hoped to get more of the design process written up since I keep talking about it but it's hard to explain some of my decisions without also going over basics like how press brakes work. But at least I hope this extra information is helpful to anyone else looking to design a case.

Stay tuned for the next update where I talk about vents! And I'll probably get distracted by talking about lasers because they're actually very related XD

Table of Contents

Next Update
 
Last edited:

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
Before I get into the vents (and lasers!), a quick [ha, that didn't happen!] addendum to Part 2.

Designing the case part 3: Sheet thickness and bend radius

I had briefly mentioned it in the Part 1, but one common mistake for beginner's is not taking into account material thickness when estimating the dimensions of their proposed case. While you can model a case like that all day long in SketchUp, in reality it's not practical to make a case from sheets that are infinitely thin.


This isn't physically possible. The thickness of the case material must be accounted for!

I'll mostly talk about sheet metal since that's what I'm familiar with but really your choice of material is only limited by your imagination and ability to work the material, or willingness to pay someone who can. I've seen cases made out of wood (from plywood to hand carved exotic lumber), nylon (3D printed), acrylic, carbon fiber, copper, magnesium, cardboard, you name it.

For small run sheet metal SFF cases though there are basically two primary materials to pick from: aluminum and steel. Even limiting the discussion to those two metals there's lots that could be talked about, like the different alloys of aluminum (or aluminium for my non-American readers), cold rolled steel (CRS) vs galvanized vs stainless, etc. but that's beyond the scope of this post. For this discussion we're mostly concerned with the thickness of the metal so for the sake of simplicity assume I'm talking typical cold rolled steel for steel and 5052 for aluminum.

---------------------------​

Steel vs Aluminum

Real quick though, you're probably wondering: "should I use steel or aluminum?"

I would recommend deciding this early on if possible because it'll have a big effect on the thickness of the material needed. Aluminum is weaker than steel so generally you'll want to use 1.5-2x as thick of aluminum as you would steel to compensate. In a tight SFF case where every mm counts this could really mess up your design if you decide to switch halfway through from 0.91mm steel to 2.0mm aluminum. On the other hand, steel is much denser and a 0.91mm steel panel would actually weigh more than a 2.0mm aluminum panel of the same dimensions!

Steel Pros: Stronger, can use thinner gauges, magnetic, cheaper, tighter bend radius, easier to laser cut

Steel Cons: Much denser (for some this can be a pro though), rusts


Aluminum Pros: Lighter, can be anodized, perception of quality, less wear on punch tooling

Aluminum Cons: Weaker, has to be thicker to compensate, non-magnetic, more expensive, can't bend as tight to avoid cracking, harder to laser cut

This is a real brief rundown of a complex subject but hopefully this will give you a starting point on deciding between the two.

---------------------------​

Intro to sheet metal gauges

Okay, back to discussing sheet metal thicknesses:


Chart from the ever helpful SheetMetal.Me

Steel thickness is specified in the confusing gauge system (at least here it is) where the bigger the gauge number the thinner the material. Not only that, but 20 gauge CRS will be a slightly different thickness than 20 gauge galvanized which is slightly different than 20 gauge stainless.

The situation with sheet metal aluminum is better. There are gauges for Alu but it's typically specified by thickness, so you'd ask for 0.06"/1.5mm aluminum instead of 16 gauge.

Something to keep in mind is there is lots of rounding going on. For example, 20 gauge standard steel is commonly advertised as 0.036" but as can be seen in the chart above the nominal dimension is 0.0359", but the actual thickness probably won't be exactly either of those numbers either due to manufacturing variation so don't bother dialing in the sheet thickness to the umpteenth decimal place in the CAD model because it's pointless.

Also, when modeling sheet metal, only include the thickness of the metal itself. Paints and powder coats are not to be included in the CAD model, exception being if the metal itself comes with a coating like galvanized steel does. The thickness of finishes like paint and powder coats are not negligible though! We'll need to account for it and I'll go over that in a later post.

Another note is that not all gauges are readily available. I removed them from the screenshot but if you go to the SheetMetal.Me link you'll see 15, 17, 19, 21, and 23 steel gauges but don't design a case using them because your manufacturer won't have them. Along the same lines, don't design using arbitrary thicknesses. Look up what thicknesses are common for the material your working with and design around those if you want your design to be manufacturable with minimal changes.

---------------------------
How thick?

So now you're probably wondering: "what thickness of steel or aluminum should I use?"

I'd say 20 gauge (0.036"/.91mm) is a good starting point for steel and 0.0508"/1.29mm for aluminum.

18 gauge steel can be used for exterior panels or where extra strength is need at the expense of weight. I would not typically recommend anything thicker than 18 gauge for steel because it's overkill but if that's the design goal then it can be used to great effect.


The Compact Splash for example uses 14 gauge (almost 2mm!) steel for that extra-rugged industrial feel.

2-3mm aluminum can be useful for exterior panels for a more premium feel at the expense of cost. I would not typically recommend any thicker than 3mm for aluminum because not all manufacturers can easily laser cut it (I'll go over why in a later post).

Whatever you decide on, try to minimize the number of different thicknesses (and material) used throughout the case. If you're case has 10 parts and each of them is a unique thickness and material, then the manufacturer will have to cut your parts from 10 different sheets and this adds a lot of extra handling and processing so it'll cost more. If all 10 parts are the same thickness and material then they can all be cut at once from the same sheet which is more efficient.

---------------------------​

Thin, stamped steel

You may be thinking: "some mass-produced cases use pretty thin steel though and it seems like it'd be a good idea for a SFF case since it saves space right?"

The thing is though, mass-produced cases use stamping, which allows the metal to be formed into much more complex shapes than is practical with the manufacturing methods used for small run production. Mainly they can press ribs and grooves into the sheet to strengthen the part, allowing them to get away with much thinner steel than is otherwise practical without excessive bending and flexing.

Let's play spot the sheet metal ribs/grooves!



Here's a Corsair 750D and just from this angle alone I see 8 of them. And that's not counting all the edges that are rolled over for rigidity (and safety) as well.

Not to say that 22 or 24 gauge steel (or even thinner) can't be used for a SFF case design, but keep in mind the limitations of the material and the manufacturing method. Also, if you use really thin steel and don't round any and all outside corners, better pack up and move into the Vatican because otherwise you're in for the haunting of the century! :eek:

---------------------------​

Implementing material thickness

So now that we know not to design cases with infinitely thin sheets and have a rough idea of material and thickness, what kind of effect does this have on case design? To illustrate, let's look at a cross-section of my SketchUp model from Part 1:



The ATX PSU form factor is 150mm wide and Mini-STX motherboard are 140mm wide. That leaves 5mm on each side between the board and the case.

I decided early on that I wanted to use aluminum because I want to leave the metal bare for that "sketchy eBay PSU" look but I don't want it to start rusting. I started with 1.63mm aluminum because I also wanted it to be plenty sturdy in case I mounted it vertically like in the NCASE M1. Plus the case will partly serve as a heatsink for the HDPLEX AC-DC.

So after adding in the basic U-shape enclosure from Part 2, how do things look?



After subtracting 1.63mm for the top cover, 1.63mm for the base, and a small gap between them (more on this in a later post), that only leaves 1.37mm between the board and the chassis!

While that could work, it's closer than I'm comfortable with so I ended up dropping down a size to 1.29mm aluminum to give myself a little more breathing room.

Not all designs will be so sensitive to material thickness, but if you're working with hard dimensional constraints, whether they're physical like fitting a 140mm wide motherboard into a 150mm (outside diameter) housing or volumetric by trying to keep under that magical x.0 liter figure you've set for yourself, then having a good handle on material thickness can make or break the design.

---------------------------​

Bend radius

Now that I've gone over press brakes, material types, and sheet thickness, another important concept for sheet metal case design is bend radius. As I hope is clear from my explanation on press brakes in Part 2, the sheet metal is bent by the punch of the press brake to form a bend and that bend has a curve to it and so has an inside and outside radius.

But this is an important concept so let me show how most beginners model their case at first and show why this is incorrect:


The sheet has a thickness to it, so that's a good start, but those perfect 90° corners are not practical to manufacture in the real world! You can get close with welding or extrusions but both of those methods aren't really suitable for the type of cases I'm talking about in this build log.



In reality, a sheet metal corner will look something like this. BTW, the bend radius is measured from the inside of the bend.

Now that I've modeled the bend more realistically, note what happens if I try to place something right up against the corner that either has practically no corner radius or one much smaller than the bend radius of the sheet:



It won't fit! This is why it's important to take bend radius into account.

That said, it's not usually important to model the bend radius exactly because sheet metal bending is a hugely complex topic, with terms and concepts like K-factor and bend allowance, formulas for calculating those, and lots of charts. It can be useful to have a basic understanding of those things but it's not actually needed for case design unless you have to create flat pattern drawings of your sheet metal parts.

The actual bend radius the physical parts will end up at will depend on the manufacturer's available tooling, material type, material thickness, etc. and most manufacturers have a proprietary formula for calculating what the bend radius should be. If knowing the exact bend radius does matter just ask your manufacturer what they recommend for the thickness and material you're considering.

For the design stage I just use 1x the material thickness for steel and 1.5-2x for aluminum.​

So if the material is 1.29mm thick then the bend radius should be set to 1.29mm for steel and 1.94-2.58mm for aluminum. Aluminum can't be bent quite as tight because it's more prone to tearing/cracking if the bend radius is too small compared to steel.

If your CAD software has sheet metal functionality then usually you can set the bend radius either in absolute dimensions or as a factor of the material thickness.

If you don't have sheet metal tools, then depending on how you have to go about modeling the bends you may need to know the outside radius as well, which is simply Bend Radius + Material Thickness for right-angle bends. Here's an example where I model the bend first as just right-angle corners and then apply a round to both the inside and outside corners with the appropriate radii (or radiuses, whatever):


And I guess that's supposed to be steel since I did a 1x material thickness bend radius :p

This is just one way of working around a lack of dedicated sheet metal tools but with any decent CAD program there should be 2-3 different ways to model the bend radius. Remove the dermis from a feline and all that.

---------------------------​

At this point I'll just admit that I can't help but going on overly detailed digressions. I meant to include it here but this is getting long enough as it is, so I'll go over parts allowance gaps in the next post and then I'll talk about the vents :)

Table of Contents

Next Update
 
Last edited:

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,780
As mentioned in the previous update, this will a (hopefully) brief overview of allowance gaps.

Designing the case part 3b: Parts Allowance (finish, tolerance, and fit)

Basically the idea is you want a small gap between the different parts of the chassis. Here's an example of what not to do:



Notice how the faces of the two parts are touching each other in the CAD model? The problem is that in the real world there needs to be a small gap between them to account for 3 things: finish thickness, manufacturing tolerance, and ease of installation.

Thermal expansion and contraction would be another reason but that's not typically something we need to worry about with computer cases.

---------------------------​

Finish thickness

As I mentioned briefly when I talked about sheet metal thickness, the finish coatings have a non-zero thickness that must be taken into account. However, this is not done by making the sheet metal in the CAD model thicker! The parts in the CAD model should only be as thick as the bare metal is.

Instead the coating thickness is compensated for with the parts allowance gap. So even ignoring manufacturing variance and ease of installation, there needs to be a gap at least 2x the coating thickness between parts. It's 2x because each face of the adjacent parts will have it's own coating.

So how thick are the different coatings? Well, to answer that we'll need to briefly (for real this time, I promise!) go over what the different coating options are:

Paint

Most common coating for metal in mass-produced cases. Not as common in small-run SFF cases though due to the perceived lack of quality and preference of small fabrication shops for powder coating. Can be used on just about any material.

I'm actually not sure what the typical thickness is, it would be dependent on the type of paint used and application method. Best to ask your manufacturer.


Powder coat


NFC Systems S4 Mini

Basically powderized plastic (hence the name) that is applied to the metal electrostically (the powder is positively charged than sprayed at the grounded metal part with special spray guns) then cured in an oven to melt the powder and form a protective skin. Available in a wide variety of types and finishes of varying protection and appearance. From what I've seen it's the preferred coating of most small metal shops. Can be used on any metal, and even non-conductive materials with special preparation.

Typical thickness: 0.08mm (0.003") according to Protocase. Another source suggest anywhere from 0.05-0.25mm (0.002-0.01") depending on type.


Anodization


NCASE M1, image source

An electrochemical process by which the natural oxide layer of aluminum is increased for increased corrosion resistance. Can be combined with dyes to provide a variety of color options. Can only be used on aluminum. This is important to note because if there is any steel hardware pressed into the aluminum sheet (like threaded studs) then the part can't be anodized.

Typical thickness: Remember when I said coating thicknesses are non-negligible? I lied, typical anodization thickness really is negligible in the context of computer cases. There are some extra-thick hard anodizations but they're rare, check with your manufacturer if in doubt.


Other coatings

Powder coating and anodization are the two most common finishes for SFF cases but there are many other coating options available: black oxide, Alodine, Parkerizing, blueing, PVD, phosphate conversion, chromate conversion, chrome, etc.


Mechanical finishes

There are also finishes that are applied to the bare metal and either used alone or in combination with a coating like anodizing. This covers things like media blasting (sandblasting), polishing, brushed finishes, tumble finishes, etc. The NCASE M1 pictured above is a combination of brushed finish and black anodization.

---------------------------​

Manufacturing tolerances

In the CAD software you may have two flanges 30.00000mm apart but the manufacturing methods we're talking about are nowhere near that precise. Every aspect of the end product will have small deviations in length, width, height, diameter, roundness, flatness, bend angle, edge perpendicularity, etc. from the 3D model.

This is where tolerances come in, which specifies the maximum and minimum permittable deviation from the nominal dimension (or other property). If you've ever looked at a drawing or spec sheet and seen this "±" symbol then you're probably looking at a tolerance spec. For example, a part length may be specified as 10.0mm ±0.1mm, meaning that the nominal length is 10.0mm but the part could be as long as 10.1mm or as short as 9.9mm and anything in between.

There's a whole system for dealing with tolerances called Geometric Dimensioning and Tolerancing (GD&T). It's not necessary to learn it for the purpose of designing indie SFF cases but it's good to be aware of the concept.

Protocase has a page on their fabrication tolerances.

---------------------------​

Ease of installation/assembly

The 3rd reason for having an allowance gap between parts is ease of installation and assembly. This is pretty straightforward, users should be able to assemble the different pieces of the case together without excessive force. Sometimes a tight fit is desirable, like bearings on a shaft, but that's not typically done with computer cases.

---------------------------​

TL;DR (So how much allowance should you have?)

Basically I use 0.25-0.40mm allowance for powder coated parts and haven't had any issues.

I would recommend doing your own research and check with your manufacturer on what they recommend though.

---------------------------
I am not an engineer so if I've mixed up the terminology please correct me.

Next update I'll finally talk about the vents! Only took 3000 words to get there :p

Table of Contents

Next Update
 
Last edited:

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,827
4,902
And again I learned so much. Not only about sheet metal design and bending, but also that the English language ruins a good word like alumin-i-um, but also needed to complicate things with "gauges". It all went wrong in the industrial age when they wouldn't adopt the Metric system I'd guess.
 

confusis

John Morrison. Founder and Team Leader of SFF.N
SFF Network
SFF Workshop
SFFn Staff
Jun 19, 2015
4,129
7,057
sff.network
And again I learned so much. Not only about sheet metal design and bending, but also that the American language ruins a good word like alumin-i-um, but also needed to complicate things with "gauges". It all went wrong in the industrial age when they wouldn't adopt the Metric system I'd guess.

FTFY
 
  • Like
Reactions: Soul_Est and Phuncz