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

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
Wait, aren't you goint to sell it? I was hoping for it to be "buyable" at some point of your R&D job... :confused:

If there's demand it's possible, but I doubt there's enough interest to justify a production run. We'll see.

PS : AMD, while presenting AM4 platform, spoke about a tiny platform...are you aware about it?

I kinda remember that but I'm not sure if it's Mini-STX, Nano ITX, or something else.
 

RobonHood

Cable Smoosher
Sep 9, 2016
12
17
Yup, the A4-SFX layout is really the only one that works to get these parts in this form factor. This thing looks pretty ugly and industrial since that's the point, but a prettier version with brushed anodization would be hilarious, it'd be like the Mini-Me version of the DAN Case :p
So you're telling me it'd be a...... FAN Case?


I'll see my self out I'm sorry... either way this looks very interesting keep us posted I enjoy reading!
 

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
I should have written this post earlier since last update I was already going over the finished case. But better late than never I guess.

Manufacturing the case part 0: Getting quotes

So this update will be about the process of selecting a manufacturer, getting a quote, communicating with engineering, etc. It's a wall of text (which unfortunately Mexico won't pay me for :p) so feel free to skip this if you're not designing your own case or otherwise not interested in this part of the process.

---------------------------​

Needle in a haystack

Finding manufacturers can be a frustrating, time-consuming process because there isn't a comprehensive directory with helpful user reviews for sheet metal shops like how there's Yelp for restaurants, Amazon for consumer products, etc.

So here are a few tips to help you in your search:


Location

China is usually seen as the way to go for manufacturing consumer products but there are several downsides to going that route: language barrier, time zone difference, IP theft, Quality Control (QC), etc. Those issues can be overcome and the lower pricing for high volume production can be well worth it but for a small run of SFF cases by an inexperienced team it may not be worth the effort.

Asian (China, Taiwan, etc.) manufacturing is a huge topic that I'm not really qualified to speak on so I'll focus on domestic manufacturing instead.

So if you've decided to look for a domestic manufacturer, then it's a question of whether to look locally or just within the country.

Local

Using a local manufacture has the big advantage of being able to visit the shop in person, whether that's to talk face to face with the account manager and engineers to discuss the project or to save time and money by just picking the cases up instead of having to ship them.

The problem of course is that you're limited by what shops are near where you live. For a big city with a strong industrial base this may not be an issue at all, but for a small suburban town there may not be any suitable manufacturers nearby at all.​


Non-local

Broadening your search to the whole country (or nearby countries) will give you many more choices, but of course that will mean the cases will need to be shipped and most communication with the company will have to be via email, phone, or even fax (you may laugh, but faxing is still common in these kind of industries).​


Finding either local or non-local (but especially local) sheet metal shops can be difficult though because not all of them have a website (or any internet presence whatsoever for that matter).

So here are a few ways to find hot singles sheet metal shops near you!
  • Google search - always a good first step.
  • Supplier directories - ThomasNet is an example of one, though annoyingly they try to get you to register to view searches now.
  • Phone book - Sheet metal shops can be pretty old-school so don't be surprised if their only advertising is dead tree format.
  • Referral - If someone who lives in your area has had sheet metal work done ask them who they used.
  • Machine distributors - This is a trick I've read about. For example: ask the Amada sales rep for your area what companies nearby have Amada press brakes in service.
  • Press releases - Equipment like laser cutters are expensive, so it's not uncommon for a machine vendor and sheet metal shop to put out a joint press release when new machines are purchased and installed.

Equipment/process requirements

This can be a chicken and egg problem, you may be willing to adjust the design to suit the manufacturer's equipment or you may have the design close to final and want to select a manufacturer with the necessary machines to make it.

For the latter though, make note if any aspects of the case design require specific equipment or processes. A few examples:
  • Tight bend radius - Almost all SFF cases will want this, so check that the shop has the appropriate tooling for the bend radius desired.
  • Lots of vent holes composed of simple shapes (circle, rectangle, hexagon, etc.) - Looks for shops with a CNC punching machine.
  • Thick (3.0mm+) aluminum - Water jet cutting may be required or a sufficiently powerful laser cutter.
  • Anodization - A shop that can anodize the parts in-house would be ideal. That's pretty rare though so you'll probably either have to find a fabricator that's partnered with an anodizer or find an anodizer to finish the cases after the manufacturer is done.
Some of these will be dependent on order volume. Cutting an extensive vent pattern on a water jet may be fine for a one-off prototype but that will probably be too expensive a method for dozens or hundreds of units.


MOQ (Minimum Order Quantity)

The CNC machines I mentioned above are expensive to purchase and operate, a state of the art laser cutter with auto sheet loading can cost $500,000+ just for the machine itself, then there's running cost of electricity, assist gas, maintenance (which CO2 lasers need lots of), etc. on top of that, not to mention the wages for the skilled operators to run the machine.

So precision sheet metal shops have lots of overhead costs thus they need to keep the machines busy fabricating parts to make a profit. This is where setup time comes in. With a press brake for example, like I showed a bit of in Manufacturing the case part 2, the correct set of punches and dies for that particular part need to be installed in the press brake, the bend profiles have to be programmed into the machine, a few test bends are done to make sure everything is setup properly, and only then can the parts be bent.

The time it takes to do this is time the press brake is sitting idle not making money. This is a part of why the price per unit goes down as the number of units ordered goes up, the setup time gets divided across more parts in the larger run.

So all manufacturers will have a MOQ, the Minimum Order Quantity. This is the minimum number of units that have to be ordered for the manufacturer to start work. For a shop that specializes in prototypes the MOQ may be 1 but for a large production-oriented shop they may not be willing to give you the time of day for less than 1000 units. Some manufacturers may be fine producing a single prototype unit, but only with a commitment of a larger order.

Sometimes the company's MOQ will be listed on their website, but if not, be sure to ask about it early on because if you only need 25 cases made but their MOQ is 500 then talking further is a waste of time. However, if business is slow and/or the design is simple and doesn't need much setup, they may be willing to work with you so if there's a company you're interested it can still be worth inquiring even if you can't meet their posted MOQ.

---------------------------​

Getting in touch

So you've tediously built up a list of potential manufacturers for your design to contact, now what? The next step is to start reaching out to the various companies to either start the process of getting a quote or eliminating them from consideration.

I'll mostly focus on emailing, but the same ideas apply to talking to a company on the phone or in person. If it's a non-local company then you may have to call at first, even if they have a listed email address or website contact form, to demonstrate you're serious. Some companies get so many "tire-kickers" that they may ignore emails from people who obviously aren't engineers.

There are many different approaches to this but in general I would not recommend copy/pasting the same email to everyone because the information you'll need from each company will differ.


Introductions

Start off with your name, company name (if you have one already, which you should if you're serious about going to production), and location.

The location can just be state/province for now, the reason to include that in the introductory email is that some companies have different account managers for different regions so including that information at the start helps you get in touch with the right person quicker. Of course if it's a local company just mention so.


Questions

You may wait to get a response to the above intro or just include these questions in the first email, up to you. But if there are any showstoppers that are unclear from looking at their website, then you'll want to clarify them early on. It's a waste of both yours and the company's time to keep emailing back and forth if, for instance, the case design requires a ~1.0mm bend radius and they only have the tooling for 5mm+ bend radius.

Some questions to ask if you're not sure on them already:
  • MOQ
  • Equipment/tooling
  • Production schedule - aka "How busy are you?". If they're suitable otherwise, but the production schedule is packed for the next 6 months, then you may have to look elsewhere if you're not willing to wait.
  • File format for quote
A few notes about file formats:

There are two broad categories, 3D and 2D.

For formed sheet metal parts, most shops I've talked to prefer a 3D CAD file so they can adjust the model to suit their tooling and bend formulas. SolidWorks is the software of choice for most small-medium shops (refer to my intro to CAD software) so a SolidWorks file is often best, but a .STEP file will usually work too.

For flat parts like laser cut acrylic panels, then 2D formats are usually preferred, usually DXF/DWG or .AI files.

Some manufacturers may only work with dimensioned drawings in PDF format. Properly dimensioning a drawing is an art in and of itself, so I'd maybe recommend giving preference to companies that don't require PDF drawings.​


Backstory

Whether to include backstory on the case/project/company will depend. If your relationship with the manufacturer is going to be a straightforward vendor/customer arrangement then I wouldn't bother. If you're wanting them to do something out of the ordinary, like a smaller run than their normal MOQ or the like, then including some context may pique their interest.

If you do feel it's appropriate to include some backstory, do NOT write a page-long narrative of the project to date at the start of the email! Account managers/sales reps are busy people so keep it brief (just a few paragraphs with maybe a link to the build log if there is one) and preferably put it towards the end of the email after the actionable items so they can read it if they want.

---------------------------​

QTY, ZIPs, BOMs, and Part #'s

So you've made your introductions, gotten some questions answered, and have perhaps whittled down the list of candidates, now to get pricing!


Quantity breaks

Like I mentioned earlier, due to setup time and other variables, price per unit will be highly dependent on the amount of units ordered. So one thing you'll want to do is request is pricing at a few different quantities. For example: 1, 50, 100, and 200.

The quantities to ask for will need to take into account the MOQ of the manufacturer and expected demand.

Gauging demand is difficult because people will be plenty willing to take a survey or sign up for a newsletter, but those things are easy to do and free so when the time comes only a fraction of them will actually follow through with their wallet. Expect only about 10-30% of those who express interest to actually buy your case.


ZIPs

A manufacturer may be able to give you rough pricing based off a sketch, screenshot, or detailed description, but to get an accurate quote will require sending them a CAD model, DXF, drawing, or whatever format they want.

When doing so, I'd highly recommend compressing them into a .ZIP file to save space since many manufacturers will have older email systems that can't handle large attachments. This is especially helpful to do for large CAD files because I've found .STEP files to be highly compressible, often reducing to just 10-20% of the original file size. If the file is still too large, then uploading to Google Drive, OneDrive, Dropbox, etc. and sending the link may be better.


BOMs

Ideally, along with the CAD model you'd send a BOM, or Bill Of Materials. This is a document listing each individual component of the product to be ordered. Depending on the company, it can be as simple as a basic spreadsheet like follows:



Pretty self-explanatory. If it matters though, get as specific as you can about the screws (and really that applies to everything). For example, if the case parts are specified as being powder coated black, don't assume the manufacturer will use black screws to match, specifically ask for screws with a black finish. Refer to my primer on screws for more info on other aspects of screws.

I'm no BOM expert though, so I'd highly encourage you to do some research on the subject.


Part numbering scheme

STX160.0 is a project I'm doing for fun so I got lazy with the part names. For a case that's actually intended to go to production, I'd spend some time coming up with a clear, consistent, and flexible part naming scheme and implement it in the CAD model and the BOM to avoid confusion about which part is which when communicating with the manufacturer.

Pages and pages have been written on the subject of part numbering systems (intelligent vs non-intelligent, number of digits, dealing with revisions, part variants, etc.) so I'm not going to cover it here.

If you're in a hurry, then a basic non-intelligent system is easiest to implement. For example: 0001 for Top, 0002 for Base, 0003 for IO Plate, 0004 for those flat-head M3x6mm Phillips screws, etc. So that's a 4-digit system, allowing room for 9999 unique parts. With any part numbering scheme, always use more characters than you possibly think you'll need because you never know.

Part numbers eliminate ambiguity, but they're not very memorable so it wouldn't be a bad idea to add a column to the BOM with a descriptive name for each part (right side panel, SFX mounting bracket, etc.) for reference.

---------------------------​

Choosing the one

Some of the manufacturers may need to clarify certain points but after a few days you should start getting quotes back from the manufacturers you contacted.

Analyzing the quotes and making a selection is tricky and there's no easy answer. My best advice is to not just make a choice based on price alone.

Consider other factors, like the timeliness and quality of the communication thus far, proximity of the shop to key markets and shipping hubs, type of equipment, payment options, turnaround time, warehousing and order fulfillment, shipping options, etc.

Also, having an account manager that's enthusiastic about the project and more than willing to answer ignorant questions is worth their weight in gold. If you're not experienced with designing and manufacturing a case, then such a rep can save you lots of time, headache, and even money in the long run.

---------------------------​

Protocase

I have to admit, even though I just wrote over 2000 words on the topic, I actually didn't go through this selection process for STX160.0 XD

The eagle-eyed among you may have noticed the Protocase logo in Manufacturing the case part 1 on the keychain of the PEMSERTER operator. Protocase is a Canadian company that does exactly what their name says, they make prototype enclosures.

The main reason I chose them is because I've worked with them before and was satisfied with the service (also, this is a personal project so I wanted to keep it separate from that other case I'm working on).

And this is a big part of why you won't find many people with experience with a diverse range of sheet metal shops. Switching manufacturers is a time-consuming and expensive process so most people, once they find a shop they are happy with, will just stick with them until they are given a good reason to look elsewhere.

So while I would encourage aspiring case designers to shop around, Protocase does have the big advantage of being a known quantity. Myself, several other people here on the forum, and Linus from LinusTechTips have used them with good results.

Pros:
  • Easy to work with (willing to answer silly questions, will work off even a napkin sketch of the design)
  • MOQ of 1
  • Wide variety of materials (steel, aluminum, acrylic, stainless, acetal)
  • Large tooling library
  • Extensive documentation on the website of their capabilities
  • Has a case designer software (I've no idea how good it is though)
Cons:
  • Can be expensive
  • No turret punch
Protocase has a lot going for them but those two cons prevent them from being suitable for most larger production runs, especially for cases with sizable vent patterns.

I'll go over the pricing for STX160.0 next but first an interesting side note: even though Protocase is located in Canada (Sydney, Nova Scotia to be exact), and Canada switched to Metric in the 1970s (though not without incident), they primarily use Standard Units since most of their customers are American companies.

Which I find a bit annoying because I prefer Metric and one of the reasons I choose them initially was I assumed they'd use Metric since they're Canadian :\

---------------------------​

1st rule in government spending James logic:

Why build one when you can have two at twice the price?


I really only need one though, whatever shall I do with the second one? ;)

It wasn't actually twice the price though. Like I've mentioned, the price typically goes down with higher quantities and that was the case here as well:



So I could have 1 case for $366, 2 cases for $250 each, or 5 for $174 each. The price per unit quickly drops as the quantity increases but it's important to note that this trend doesn't continue forever though. Protocase, and most similar sheet metal shops, don't do stamping so the price will level off after a point.

Anyway, I could have one case for $366 or two for just $135 more, so I went with 2 'cuz why not. Plus, now I can claim that STX160.0 had a production run, a production run of only two units, but it's not just a one-off prototype nonetheless :p

That $70 setup fee is an administrative-type charge that Protocase does for all orders. Notably they did not charge me a design fee since I provided a STEP file with bend reliefs, PEMSERTs, and such details all accounted for.


While they're plenty willing to design a case from even a napkin sketch, you'll have to pay an engineer to model the case from scratch in SolidWorks and that can cost $300-$1000+ depending on complexity. So this is where it pays to learn a proper CAD program.

Overall I felt the price for this project was reasonable, but on the other end of the spectrum is Project Orthrus, the prototype of which was also made by Protocase but costed $4026. $800 of that is the design fee and I'm not sure if the setup fee is included. I'm pretty sure that's in Canadian dollars and not USD but it's obviously expensive either way.

Value is subjective though and he felt that he got his money's worth but there are certainly other sheet metal shops out there who could have produced just as good an end product for a lower price. On the other hand, like I said, Protocase is a known quantity and if he had gone through 2 mediocre shops before finding a good one at $1200 a pop plus months of frustration then the $4000 for Protocase wouldn't seem so bad in comparison.

---------------------------​

This post has gone on long enough so I'll end here and write up an addendum later going over communicating with the engineer in prep for production.

Table of Contents

Next Update
 
Last edited:

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,839
4,906
Awesome write-up, I again learned a lot !

So the question on most people's minds is: what will you do with the second case ?
 
  • Like
Reactions: Commissariat

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
So the question on most people's minds is: what will you do with the second case ?

I'll sell it, but you may want to wait for my post on mistakes made before PMing me :p
 

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
Coincidence. I don't really like it, it looks like cheap spraypaint instead of anodized aluminum.
 

CC Ricers

Shrink Ray Wielder
Bronze Supporter
Nov 1, 2015
2,233
2,556
Oh, other pics make the heatsink look like bare aluminum. I do like the cables, though.
 

alamilla

SFF Lingo Aficionado
Feb 11, 2016
120
114
This is such an interesting project and thanks for such a thorough run down of the whole process.

What is the CPU cooler height limit out of interest?
As although CoolChip have yet to release their low profile cooler (in spite of teasing a finished product) it would be interesting to use that in place of the NT07.
From the PDF on their website it measures 26.5 mm
 
Last edited:

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
Oh, other pics make the heatsink look like bare aluminum. I do like the cables, though.

I'd rather have bare aluminum over whatever's on it now.

This is such an interesting project and thanks for such a thorough run down of the whole process.

Thanks!

What is the CPU cooler height limit out of interest?

I don't have the CAD file in front of me right now but it's around 24-25mm.
 
  • Like
Reactions: Soul_Est

jeshikat

Jessica. Wayward SFF.n Founder
Original poster
Silver Supporter
Feb 22, 2015
4,969
4,781
In the pics the spindle sticks out past the fins but in the spec sheet it's flush. Either way though, I don't think grinding down the top of the spindle is a good idea.
 

alamilla

SFF Lingo Aficionado
Feb 11, 2016
120
114
In the pics the spindle sticks out past the fins but in the spec sheet it's flush. Either way though, I don't think grinding down the top of the spindle is a good idea.
You're probably right.
Guess you'll just have to add a couple more millimetres to the production run ;)