The key to making this project happen is efficient manufacturing. There are many ways to peel an apple, but not every way will make you the leading pie distributor in the area. Since IBM took care of all the metal work required for a fan enclosure, I can focus my efforts on PCB design, part optimization, connector selection and filter tray production. For this project, I decided to minimize the amount of hand soldering at all costs. Hand soldering is a time consuming and fiddly process that eats up precious time and creates inconsistent assembly times. Solder may not wick into a hole as nicely from one wire to the next. A pad may burn off the control PCB if the soldering assembler applies too much heat. Heatshrink is a must if wire-wire soldering is needed. These are all nuisances to the production workflow and I didn't want anyone to have to spend time mitigating such issues.
I'm designing the PCB around common through hole connectors. I am planning on reusing the fancy molex connector on the back of the fan, having a JST connector on the PCB for the potentiometer and a mini-fit Jr like connector for the battery input and motor tone generation. The enclosure features two threaded binding posts for mechanical mounting, which is an absolute godsend for easy modifications. I originally was planning to put my custom PCB between the two binding posts. The rough size was 130mm x 20mm, plenty of area to fit the control circuitry. I started board design in Kicad, but a problem came up with physical assembly that made me reconsider my design.
The binding posts are 1/8in away from the side of the top shell of the fan enclosure, leaving little means of egress for wire routing and wire placement, let alone the use of connectors. The PCB would have had to sit below the fan and be soldered only. That would create a very frustrating assembly process and frankly one that no one should bear. Moving the PCB to the back right corner alleviated all of the problems I had with the first PCB and allowed me to reuse the incorporated molex connector. This saves me precious time as now I don't have to cut the connector off, wire strip each of the 9 leads and manually solder them to the board.
This back right corner had some challenges though. For one, there was only one binding post accessible in this corner. A quick and clever solution was to use one side of the the PCB as a toe-in mount, where that corner would feature a fork designed to fit around an internal fan mounting post. The above kicad render is a placement mockup, the design has not yet been finalized but is roughly what it will look like.
An important consideration for any PCB in a metal enclosure is isolation. Since there will be through hole connectors on the PCB, the board needs to be offset from the shell. Plastic washers and adhesive backed rubber spacers will be used to mount the PCB into the case. clear plastic adhesive backed sheet, commonplace in most PC power supplies, will also be used to line the back area of the PCB mount area.
Since I'm working with ewaste / old parts, some of these fans will be used, some will be new old stock. The used ones will have to be cherry picked and analyzed for dents and defects. Used fans will also need cleaning. I plan on using compressed air to thoroughly blow out each used fan, but final hand cleaning with a damp cloth will be up the the end user. Cleaning by hand is easy and harmless, but it is time consuming. Compressed air is a good common ground and at the same time, by the end user cleaning their unit, it can provide them with more of a sense of value for their fume extractor (like the economics of a picked apple at the grocery store).
Another major manufacturing challenge is the mass production of the battery adapter plates. More often than not, Drill manufacturers make intentionally overly-complex battery connector geometry in effort to mitigate knockoff production. Regardless of how absurd the battery mount CAD looks, the major problem is finding an easy way to mass produce high quality battery mounts that will work every time and fit genuine connectors.
So far, I have resorted to 3D printing, but I have found that 4 hour print times are a bit much. I am in the process of redesigning the cad to optimize for machining. The good news is that the clam-shell nature of the dewalt 20v mount should make it easier to machine the halves out of Delrin or HDPE. I could just rely on a series of end mills instead of needing a symphony of woodruff cutters to complete the oddball geometry.
The filter tray design was a bit tricky. This is not because of unfriendly geometry, but rather because of the size of the part and the nature of the item. The filter tray measures 8.25in x 7.95in x 1.74in tall. This size pushes the item into injection molded plastic territory, because my FDM 3D prints had a build time of over 14 hours! That's with optimization and ruggedness factors built into the print.
However, if the demand for this unit is under 5000 units, injection molding is totally not an affordable option either. At ten thousand, it may make sense. An easy way to buy time and thwart the high initial financial burdens of injection molding is to simply use another material altogether! Wood is a cheap and easy to source material, can be easily cut and glued together. A laser cutter, titebond and a wood band clamp can go a long way.
The initial production run will rely on a wood based filter tray. I can easily enough farm out the laser cutting to a service or job shop and do the gluing and filter assembly in house, or with a small discount, pass that assembly burden off the the consumer. Everyone has wood glue. It has been an essential school supply since the dawn of time. There is no reason why the enduser could not glue together pieces of supplied wood unless they simply did not want the hassle. That is why I am offering different levels of assembly in terms of buying options. Here is my approach to assembly of the fume extractors in a convenient diagram:
I am have decided to select option number two because my aim is to make assembly as fast and efficient as possible. By removing hand soldering from the assembly process, I can optimize the assembly through specialization. This way, I can contract out the wiring harness soldering to multiple contract workers and have multiple sources for the assembly team in order to prevent assembly bottlenecks.