Next in my series on builds for my new desk, is a under desk mounted USB hub. During the planning phase for my desk setup, one requirement was easily accessible USB 3.1 ports with at least one type C port. This proved to be much more complicated than I originally thought.  The search for a regular USB 3.1 hub ended with a single candidate. Coincidentally, Linus Tech Tips had reached the same conclusion in their Project Zero Cables desk.

The EZDIY-FAB USB 3.1 Gen2 Hub and Type-C Port Front Panel USB Hub is meant to be mounted in a 5.25" drive bay and powered by an floppy connector, but was the only real solution I could find. To power this hub externally with +5V and +12V, I found a 12V 150W Pico ATX PSU and a generic 12V wall adapter. A female to female USB type C adapter would also be used to connect to the host computer.

The hub itself is a pretty compact board with three USB 3.1 Gen 2 type A ports, one USB 3.1 Gen 2 type C port, and two 18W QC3.0 type A ports. I took a bunch of measurements of the board and created a CAD model to help with designing the new under desk enclosure.

To reduce the bulk of the wiring needed to power the hub from the PSU, some cable rewiring was needed. The power output on the power supply uses a JST VH series connector. The pins can be removed from the connector on the cable and newly crimped pins can be inserted.

The hub is powered with 5V, 12V and ground via a Berg floppy power connector. I clipped off one end of the extension cable and the wire gauge was just right for the VH crimp pins I had on hand.

I stripped and crimped on the new VH pins and inserted them into the connector. with pin 1 being 5V, pin 4 being 12V, and ground in the middle pins.

I also reduced the length of the 12V input cable for the PSU.

The enclosure design went through several design iterations. The print geometry was more conducive to being three pieces. The geometry was optimized to minimize support material. By making the front face a flat part, it allowed me to choose a different material.

I choose 3mm thick bamboo plywood to match the Fully Jarvis Bamboo tabletop. The face plates were laser cut, then sanded and varnished.

The final design uses M3 x 5mm Phillips head screws to secure the PCB and the top lid to the main body. The small bulge and cutout was designed to fit the ATX connector's latch, this helped to minimize the height of the entire assembly.

The rear and front face plate was glued on to the main body with CA glue.

The hub PCB was screwed into place. The barrel connector was secured to rear of the enclosure.  

With the PSU installed with double sided tape, the input and output cables can be connected. The type C adapter is used as a extension and to allow a regular type C cable to plug in.

With the lid closed, the assemble is about 30mm thick with the small bulge adding just 4mm to the thickness.

The under desk hub was secured to the bottom of my desk with some wood screws. I ended up using a 10 feet COSEMI Active Optical USB 3.2 Cable to connect the hub to my computer due the length of cable management and the fully raise standing desk. Overall this turned out quite well in both performance and aesthetics.

As part of my continuing series on my new desk, this post covers my custom CPU cart. Although CPU holders and carts seem like a throwback to the 1990's and the beige boxes of the era, they still provide a vital function. My current rig is housed in a Corsair Obsidian 450D and needs plenty of air flow from the bottom fans; placing the case directly on the carpet would starve it of the much needed airflow. I also wanted space for my UPS and Drobo to simplify the sit-stand desk arrangement.  

The cart was designed with 3030 aluminum extrusions, laser cut plywood, 3D printed parts, and casters. The extrusions were designed to be assembled with blind joints and some 3d printed parts to hold on the plywood tops and casters. The two tier design allows for better use of the footprint to maximize usable area for the UPS and Drobo.

Since 6mm birch plywood was used in the design, I was worried about the stiffness of the upper surface top. I opted to have two pieces laser cut to be laminated together.

The plywood pieces were sandwiched together with Titebond II wood glue and clamped together. I should have used more clamps as the clamping pressure didn't seem very even after the glue dried.

The tops were sanded with 120 and 240 grit before being coated with wipe-on poly. Another pass of 240 grit was used to knock down the grain rise after the first coat. I went thick on the poly with 10 coats to get a nice smooth and shiny surface. The mounting holes were countersunk to allow the bolts to sit flush with the surface.

The 3030 aluminum extrusion was cut to length by Misumi. The horizontal section extrusions have a single smooth face while the vertical ones have two smooth faces. The ends were tapped for M6 screws and access holes were drilled.

The 3D printed parts were printed in white MatterHackers Build Series PLA. Two types of heat-set inserts were used. The casters used M5 screws while the rest of the assembly used M6 screws.

The corner brackets add rigidity to the frame and provide mounting points for the top and caster. On the bottom face, roll-in T-nuts were used for securing the caster plates and he toppers were secured in place with M6 flat head screws.

The corner trim prices protect the corners of the topper and cover the ends of the extrusions. The UPS lives on the lower level, while the Drobo sits on the top. The fully assembled cart provides a clean solution for keeping my case and accessories organized.

This project involved mains wiring. I don't recommend trying anything mentioned in this write up unless you are sure you know what you are doing. Mains voltages can KILL and severely INJURY you.

Finding the right lamp for your workspace is not always an easy task. Searching through the Ikea catalog offered up some options, but none were quite right. The Skurup Wall Lamp was a close fit, but only accepts E12 bulbs. I have a full Philips Hue setup for all the lights around my place and as such had a couple extra E26 bulbs that I wanted to use.  The Skurup seems like it could be adapted to the E26 bulb, but the standard E12 to E26 adapters on the market made the bulb stick out way too far from the lamp shade.

The first step of any adapter project is to see what you have to work with. Inside the metal lamp is a plastic housing containing the switch and bulb socket. I had two lamps to work with so I wasn't too concerned about breaking anything.

Removing the threaded nut on the bulb socket allows the covering plate to be removed. Line and neutral come in from the base of the lamp, with the line side going to the tiny switch at the back. A loop molded into the plastic housing provides strain relief for the AC wiring.

The bulb socket is removed by prying out the two clips, one on each side of the socket.

The socket uses a push-in type connector for the wiring. A crimped wire nut under the heat shrink completes the circuit from the switch back to neutral.

Removing a plastic washer and screw allows the cable to be freed up. I clipped off the switch as I would not be using it. The new bulb socket will be directly wired to the cabling.

After some sizing and fit checks, I decided to remove the two socket clips. A pair of side cutters made short work of the thin plastic. As an experiment I swapped out the wiring on one lamp and I wouldn't recommend trying to thread wiring through the tubes of the lamp.

It took a few iterations to work out the optimal sizing and hole positioning. The new socket adapter would take advantage of the strain relief plate mounting point. Having this fastener connection really improved how robust this adapter would be. The E26 bulb socket came out of a set of Aplstar Black Retro Edison Lamp Holders. They are pretty basic., but work just fine.

The initial fit checks and prototypes were printed in PLA, but the final product was printed in ABS for better heat deflection.

The final iteration was a two piece design that provide access to the screw terminals on the E26 bulb socket.

Once the bulb socket was in place, some CA glue pressed into the top cap keeps everything secure.

The completed adapter press fits into the existing recess and the screw provides an secure attachment point. The wiring is passed through the original strain relief loop, but it's a tighter fit.

A Philips Hue E26 bulb screws in perfectly. The original lamp was only rated for 8.9 Watts and the Hue bulb pulls 10W, but it shouldn't be a problem. It seemed like the switch was the limiting factor.

It's a pretty close fit with just a few millimeters of bulb sticking past the rim.

Two modified and installed Skurup Wall Lamps with Philip Hues Color bulbs above my new desk space. This was a pretty quick weekend project, but ended with excellent results. I may post the adapter models at a later date if long term use doesn't reveal any problems.

When people find out you have a 3D printer, one often gets ask to print random things. This is one of those stories. My sister recently release a kids book called Baby Snack Time and wanted a giant sliding puzzle as a promotional tool. It's like those sliding puzzles you got a child but much larger, about 680mm by 680mm . The individual pieces slide along other pieces and are captive.

I started the design with a tessellating square that was 152mm by 152mm. Two sides have 45-degree cutouts and the other two have 45-degree projections. The 45-degree angle allows for the puzzle piece to print with minimal supports.

When assembled the projections ride in the cutouts, allowing movement in one axis but captive in the other. The frame pieces replicated the projects and cutouts, making all the pieces fully captive in the puzzle. The dimensions of the assembled puzzle is 682mm by 682mm. The frame is held together using socket head screws. There is about 1mm of built-in tolerance for all the pieces.

The first few pieces printed with no issues, but the large area makes it very sensitive to any issues with the first layer. Everything was printed in MatterHackers Build Series PLA on my Railcore II.

The frame rails, being nearly 700mm long, required the use of the new Creality CR-30. Naomi Wu did a great job getting this printer out and the quality is excellent. Each frame rail took about 30 hours to print in PLA.

All 15 puzzle pieces and the 4 frame rails took over 10 days of continuous print time and used over 4kg of filament.

The frame rails came out less than 1mm out of spec on the length but were slightly curved. A quick blast with a heat gun to soften the PLA allowed the rails to be straightened.

The frame was square after the straightening process. It was at this point the size of the puzzle really hit me.

Heat set inserts provide the threads for the socket head screws. These are M5 inserts.

I designed the hole pattern on each rail to be specific for one side of the assembly. This makes it impossible to assemble incorrectly.

This closeup of the rail really slows the layer lines and minor imperfections, which isn’t great for a presentation piece.

A test fit with 4 of the puzzle pieces shows the slight color mismatches from different spools of filament.

Since this was a presentation piece, it would require reinforcement as well as a high-quality finish. The frame rails were coated in Total Boat High Performance Epoxy. Originally, I had planned to add a strip of fiber glass to make the rail stronger, but testing pieces showed it wasn’t necessary.

After a few coats of epoxy, the rails were allowed to cure for 48 hours.

A layer of glazing putty filled in any low spots and layer line inconsistencies. It’s critical to allow the putty to fully cure before starting to sand.

The rails were sanded smooth using 240 grit followed by 400 grit. The resulting finish was smooth enough for the primer.

Fully assembled, the heft and size of the puzzle was unexpected. Four kg of PLA is quite a bit.

A couple coats of filler primer fills in any minor imperfections. After a light sanding with 600 grit sandpaper the rails are ready for paint.

Rust-Oleum Gloss Deep Mint provides a great shiny coat to the frame rails. The paint was allowed to fully cure before the next step.

To allow the puzzle pieces to move smoothly against the rails, a layer of felt was glued into place. The compressibility of the felt takes up some of the looseness from tolerance stacking.

The final assembly of the frame requires an exact order to allow all the pieces to fit into place and the rails to bolt together.

At this point over 3 weeks have gone into the puzzle, but most of the time was the 3D printing and waiting for things to cure or dry.

Water slide decals was the first method to apply a graphic to the puzzle. The results weren’t great, so a set printing in vinyl was ordered.

The completed puzzle looks quite good despite the issues with the graphics.

One issue, if you’re not great at puzzles, is trying to put the puzzle back into the correct order.

Eventually the printed vinyl arrived and was applied to the puzzle. The new printed vinyl looks much better than the water slide decals. So far, the puzzle had held up to the abuse of being a promotional tool.

I find that the regular Creality Splice Kit for Naomi Wu's CR-30 3DPrintMill is a bit long for everyday use in my space limited apartment. This 3D printed version gives you an extra 180 mm for parts to roll out.

Hardware

• 4x - M5 x 20mm
• 6x - M5 x 40mm
• 6x - M5 Washer
• 6x - M5 Hex Nut
• 6x - 605ZZ Bearing (Inner Dia.=5mm, Outer Dia.=14mm, Thickness=5mm)

The rollers are held in place by M5 bolts on each rail. The vertical height of the roller are adjustable via tightening the M5 bolt with M5 hex nuts inserted into the rails.

It is best to just have the bolts finger tight to prevent any binding during assembly. The image shows shorter bolts, but you will want longer ones, at least 30mm, to make installation easier.

Rollers should be printed with the asterisk facing up at 0.3mm layer height.

The rollers have internal geometry so that supports aren't needed.

The 605ZZ bearings should pop right in, but some adjustment might be needed depending on your printer.

I found that it was easier to install fully assembled, with masking tape holding the rails and rollers together. Four M5 x 20mm bolts secure the rails to the tapped holes on the printer. You will have to remove the two extrusion covers that block the hole. Once the rails are secure, I set a straight level across the print belt. With the rollers still loose, I pushed them up to make contact with the level before tightening the M5 bolts. Repeat this process for all the rollers.

You can find the STL files on Thingiverse.

I rarely work at my electronic workbench, even though it's complete with displays, power supplies, oscilloscopes, and soldering stations. I prefer to work on firmware and testing on my main desktop. Having a handy source of power, other than the 5 volts that USB provides, is very handy. I didn't want to spend too much time building something complicated and decided to go with parts I could get off Amazon.

The casing and other 3d printed parts took about a day to model and tweak, but are pretty basic. All the other parts are available via Amazon

• USB-C PD 20V Trigger Module (Amazon)
• DSP5005 Variable Power Supply (Amazon)
• LM2596 DC/DC Voltage Regulator (Amazon)
• LED Digital Voltmeter Gauge (Amazon)
• Banana Plug Socket Connector (Amazon)
• Latching Push Button Switch (Amazon)
• Manual Reset Low Profile ATC Circuit Breakers (Amazon)
• 40mm 24V Fans (Amazon)
• 18 AWG 6 Colors (Amazon)
• 18 AWG Orange Wire (Amazon)
• Quick Splice Terminals Connectors (Amazon)
• WAGO 221 Lever-Nuts (Amazon)
• M4 Self Tapping Countersunk Screws (Amazon)
• M3 Self Tapping Phillips Pan Head Screws (Amazon)

The USB PD module would provide 20V at up to 5A to power the whole thing. The 24V fans are hooked up directly to the 20V source so that the fans will run when power is connected just to make sure things stay cool inside. The 5A circuit breaker provides some safety incase of a short. The latching switch allows the voltage regulators to be turned on and off. I'm exceeding the rating of the switch, but for this application it shouldn't be too much of an issue. The LED indicator and the voltmeters are powered by the 12V output from one of the LM2596 regulators.

The DSP5005 and LM2596 regulator are all powered by 20V. For the DSP5005, this limits the maximum output voltage to around 18V. The individual LM2596 modules are adjusted to output 3.3V, 5V, and 12V. The final outputs for each regulator are connected to a banana plug connector.

The maximum input current is limited to 5A by the USB PD standard, so 18 AWG wire is more than sufficient. The only soldering required for this project is to attach the output on the USB PD module and the input/outputs on the LM2596 modules.

Mounting the USB PD module is a pain. As the module has no mounting holes, the 3D printing mount uses the edges of the PCB for alignment and clamping force to keep it in place. A single M3x6mm pan head screw is used to keep the two halves together.

The ATC circuit breaker holder is just a couple of right angle quick connectors held together. Two M3x12mm pan head screws are used to keep the connectors in place.

Two more M3x6mm pan head screws secure the USB PD module mount to the faceplate.

The three LM2596 are mounted to the regulator tray with six M3x6mm pan head screws. The tray has some hole designed for cable ties to keep the input wires organized.

The three voltmeters are secured to the face plate with six M3x6mm pan head screws.  

Extra wires were removed the latching switches harness, as space was going to tight inside the case. The switch is mounted with the hardware that was included. The banana plugs are screwed in place with the connector spade orientation retained by features on the face plate.

The DSP5005 pushes into the opening on the face plate and should snap in to place.

The banana plug connector are paired with the corresponding voltmeter. Output is above and ground is below. The ATC circuit breaker fits perfectly in the open slot.

To keep the project quick and easy to troubleshoot, I went with lever-nuts instead of soldered or crimped connections. The lever-nuts are larger but some features on the face plate keeps things organized.

The regulator tray is mounted to the face plate after all the wire lengths were checked and adjusted. I ended up only using two  M3x6mm pan head screws with the mounting holes on the regulator tray. The input cables on the regulator tray made putting screws there too difficult. With all the wiring connected, the face plate is jam-packed.

The first power up test worked perfectly. The refresh rate of the LED displays on the voltmeters is a bit low to capture correctly. On center left is 5V, center right is 3.3V, and lower left is 12V.

The 40mm fans are mounted the back casing with eight M4x12mm flat head screws. This is not the greatest mounting solution. There isn't much material on the case for the chamfer. I also ended up using six M3x35mm pan head screws to secure the face plate. I didn't have the correct M3x35mm flat head screws on hand.

The fully assembled unit is quite dense. The use of the pan head instead of flat head screws acts as little feet giving the fans some room to breathe, but also scratches up my desk if I move it around.

I mostly use it in the vertical orientation with banana plug to male jumper wires. It works quite well for powering breadboards that needs the whole range of voltages.

The back isn't optimal and some redesign work is probably needed to fix the minor issues. At the moment though, it's doing its job well enough.

Having this little guy that can be powered by any 100W USB PD wall adapter or compatible power bank is quite convenient.  The existing portable power supplies on the market are probably better in every way, but for a quick and dirty 2 day project, this exceeded expectations. I've released the STLs for the 3D printed parts on Thingiverse.

I've been keeping my tools in drawers or strewn around my work area for years. This isn't the best system, especially if the tool collection continues to grow. I was inspired by Adam Savage's custom tool storage stands that he talks about in this Tested video.

I love the idea of "first order retrievability" and how much more efficient working with easy access tools is. I made a list of the tools I used the most and started to think about how to build something compact but full featured like Adam's tool storage.

I lack the ability to rip down large sheets of plywood. I just don't have the space to build custom drawers and shelves from plywood. The storage also had to be compact, I don't have much extra square footage to loose to more storage. Looking around for a quick solution, I came upon DeWalt's TSTAK storage system.

I decided on using two deep single drawers (DWST17803), three double shallow drawers (DWST17804), and one 4 wheel cart base (DWST17889). This gave me a good base to store the loose tool and bits. The TSTAK connection system was also simple enough that I could build my own adapter and have a custom rack on top.

Using a combination of aluminum extrusion, laser cut plywood, 3D printed parts, and a couple Poppin medium accessory trays, I came up with a design I was happy with. While I waited for everything I ordered to arrive, the 3D printed parts were printed in Matter Hackers Tough PLA.

Hardware

• 2x - 220mm 1515 Extrusion
• 2x - 250mm 1515 Extrusion
• 6x - 350mm 1515 Extrusion
• 2x - Poppin Medium Accessory Trays
• 2x - DeWalt TSTAK Deep Single Drawers (DWST17803)
• 3x - DeWalt TSTAK Double Shallow Drawers (DWST17804)
• 1x - DeWalt TSTAK 4 Wheel Cart Base (DWST17889)
• Bunch - M3 Nuts and Bolts

Laser Cut

• 1x - Laser Cut 6.2mm Birch Plywood

3D Printed

• 2x - TSTAK Adapter
• 2x - Right Frame Corner
• 2x - Left Frame Corner
• 4x - Ladder Connector
• 1x - Hammer Rack
• 3x - Tool Rack
• 1x - Partition
• 1x - Lower Container Spacer
• Bunch - Wiha Spacers

The top of the stand was laser cut from 6.2mm birch plywood. The parts are sized to give a nice interference fit when slotted together. The parts are assembled with some wood glue and a mallet. It should only go together one way, working from the left of the assembly to the right.

The assembled top was nice and rigid with just the wood glue. The 100 holes in the center section were sized for Wiha Precision drivers, so you may need spacers for other tools.

The plywood was sanded down and a few coats of TotalBoat Lust varnish were applied. The vanish provides a nice thick coat to protect the wood from the abuse this thing is going to have to take.

The top section is assembled with M3 hardware and the TSTAK adapters just snap right in to the top of the drawers. The two Poppin trays are held in by the center spacer. They provide nice storage for the various adhesives, lubricants, and solvents I use. With the partition inserted into the pocket, you can put all sorts of tools in there for quick access.

The hand tools drawer has hex wrenches, driver bits, etc. The cutting drawer has a variety of blades, saws, and handles. The finishing drawer has sandpaper, polishing cloths, brushes, etc. The measurement drawer had calipers, micrometers. squares, and rulers. The drill and taps drawer has taps, drills ,and drill guides. I have a drawer for lathe tools, as my micro lathe hangs out nearby. The two deep drawers are for electronics tools and specialty tools. The electronics drawers has wire strippers, crimping and soldering tools. The specialty drawer has a mix of tools that I use occasionally, such as an airbrush and label maker.

The whole tool cart has a compact footprint while holding a lot of tools for easy access. I've posted the design and CAD files on Github, for those that want to build their own.  

When I first ordered the Prusa Mini, I did so with full confidence in the product. Prusa printers have a reputation of being reliable and easy to use, which was exactly what I was looking for. Having a workhorse printer that did not require constant adjustments and tuning seemed great. The Mini's build volume would cover most of the parts I print often. With all this in mind, I placed my order.

It was a few months before the printed was shipped. In that time, Prusa had upgraded a few parts and it became the Prusa Mini+. It was a nice gesture to update the printers that had been ordered, but not yet shipped. However, once it arrived I immediately started having problems.

The bowden tube between the extruder and the hotend is actually two tubes. The first tube is connected via compression fittings to the extruder assembly and the hotend, with ferrules crimped onto the PTFE tube.  A second tube guides the filament from the compression fitting down into the heatsink. I was getting this clog pictured above. It sat at the interface of the second tube and the heat-break, which brings the extrusion to a halt.

After replacing tubes, tightening screws, and hours of tinkering, I made no progress. It wasn't a great initial experience. Eventually, the compression ferrule came off the tube. It wasn't worth the effort source a new ferrule and crimping tool. Instead, I ordered the Bondtech extruder for the Prusa Mini, the Bondtech heat-break, PC4 M8 x 1.25 quick insert connectors, and some Capricorn PTFE tube.

The new quick insert connectors would allow a single tube to go from the extruder down to the heat-break, bypassing the need for two separate tubes. This proved to the best solution, and with auto bed leveling issues aside, the printer could now print reliably.

To make the Prusa Mini+ a nice complete package I added an attached filament spool, tool holder, handle, and feet. The parts were designed to fit inside the 180mm by 180mm build area. Prusament PETG was used to match the colors and materials already used on the Mini. Roll-in M6 T-nuts and M6 bolts are used to attach the parts to the extrusions.

The spool holder has a bayonet style lug that twists into the frame.

The handle and tool holder have access holes for the M6 socket cap bolts.

The feet are attached to the bottom and rubber or foam feet can be stuck on. I also made a shorter version of the feet, but I opted to have a wider base footprint.

After the upgrades and some adjustments, the Prusa Mini+ seems to be working well. It has some limitations, but for what it is it certainly performs well enough. The models are available on Prusa Printers or Thingiverse.

Every time I browse through the selection of Raspberry Pi cases on Thingiverse or MyMiniFactory, I feel a bit overwhelmed by the options. I rarely find exactly what I'm looking for, so I often end up designing something new.

This retro-cyberpunk inspired Raspberry Pi 4 case was specifically designed as a case for a MotionEyeOS camera installation. It supports a Raspberry Pi 4, one end of the Arducam CSI to HDMI Cable Extension Module, and a 0.91in 128x32 OLED display.  A complete list of materials are

• 1x Raspberry Pi 4
• 1x Arducam CSI to HDMI Cable Extension Module
• 1x 0.91in 128x32 OLED
• 2x M2.5 x 6mm Flat Head
• 2x M2.5 x 8mm Flat Head
• 2x M2.5 x 12mm Flat Head

A M2.5 tap is needed to tap some of the holes or you can get self tapping screws. Two small sections of 1.75mm filament are used as pins to secure the Cable Extension Module. Adafruit has a great tutorial on how to use SSD1306 OLED Displays with Raspberry Pi.

You can find the STLs on MyMiniFactory or Thingiverse.

I had started writing this post more two years ago before abandoning the project. The printer just wasn't mechanically capable of what I was looking for.  I'm publishing the draft of the post just incase some of the images prove to be useful for others.

Minor improvements.

I had started writing this post more two years ago before abandoning the project. The printer just wasn't mechanically capable of what I was looking for.  I'm publishing the draft of the post just incase some of the images prove to be useful for others.

Check out Part 1 for the start of the story.

Check out Part 1 for the start of the story.

Hotend Replacement

The objective of this hotend replacement is to be able to accommodate a E3D V6 hotend and to improve the efficacy of the part cooling fan. Why a E3D V6? In my opinion the E3D's V6 is the best value/performance option on the market. I prefer to use genuine E3D parts, but the market is full of super low cost clones, which potentially offer a even better value/performance ratio. I actually use some clones for general sizing an positioning as they are a perfect mock up of the final parts.

Measuring

The first step of a project like this is sizing and understanding the constraints. The existing Mini Delta hotend was removed from the effector plate by removing just 3 screws.

A single 30mm fan provides cooling to the heat sink and acts as a part cooler. With not much airflow to start with, the fan really isn't capable of being a good part cooler in this configuration. There isn't much space to work with to get air down to the part, as the effector arms range of motion severely limits this. Best bet to reduce sizing problems was to stay inside the existing footprint of the old mount.

The footprint is pretty straight forward to work from, but it was also important to get the measure of the effector plate to see what could be done routing air to the part.

With all the measurements done, the design work can be moved into CAD.

Design

The highest priority requirement was to get a E3D V6 mounted and in the exact same nozzle position of the stock hotend. I had done some design work for the V6  for another earlier project, so I was able to reused the groove mount design.

This was the first iteration of the design. It consists of three parts, the base, heat sink mount, and grove mount. Both the base and heat sink mount parts contains embedded 3mm hex nuts. I opted for the embedded nut over a heat set insert because of the depth required for the fasteners. Due to space constraints the front mount point would have to be tapped directly into the plastic, not ideal, but the loads involved are low.

The triangular cut out in the base provides wire pass-through and access for the heater and thermistor cartridges. The heat sink mount can accommodate a 30mm fan and channels the airflow across the V6 heat sink. There's is mount feature on this section for the part cooling fan for later design iterations.

I happen to have a bunch of old blower fans in my parts inventory from a previous project. It's a Sunon Fans GB1205PHVX-8AY.GN, which is a blower style 50X50x15 Maglev fan running at 12V. Although this particular part is obsolete, Sunon has a direct replacement available. It can deliver 4.7 CFM of airflow, which should be more than enough for a part cooler.

After a few failed and dead end iterations of trying to route airflow from the fan to the part, I decided to change strategies. This lead to the second major revision to the design.

The ducting for the part cooling fan was integrated into the base and heat sink mount. This reduce the parts count and increased space efficiency. This design also reduces the complexity of the fan mount. The air exits the base via an integrate nozzle that is aimed toward the hotend nozzle. This was the most promising iteration and was worth further refinement.

Version 1.0

This will likely be the final version that will make on to the printer.

A hole to capture a hex nut was added to the base for the front mount hole. There wasn't space lower down close the effector plate, but higher up it would fit.

The cooling nozzle, visible in the back, was adjusted to improve flow to the target area near the nozzle. Another triangular cut out was added to ease the installation of the thermistor cartridge. The hole for the heat sink was pushed forward a few mm to allow the use of a silicone cover on the heater block.

Cable management hooks were added to help with the wires for the heater and thermistor cartridge. The geometry was improved on this part to decrease the print time.

The embedded hex nuts are visible for the V6 groove mount. Overall, I'm pretty happy with this design. It's compact and relatively easy to print without needing supports.

Integration

Since the stock hotend used a single fan for the heat sink and parts cooling, there is only a single fan connector in the wiring harness. This wire is on a PWM line so it works fine as a parts cooler fan controller. The heat sink fan really just need a straight 12 volts. The stock hotend also used a glass bead thermistor which is a pain to mount on the newer V6 blocks. Some rewiring will be need to swap that to use the standard thermistor cartridge.

The models for this hotend mount will be released with a Attribution-NonCommercial 3.0 United States (CC BY-NC 3.0 US) license after this series is completely published.

The STL and CAD files can be found on Thingiverse.
Thingiverse

Since I need to do some rewiring anyways, I opted to reevaluate the use of the external power brick and the lack of a power switch.  More to follow in Part 3 of this series.

The Monoprice Mini Delta is a neat little delta printer that can be had for about $160 USD. On paper, it is the perfect small printer: compact, heated bed, auto-leveling, color LCD. I was an early backer of the printer, and received one of the earlier printers. After using the printer for a few weeks, the flaws were quite evident. The biggest issue I had with the printer was the bed material and the auto-leveling. Parts of the print would have trouble sticking to the bed and print quality suffered as a result.  I shelved the printed for nearly a year before taking another look at it.

The Idea and the Problems

The idea of a portable 3D printer was an concept that has been bouncing around my head for a while, but never really explored. After staring at the Mini Delta for the better part of a year, I finally decided to see if it was viable. The first issue was to attempt to level the build correctly. After following the instructions found on various forums and wiki, the printer still wasn't quite right.

Overall quality wasn't spectacular, but was serviceable. Some tuning on the slicer side improved the print quality, but issues with the first few layers persisted.

The quality and reliability wasn't good enough for my needs. I started looking at replacement firmware options.  The board is actually pretty good for being in a budget printer. The microcontroller being a STM32F070CB, removed mainline Marlin as an option.

Solution

Then I found Marlin4MPMD on Github. After some investigation I discovered that my particular printer didn't ship with a UART bootloader. In order to load a fresh image on to the microcontroller, I had to breakout my Segger J-Link, but any Serial Wire Debug programmer would work.

The pinouts for the STM32F070CB yielded the appropriate pins for SWCLK (PA14) and SWDIO (PA13). The pins should be 5V tolerant according to the documentation, but without having the schematics for the board, 3.3V is a safer move.

Jumper wires were soldered to the through holes to make future use easier.  

Following the instructions on the Github wiki, the chip was reflashed with a prebuilt image using Seggar's J-Flash Lite. Since I have an pre v43 version of the Mini Delta and the provided power brick, I went with the standard image with a 5A limit.  

Before going into the calibration phase, I felt that replacing the bed material would be a good idea. I wasn't able to find a 120mm circular magnetic build plate, so I went with borosilicate glass.

The original bed material was pulled off with the help of a putty knife. The new glass bed was attached using thermally conductive double side tape. It's not the best way to do the job, but it will work for now.

Results

After going through the calibration steps, the printer was again ready to print. Mesh auto-leveling was used to probe the bed and results saved to the SD card. The baseline before the firmware update is on the left, first pass calibration in the center, and completed calibration on the right. The cat on the right had accurate dimensions within 0.08mm, so it's doing quite well. There were no adhesion problems when using some glue stick on the bed. The hotend on the Mini Delta uses a single 30mm fan for both heat sink and part cooling. From the print quality it's obvious that the part cooling airflow is insufficient to cool the PLA.

Next Steps

It seems that the mechanics of the printer are solid and are more than capable of delivering the accuracy I'm interesting in. However, the hotend falls far short. I wasn't impressed by the existing designs people have come up with for replacing the hotend on the Mini Delta, so I set out to design my own. More to follow in Part 2 of this series.

I've recently become the co-owner of a 1972 Porsche 914 with a 5.7L LS1 swap. As a result of the shift from air cooled to water cooled, the car required the addition of a gauge for coolant temperature. An oil pressure gauge was also added and the two gauges sat together, mounted on the floor near the shifter. It didn't look great.

Quad gauges were a popular alternative to the stock fuel combo gauge, but were hard to find or quite pricey. I set off to build my own using a salvaged Porsche 944 fuel gauge and a spare 914 combo gauge.

Rather than just stick a bunch of VDO gauges together in a shell, I wanted to preserve the indicator lights for the generator, low oil, low fuel, and parking brake. This resulted in a custom PCB that would plug into the existing wiring in the car and connect to the VDO gauges. All inputs from the car are optoisolated. An ATMega328PB at 16Mhz powers the operation. Overkill for this application, but it's just so damn easy to use. A LDO voltage regulator provides 5V power by stepping down the 12V from the car.

In order to light the gauge and provide indicators, 12 SK6812RGBW LEDs were used. They are controlled by the microcontroller via a 1 wire serial protocol, which allows for full RGBW control. The ability to individually address the LEDs allows them to act as both lighting and indicators. The VDO gauges insert directly into the PC pins on the board, and the board uses two styles of VDO gauges. An older style gauge, which was salvaged from the 944 fuel gauge, and a new style gauge which was purchased off the shelf.

The gauge panel was laser cut and etched from painted acrylic. The final result looks excellent. The VDO gauges were attached to the gauge panel with the factory screws.

The gauges are packed in as tight as possible and it is obvious which gauges are new and which were salvaged.

The gauge assembly was mounted on the PCB.

The spacing between the gauge panel and the walls of the enclosure allow light from the LEDs to shine through.

The gauges are secured to the PCB with hex nuts that perform dual functions of securing the gauge and connecting the gauge to ground.

This was the critical piece salvaged from the 944 fuel gauge.

In order to mount the button on the gauge to match my existing gauges, a 7.5mm hole needed to be drilled. Drilling thin sheet metal without damaging the surface is challenging and this jig makes the task much easier.

There are two upper parts of the jig. One with for a 2mm pilot drill and a 7.5mm for the final hole.

The hole was drilled and the button installed without a scratch on the surface.

The gauge body is made from 3D printed High Impact Polystyrene. A support inside provides rigidity and a place to mount the gauge face.

Heat set inserts are used to provide mounting points for the PCB.

The original two screws from the 944 were trimmed down to 4mm of length and used to attach the gauge face.

The button is secured with hot glue, which is reversible, if it ever needs to be removed.

The front face glass and button from a 914 combo gauge were used to complete the gauge face.

A few taps around the back bends the metal on to the lip of the gauge body. The internals are inserted from the rear, which allows for easy maintenance or firmware changes.

The completed gauge was difficult to photograph as the glass was highly reflective making the markings hard to see.

The PCB provides pogo pin programming port to allow for quick firmware changes. The serial port on the microcontroller was also broken out, but was not populated in this version. The total cost of the project was around $240. The price will vary as the price of 944 and 914 gauges on eBay seem to fluctuate.

The completed gauge in action looks quite good. This iteration of the design is suitable for very low volume production, but the reliance on salvaged parts complicates the process.

The original project can be found on HackADay.io

Change of Plans

The initial plan to just refurbish the mill was scrapped, in it's place is a new plan to build the mill into a more robust VMC type tool. The new setup has a flood coolant system, and more integrated electronics. The plans for the enclosure/stand will be made available soon. The following is a brief log on the fabrication of the enclosure/stand.

The mill enclosure/stand is made from plain steel tubing welded together.

The tubing was squared up using magnetic squares and welded with flux core steel wire. I didn't have a gas bottle for proper MIG welding.

This is the upper section of the stand. Its made from 1.5" x 3" 16 gauge steel rectangular tubing.

A plate was welded to the upper section as a mount for the display and keyboard arms.

The completed stand before coating.

0.75" square tubing was used to build the enclosure section.

Completed enclosure.

Welding in the coolant tray.

Coolant drain holes.

Attaching the heavy duty casters after being painted.

Mounted a 4U drawer on the left and a 9 channel power strip and a 4U server on the right.

LinuxCNC running on the server, with the display and keyboard mounted.

The arms can swing and be adjusted vertically.

The enclosure is still missing the acrylic sheets and the coolant system has yet to be installed. The mill was disassembled completely and cleaned. It will be reassembled directly on the new stand pretty soon.