Sunday, September 17, 2017

Restoring the Ride (revisited!)

For the last few weeks, I've been totally obsessed with a new project, fixing, building, and revising the ghost of some of my past Maker failures at the same time.

I decided to update my aging Mitsubishi Eclipse Spyder, spiffing it up and reversing some of the ravages of time.  It's still a daily driver, so anything I could do to make it better would make my commute more enjoyable.

This is my car on the day I drove it home twenty years ago.  It has been my constant companion, having followed me through many happy times and countless life changes.

Yet, the years have resulted in a fair amount of wear and tear -- for both me and the car -- but at least for the car some of that can be undone with a little skill and significantly more effort.

I decided to give the Eclipse a full makeover both inside and out.  

For the interior, this included stripping it down and replacing the stanky carpet (top picture), cleaning up and touching up all the plastic surfaces, fixing and covering the upholstery, upgrading the gauges and lighting  (LEDs!), and cleanly rewiring the radio and all the accessories into a separate fuse box case I made on the laser cutter because I could.

For the engine bay, I got inspired by some online modding communities to do some largely unnecessary upgrades.  I cleaned everything up, replaced all the belts, upgraded the hoses with silicone and stainless steel, detailed some items, and upgraded the intake (little more pick-up, nice vroom-vroom sound!) just for fun.

The toughest challenge, however, was the exterior, specifically the hood and front bumper.  They were in poor shape, with cracking and chipping paint, making the Eclipse a 20/20 car -- it looked good only if you watched it going 20 miles per hour from 20 feet away.

The most painful part was that it was all my fault, and the latest example of three failed previous hood paint jobs:

Failure #1 (2004) - After touching up some chips, I clearcoated the hood with an underpowered compressor, resulting in a generally acceptable but slightly uneven clearcoat in sections.

Failure #2 (2008) - To cover up the uneveness, I repainted the hood with custom mixed paint, but decided to save money using inexpensive clearcoat from OReilly's.  It looked good at first but developed spider cracks after a few years.

Failure #3 (2012) - The spider cracks wouldn't make a good surface to paint on, so to avoid stripping the hood, I purchased a new aftermarket "primed" hood I found for sale on Amazon.  Since it was already "primed" I assumed I could just scuff it and paint it.  I also used extra materials to repaint and clearcoat the front bumper.  Neither surface was apparently prepped well enough though as they both started cracking and chipping after a few years.

So this is where I began try #4.

Since I still had the original factory hood around, I decide to strip it down and start from scratch, doing everything possible to help the paint stick.  This was a painful process that involved paint stripper and a lot of scraping and sanding with a dual-action sander (thanks Harbor Freight!).  Once all the old cracked clearcoat was gone, I wet-sanded it smooth.

The bumper got a similar treatment with sanding off of all ugly bits and really scuffing up the rest before wet sanding to smoothing it back out for primer and paint.

Ignore the hood in this pic.  It's actually the bad one and while it kind of looks OK in this picture, it looked horrible from any other angle.

With that done, both parts were ready for paint.  Time to put on my Breaking Bad costume.  Safety first.

First, I primed both parts using a two-part primer that you mixed together before spraying.  It was expensive, so surely it must be good and give maximum stickiness.  I used a $10 spray gun, also from Harbor Freight.

Then came the color coat.  Looks better already, just not shiny.

Now the all important clearcoat.  This time I used a pricey urethane clearcoat that, like the primer, comes in two parts that you mix together, which starts a countdown clock for you to apply it before it hardens by chemical reaction.  It's supposed to yield a super hard clearcoat, which I needed because I don't want to do this again...again.

Fingers are crossed.

Incidentally, I purchased all the paint materials together at 

It was getting dark by the time I did the bumper. Even the dark can't hide that shine though!.

The next day, everything was looking promising, but the hood had picked up some bubbles and specs of stuff while it dried.  I really should have sprayed it propped up.  Oh well.  Good to know for the fifth time.

There better not be a fifth time..

To smooth it out I ended up doing a WHOLE LOT of wet-sanding with successive grades of sandpaper (1000, 1500, 2500, 3000), then buffing with rubbing compound and polish.  The good news is that this got rid of any orange-peel like bumpiness in the paint at the same time.

This was a PITA.   I knew I needed to wet sand enough with each grit for the buffing to work, but was constantly afraid of accidentally sanding through to the color coat.  That would have force me to repaint everything again.  In the end, I had to go back a few times over the course of two afternoons of messing with it to re-sand everything before the hood would polish up well.

This is the result, with the hood swapped back onto the car.  Not too shabby, if I say so myself (though still waiting on a new logo badge)

 Now no longer a 20/20 car.

Maybe even a 5/5 car now.

Wednesday, September 14, 2016

DIY Faux Ceiling Beams and Laser Cutter Exhaust

Ever since I finished constructing the Monkey Cave, I've thought about adding ceiling beams to complete the look.

Not only could they help even out the difference in height due to the drop ceiling for the garage door, but they could give the Cave some interest and rustic appeal, kind of like Tuscany, minus Diane Lane and plus 3D printing and laser cutting.

Exposed ceiling beams, though, aren't really native to construction in this part of the country.  Instead, I found that most beams are really "faux" versions made of lightweight foam.  Light on weight doesn't mean light on the pocketbook, however as I'd have to spend over $500 to get them shipped, which seemed too much to me to pay for a little Tuscanification.

I also considered making "Box" beams from separate planks nailed together.  While hollow, they would still be quite heavy compared to foam beams.  More importantly, they wouldn't look solid, as it would be very difficult to hide the seams and finish the planks well enough to provide the illusion of the one piece beams I wanted.

Because no solution seemed ideal, for a long time I shelved the idea entirely of adding ceiling beams at all.

Recently, though, the opportunity to revisit ceiling beams came when pondering upgrades to my laser cutting setup.  Up to now, I used a flexible hose to route the fumes out a vent opening in a nearby exterior wall.  But because the hose crossed a walkway and was inconvenient to connect and disconnect, I wanted a more permanent solution.  I came up with the wacky idea of ducting the fumes across the ceiling instead, perhaps hidden within a faux ceiling beam.

Once again, I looked into foam beams, but found that their walls were simply too thick.  To get a big enough opening inside to provide adequate airflow, I'd need at least a 7x7 beam, which would be both too big and even more expensive than before.

Brainstorming for other materials to use, I came across a solution I'd never seen used before: PVC fence posts.  They're light, hollow, come in wood grain finishes, and Home Depot had some "end posts" for sale at a really good price.  They could work perfectly if I could tape up the side openings, connect some of together to extend their length, and perhaps add a faux finish on top to make them look more like stained wood.

To reach the ceiling, I also needed to add a vertical section.  For this I used some oval galvanized ducting.  I added pop-riveted sheet metal to close the ends and outlets at the top and bottom.

I mounted the assembled unit in a corner where there was an offset in the wall, allowing me to cover it up later by extending the front-most section.

To pull air through the duct, I mounted a 12V bilge fan into a short length of post, laser cutting an acrylic plate to seal the fan in place.

The fan would be mounted at the end of the ducting run just inside the vent opening in the wall.  The wiring to power it would run inside the ducting to the laser cutter.

To finish off the beams, I cut up the edge of a chip brush to make it ragged and used it to dry-brush two coats of mahogany gel stain onto post segments.

Despite being fairly loose and random with the paint brush, I was pleasantly surprised by the final effect.  Even with minimal effort, it looked very convincing even close up.

The vent opening in the wall was located a few inches below the ceiling.  To match it, the ceiling beam duct couldn't be mounted directly to the top.  Instead, I decided to use short standoff blocks--fastened to the ceiling with toggle bolts and wrapped with short sections of post--to hold the beams at the proper location.  Besides adding the needed spacing, the standoffs also provided a way to hide the joins between separate sections of pipe.

Designed specifically for faux ceiling beams, Home Depot sells rubber strips made to resemble distressed bronze straps with rivet heads.  Wrapped around the posts at the standoffs, they hide the seams while also providing the perfect accent detail.

I added framing around the vertical duct, covered it with drywall, and then taped and buttered the seams.
I sprayed it with a texture gun and let it dry. When painted, it nicely matched the adjoining wall.

Finally, I added two non-functional beams to complete the installation.  Here is the completed project.

Sunday, April 3, 2016

Hacking the K40 Laser Cutter

Bargain Chinese laser cutter/engravers typically come with funky proprietary controller hardware and software.  While more-or-less functional, the software tends to be buggy, and limits future support availability and upgrade options.

This post details my experience converting my K40 Laser Cutter to use open source controller hardware.

I chose an AZSMZ Mini to use for the new controller.  A variant of the Smoothieboard, the Mini is a compact unit whose built-in features make it very easy-to-use for this application.

I bought an AZSMZ Mini with stepper drivers and LCD display for $80 on eBay.  As my stock control panel was stark and minimalist, I only had to rearrange some of its controls to make space for the new LCD display.

The first step was to build a support for the Mini and LCD board to mount beneath the control panel.  I cut a bracket out of clear acrylic sheet and attached the controller and LCD to it with nylon standoffs.  In front of this, I mounted a new control panel that I cut out of two-color plastic sign-making laminated sheet.  Fortunately, a laser cutter is the perfect tool for doing this kind of work!

With the new control panel mounted in place, the next step was to hook up the controller board to the laser cutter power supply and motion control hardware.

Most others who have done similar upgrades have simply thrown out the old controller (in my case a Moshiboard) and painfully wired up the replacement directly to the existing power supply, stepper motors, and endstops, either directly splicing cables or using a Middleman board to aid in the process.

I, however, decided to take a different approach.  Instead of tossing all the old hardware entirely, I'd make my own adapter board that would allow me to switch back and forth between the old and new.  This would give me the flexibility to use the best software for a particular job, as well as options in case of an incompatibility or breakdown occurred in one specific software/hardware combination.

While power supplies tend to come in many variations, Moshiboard controllers do not.  By designing an adapter board to replicate the Moshiboard connectors, I could simplify the wiring and reuse the existing cabling as well.  While some laser cutters have separate cables for endstops and the X stepper motor, mine combines them in a single flat flexible ribbon cable, so I made an adapter for this configuration.

I was surprised to find out how cheap custom circuit boards are to produce if one is willing to wait a month for shipping from the Far East.

I downloaded and learned a free design program called FreePCB and used it to layout my custom adapter board.  It uses standard connectors and three CTS 206-125 dip switches to choose which controller to enable.  The idea is to switch all dip switches to one direction (with the power off!) to select which controller board to use.

I had 10 boards manufactured at SeeedStudio for a total of 20 bucks.

Wiring up everything was super easy.  The connectors on my adapter board line up in position and orientation with the ones on the Moshiboard, and are attached with short jumper cables.  The only exception is one of the two FFC ribbon cable sockets (the one that connects to the Moshiboard), which I designed to be rotated 180-degrees around so that it could be connected with a very short FFC bent neatly back over itself into a U shape, as pictured below.

Cabling Diagram

The AZSMZ Mini has a built-in voltage regulator, so a single 24V power connection powers both the stepper motors and electronics.  A single wire connects each of the X and Y endstops, and the stepper motors connect with simple 4-wire jumper cables.  The fire line on cutter is active low, so a feature of the Mini simplifies its connection.  The board comes with Mosfets that switch to ground for heater and fan connection, so I simply wired the fire line to the D8 bed heater terminal which is controlled by the PWM1 line.

With the hookup complete, the only thing needed to be done is configuration.

Smoothieboards support easy configuration via a config.txt file that is loaded at boot time from a MicroSD inserted in the main board.

Here is a copy of my configuration.  I had to make adjustments from the defaults to calibrate my motor movement and direction, as well as lower the acceleration and fire PCM rates to get the behavior I wanted.

And here it is the result; all connected up and ready to go.  I've made a few test cuts so far, and all is looking good!