View Full Version : [Project] DREGS, MIRTH, etc. (Reworked Desk)

11-08-2016, 09:21 PM
So, making myself write and post this to distract from my rather severe anxiety about the prospect of chumpchange winning. Planned to stop by Home Depot to pick up additional supplies, but could not motivate myself to go out today (early voter on 10/26). The desk I've previously mentioned when in awe of CasperH's workspace (http://forums.trossenrobotics.com/showthread.php?7875-Z-363-Spectre-3D-Printed-Body-for-Bioloid-Premium/page2) is two pieces of 5/8" MDF on a round steel tube frame made from three tubes bent into a C-shape and bolted together.

The larger/lower level is 35.875" wide and 23.875" deep with 1/16" bent steel brackets providing additional support to the MDF instead of just relying on the ~2" wood screws going horizontally into the 5/8" thickness of the MDF. Going to finish stripping/sanding the bubbled surface and remove the melamine from the borders then spray the MDF and tiles with 3M High-Strength 90 Spray Adhesive to bond them together and use silicone caulking instead of grout for a non-conductive, heat resistant, waterproof work surface. The edges will be finished with 1" to 1.5" 6063 aluminum angle (1/16" thick) that will sit flush with the tile surface at the front half and provide a slight barrier at the back half (stuff often slides off the back of the MDF as-is). The aluminum angle will also help correct the slight sagging of the lower level MDF that is not significant, but is noticeable.

The upper level is 35.875" wide and 9.875" deep with additional 1/16" bent steel brackets just like the lower level to keep the 5/8" MDF from splitting open because of the ~2" wood screws. The sagging is extreme (~0.5" at center) and normal height only leaves ~8" between the two levels, so not going to reuse that piece of MDF. I've not finalized the design of the upper level, but going to be one or two level cabinet with frosted polycarbonate front cover (peg board) that provides at least 12" of clearance from most of the tiled work surface. The pegboard is going to be spaced 0.5"~1" out from another opaque board dotted with LEDs for back-lighting and sealing against dust intrusion into the lower cabinet through the pegboard holes. The important part is a 24"x24"x1" HVAC air filter chopped down to 12"x24" and installed in a cutout centered in the sloped bottom of the lower cabinet which feeds a 24V/1.1A/~280CFM axial blower that will discharge through a vertical duct running up the back center of the cabinets. The large filter area should mean limited pressure drop to the intake of the blower that has a blade diameter of just under 6". Will probably have some cooling fins protruding into the vertical duct for electronics in the lower cabinet.

11-09-2016, 01:20 AM
Something like this?


Will it be watertight enough for all the tears?

11-09-2016, 02:11 AM
More worried about it surviving impacts from my head.

11-09-2016, 05:57 PM
Decided to be lazy, so just flipped and painted the bare underside of the 2'x3' MDF panel since the small screw holes will be covered by the tiles and it inverts the slight sagging of the MDF. Used up the last of the satin (almond?) can of spraypaint since it did not quite match the existing color of the basement door and wanted something a bit tougher than latex primer. Also going to try using only clear Silicone II (diluted with glycerine and/or acrylic paint) (http://www.myheap.com/casting-molding/my-heap-mold-book/chapter-8-silicone-caulk-molds.html) to attach the tiles to the MDF instead of the spray adhesive. Have to wait until tomorrow for the paint to be fully cured, but can still cut and attach the aluminum angle to create a basin of sorts before trying to attach the tiles. One leg of aluminum angle gets attached to bottom side of MDF with short woodscrews and a bit of silicone with the other leg pointing up to create a 5/16" or 9/16" rim/lip around the perimeter of the MDF. Going to take some measurements and do a bit of math to try to get the volume of silicone applied to the MDF to fill all the gaps between the tiles in one go instead of attaching the tiles with some silicone then filling in the gaps one at a time.

11-14-2016, 03:52 PM
Had a yard sale over the weekend, so no progress except to verify that my first test of just silicone and acrylic paint will never fully cure when squished too thin between two pieces of scrap polycarbonate. Likely used way too much of the cheap, thick, gloopy acrylic paint for the amount of silicone that I added to the mix (not so much one drop as one thick blob of paint). The silicone requires the addition of water to cure, while the paint requires evaporation of existing water to set/dry, so too much paint means it never loses enough water when squished too much between two impervious surfaces (like glass tile and enamel sealed wood).

Thinking I need between 800 and 1200mL of silicone mixture to bond the tile and fill all the gaps depending on how thick to make the layer between the tiles and wood.

11-14-2016, 05:29 PM
DIY adhesive chemistry. I love it :-)

I presume this is "because you can," given that the cost of tubes of industrial adhesive in tubes from the Depot is pretty low?

11-14-2016, 06:15 PM
Depends on the adhesive. Lots of them (like contact cement and even 3M 90 spray adhesive) require at least one pervious surface to permit the solvents in the adhesive to properly evaporate, otherwise you have to get both surfaces coated in a thin layer of adhesive and let them sit apart for a while to let the solvent evaporate before they can actually adhere together. Single component RTV silicones (acetic acid and alcohol-based cure processes) require some amount of water added to the volume of goo to properly cure, so face the same problem as other adhesives when placed between two impervious surfaces (for opposite reasons) but water-bearing additives can largely solve that problem for silicone.

Want something with a bit of flexibility, lots of heat resistance, and an 'all at once' assembly process so I don't have to go and squirt another sealant between the gaps of 1176 0.75" square glass tiles. Those requirements pretty much eliminate most adhesives and grouts. Casting Silicone II modified with glycerine and acrylic paint lets me just mix the volume needed in the correct proportions and pour it onto the table, spread it out evenly, stick down the mosaic tiles, and gently press the tiles into the silicone until it oozes into the crevices between each tile. The paint also lets me make the silicone be any color I want instead of being stuck with whatever color the adhesive ends up being after curing. This is somewhat important for the desk surface, and also nice for future plans to start casting some small, grippy robot wheels onto 3D printed hubs once I get the mix right.

11-14-2016, 07:39 PM
Grouting when done isn't THAT hard ...

Anyway, interesting idea on the small wheels. If the adhesive doesn't need a permeable surface, how will you make it release from the mold, though?

11-14-2016, 08:09 PM
Grout is not a flexible material, except for latex based grouts which do not have much heat resistance. Grout also requires additional sealing against moisture and a bit more difficult to color match.

As for molds: coat it with a release agent as you have to do with just about any other mold. The silicone is intended to be the tread and cast-in-place onto the 3D printed hubs held in the 3D printed mold. The mold gets made from two or more interlocking segments that are clamped together by a band on the exterior with a peg/mini-shaft to hold the hub in position while pouring and curing.

Surface bonding and curing are independent. Curing is dependent only on getting enough moisture (or hand-wavy magic when using two-part platinum cured silicones). If the surface is pervious, the silicone can seep into the pores and improve bonding because of the very high surface area and creating 'tendrils' of silicone partly locked inside the pores (also makes it nearly impossible to create a mold for casting unless the surface is properly sealed and covered with release agent). If the surface is impervious/sealed, the silicone seeps into any surface texture but does not penetrate into the surface material. This results in a lower bond strength but it should still be more than sufficient for bonding the glass tile to the highly textured wood surface as long as the surfaces have been cleaned of any contaminants (dust, oils, detergents, etc.) before application of the silicone.

11-17-2016, 06:25 PM
Finally settled on some project names for parts of this collection of desk mods, but forgotten how to rename the thread itself without simply moving it to a different sub-forum then back again. Meh.

DREGS - Desk Refurbishment and Expansion: Glass-Siloxane Surface
MIRTH - Mixable Illumination Reconfigurable Tool Holder
TRAFFIC - Total Room Air Filtration and Forced-air Internal Component Cooling;

Currently testing a new mix using a lot less paint and no glycerine. Also used only manual pressure for a 1~2mm gap between surfaces like the tiles will actually get, instead of using the locking clamp like last test for curing very thin sections of silicone between two impervious surfaces. Decided to go with a 12V/1A 120mm axial computer fan instead of the bigger 24V fan. Still moves up to 151 [CFM] of air, but quite a bit less expensive and easier to replace should it be necessary.

01-03-2017, 03:35 AM
I've been oddly productive the last few days while watching the Yowamushi Pedal series for the second time on Crunchyroll (starting the third time right now). I've finally gotten the IMU and sensor fusion code for Teensy-ShInDy compiling and operating correctly, although it did take a bit of time since I suck. Was only getting two axes from the accelerometer because I sent a '3' instead of '7' to the enable register and lots of little typos / copy-paste errors that caused some issues. Took a while to find my error in the sensor fusion code: I isolated all of the filters from the rest of the code and had to pass everything in a few pointers/arrays but forgot to update the delta_time variable prior to passing it to the filter update function. Also managed to convert everything to ROS-like standards with leading underscore instead of trailing underscore for class global variables. Still do not have the SSD1306 OLED code working correctly, but hoping it is just the lack of external pull-up resistors. Tomorrow is the dynamixel library and more testing of the OLED code, so maybe I will finally get Ripley operating again.

That little bit of progress combined with the looming deadline for the 'tables and desks' contest on instructables seemed to be sufficient motivation to finish DREGS. The tile and silicone surface has been finished, but have to remount it on the frame and fill in some of the gaps between the tiles that did not fill because I used too little silicone. MIRTH and TRAFFIC have been cut down to just a filter and surplus 120mm axial blower mounted in an Instamorph housing and powered off a 12V power supply, but will be revived sometime in the future with proper aluminum extrusion frame instead of the crap wood I would be forced to use right now.