Additive Manufacturing Machines are Hot &How We Solved Problems Plaguing Our 3D Printer

By Mark McCrate       Edited by Clayton Smith

3D printers offer a bevy of manufacturing options, ranging from hobbyist tinkering to government agencies building dissertation worthy devices.  Thankfully many designs are open sourced making proliferation easy.  The “shoebox” case and control board are a great example of liberal copying, Figure 1 and Figure 2, respectively.  These have been widely adopted for good reason.  This older design architecture is very compact, easy to construct, lightweight with minimal external cabling, and has an intuitive interface to boot.  However, those benefits come with a very real detriment—poor thermal management.

Figure 1  Widely adopted “shoebox” enclosure case.

Figure 2 Common control board.

To prolong part life, prevent failures and perhaps improve performance, this article details several design modifications our hackerspace/makerspace implemented to remedy woes.  They are broken into sections:

• Essential – for preventing problems and prolonging life
• Optional – to extend operating margins
• Troubleshooting needs – for our specific case to regain functionality of a burnt and baked PCB board

Feel free to focus on areas most pertinent to your situation.  By following our hard learned lessons, hopefully you will be raving about saving some effort. 

Note, the electronics board shown herein is based on the RepRap Melzi V2.0 documented here: https://reprap.com/wiki/Melzi and installed in the shoebox that is part of many products.  If you have any of the growing list of products at the end of this essay, bookmark this page!

Essential 1: Check all fan air flow directions.

The goal of fans is to move heat as efficiently as possible.  Whether your shoebox has only one fan or multiple, all air must flow in concert to move heat from inside to outside the enclosure.  If fan flow direction needs altered, remove all supporting fasteners, flip fan(s) and reattach. 

In our case (pun intended), the power supply exhausted directly onto the control board and the lone case fan pushed every warm air molecule on a collision course with the LCD screen, a fact easily verified by looking at the flow arrows molded into the fan’s plastic shroud; in lieu of finding visible arrows similar to those shown in Figure 3, hold a tissue near a running fan to reveal which way air is blowing.  By reversing our fans, cooler ambient air is now pushed into the center of the box, impinges onto hot parts, gets directed toward the power supply inlet, travels through the power supply, and is then immediately whisked away by the case fan breeze.  This change dropped the internal box temperature from sauna to only slightly above ambient!

Figure 3  PC Fan with rotation and flow direction arrows, from computerinfobits.com

Essential 2: Better heatsinks are a good way to get more heat out of components and into convecting air. 

Figure 4 shows there were some rather squat anodized aluminum heatsinks with a questionable thermal compound interface.  These had to go!  To remove the old heatsinks we used pliers to gently twist and pull; another quick and easy method to remove heatsinks is by simply nudging them with a flat blade/head screw driver – they should pop right off.  Once the old metal is out of the way use an alternating combination of concentrated rubbing alcohol, acetone, and a cotton swap to scrub and clear away any remaining adhesive from on or around the chip(s).  We installed taller copper pin-finned heatsinks bonded with 3M 8815 thermally conductive tape.  Thanks to this change, the heatsinks are no longer uncomfortably warm. 

3M did not sponsor this work; it just so happens we used the same 8815 tape for another project years ago, with great success, and had extra.

Figure 4  Original inadequate heatsinks were black anodized low profile Aluminum.

Completing just the two essential tasks basically guarantees a long life and problem-free user experience, however, further optimization will expand operating margins.  Two optional modifications are detailed below.

Optional 1: Improve overall airflow by reducing restrictions. Most shoebox designs have very few openings thereby restricting airflow which is odd because many tower PC cases devote five, six and sometimes more facets to air flow; as an example, look at Figure 5 which is basically a Swiss cheese structure!  To further stress this point, when designing their ‘towers’ Apple has devoted design time to thermals as far back as always [citation needed] with their fan-free designs, asymmetrical blisk, zoned volumes, unified heatsink, etc.  Apple’s 2021 white paper “Mac Pro Technology Overview,” archived HERE, highlights open chassis lattice design, pg. 21 or Figure 6, and devotes several paragraphs of the 45 page treatise to thermals.  While it is possible to copy baffling structures, a more practical approach is to attack the outside with a drill and, like a deranged affineur, persist until the structure more closely resembles a fine Swiss cheese.

Figure 5  Cooler Master’s example of a fully perforated chassis.

Figure 6  Apple’s lattice structure.

In our example, to increase air flow we started by removing the back panel (where all the heat was previously coaxed), laid out a grid pattern and hole size to maximize material removal while maintaining structural integrity, drilled and deburred.  Our work is shown in Figure 7…increased exits points with nearly infinite more area and subsequently reduced backpressure.

Figure 7  3D printer control box rear panel with added perforations for increase air flow.

Optional 2: Move all fans to places where they impinge directly onto high power / hot components. 

Many 3D printer cases are designed with fans that suck air out of enclosed volumes, however, any pinholes short circuit this flow.  A better arrangement, to maximize heat transfer, is to place fans proximal to and blowing directly on susceptible components, by doing so, warmed air can exit any aforementioned pinholes.  Apple and others take it a step further by partitioning the tower for multi-fan multi-zone cooling, see white paper pg. 22 or Figure 8. 

Figure 8  Apple’s multi-fan multi-zone cooling.

Moving the fan locations takes careful planning and it is an obstacle many will choose to avoid because it demands answering questions such as: will it clear all existing parts, are all necessary tools easy to acquire and does a new location require rewiring?  We reasoned the effort was worthwhile, spent an evening moving a fan from the rear to the side, Figure 9, and measured a roughly 10*F lower box temperature despite forgoing the pains of installing baffles.

Figure 9  Test fitting a new fan location.

For most people those steps detailed above are the whole ordeal.  Hopefully, your hardware is healthy and the essential and optional steps got implemented in time.

For the unfortunate, including our hackerspace, the journey started from a need to troubleshoot a loss of functionality (namely, no bed heating) and lead to a two-step recovery process.

Troubleshooting need 1: Fix PCB traces.

Upon removing our control board, a quick look at the its top, Figure 10, revealed discolored connectors complete with melted and charred plastic, while the bottom showed more signs of thermal stress including paint bubbles and reflowed dull solder joints, Figure 11.  Probing around this troubled area with a multimeter, we discovered the root cause of no bed heating was a burnt trace/open circuit to the bed connector; apparently a concentration of current and heat was enough to burn the thin trace off the board.

Figure 10  Printed Circuit Board Assembly has a discolored charred pin.

Figure 11  Printed Circuit Board has both paint bubbles and reflowed solder.

A cool tip to both fix our open circuit problem and simultaneously increase trace conductance is to leave traces bare, then coat them with solder.  Examples of this are prevalent especially in power supplies.  EEVblog demonstrates this wonderfully on YouTube, EEVblog #317 – PCB Tinning Myth Busting. 

To scrape off the solder mask and expose fresh copper traces, ripe for absorbing solder, we used a sharp razor blade.  Going a step further, and to provide extra reinforcement to the pin, we applied solder generously around the pin base, much like mulching around a tree.  At the time of writing, this abundance of solder has survived hours of “burn-in” testing.

Troubleshooting need 2: Install higher current connector(s).

Prior to executing the PCB trace fix, we removed the original charred pins, enlarged the paltry through holes and then installed larger pins from a new “Phoenix connector” with a higher current rating.

Having read this far is a testament to how much research and effort you are willing to devote to modifying your 3D printer for preserving performance and extending operational life.

List of Products

After gaining widespread adoption the common “shoebox”, Figure 12, is used in the following growing list of products. 

IF YOU KNOW OF OTHERS PLEASE COMMENT TO HELP BUILD A MORE COMPREHENSIVE LIST!

Alfawise U20

Anet E10

Creality CR 10 mini

Creality CR 10

Creality CR 10 S4

Creality CR 10 S5

Creality CR 10 V2

Monoprice Maker Select 3D Printer v2

Tevo Tornado

Wanhao Duplicator i3

Figure 12  Shoebox from another angle.

On SpaceX’s Starship Flight Test #1

by Mark McCrate

In a genius move SpaceX and The Boring Company are teamed up to do incredibly rapid excavating and tunneling with an innovative star [tipped] ship, complete with a less dense liquid TNT replacement called methane.

Hyperion EOS 0720i SuperDuo 3 manual

I have owned and used a Hyperion EOS 0720i SuperDuo 3 for several years. It has worked great since day one! Eventually I needed to use their Control and Data Suite software and I have already read its online manual, twice, and tried each function at least once. After reading all documentation, it became apparent having a downloadable and searchable manual would be beneficial. I want to contribute by assembling a new manual; it is based on what is available plus a few more screen shots. [DOWNLOAD PDF BELOW] Hopefully this endeavor is worthwhile. Please let me – everyone – know if you found it useful by commenting: we can work together to create a better pdf if people suggest enhancements or find errors.

hyperion-eos-control-data-suit-manualDownload

Eleventh Hour Post

Since 2013, I have posted to this nook of the net at least once per annum; in 2021 this short sentence is sufficient to continue that cadence.

Tesla’s New 4680 Battery Geometry

On 2020 Sept 22, Tesla held a long awaited “Battery Day” event. This event was long awaited for at least two reasons: first, Tesla announced – or leaked – its 2170 battery plans way back in 2016 Quarter 3 and second, Tesla completed acquisition of Maxwell Technologies on 2019 May 16 for reasons many speculated but were waiting to fully understand (we know Tesla innovates rapidly so any delay in official news becomes long awaited!). During the event Elon Musk and Drew Baglino talked through a 74 page slide deck detailing many advancements they have worked on to turn production hell into an almost generational advantage. Thankfully Tesla has made the slide deck available for download. A truncated list of some new ideas include:

cell design
cell factory
anode material
cathode material
cell vehicle integration

There are many innovations! One could argue ideas such as tabless, dry cell or thermals are the most compelling but how many are aware of current cylinder construction, care about manufacturing processes/chemistry or bother to understand thermal management? Looking at the aforementioned topics, it is prudent to forgo them to focus on the only thing people will evidently remember, that is, the new geometry. So borrowing from the succinct 1865, 18650 and 2170 titles, this article will focus on Tesla’s new 4680 battery geometry.

These new cells remain cylindrical. This is a good choice for building battery packs because the packing efficiency remains 91%. To achieve a 91% packing efficiency assumes placing circles on an infinite plane and the floor pan of an automobile is essentially an infinite plane.

“Can” dimensions are much closer to optimal. A sphere maximizes volume while minimizing surface area, but it is impractical for building batteries and does not pack well. A cube packs well, and is close to optimal as far a surface area is concerned, but has painful internal corners. A cylinder is a great compromise. With calculus, it is easy to calculate the dimensions necessary to maximize the volume of cylinder while minimizing surface area and not surprisingly the ratio is h = d or 1:1; basically a sibling to a cube. Looking at all the battery sizes Tesla has built into vehicles and their height to diameter ratio shows, h/d:
1685 = 3.61
2170 = 3.33
4680 = 1.74
Given 1 is optimal the 4680 battery is almost there.
A 4680 cell has a volume ~5.5X that of a 2170. Slide 28 states 5X energy and 6X power so it appears much of the inside chemistry is close to constant; well, changing but not as rapidly as from 1865 to 2170.

Wall thickness is currently an unknown so assume a uniform 1 mm wall thickness for all battery sizes Tesla has used/uses – a gross over exaggeration and simplification – to keep things easy. Taking a 2170 as the new baseline, adding both the can and cap, an 1865 requires only 78% as much material as a 2170, whereas the 4680 needs 289% more material; however, looking at a whole pack constructed using 4680s, it only uses 60% the material of a pack built with 2170s. In reality, since slide 63 shows the cells will be part of a honeycomb like structure the weight savings is even more. It gets better. With 4680s a vehicle will only need 18% of the current cell count. Producing, quality controlling and building with hundreds rather than thousands of cells is superior because it enables saving both time and money. True, a brave person could gather five and a half 2170s, cut and tape them together to build an “equivalent” 4680 today however that person might not survive the resulting fireworks so it certainly will not be me!

Comparing past, present and future is interesting because Tesla has manufactured, currently manufactures and will likely continue to manufacture all of the aforementioned battery sizes for a long time. Figure 1 and Figure 2 are side-by-side size comparisons of 1865, 2710 and 4680; they are all additively manufactured using Fused Filament Fabrication (FFF, formerly Fused Deposition Modeling, FDM); the color is simply a product of material availability. Smaller models were downloaded from Thingiverse, whereas the new large cell is modeled from scratch.

Figure 1

Figure 2

A single rendering on slide 28 shows what Tesla’s new 4680 battery may look like. The rendering is complicated with perspective, shadows and lighting – thankfully it shows the intricate cap geometry. Having only a single angle to work with provided a fun challenge for recreating every fillet, flat and facet in 3D – patience is key! Some features are probably shown more pronounced and severe than they will eventually be, nevertheless, included with this article is a downloadable stereolithography file (.stl), that replicates the rendering to within a millimeter and with a high degree of confidence. Other models exist so it is a good idea to compare and I encourage everyone to create their own.

https://grabcad.com/library/4680-1

Tesla’s 2170 specifications are unpublished or aloof (they are likely available inside academia or hidden behind paywall protected journal/conference papers). Having the 2170 specifications would make reporting 4680 energy storage and power delivery capabilities easy, slide 28 shows they are 5X and 6X, respectively. In lieu of published data, it is possible to do back-of-the-envelope calculations and compare them geometrically to a fairly large pouch/prismatic cell, the SPIM08HP, Figure 3.

Figure 3

This is apropos since the features of a 4680 serve to blur the chasm between cylindrical and prismatic batteries. Cylindrical cells are more geometrically stable, better optimize surface material and are easier to assemble into battery packs. The volume of a 4680 is 132952 mm^3 whereas a SPIM08HP is 128800 mm^3 with a capacity of 8Ahr meaning the former can – theoretically – hold ~3% more electrons; new chemistry will likely enhance this further. By design, electrons only liberate via the ends of batteries. The cap area of a 4680 is 1662 mm^2 whereas a SPIM08HP “cap” area is 920 mm^2 with a published current delivery of 200 amps meaning the 4680 might deliver 361 amps, whoa! Heat must be removed. The max distance heat must travel in a 4680 is 23 mm, however 40 mm is more realistic, contrasted to the 4 mm heat must travel to escape a SPIM08HP. Thermal management looks to be a gating factor.

Notes on SPIM08HP: they appear to be discontinued; not much information is available; while the energy storage number was easy to verify, the power delivery number came from a seller’s website and has not been verified; burning a battery was not in the cards!

Combining past and current innovations means Tesla’s 4680 battery cell retains advantages of cylinders (handling, packing, safety, etc.) and gains advantages of prismatic cells (thermal and electrical flux, etc.) and will help accelerate the world’s transition to sustainable transportation, millimeter by millimeter.

On battery day they touted a new cell,
everyone watching thought that looks swell.

Made in an entirely different kind of cell factory
like printing or bottling oh so fast you’ll see.

Times are a changin’ for the anode,
Silicon got no time to erode or expand and explode.

Cathode with cobalt is sometimes bad and sad, a little,
more iron and nickel, now that’s plain mad, a radicle.

A tantamount pith to thum,
everyone wants more Lithium.

While many incumbents lack will and start whining,
by recycling, refining and mining we’ll start shining.

As a blast from the past, they are looking to cast,
a structure for fuel, don’t be a fool,
take the cell, a functional member, make the sell, a structure you’ll remember!

TS100 Soldering Iron Easter Egg

by Mark McCrate

The TS100 is a pencil style soldering iron featuring a mini form factor, OLED display, maximum 65 Watt output and user customizable bits. Its thermal performance appears to be on par with established greats such as Weller however being a new player, the durability has yet to be proven. This much many people – including you – probably already know since you are reading this to learn a little more.

I have seen several people successfully using the TS100 as a daily driver but my personal experience is a paltry one pad demo so when my 20 year old donated soldering iron kicked the bucket R.I.P. it seemed logical to test a paradigm shift and go ultra-portable. There are a multitude of sites where you can purchase a small kit, mine shipped with impressive speed and configuration only took tens of minutes. The remainder of this article will cover setup, manuals, system parameters and an Easter egg; it is not a review for two reasons: first, plenty of unboxing/first impression articles and/or videos already exist and second, it will take a year of use/abuse in various setups and a plethora of environments to accurately assess.

A user/instruction manual should be included in most kits. In the box I received, was a child sized pamphlet measuring 3×5 inches or 8x13cm, written in English and Russian? Inside there are plenty of line art drawings; online manuals versions sporting full-color pictures exist as well. The translation errors are forgivable because the meaning gets conveyed, however, errors such as swapping short with long, both with one, and act with wait are inexcusably heinous and make initial configuration attempts painful. Luckily, years of smashing buttons on video games or the more ancient art called percussion calibration has trained us that random input paired with careful observation are indispensable skills. Expecting firmware post 2.18 is immune I won’t dwell here…if at first you don’t succeed try, try again, differently!

When it comes to versatility, the manual explains 8 different user configurable system parameters. Upon deciding which settings are best for me, I plugged in the micro USB and was surprised to find 9 parameters and that none of the names matched the manual text. Serendipitously T_Step and Step_Temp are virtually identical. The ninth hidden parameter is an unexpected Easter egg: HAND! It is boolean and defaults to zero.

To figure out what HAND does I used the aforementioned method of applying random inputs mixed with careful observation. And it did not take long to discover the parameter rotates the display 180 degrees turning default Fig 1 into Fig 3 and default Fig 2 into Fig 4.

The ninth – hidden – parameter is a cool Easter egg, especially for ambidextrous builders! I recommend reading the instruction manual completely wherein you’ll learn how spice up the bootup using this attached .bmp “artwork” or your own creation.

Update: WordPress security does not allow uploading even monotone .bmp files so please download the .png Fig 5, and simply change the extension.

Fig 1 default

Fig 2 default

Fig 3

Fig 4

Fig 5

Dell Thunderbolt Docking Station WD19TB or WD15, WD19 and WD19DC, too!

The purpose of this post is to shine more light on mounting any of the above docking stations since there is a gap in available information on the topic.  To be brief the bottom or underside of a dock is shown in Fig 1.

Figure 1  WD1XX bottom or underside.

As far as I can tell this is the only image of these models.  Neither the image nor any documentation I have read contains useful information on mounting geometry.  Therefore I have overlaid the dimensions in Fig 2.

Figure 2  Bottom including dimensions.

The two screw holes are tapped at M2.5 and are 5mm deep, they are on a 63mm square or ~89mm apart.  Armed with this knowledge you can use fasteners only or make your own mounting bracket.  Some people will appreciate not having to spend $30 for a factory bracket with its limited VESA-only geometry.

Enjoy!

Fastener Foible and Software Gore?

If I were a gambling person it seems reasonable to bet (without researching) archaeologist found nails predate screws by centuries or millennia.  We could postulate many reasons why.  Working together a good anthropologist and engineer could arrange a list of reasons discussing the various pressures and availability of materials.  Ultimately they would have generated an organized list.  Ironically, arranging a list of fasteners into a logical listing is astonishingly impossible for most superstore websites.

To ensure this is a systemic problem not limited to my local big box chain, I need a list of all superstores located in the United States meaning a quick search at the first stop of any project: Wikipedia!

https://en.wikipedia.org/wiki/List_of_superstores#United_States

According to this list four stores sell hardware in noteworthy quantities so all future data will be both extended and/or limited to these: Ace Hardware, The Home Depot, Lowe’s and Menards.  Since saying fastener can mean a vast array of items, I will filter to only include machine screws.  Judging by the layouts, selecting a diameter is one of the most significant criteria and I completely agree.  Selecting a desired diameter should be a simple, choose Metric or Imperial and a computer should be provide a list from smallest to largest or vice versa so a patron can quickly choose what they need.  Alas, this is where things break down because none of the four stores listed can figure out how to count up or down, and most struggle to differentiate between measurement systems.  Below are screenshots from four different websites demonstrating the problem; they have been cropped to show only the salient points.

Ace Hardware’s inability to sort.

 

Home Depot’s inability to sort.

 

Lowe’s inability to sort.

 

Menards’ inability to sort; scrolling, on the actual page, reveals the list alternates b/w Imperial and Metric.

Thankfully McMasterCarr, the absolute golden standard for online catalogs, knows how to sort a list of numbers with ease and I am hoping others will master the skill eventually.

Anyway, while discussing the funny frustration with a friend they started snickering and suggested rather than bickering I substitute with string or epoxy.  True being bound or bonded to any one method is folly, however, sometimes we must eschew glue where only a screw will do!

Full Comprehensive Complete wiring guide and review for a 48 Volt 1800 Watt Brushless DC motor (BLDC) with hall sensors and 33 Amp 1800 Watt sensed controller ● motor model my1020 BLDC ● controller model BY15WF01-A

To maintain formatting this entry is published as a PDF, click below.

Wiring Guide.pdf

 

MakerFaire NYC exhibit

Our ELECTRIC VEHICLE TEAM has been accepted to exhibit @MakerFaire September 22nd and 23rd http://makerfaire.com/new-york/.

Come see us there!