[Arm-netbook] Existential 3D Printing Moments

[Luke Kenneth Casson Leighton]

On Fri, May 19, 2017 at 12:35 AM, Neil Jansen <njansen1 at gmail.com> wrote:

> Notes / minutes from today's voice call are below.

 neil, really really appreciated here, this is amazingly
comprehensive, how the heck did you remember all that we said? :)

 btw apologise i should have said, the conversation with aleph objects
was private and not to be announced, they haven't given permission to
do that, as they are *considering* and we are *assessing* - privately
- the feasibility.

 a couple of things i forgot to mention, one is to emphasise the
"bang-per-buck" part.  i did this kind of thing back in... 2003 i
believe it was, where i was designing an ultra-grade encryption
symmetric key algorithm.  key sizes of 16384 bits, block sizes of
32768.  i used csrc.nist.gov's STS (statistical test suite) to assess
it, but it was *literally* taking days to run tests of 1000 groups of
100,000 bits so i decided to buy some computers plural.

 now, i set myself a budget of GBP 1,000 and started looking at what i
could get.  i found a motherboard for GBP 11, a processor for 25 that
was a 1ghz Athlon, 256mb RAM for 25 or something, case for GBP 15, no
hard drive, PXE-boot them all, bottom line it all came to GBP 125, so
i could buy 8 of them.

 now, when i talked to a supplier he said, "um, why such slow
machines?? why don't you get these 2ghz Athons" and i asked "ok, how
much are they?" and he said "200 GBP each processor and you have to
have a GBP 75 motherboard" so i said, "ok so i'd only be able to get 3
of those, which gives me a total of 6000 Mhz processing speed.  i can
make that *8000* Mhz by using the 1ghz processor, for the same money.
and that's when the lightbulb went on in the supplier's head.

 this kind of design assessment trick i've only ever heard being used
by people who make beowulf clusters, the word "cluster" being the key
word.

 unfortunately, most 3d printers are simply not designed in the west
around "clustering".  they just aren't designed *and marketed* as
"maximising the print output for the money".  in china that's probably
a totally different matter, so i'll look up the wanhao duplicator
later (lead appreciated, neil).

 the numbers i did *after* i wrote the update, so haven't published
them.  here's some of them i worked out, based on a budget of say $USD
2500.

 * mendel90 - i've had mine running at 200mm/sec (yes, really,
200mm/sec *print* speed and a 250mm/sec travel speed).

 cost is around $500 so 2500 / 500 = 5.  5 x 200mm/sec = 1000 mm/sec

 * lulzbot taz 6 - we'll find out the quality on this soon enough but
let's assume they run them at 50mm/sec

 cost, if sourcing parts instead of buying retail (which is $1250)
might be as low as $700, so 2500 / 700 = 3.5 call it 3.  3 x 50 =
150mm/sec.  even if you pushed it to 200mm/sec it's still only 3 x 200
- 600 mm/sec

 * cheap and cheerful taobao-sourced china clone.  would not wish to
run it faster than 40-50mm/sec.

 cost, maybe $200.  2500 / 200 = 12.  12 x 50 = still only 600 mm / sec!


so there's this strange trade-off between going too cheap so that the
print quality is adversely affected by the low quality of the
components, and going so expensive that you simply can't multiply them
up for the budget.

what i am looking at therefore is parts which will get me sustainable
speeds that the MendelFlex can reach, but without the pricetag of an
Ultimaker-2, MendelFlex or Lulzbot Taz 6.  here's a video of mutley3d
running a MendelFlex at 350mm/s printing and 400mm/s travel:

 https://www.youtube.com/watch?v=pO4qp_m5Y-E

the sound is frickin hilarious :)

now, neil, this is the kind of speed at which an arduino 2560 *cannot
cope*, and, also, where the design flaws inherent in RAMPS - using
prototyping Evaluation Boards (polulu-style drivers) - start to show
up.

david crocket (dc42) has specifically designed the duet series so that
you can consider exceeding PWM rates of 150 khz, which is what you
need if you want to sustain 500mm/sec for example (and do not want
problems to occur at speeds well below that).

also i forgot to mention that i've been using the E3Dv6 "volcano"
which, when you use a 0.5 to 0.8mm nozzle can easily do flow rates of
something mad like.... 200c^3/min - particularly when combined with
the huge 40:1 gearing of the flex3drive which can *accurately* deliver
the kinds of torque needed.  the volcano basically turns the heat
chamber round, so that the heat area is 20mm long instead of the usual
10mm.

none of this stuff i knew about, the 18 months before i began
designing the laptop's parts.  oh well.


> settings have been tweaked, it's still hit or miss.  lkcl mentioned that he
> recommended some sort of worm drive extruder that had a remote mounted
> stepper or something like that.  As long as the extruder gear is mounted
> near the hotend,

 it is.  the automotive-grade flexible driveshaft is 750mm long and
would snap (or twist) if used "direct-drive" style.  hence the reason
for the 40:1 wormdgear.  the end result: WAY better than the COMBINED
best characteristics of *both* bowden *and* direct-drive extruding.


> * On the subject of controllers (not mentioned on the call), the cheap
> Arduino 2560's and RAMPS 1.4 with Marlin are just fine for production use.
> Really, any controller is fine as long as it's a Cartesian machine.  The new
> ones with ARM Cortex and fancy drivers are nice, but not required.  I did it
> without the fancy stuff just fine.

 ... but not at 200 to 350mm/sec print speeds :)

> * On the subject of designs / frames, building from scratch isn't that
> difficult, this is the option that I took with our 3D printer farm.  I
> recommended to take a puddle jumper flight to Shenzhen, go to the markets,
> buy extrusion, screws by the bag, all that, and use Taobao for what's not at
> the market.  It can be done, and literally everything on the BOM can be
> delivered to your door in less than a few days, tops.  That's one thing
> Asia's great for.  Cheap, fast, and good enough when it comes to RepRaps.
> Our 3d printer farm costed less than $200 USD per machine, built from
> scratch.  They worked admirably.

 that's one of the reasons i'm here.  to be able to get parts quickly,
without tariffs, and at lower cost.

> * On the subject of Chinese clone 3d printers.  The Monoprice machine is a
> rebranded WanHao Duplicator i3.  Wanhao is a huge company in China, they're
> making a TON of these things, mainly intended for the Chinese and Asian
> markets.  Companies like Monoprice and others are approaching them to setup
> distribution in Western countries.  They're actually not that bad!  I'm not
> sure that they could be used for production, but they're better than you'd
> give them credit for, and they're cheap.  Very cheap.  Not as good as a
> Lulzbut Taz or an Ultimaker by any means.  But good enough that I'd buy one
> just to have around.

 i took a quick look: the vertical x-rod arrangement is one i know can
cause "head bounce" during higher-speed operation, and any "play" in
the bearings are amplified.... i explain it here:
http://reprap.org/wiki/Mechanical_Rigidity#Vertical_vs_Horizontal_Carriage_Mounting

> * On the subject of whether or not Western 3D printer OSS companies are
> dying or not.  The ones that are innovating are doing just fine.  Others,
> like RepRapPro, are dying because they're not innovating.  Adrian Bowyer,
> the founder of the RepRap project and was the owner / founder of RepRapPro,
> is really solely responsible for his company going out of business.  While
> all of the innovation was happening in the RepRap scene, he sort of turned
> his nose at it and continued to sell machines with outdated technology.

 .... oooops....

 i knew i could feel that something wasn't quite right on the reprap
forum, i just didn't know exactly what. appreciate the heads-up very
much.

> as 'trusted' in the Western markets ... What's crazy is, the most trusted
> brand in 3d printing right now isn't Makerbot or some other closed source
> company.  It's an open source Libre company.  That's insane but true.

 it's no surprise.  makerbot was secretly patenting public domain
discussions from forums and pissed *everybody* off.  the engineers
know it, and it's not that uncommon for employees to subconsciously
"self-sabotage".

 but also this stuff is *really hard*.  working in isolation just
doesn't cut it, and that's why libre collaboration is pissing all over
proprietary companies.

> just fine.  Faberdashery's filament are not magical.  There's nothing that
> separates them from other filament manufacturers, other than their Pantone
> color accuracy.

 i loove their colours :)

>  Other than that, it's pretty run-of-the-mill virgin PLA
> that's been run on a decent extrusion machine.  The first step to good
> filament is to start with virgin PLA.

 i'm really really happy to hear of (and then test) known filaments
that are of the same quality... particularly if they have the same
kinds of eye-popping colours.

> I also expressed my interest in the EOMA68 standard and how it may fit in to
> several of my ongoing projects.  Mainly an open source laptop with a
> mechanical keyboard and 20+ hour battery life, and a portable digital radio
> system intended mainly for amateur / ARRL type stuff, but could also be used
> by security researchers or anyone else wanting a powerful yet portable SDR.

  love it.  well let's get you on the list for a pre-production prototype ok?

 welcome to the list neil.  really good talking with you.

l.