Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:02
Give me a break here, man. Get
0:04
this show on the road. Hello
0:08
and welcome to the Hackaday podcast. I'm
0:10
Elliott Williams. And I'm Tom Lardie.
0:13
This is episode 275. Mud
0:16
pulse telemetry, 3D printed
0:18
gears in detail, and display
0:20
hacking in our future. This
0:22
week in the news, Raspberry Pi goes
0:25
public. If you were clicking on the
0:27
London Stock Exchange website over and over
0:29
again to see when Raspberry Pi finally
0:31
showed up, it finally happened what? I
0:34
guess it was Monday last week? We'd
0:37
heard rumors of their IPO for a
0:39
few months and it looks like it's
0:41
finally happened. 39
0:44
pounds cents, which I guess are called
0:46
pence. A share will get you a
0:49
share of Raspberry Pi holdings, PLC. Yeah,
0:51
I mean, I feel like we've been
0:53
talking about this so long. Part of
0:55
me thought it already happened, I guess,
0:58
because we've definitely done like several articles
1:00
on it. I, Jenny's been on top
1:02
of this, like you said, for months
1:04
now. It was all when I, when I
1:06
saw the headlines, I was almost like, wait, I thought
1:08
we did that already. Like, were
1:10
they not already? But now it's
1:13
official, official. Everybody seems
1:15
not super happy in the comments about
1:17
it, I gotta say. I didn't see
1:19
a lot of positive sentiment. The
1:23
comments are off the rail
1:25
negative about capitalism in general,
1:27
and I think are full
1:29
of misunderstandings and I'm
1:32
sorry, I don't mean to call you
1:34
all out, but my God. So
1:37
I dug into this a little
1:39
bit and the way that Raspberry
1:41
Pi has been run since forever
1:44
actually is the nonprofit part, the
1:46
Raspberry Pi Foundation, spun off and
1:48
wholly owned. The Raspberry
1:51
Pi Limited, which was the part
1:53
that actually did the money stuff,
1:55
made the computers, sold them and
1:58
did the marketing there. Whereas
2:00
the Pi Foundation is basically
2:03
makes educational stuff and tries to
2:05
get computers into schools and in
2:07
the hands of students. It's a
2:10
real legit non-profit charity. And this
2:12
is apparently a normal thing to
2:14
do in the UK. If
2:17
there's a business-y aspect to a
2:20
charity, it lets the charity earn
2:22
money through this one part of
2:24
the company, support itself by owning
2:26
kind of a company that then
2:28
makes money to be funneled into
2:30
the charity. And that's the way it's
2:33
been working since forever. They've spun it
2:35
off now, right? And so the Raspberry
2:37
Pi Limited is now Raspberry Pi Holdings,
2:39
PLC, and you can buy shares in
2:42
it. And the foundation still owns it.
2:44
And in fact, what happened here is
2:46
the foundation sold a bunch of its
2:49
shares. They were saying that the foundation
2:51
still owns 70% of the stock,
2:53
although I'm not 100% sure of this. But
2:56
they basically made a crap
2:58
ton of money for the charity-slash-nonprofit
3:01
part by selling the stock off.
3:03
And now they've got a big
3:05
old fat endowment with which to
3:08
keep doing their public good stuff.
3:10
So I think that's really cool,
3:12
actually. Whether this will
3:14
change the business going on with the
3:17
Raspberry Pi making, who knows, right? The
3:19
foundation's still the majority shareholder. So they
3:21
kind of have to do what the
3:23
foundation tells them to do. There
3:26
were some people in the comments saying, you know,
3:28
now they have a profit motive, they'll something, something,
3:30
something, but I don't think they have
3:32
any more or less of a profit
3:34
motive than they have before. And their
3:36
profit motive was to make more money
3:39
so they could give it to the
3:41
charity. Like, that's actually the goal of
3:43
Raspberry Pi Limited. So I don't know.
3:45
A lot of the commenters said, oh, you know,
3:47
they're moving away from the makers and they're
3:49
going on a more corporate route or whatever.
3:52
But like, it was never really about makers.
3:54
We just sort of co-opted it. You know,
3:56
it was always about being an educational computer
3:58
that was cheap enough that you could... give
4:00
away, right? Like we saw, oh
4:02
wow, cheap Linux computer with GPIO, we could do
4:04
stuff with that and kind of ran with it,
4:06
but it was never really meant for us. So
4:09
I always thought this was kind of weird. Whether
4:11
this, like they, there's this feeling in the community
4:13
that they, the raspberry pie foundation like Oza something,
4:15
you know, like they've always been pretty clear about
4:18
what they're trying to do and
4:20
they're just continuing to do that. And you
4:22
know, undoubtedly it's been a boon for them
4:24
to have, you know, hackers and makers buying
4:26
these things, right? But that was
4:29
never, it was all, it was just like a
4:31
nice side effect. I think it was, it
4:33
was never the goal. It was not made
4:35
for us. We just, you know, saw it
4:37
as an opportunity. You know,
4:39
when we were pushing even Upton
4:41
about this in interview with Floss
4:43
Weekly, he ended up saying, look,
4:45
I started this whole thing trying
4:47
to make computers cheap and ubiquitous
4:49
and get Linux into the hands
4:51
of school kids everywhere. And that's
4:54
how I started it. And that's
4:56
what I'm still doing. And anyone who doubts
4:58
that can just, you know, wait 10 years
5:00
and see how it turns out. And
5:03
now that it's happened, I think that's where we're at.
5:05
I think we're at the no
5:07
more time for guessing what's going to happen. It's just
5:09
now the wait and see time. This
5:14
is a journey into sound.
5:16
What's that sound? What's
5:19
that sound? All right.
5:21
Well, let's head on off to what's
5:23
that sound. This week is a guest,
5:25
the sound week, which means you listeners
5:28
and special guest, Tom Nardi get to
5:30
listen to the sound. Guess
5:32
what it is. The difference of course, is
5:34
that you will be eligible to enter your
5:36
guests and hope to win a Hackaday podcast
5:38
t-shirt if you get it right. Whereas Tom,
5:41
I just gets fame and fortune. Yeah. Right. I'm
5:43
starting to think, you know, what's that like face
5:45
blindness where you can look at somebody,
5:47
you know, but you don't recognize them. I think I have
5:49
that book for sound. I have sound blindness.
5:51
Like I can hear them, but I can't
5:53
identify them. It's hard. Let me
5:56
listen to this one. Oh,
6:02
come on. And,
6:07
uh, what the hell is
6:09
this? Like
6:15
I recognize it, but I just can't.
6:18
Especially that... You
6:22
can blame Christina Panos for this one, by
6:25
the way. She sent this one in. It
6:27
is certainly some game thing
6:29
of an age. Yup.
6:32
Yup, exactly. Game thing of a certain age. Correct.
6:35
Is it an arcade? It is. Oh,
6:37
man, why can I not? It's not... It's
6:39
not... Congratulations, Tom Nardi. Oh, there we go.
6:42
Oh, man. There you go. Good job.
6:45
I instantly, I remembered, you know,
6:47
it unlocked like a core memory,
6:49
but I have trouble drawing the
6:58
connections. I had to let that one percolate
7:01
long enough and it'll, it'll come up. Well,
7:04
congratulations, Tom. The rest of
7:06
you head on over to hackaday.com/podcast. Scroll
7:08
on down to what's that sound. Click
7:11
on the form, enter your handle and
7:13
your best guess. And
7:15
if you get it right, we'll put you in the
7:17
drawing to get a Hackaday podcast t-shirt next week. I
7:20
think we'll get a lot, a lot of ones
7:22
for this. Last week we
7:24
were looking at a hack from Marco
7:27
Reps, who's a little bit of a
7:29
volt nut. And it was how to
7:31
make a switch that is all copper
7:33
so that it didn't have any temperature
7:35
sensitive thermocouple like junctions in it that
7:38
would introduce like nano volts of imprecision
7:40
into his measurements. Reflecting on that, I
7:42
realized that what I really liked about
7:44
that problem was not his solution to
7:47
it, although it was totally awesome. But
7:49
what I really liked most about that
7:51
was like the problem itself was a
7:53
crazy and interesting problem. You
7:56
could just kind of spin your mind around for
7:58
a little bit. And that is the. with
8:00
this hack, but in spades.
8:03
This is bi-directional data
8:05
transfer through mud. This
8:08
is a review article, an
8:10
academic review paper by Saleh
8:12
Mochaka, Ai Ping Wu, and
8:14
Ching Ching Fu. It
8:16
is about mud pulse telemetry.
8:18
And you're all saying, what?
8:21
But here's the problem, and it's
8:23
a really interesting problem. If you're
8:25
drilling a really deep well, like
8:28
for oil or whatever, kilometers
8:30
deep into the ground, you
8:32
wanna know the temperature, the
8:35
pressure, radiation, speed
8:37
of rotation, all sorts of information
8:39
about what's going on at the
8:41
drill head way down there, but
8:44
you're stuck up at the surface.
8:46
How do you transmit this up?
8:48
They've tried all sorts of things,
8:51
including but not limited to direct
8:53
electrical and electromagnetic transmission, but the
8:55
borehole is filled with a mud
8:58
fluid, basically. And you could think
9:00
of it as being a tube
9:02
with a slightly wider than the
9:04
tube drill head at the bottom.
9:06
And so then there's all this
9:08
kind of extra space there, and
9:10
they fill it, they pump pressurized
9:12
mud through there, both to kind
9:14
of excavate out the junk that
9:16
they're drilling up at the time,
9:18
but also to keep everything cool.
9:20
And mud pulse telemetry is modulating
9:23
the pressure wave in this
9:25
mud that goes back up to send signals
9:27
from the drill head. If you haven't read
9:29
the article, take a wild guess
9:31
at what kind of baud rates
9:34
they're getting out of this, that
9:36
six kilometers underground by modulating the
9:38
amount of mud flowing
9:40
through this crazy pipe. The answer is
9:43
kind of in the two to 10,
9:46
maybe the highest baud rate systems are
9:48
kind of up around 20 baud. And
9:51
so there's not much data you
9:53
can get out of this. And
9:55
the reason is it's just an
9:57
incredibly difficult problem. The
10:00
paper goes into the kind of three ways
10:02
that they do this. The first two are
10:04
basically by opening and closing a valve, and
10:06
so you either let more mud slip through,
10:09
or you push some extra mud through
10:11
with kind of a secondary little pump.
10:13
And the third one is called the
10:15
siren, which is my favorite of them,
10:17
which is letting more and less mud
10:20
go through. So you're making kind of
10:22
a sinusoidal pressure wave in there and
10:24
then opening and closing the veins on
10:26
this. So you're actually doing amplitude
10:29
modulation on the carrier
10:31
of varying mud pressure. I just,
10:33
that's the craziest thing I've ever
10:35
heard. So the paper takes this
10:37
as kind of a signal processing
10:39
challenge and looks at what the
10:41
different sources of noise are, which
10:43
include, you know, mud isn't just
10:45
mud and the pressure wave like
10:47
disperses as it moves up the
10:49
mud. There can be more or
10:51
less air in the line, more
10:54
or less gas, more or less
10:56
rock that changes the transmission characteristics
10:58
all the time. Well, I mean,
11:00
there's the pump that's pumping this mud
11:02
through the system. It's got to have
11:04
its own kind of noise. And so
11:06
then they look at all the kind
11:09
of frequency domain ways you can filter
11:11
this noise out and try to recover
11:13
the signal from this weird, weird, weird
11:15
signal. Anyway, really, really, really neat research
11:18
paper because it just opened
11:20
my eyes to something I'd never even thought
11:22
about before. You know, when I first read
11:24
this, I immediately like, wow, this is like
11:27
a Dan Maloney thing. So I
11:29
liked when we did our weekly call and
11:31
Dan was like a little upset that
11:34
he didn't get dibs on this because
11:36
he's kind of the resident, like, here's
11:38
a weird thing that your life depends
11:40
on in a way, but
11:42
you didn't know about it, right? Because, you
11:44
know, as we saw in the comments, the people
11:47
who work in, you know, the oil field, this
11:50
is something they deal with daily that none of us
11:52
have ever come across, right? Unless
11:54
you have a borehole in your backyard that
11:56
you've been working on. This is just well
11:59
outside. of our wheelhouse. But fascinating
12:01
stuff. To me, I think the
12:04
most interesting part, and Potslough have
12:06
misunderstood this entirely, but when
12:08
they're drilling these things, they're just sections of pipe,
12:10
right? So you go as far as you can,
12:12
and then you just put another section of pipe
12:14
on and put another section of pipe on. That's
12:16
how you get farther and farther down. So there's
12:18
no wires or anything, right? So they needed to
12:20
use the medium to transmit,
12:22
but that also means there's no power.
12:24
Is it powered or running on a
12:26
battery? Maybe. Maybe, but then you're not
12:29
pulling it up to charge it. Once it's
12:32
down there, it's down there. Right. So unless
12:34
I'm wholly misunderstanding it, and
12:37
I'm sure someone will be happy to
12:39
correct me. That's funny. This whole article
12:41
is all about the data transmission and
12:43
the signals problem here, and
12:45
not at all about what the sensor's
12:47
like. I was looking for that, an
12:50
explanation of the actual sending part, and they do
12:52
kind of skip over that. But there could be
12:54
some little turbine or something down there. I bet
12:56
it's powered with a big trunk and battery. Maybe
12:59
it's a turbine. Maybe. I mean, if you're
13:01
got working fluid, it could be spinning off
13:03
of that. Well, I mean, in
13:05
any case, it's certainly, it's its own little
13:07
island, right? Whatever it's doing, it has to
13:10
get its own power and send its own
13:12
stuff. It's not physically connected. And that kind
13:14
of made me think of one
13:16
of the ideas for like
13:19
a Venus probe was because
13:21
it's so ridiculously hot and
13:24
high pressure that you would send like an analog
13:27
kind of thing that would, you know,
13:29
based on the temperature, it would like
13:31
flap a reflector and then you would
13:33
use radar to see that. You
13:36
know what I mean? So you'd be able
13:38
to get sensor data from the surface, but
13:40
there would be no electronic components. It would
13:42
be all this like mechanical analog stuff that
13:44
would, that would be able to survive. And
13:46
it just made me think of that where
13:48
like you're trying to get data from an
13:51
incredibly hot style environment with
13:53
no traditional link. And you kind
13:55
of have to think outside of
13:58
the box. But what I
14:00
love about it, about it is that
14:02
a number of people in the comments,
14:04
of course, have worked in the oil
14:06
and gas drilling industry. And your comments
14:08
range from mud butts, I
14:10
cannot express how hilarious it is to
14:12
have a single bit per second data
14:14
transfer rate with 2024 tech, but
14:17
that is the reality on some nightmare
14:19
wells. He's like, if you could make
14:21
a Pulsar that worked reliably, even at
14:24
1k per second, you'd be able to
14:26
print money. And I guess he's right,
14:28
you know? Oh, look at that,
14:30
there you go. Another interesting bit of this tech
14:32
is the tools strive to get the sensors as
14:34
close to the drill bit as possible. Some of
14:36
the designs are working to push the temperature ratings
14:38
of equipment to 200 degrees C, including
14:41
the mud pulsors. The batteries for
14:43
these devices are designed for extreme
14:45
heat and don't work at all,
14:48
even at room temperature. What? Yeah,
14:50
so totally crazy. And of
14:52
course he's right, like the
14:54
oil and gas drilling industry
14:57
is trillions and trillions of
14:59
dollars, right? And it's super important
15:01
for them to get good data about
15:03
what's going on way down at the
15:05
drill head. So this does have to
15:07
be one of those things where, you
15:09
know, if you could actually double or
15:11
triple the data rate they'd get out,
15:13
it would actually be worth tons and
15:15
tons and tons of money. And it's
15:17
pretty funny that a number of people
15:20
in the comments worked in the oil
15:22
and gas drilling industry and yeah, worked
15:24
at these holes and all came up
15:26
with this. And everybody's like confirming this
15:28
to be a super difficult problem and
15:30
be a really interesting problem domain. So
15:32
really, really cool. I have to say
15:35
19 years of reading Hackaday, how have
15:37
I never heard of mud pulse transmission?
15:39
All right, now I picked this next
15:41
one because it caused quite a bit
15:43
of debate over the weekend, but also
15:46
it's something that I've kind of looked
15:48
into personally. So I found it fascinating.
15:50
This is gas-tight FDM 3D printing is
15:52
within your grasp. And I got to
15:54
say, perhaps if Jenny hadn't
15:58
picked that particular title. Because
16:00
if you are making 3D
16:03
printed little pressure tanks, you probably shouldn't
16:05
be holding them. They should not be
16:07
within your grasp. They should be safely
16:09
separated from your personage. Because
16:12
obviously there's a high likelihood that
16:14
they're gonna delaminate and explode, right?
16:16
So this is a whole experimental
16:19
thing that the person's name is German Engineer.
16:21
At first I thought that was like a
16:24
placeholder, but apparently that is the
16:26
name, German Engineer. And they
16:29
wanted to 3D print little
16:31
propane tanks for reasons
16:33
that we won't worry about. But when the
16:35
propane is liquefied, it's between like 150 and
16:38
200 PSI, which is considerable. And
16:42
when attempting to 3D print a little tank
16:44
for it, it just, forget about it. Like
16:46
there's no way. It just
16:48
permeates through the layers. Like
16:51
I said, I've tried to do something similar
16:53
with like compressed air and it just doesn't
16:55
work. That's not what 3D
16:58
printing, desktop 3D printing is designed for, right?
17:00
But what this person found is that they
17:02
were able to seal the print,
17:04
in this case was able to
17:06
create a gas-tight little tank that
17:09
successfully held the pressure. You know,
17:11
even though Jenny very clearly said
17:14
in there that this is probably something
17:16
you shouldn't take lightly, it was
17:19
like the topic of the day on the
17:21
comments was everybody saying, oh, this is crazy.
17:23
And you're basically making a grenade and never
17:25
do this and all that kind of stuff.
17:27
I mean, I don't disagree. I
17:29
would probably, I would say I would definitely never
17:31
put 200 PSI into a 3D printed tank,
17:35
but I do think there's, you know,
17:38
the technique of sealing the print to
17:40
the point that it can hold pressure,
17:42
I think is interesting. And there's certainly,
17:44
you know, lower pressure applications that are
17:46
less exploding than I think. Like I
17:49
was saying, my thing that I wanted
17:51
to try to do compressed air for
17:53
like a horn that
17:55
I was thinking about doing something with, you
17:57
know, would not need to be.
18:00
that kind of pressure, right? So I think
18:02
that the idea has merit mainly
18:05
because you can make these bespoke
18:07
tanks that could be, you know,
18:09
fit into whatever shape is necessary
18:12
for your project, you know, if
18:14
you had something that needed a little bit
18:16
of pressurized something to be able to do
18:19
like a conformal tank is pretty interesting. But
18:21
yeah, it's definitely a, you need to be
18:23
very cautious with this kind of thing for
18:25
obvious reasons and kind of keep in mind
18:27
that you know, you're well outside of the
18:30
intended use for 3D printed
18:32
plastics. Is he doing this
18:35
for propane or just for? Yes,
18:37
but specifically for propane and I
18:39
don't 100% sure
18:41
why. So it's a really double whammy
18:43
where he has a pressurized gas that
18:45
may explode and also happens to be
18:48
hugely flammable. Because part of me wouldn't
18:50
hesitate to do this with pressure air
18:52
or pressure water if you had tested
18:57
it enough beforehand. What I mean
18:59
by that is you, you know,
19:01
do what they do when they test metal
19:03
pressure tanks. You fill it with water, you
19:05
put it up to that pressure and you
19:08
make sure it doesn't explode and then you
19:10
put it up to twice that pressure and
19:12
you make sure it doesn't explode, then you
19:14
can be pretty sure that it won't explode
19:17
at the, you know, rated pressure. Ironically, you
19:19
don't fill it with air at that pressure
19:21
because the air wants to keep expanding and
19:23
it makes it into a very dangerous little
19:26
bomb. Water is non-compressible, it doesn't store any
19:28
energy, it just pressures the tank up.
19:30
So there are totally safe and
19:32
reasonable hobbyist ways to test these
19:34
pressure tanks, but man, I don't
19:36
know anything about propane, but it
19:38
wouldn't surprise me if it were
19:41
a solvent that worked on plastic.
19:43
And so... Oh yeah, that
19:45
seems reasonable, you know. He's coating it
19:47
with something called Dichthal and I mean,
19:50
they must tell you what it's resistant
19:52
to, etc, etc, but I don't know.
19:54
And that's true, somebody points out in
19:56
the comments that, you know, the whole
19:59
phase changed. thing with the propane
20:01
makes it extremely cold too. So not
20:03
only do you have whatever dissolving properties
20:05
the propane probably has to begin with
20:07
on the plastic, now it's now it's
20:10
cold. So it's like really just daring
20:12
this plastic to blow up in your
20:14
face. And brittle. I
20:16
should say for anybody who's just listening that
20:18
like when we say tank, I
20:21
mean the things he's making are very tiny,
20:23
you know, so it's not like he's trying
20:25
to put this thing on a barbecue grill
20:27
or something. I mean these things are like
20:29
golf ball sized. I mean they're super, right?
20:31
Super small volume. Again, not saying I mean
20:33
there's still plenty of danger to be had
20:35
here, but I mean I think it is
20:37
important to just say we're not talking about,
20:39
you know, a 15 pound tank of propane
20:41
like you'd get, you know, for your grill.
20:43
I mean this is a much much smaller
20:45
thing. I would be really
20:47
sure to dot the I's across
20:49
the T's on this one before
20:51
making even tiny little bombs like
20:53
these. Remember folks, even the tiny
20:55
bombs you need to
20:57
be careful of. See what we got there.
21:00
Two, five, six, eight. I mean
21:02
I'm looking at this the sheet
21:04
for it doesn't really mention. There
21:06
we go. Your advantage is efficient
21:08
use of materials, drinking water approved,
21:10
gas and liquid type components, high
21:12
chemical resistance, including oil, improved
21:15
UFEE resistance. I
21:17
mean it may be. So they got
21:20
a phone number. It says there'll be
21:22
happy to answer any questions. We should
21:24
do a live call. Call them right
21:26
up. Yeah, you ask them in German
21:28
like hey if I wanted to make
21:30
some small propane tanks. I
21:32
mean if this stuff is
21:34
resistant to propane then maybe?
21:37
Oh god I'm so
21:39
scared. None for me.
21:42
Good luck Lewis Marks. Yeah probably
21:44
don't do this one at home.
21:46
Well from totally sketchy 3D printing
21:48
to completely normal entirely sane 3D
21:51
printing, HowToMechatronics has a video studying
21:53
the finer points of 3D printed
21:56
gears. And here he's going through
21:58
how noisy they are. are,
22:00
how efficient they are, how
22:03
easy they are to put together, how
22:05
much backlash they have and then in
22:08
the end kind of how strong you
22:10
can make a gear train out of
22:12
3D printed gears and he's looking at
22:14
the three main contenders which are kind
22:16
of flat gears, helical gears and herringbone
22:19
gears and if you've never seen the
22:21
kind of two weirder types, you know
22:23
flat gears are what you normally think
22:25
of as gears. Helical gears,
22:28
the gears themselves, the gear teeth
22:30
themselves are cut on an angle
22:32
and then herringbone of course is
22:34
cut in a V shape so
22:36
that they angle in from both
22:38
sides. The advantage of helical and
22:40
herringbone should be that they have
22:43
less backlash than flat gears because
22:45
instead of the teeth mating in
22:47
one plane, they actually mate across
22:49
the slant or the double slant
22:51
in the case of the herringbone
22:53
and that should help with the
22:55
backlash. Conversely, the more complicated gears
22:57
should also have a little more
22:59
friction, should be a little bit less
23:02
efficient and that's one of the things he
23:04
actually tests here. Not entirely
23:06
surprisingly, he does get that result. What's
23:08
interesting to me is that the flat
23:10
gears run the loudest and I guess
23:13
that's because they don't have a continuous
23:15
surface of pressure on each other whereas
23:18
the helical and herringbone gears run quieter
23:20
but then what he does is he
23:22
runs them at a certain speed and
23:24
figures out how much power he has
23:26
to put in to have it run
23:28
at that speed and that's kind
23:31
of a measure of how efficient they
23:33
are and he of course also gets
23:35
the expected result there that the flat
23:38
gears are the most efficient. What I
23:40
thought was really weird here is that
23:42
although when he does his backlash test
23:44
and he finds that as
23:47
you'd expect the flat gears are the
23:49
worst, the helical gears are the best,
23:51
I would have expected the herringbone gears
23:53
to do at least as well. Maybe
23:55
his V angle isn't steep enough or
23:57
something. When he repeats this with a
24:00
a four gear gear train, he actually
24:02
gets a lot less backlash on the
24:04
flat gears than he does on the
24:06
helicals. And I think that must be
24:09
due to some kind of precision in
24:11
manufacturer or something. I don't know. I mean, these
24:14
are all 3D printed, so it must be some
24:16
kind of difference across the gears in that way.
24:19
I really can't explain that result. The
24:21
thing that I think ends up being
24:23
a totally useful take home if you
24:25
end up having to do your own
24:27
gear train is when he's doing the
24:29
strength tests, the wider and bigger
24:31
the teeth on the gear are,
24:33
especially with 3D printed ones, the
24:36
stronger they are in general. And
24:38
so there's kind of a ratio
24:40
of the tooth width to the
24:42
diameter of the gear that's called
24:44
the modulus. And so if you
24:46
are picking gears for this, design
24:48
your gears to have fewer, bigger
24:50
teeth and they'll be stronger and
24:52
more infill if you need it,
24:54
more perimeters if you don't wanna
24:56
change the whole infill. But that's
24:58
kind of the big take home
25:00
on this for me. You will never, ever
25:02
get me to give up my herringbone gears
25:04
though. Yeah, I knew you were gonna go
25:06
for this one. There you are, I confirmed
25:09
herringbone fuxiano. And I think that
25:11
is his end conclusion, right? Is that if
25:13
you're gonna be printing them, you might as
25:15
well just print the herringbone because they are
25:17
uniquely suited for 3D printing because of the
25:20
shape. And it turns out, hey, they actually
25:22
work pretty well, which I don't think is
25:24
any great surprise. I remember
25:26
some of the old school 3D printers
25:28
had like the big printed herringbone gear
25:31
for the extruder. As long as we've
25:33
been squirting plastic out on desktops, the
25:36
herringbone gear has been a part
25:38
of it, right? And that's cause when you'd
25:40
have to do those retracts, the lower backlash
25:42
there makes them respond better. I have an
25:44
Wade's herringbone gear extruder on my old Prusa
25:47
2 printer, which has gotta be like 12
25:49
years old at this point. But no, it's
25:51
always cool to see some
25:53
real data to back it up
25:55
and some of these things we just take for granted,
25:57
right? They were granted and maybe not
25:59
know why. And so
26:01
I did appreciate this one. I also thought it
26:03
was like when it goes into stuff like the
26:06
noise that they make, it's not something I normally
26:08
kind of consider, right? It's interesting to see like
26:10
the causes of the actual gear noise and how
26:12
they mesh together and why that might be. If
26:14
you need something that needs to be quiet for
26:17
whatever reason, you know, that's the design of the
26:19
gear can impact that. And it
26:21
wasn't something that immediately comes to mind
26:23
when I think about designing gears is how
26:25
to, you know, reduce noise, you know,
26:27
the more gradual engagement, he says, like the
26:30
helicals and stuff, instead of just, I guess, basically,
26:32
instead of smashing into each other full
26:34
force, you know, when both faces hit each other
26:37
simultaneously, you have this kind of gradual thing.
26:39
He says it's not only smoother, but you
26:42
can just hear it. You can hear that it's
26:44
quieter. So if you were looking to have something
26:46
that's sitting next to you and needs to be
26:49
operating without disturbing you, you know, that's an interesting
26:51
takeaway that, you know, using a more
26:53
angled gear face can make it quieter. And
26:55
not entirely irrelevant. In the end, he wants
26:58
to do a wear test and he runs
27:00
these for an hour or two and doesn't
27:02
see any change in his first sample. And
27:04
he's like, this is too loud. I can't
27:07
run these. You know, he's got a weight
27:09
on the end of one of these arms
27:11
and he's running it at high speed, hoping
27:14
to wear the gears out and doesn't wear
27:16
them out within an hour and gives up
27:18
because it's just too noisy in his apartment,
27:20
bails out on that one. And so we
27:23
don't get the results on that one at
27:25
all. The PLA seemed
27:27
to work fine, you know, whereas
27:29
I kind of think of, I don't know,
27:31
maybe because it's stuff like PETG is
27:34
kind of gained popularity. I think a PLA is sort
27:36
of the, you know, if I'm printing like some little
27:38
trinket or something, it's PLA. I don't really think of
27:41
it as a, if I'm thinking like a practical functional
27:43
print, I'm usually thinking of a more
27:45
robust plastic. So it was interesting to
27:47
see that even with like the bare
27:49
minimum PLA that any printer, no matter
27:52
how janky it is, should be able
27:54
to square it out PLA can
27:56
still produce viable gears is a cool
27:58
takeaway too. need some abs
28:00
or something just to make a working gear,
28:03
at least not at this scale. These are
28:05
pretty chunky gears. The other takeaway I got
28:07
out of this, and this is one that
28:09
kind of undoes some of the results, I
28:12
guess, in a way, like another takeaway that
28:14
you can get out of this is that
28:17
he had some gears printed in PLA,
28:19
and then he had some gears printed
28:21
in PLA that was an older PLA,
28:23
and it was brittle and it broke
28:25
at a different strength and stuff. And
28:27
he had nylon, and he had nylon
28:29
gears fail at a really low
28:31
force, like lower than the PLA and
28:34
lower than the PTG. And like you,
28:36
I would expect the nylon gears to
28:38
hold up better than the kind of,
28:41
you know, the pedestrian PLA, right? But
28:43
I think that, you know, I don't
28:45
know if it was difficulty in printing
28:48
nylon. Nylon's a bitch to print with.
28:50
I don't know if it absorbed water
28:52
because it likes to do that too.
28:54
I mean, I don't know what the
28:57
reasons for this variation are, but it's
28:59
worth noting that he had a
29:01
bunch of variation across these different
29:03
plastic types, and from gear to
29:06
gear even, that he couldn't really
29:08
explain, and doesn't really make sense
29:10
with kind of the theory and
29:12
the fundamentals here. And so I
29:14
think that's another, I mean,
29:16
on one hand that kind of clouds his
29:19
scientific results a little bit. It puts
29:21
a little more fuzz in there, but
29:23
on the other hand it's useful to
29:25
know that if you're worried about printing
29:27
these gears yourself and all you've got
29:29
is PLA, just make the cleanest,
29:31
bestest, thickest, most infilliest print
29:33
you can, and that'll probably do
29:36
it. Given variation across plastics and
29:38
everything, if it really
29:40
matters, you're just gonna have to test
29:42
it. Well, to briefly pull this away
29:45
from the trials and tribulations of 3D
29:47
printed plastic, this one I want to
29:49
talk about is ESP32 powered crunchy makes
29:51
beats on the go. I thought this
29:54
was a really cool project, so much
29:56
that I wrote it up when I
29:58
saw it. I'm not... necessarily musically
30:00
inclined, but I thought it was
30:03
really cool. This is just described
30:05
as like a key chain
30:07
form factor music making platform
30:09
where it's, you know, it's
30:12
ESP32, it's an audio
30:15
amplifier, it's some buttons and LEDs and not
30:17
a whole lot else. It can be built
30:19
with like off the shelf modules on a
30:21
little perf board or you could do a
30:24
proper PCB. Regardless how you
30:26
get there, you have an array of, you
30:28
know, four by four tactile buttons and
30:31
there's a little, you know, printable
30:33
legend that shows what they all
30:35
do and they play different electronic
30:38
sounds and instruments and you can
30:40
layer them and mix them. And I
30:43
guess a tracker would be the
30:45
proper term for this sort of
30:48
does a lot of stuff without
30:50
requiring a lot of hardware, which I thought
30:52
was really impressive. A lot of the, a
30:54
lot of the heavy lifting is being done
30:56
in the software, which is open source and
30:59
kind of the hardware here
31:01
is my interpretation of it is that it's
31:03
a little bit free form where there's a
31:05
few different ways you can arrive at the
31:07
end goal and it's not super important how
31:09
you put it together. So long as the
31:11
key players are there, you know, so long
31:13
as you got the ESP32 and you got
31:16
the audio, the I2S audio
31:19
chip that the code is written for
31:21
and you have enough buttons and LEDs,
31:23
the rest kind of just takes care
31:25
of itself. So there's room
31:27
for interpretation in the hardware side of
31:29
things to get this code running on it.
31:32
There's a little demo video that I said to me,
31:34
it sounds awesome. Maybe I don't
31:36
have the trained ear necessary for this,
31:39
necessary to pass judgment, but I was
31:42
very impressed with the sounds it was
31:44
making for how little hardware was involved.
31:46
And I think just how cheaply and
31:48
easily it can be put together is
31:51
really interesting. I think it's
31:53
very approachable. I could see this
31:55
being something you could do it
31:57
like a hacker con or something
31:59
like Like it could be like a
32:01
little workshop that people put these things together.
32:03
You could do it even as like the
32:06
badge for an event, honestly, and still have
32:08
the per unit cost pretty low. This was
32:10
just released right before we wrote
32:12
about it. I think what I was writing
32:14
the article, like the, the GitHub wasn't even
32:16
like more than a day old. So cutting
32:18
edge stuff. And I'm interested to see where
32:20
it goes from here. I think it's a
32:22
really cool little platform to play around with.
32:24
So if you're, you want to mess around
32:26
with electronic music but you don't want to
32:28
spend big money for like a synthesizer or
32:31
like real stuff. And you just want to
32:33
say, Hey, am I into this? Am I
32:35
into pushing little buttons and making
32:37
sounds? Like this
32:39
is a cool way to get
32:41
started. This is super similar to
32:43
like the teenage engineering pocket operator
32:46
rhythm one. Like they're like drum
32:48
and bass sequencer machine. It's got
32:50
like 16 buttons. So you
32:52
can do the 16 step thing and you just
32:54
hit them when their time comes and it plays
32:57
in loops and looks like a
33:00
total lot of fun. It's like a
33:02
pocket operator, but it doesn't have a
33:04
screen on it. And ooh, maybe that's
33:06
the next thing for people to work
33:08
on here. It's funny because I've heard
33:10
a bunch of people talk about making
33:12
a device like this but never actually
33:14
seen one show up. So I saw
33:16
this and I'm like, Oh, who did
33:18
this finally? I was stoked to see
33:20
it done. This is like you say,
33:22
the GitHub is three days old and
33:25
the code commits half of them
33:27
say initial commit code is a
33:29
bit unorganized but fully functional, which
33:32
is totally my kind of commit
33:34
message. So, you know, you know
33:36
what you're getting here folks, but
33:38
like you, I really like that
33:40
the hardware is kind of what
33:42
it's not anything special. It's entirely
33:45
in flux. You could just totally
33:47
take this run with it. You
33:49
know, A just wire up
33:51
a couple of modules, B make a
33:53
super simple PCB and do it that
33:55
way. You know, there's a million different
33:57
ways to do this. And I think
33:59
that's... pretty cool because of
34:02
how accessible it's going to be. You
34:04
know, maybe I would add some more
34:06
blinking lights to it. I want a
34:08
couple more. I want an LED per
34:10
button, I think. But other than that,
34:12
I can't think of anything I'd change
34:14
on that side. So then it's all
34:16
just software refinement. And I'm hoping that
34:18
a bunch of other people will help
34:20
him in his project start sending him
34:22
commits with some really cool either sounds
34:25
or modes or whatever else. This
34:27
could turn into something really fun. And I
34:29
do like that the hardware side of the
34:31
platform is just so accessible. It makes it
34:33
easier for people to hack on the software side. Yeah,
34:35
I mean, I think there's a good chance a lot
34:38
of people could build this with what they have laying around
34:41
right now. Like as they're listening
34:43
to this, you know, I mean, maybe the audio
34:45
chip is, you know, you might
34:47
have to go out and get that. Although
34:49
I'm fairly sure I even have one or
34:51
two of those. Yeah, I think it's a
34:53
pretty common like Adafruit audio module. This
34:56
is sort of, you know, boilerplate
34:58
minimalist hardware. It could definitely stand
35:00
to be like with a screen
35:02
and more LEDs. But of course
35:04
that could be further variations. You know,
35:06
like you could you I like that you could
35:08
still have this very basic thing, but then you
35:10
could you could expand it if you wanted to.
35:13
Right. Just like you could
35:15
get it done on a proper PCB or you
35:17
could just kind of dead bug it. You know,
35:19
if those kind of improvements could be made, you
35:22
know, without compromising the low bar of entry. I
35:24
think that's cool, too. Right. Like
35:26
you could have if your board has a screen and
35:28
great, then it uses it. If not, it just, you
35:30
know, it just doesn't. You know, I mean, I think
35:32
that that would be cool to see that you could
35:35
get in in the door with with a very minimal
35:37
hardware. And then if you want to add that screen,
35:39
if you want to add more LEDs and that sort
35:41
of thing. All right. Well,
35:43
that was it. If you enjoyed us
35:46
not talking about 3D printing for 10
35:48
seconds, that time is over because here
35:50
comes Thomas and Latterer. Can
35:52
a toy 3D printer be made
35:54
great? And a few videos ago,
35:56
he was looking at a $75.
36:00
3D printer, the EZ3D. As
36:02
he says, it's a toy 3D
36:04
printer. It's a pretty crappy 3D
36:06
printer, but it has some neat
36:09
design features. One of them
36:11
in particular is that the three movement
36:13
axes are identical to each other, and
36:15
they're just kind of stuck on to
36:18
each other to give it movement in
36:20
XYZ. That's kind of this device's most
36:23
obvious design feature, but it's also kind
36:25
of its Achilles heel, because none
36:27
of them are particularly stiff, and
36:29
in particular they're based on
36:31
kind of C-shaped sections of
36:34
molded plastic. And although
36:36
they're pretty stiff in two directions, they're
36:38
not very torsionally stable, so you can
36:40
twist them really easily. And this thing
36:42
is a wobblebot because of
36:45
it. And so that's the first thing
36:47
he is trying to fix in this
36:49
video by printing kind of a really
36:51
neat insert that goes inside the C-channel
36:54
and makes it torsionally stiffer. He concludes
36:56
that the motors are crap and adds
36:58
better motors to it. Not much
37:01
of a surprise there. And as
37:03
Jenny points out in the article,
37:05
this video is sponsored by a
37:08
company that makes motor controllers, so
37:10
he throws these BQ... What
37:12
is it? Big Tree Tech Clipper
37:14
Motor Driver and Controller Suite at
37:17
it. And that's kind
37:19
of funny to do. For those of
37:21
you who aren't clipper nerds, it moves
37:23
the motion control over to a bigger
37:25
computer, over to a full-fledged Linux computer
37:28
rather than a tiny little microcontroller. And
37:30
so it can do all sorts of
37:33
real-time processing things, like compensate for the
37:35
way the frame wiggles and stuff. And
37:37
so actually if there's anything that clipper
37:39
would be best for, it's kind of
37:42
a crappy little printer like this. Because,
37:44
you know, if the frame's wobbling but
37:46
you can compensate that out in software,
37:48
that's a winner. It's interesting to see
37:50
he actually gets incredibly fast speeds out
37:53
of this thing in the end by
37:55
stiffening it up and throwing good motors
37:57
at it. On the other hand...
38:00
It's a tiny little print bed,
38:02
and there is kind of the
38:04
square cube law where the smaller
38:06
your printer is, the easier it
38:08
is to make it stiffer. And
38:10
at this size, I think you
38:12
should be able to make a
38:15
printer that's really, really, really fast.
38:17
The size-speed relationship is taken super
38:19
advantage of by the 100 printer.
38:22
I'll throw the link for that in
38:24
the show notes, which is a mostly
38:27
3D printed frame. And you'd think 3D
38:29
printed frame, it's gonna wobble like
38:31
crazy. But again, it's printing on
38:34
such a small scale that it
38:36
can do outrageous speeds and accelerations.
38:39
And this is
38:41
a stunt printer. It's designed to
38:43
print a benchy, so a
38:45
small print as fast as it possibly
38:47
can, and everything else be damned. But
38:49
it's really kind of cool. And
38:52
if you wanna see how far you could
38:54
press this kind of make it stronger just
38:56
simply by making it smaller, I think this
38:58
is the way to go. This
39:00
is really interesting to me because if you kind
39:02
of wind the clock back, there was this feeling
39:05
that the $100 3D printer was
39:07
some goal that had to be hit
39:10
for mass market adoption, right? It
39:12
seemed like at the time, unless
39:15
3D printers could be as cheap as
39:17
2D paper printers, they'll never
39:19
be in everybody's home and they'll never be
39:22
this transformative technology. And obviously that never happened
39:24
for a lot of reasons, but I think
39:26
the cost of the printer ended up not
39:28
being nearly as much of a problem as
39:31
the fact that these are just not meant
39:33
for everybody's home. I think
39:35
that in the early buzzword days where everybody
39:37
will have a 3D printer and it's gonna
39:39
hyper-local manufacturing and all this stuff, you won't
39:41
have to wait for things to be shipped
39:43
to you. You'll just print them out overnight
39:46
and all the crazy hype that was surrounding
39:48
the printers. And it's interesting to see that
39:50
now we can do it. We
39:52
can make a sub $100 printer, but
39:55
nobody wants it because we
39:57
realized that that was never actually the goal.
40:00
and goose that we thought it was.
40:02
People are happy to pay, not a
40:04
whole lot more, maybe like $250. They're
40:07
happy to pay that, $250, $300
40:09
for a machine that is not
40:11
made entirely of plastic and can
40:13
function without these kind of systemic
40:15
modifications. So it's an interesting
40:17
kind of example of something that we waited
40:20
so long for and then in the end
40:22
didn't really want or need. It's one of
40:24
those things where like the anticipation is better
40:26
than having it, right? I just
40:29
bought a $100 3D printer and it's actually the
40:31
second $100 3D printer I bought and it's because
40:33
of mass production
40:37
and kind of cheap and cheesy mass
40:39
production that I was able to do
40:42
it. It's normally like you say a
40:44
$260 printer, but
40:46
they take returns and then
40:48
they sell their returns unchecked
40:50
to unsuspecting suckers for a
40:52
hundred bucks. Al Williams told
40:54
me to do this and I did it
40:56
anyway and it turned out to be
40:59
fantastic. I picked up a printer
41:01
that was unchecked, returned
41:03
to the manufacturer for a hundred bucks,
41:05
bolted it together, it worked straight out
41:07
of the box. My son is the
41:09
happiest son with a 3D printer in
41:12
the world and after watching
41:14
this go jealously for three, four months
41:16
now, I just ordered myself one last
41:18
week. No lie, $100 3D printer coming
41:21
to my door
41:23
and it's not even a piece
41:25
of crap. It's a one-year-old Creality,
41:28
right? It's a $260 printer,
41:30
but because I'm willing to take the
41:32
risk on it requiring possibly some work
41:34
on my part, a hundred bucks, no
41:37
problem. And it's like a hundred times
41:39
better than this $75 crappy printer with
41:43
any kind of work done to it. But then
41:45
again, I also agree with you and we're
41:47
at a point now where the
41:49
price pressure on 3D printers isn't
41:51
such a big deal. You can
41:53
get a totally reasonable 3D printer
41:55
for 200-300 bucks and I think
41:57
there's just not much incentive.
42:00
to push the price lower and I
42:02
think the horrible trade-offs you have to
42:04
make to push the price lower like
42:07
is evidenced here it's just not worth
42:09
doing and you shouldn't buy one. There
42:12
was a time it was it was
42:14
hype and buzz and all that where
42:16
you know everyone was breathlessly reporting on
42:18
the 3d printing it was gonna change
42:20
the world you know it was the
42:23
Bitcoin crypto blockchain
42:25
of its of its time for a while
42:27
right that is not the case for many
42:29
reasons and it now has become a tool
42:32
like any other right so now the
42:34
3d printer I think that the the
42:36
reality has more than kicked in and
42:38
if you want one you want a
42:40
good one and you're willing to spend
42:42
a couple hundred bucks and it's like
42:44
just like a drill press or you
42:46
know a table saw or something nobody's
42:49
chasing the $99 table
42:51
saw to put them in everybody's house
42:53
because the other day not everybody needs
42:55
table saw right you know it's so
42:57
it's it's made the transition from buzzword
43:00
and hype to practical tool and
43:02
the market has kind of found
43:05
the price people are comfortable with you know
43:07
of course if you could pay unless you
43:09
would but I think there's there's
43:11
an understanding that like there's too many too
43:13
many trade-offs and and the average person who
43:15
wants to be able to print stuff at
43:17
home is comfortable with paying you know two
43:20
to three hundred bucks and getting something that
43:22
works out of the box and and we've
43:24
sort of hit equilibrium now this last one has
43:26
nothing to do with a 3d printer I hope that you'll
43:28
forgive us it is fantastically
43:30
impressive accomplishment this is Baldur's
43:33
Gate 3 comes to the
43:35
TRS model 100 and that
43:37
is not
43:39
as exaggerated as you might
43:41
think this is a demake
43:44
of at least the beginning part of
43:46
Baldur's Gate 3 I mean it is
43:48
a TRS 80 right like you have
43:51
to have to give it some some
43:53
leeway this is created by Alex Bowen
43:56
and the project itself is
43:58
is really well documented and
44:00
it's all the sources out there should you
44:03
want to fork it. Alex
44:05
actually developed an engine that is
44:07
flexible enough you can load your
44:09
own levels and stuff
44:11
into it. And then kind of forked that
44:14
into the Baldur's Gate 3 DMake, which
44:17
is really cool instead of it making like
44:19
this monolithic thing. It kind of has now
44:21
set the stage for potentially other games to
44:23
use it for the hot TRS-80 Model 100
44:25
gaming scene. For
44:30
those in the audience who don't have a
44:32
TRS-80 Model 100, probably
44:34
roll back a little
44:37
bit. This is Tandy's
44:39
1983, I think,
44:42
foray into portable computing. I mean, it's
44:44
been called like the first laptop sometimes
44:46
and I think there is some debate
44:48
a little bit about who was first.
44:50
I think the not to get
44:53
too lost in the weeds. I think Tandy licensed
44:55
the design from a Japanese company or something that
44:57
was already out, but whatever. It's
44:59
a super early portable computer, one of the
45:01
very first. Certainly one of the
45:04
very first successful ones. And part of
45:06
that was that it came out of
45:08
the box with like a basic interpreter
45:10
and tools and stuff installed on it.
45:12
So it was easy to make your
45:14
own software and round the machine. Alex
45:17
actually did attempt to make
45:19
this game with the built-in
45:21
basic. If you see in the video, there's
45:23
a little bit of talk about that, but
45:26
it was just too limited to do
45:29
something of this scale mentions like
45:31
variable names can only have two
45:33
characters or something. So you very
45:35
quickly run out, you know, if you're trying to
45:37
make a game of this scale, you very quickly
45:40
run out of variable names when they can only
45:42
be two characters long. So the end
45:44
result had to switch it over to doing it
45:46
in assembly and there's all
45:48
kinds of crazy optimizations that had to
45:50
be made to pull this off. And
45:52
it's, you know, obviously it's ASCII art, to
45:56
do real graphics on a machine in this age
45:58
would be pushing it, but it looks really... impressive
46:00
with the ASCII world off to the left. And
46:05
then you have the text menus
46:07
and stuff on to the right.
46:09
And there's also a neat trick
46:11
where using Unicode characters in the
46:13
source code lets you get a
46:15
visual preview of what the game
46:17
will look like when it's running
46:20
on the hardware, because
46:22
the levels are made up of these,
46:24
basically, strings of characters. So you can
46:26
actually look at the strings of characters
46:28
in the source code and see what
46:30
the game would look like running on
46:32
it, because it's just printing out lines
46:34
of effectively text. There's a lot of
46:37
cool optimizations to make this even remotely
46:39
possible. And one of the ones I
46:41
really liked is there was no
46:43
way to put all the text and stuff.
46:46
This is an attempt to recreate a Baldur's Gate
46:49
3, which is a modern RPG. And so there's
46:51
a lot of text and story and stuff. And
46:53
there was no way to just fit all that
46:55
as is. So Alex came
46:57
up with this routine with a Python script
47:00
that ran through all the
47:02
text in the game and
47:04
identified repeating character patterns and
47:06
then mapped those to a
47:08
code that you could then piece
47:13
together sentences and words with
47:15
the code that corresponds to the
47:17
character, if that makes sense. So
47:21
instead of storing whole lines of
47:23
text, you'd be storing these strings
47:25
of digits, basically. It would
47:27
piece them together, which is a
47:29
really cool way to save a lot of
47:31
space, but it isn't computationally
47:34
complex or anything. It was relatively easy
47:36
just to look up these strings when
47:38
the text needs to get printed out.
47:41
So a lot of cool stuff about working on
47:44
extremely low-end hardware.
47:48
The fact that it runs on the real,
47:50
you could play it right now in an
47:52
emulator. But if you have
47:54
a Model 100 kicking around that you can run
47:56
this right on the real hardware is super
47:59
impressive. And I do actually
48:01
have a model 100 somewhere that
48:03
I'm very tempted to To
48:06
break out and try this If
48:09
anyone in the audience has one I'd love to I'd
48:12
love to hear about how it goes for you
48:14
And you know how far he got playing this
48:16
game All
48:21
right my first quick hack this week
48:23
from Nissan ICE pickup to Bev
48:26
with Nissan leaf heart is
48:28
the story of EV swap
48:30
conversions taking apart a normal
48:32
pickup truck and Reassembling
48:35
it and converting it into a
48:38
battery electric vehicle If you are the
48:40
kind of person who just clicks on
48:42
the video Don't do that here by
48:44
all means click on the first link
48:47
which goes into a long Forum
48:49
post about how this was
48:52
done That's where all the neat nitty-gritty
48:54
details are and that's where kind of
48:56
all the differences between Gas
48:59
car and an electric car are
49:01
hidden and it's of an incredibly
49:04
interesting look at Everything that's missing everything
49:06
that needed to be changed everything that
49:09
needed to be added in Taking
49:11
a crashed up leaf and making
49:13
it live again in a pickup
49:15
truck body next up unkulee Business
49:18
card challenge entry who do
49:20
you love this is his business card
49:22
for a Made
49:26
up doctor love divorce attorney
49:29
It's a galvanic skin sensor
49:31
love tester game And
49:33
he says he dreamt us up with
49:35
his wife and they had a good
49:38
time making it They've played it with
49:40
other couples who have either gotten into
49:42
fights or laughed themselves silly about it
49:44
so this is a lovely entry into
49:46
the business card challenge for doctor love
49:49
divorce attorney and A reminder that
49:51
it's running for another couple weeks if you've
49:53
got a business card project out there get
49:55
it on in And
49:58
last up AI kayak controller the
50:00
paddle show the way by Braden Sunwald.
50:02
This is the story of him building
50:05
the controller for an electric kayak. And
50:07
I think the electric kayak by itself
50:09
is really cool, but the idea that
50:12
you want to make it like an
50:14
electric bike, where you paddle with the
50:16
kayak paddle and it just assists you
50:19
on each stroke, seems like a lot
50:21
of fun. And so he's
50:23
got a little module that sticks on
50:25
the kayak paddle and it's got an
50:27
accelerometer in it. A lot of the
50:29
story here is about making it waterproof
50:31
and solid of course, because it's going
50:33
in the water. But the rest here
50:35
is about collecting data from the user
50:37
so that he can later figure out
50:39
how to best add some motor assist
50:41
to this. He gets a lot of
50:43
flack because he's going to use a
50:45
neural network for it. But I don't
50:47
think this is actually AI hype. I
50:49
think this is actually a totally legitimate
50:51
application of neural networks to figure
50:54
out how much motor impulse to
50:56
give based on accelerometer readings. Well,
50:58
my first quick hack is Donkey
51:01
Kong Bongos Ditch the GameCube Go
51:03
Mobile. You may but probably may
51:05
not be aware that in the
51:07
early 2000s, Nintendo released a bongo
51:09
for a few for their GameCube,
51:11
along with a couple rhythm games
51:13
that had you playing along in
51:16
front of your television. The
51:18
bongos have faded into
51:20
obscurity, probably not surprisingly. But
51:23
this particular hacker BL3i really
51:25
wanted to bring them
51:28
back to life and went through the
51:30
trouble of coming up with a add-on
51:33
that the bongos plug into and
51:35
the original bongos aren't modified in
51:37
any way. Like they are still
51:40
intact for bongo collectors out there.
51:42
Like it even still has the
51:45
same controller cord. This is just like a
51:47
box that clips onto it and
51:49
has an Arduino to interpret the
51:51
control signals coming from the bongos
51:53
because they looked like a regular
51:55
GameCube controller to the console. It
51:57
reads the control signals, plays, wave
52:00
samples off an SD card, depending on what
52:02
part of the bongos you hit, and
52:05
has a battery and all this stuff so
52:07
you can go mobile. It's really cool. I
52:10
would say almost like
52:12
a preservation of this little
52:14
known accessory while still keeping
52:17
it intact and avoiding that. The urge to
52:19
just gut it and put a Raspberry Pi
52:21
in it. Kudos
52:23
to keeping them in their original
52:25
form while still being able to
52:28
accomplish the goal of playing them
52:30
without the console. Next
52:32
up is Baffled the Normies
52:34
with this binary thermometer. This
52:36
is a really cool, simple project
52:39
from Clovis Fritzen. It
52:41
is just a little
52:44
DigiSpark ATtiny board, temperature
52:47
sensor, and six LEDs.
52:49
The temperature value is
52:52
displayed in binary on the LEDs.
52:54
We've seen binary clocks plenty of
52:56
times before, but it's a little
52:58
bit harder to decode that
53:00
in your head than the temperature. So if
53:03
you're looking to get started on showing
53:05
information in binary around the
53:08
house, it might be good to start
53:11
with a project like this where it's
53:13
a little bit easier to digest and
53:15
work your way up. Finally, OpenSCAD cranks
53:17
out parametric CNC clamps. I'm
53:20
not alone in my love of OpenSCAD.
53:22
I know Elliott will agree with me.
53:24
It is an awesome way to do
53:26
parametric designs and make something where you
53:28
can just plug in a couple of
53:30
values and have the 3D model
53:32
change when I haven't to literally
53:35
redesign the model. It lends
53:37
itself to projects like this where
53:39
you can have a
53:42
base clamp design and just plug
53:44
in length and the width
53:46
and the height and the angle and
53:48
all that kind of stuff and get
53:50
a 3D file to print out. OSSTAT
53:53
released this script
53:55
that you can run locally on
53:57
OpenSCAD, but it's also... compatible
54:00
with Makerworld's online customizer, which is cool. I didn't
54:02
actually know they had kind of dusted that off.
54:04
That was like an old thing a verse thing
54:07
that I don't even know if it still works,
54:09
but you'll be able to, you know, create a custom
54:11
design right in the browser and download
54:14
an SCL that's fit for printing. You
54:16
know, some people would say, Hey, I
54:18
don't know that I want to hold
54:20
down my CNC work piece with 3d
54:22
printed clamps. But to that, I would
54:24
say, you know, they're 3d printed clamps.
54:26
You can just print more. Like if
54:28
one or two is a little sketchy,
54:30
I'm sure a dozen will, will be able
54:32
to get there. All
54:40
right. That brings us to our
54:42
campus articles. These are long form
54:44
pieces written by our fantastic Hackaday
54:47
writing staff. This week I picked
54:49
Dan Maloney's scrapping the local loop
54:51
by the numbers. And this is
54:53
started off as a thought experiment.
54:56
If we're switching all of our
54:58
telephone networks to fiber optic, what's
55:00
going to happen with all of
55:02
this buried copper with the prices
55:05
of copper going through the roof?
55:07
Maybe it's worth digging that up.
55:09
Or is it just too expensive to
55:11
do so? Should we leave it in
55:13
the ground? And so Dan's trying to
55:16
ballpark that here and comes up with
55:18
about 800,000 metric tons
55:20
of copper in the telephone, whatever
55:22
it is, the public switch telephone
55:25
network. And again, that's just amazing.
55:27
If you think about that in
55:29
terms of, you know, it's stretched
55:31
pretty thin in all those wires,
55:33
but then again, think about a
55:35
country the size of the U.S.
55:38
There's a wire that goes to everybody's
55:40
house. It's insane to think of the size
55:42
of the system. So you, you know, so
55:45
you add all that copper up and it
55:47
ends up being at today's rates about $7
55:49
billion worth of copper. And
55:52
now the question is, is that a lot? Is
55:55
that a little? It's spread pretty thin if you
55:57
think about it. And, uh, you know, that's a
55:59
lot. of digging to get it up.
56:01
So Dan Ballparks, what else you could do
56:04
with this copper wire, right? Say you took
56:06
all this wire and you melted it down,
56:08
reformed it into thicker wires, thinner wires, whatever.
56:10
What could you do with it? And so
56:12
he starts with, you know, an electric vehicle
56:15
has about 60 kilograms of
56:17
copper in its windings. If you
56:19
pulled all of the telephone network
56:21
up, you'd be able to build
56:23
13 million cars. That's kind
56:25
of cool. I mean, just the copper
56:27
for the windings on 13 million cars,
56:29
but 13 million cars is pretty good,
56:32
except in 2021, people globally bought
56:36
80 million cars, you know. So
56:38
if you did this one time, dig up
56:41
all that copper, it wouldn't even really cover
56:43
the number of cars produced in one year,
56:45
much less in forever. So he tries to
56:47
think about it in a different way, which
56:49
is to wind it up
56:52
into generators and use it in wind
56:54
farms. And there he kind of gets
56:56
a little bit of a bigger result.
56:58
So if you wound it up
57:01
into power generators, you could probably build 267,000
57:03
megawatts of online wind power, which is pretty
57:10
cool. And so when you compare that
57:12
number to how much it would take
57:14
to like completely decarbonize the world's economy
57:16
using wind power alone, it ends up
57:18
being about 0.05 percent of the amount
57:23
you'd need. So again, drop in the
57:25
bucket. And so that's what's weird here
57:27
is that somehow $7 billion
57:31
worth of copper ends up on a
57:33
global scale or on a national scale,
57:35
even just not being that much. And
57:37
it's one of these weird kind
57:40
of paradoxes. The scrap price
57:42
of the metal might be $7
57:44
billion, but if it costs $10 billion
57:46
to get it out of the ground, you've not
57:49
exactly made yourself a profit. And
57:52
I imagine you could do it for less
57:54
than $10 billion, but still, it would be a ridiculous
57:57
effort to pull it all out. Like
58:00
I said earlier, Dan always kind
58:02
of brings the receipts for these
58:05
technologies that are going on
58:07
around you and the copper
58:09
network that is strung
58:11
above and below us, is a
58:13
great example because
58:16
he describes it as one of the contenders
58:19
for the largest machine ever built, but how
58:21
much of us really think about the copper
58:23
lines that we probably don't use anymore because
58:25
we now have fiber optic internet and cell
58:28
phones and it's just a
58:30
thing that's sitting up there,
58:33
but still has that value because of
58:35
the raw material. So it is kind
58:37
of a unique, usually when a
58:39
technology kind of goes to the wayside, it's
58:41
just like, that's it because it's been
58:43
replaced by something else, but this one has a
58:46
material value that kind of makes it unique. And
58:48
I appreciate that he did do it by the
58:50
numbers, like the article says, and
58:52
kind of breaks down how many things you
58:54
could do and what the reality of it
58:56
is, we need a lot of copper. Like
58:59
we need a lot, a lot of copper
59:02
and even if you pull, even
59:04
if you can snap your fingers and get everything out
59:06
of the ground and off the poles and you had
59:08
it in this big pile for
59:10
you, that's nothing compared to what it
59:13
would really take to
59:15
get us where we need to be in
59:17
terms of electrification of everything. It's
59:19
kind of a sobering thought that, you know,
59:21
all that, like I said, all that effort
59:24
and all the technology and the labor,
59:26
the sweat equity to put all that
59:28
stuff out there is still nowhere
59:30
near what we would need to get where
59:32
we're trying to go. The
59:34
amount of copper we're mining annually
59:37
now makes this small amount that's
59:39
actually locked up in the telephone
59:41
network not look like all that
59:44
much. And you know,
59:46
Dan's baseline here is how much
59:48
copper do we need to completely
59:50
electrify the economy and that is
59:52
daunting and terrifying. But
59:54
then if you look at this as a
59:56
fraction of, you know, kind of annual copper
59:59
mining in copper production, it ends up not
1:00:01
being all that much. And so maybe it
1:00:03
doesn't make sense to pull it up. I
1:00:05
don't know. Dan's conclusion is,
1:00:07
you know, it's money sitting there in
1:00:10
the ground and it probably will eventually
1:00:12
be worth it to dig it all
1:00:14
up. And he sent
1:00:16
a link out that where it points
1:00:18
out that the mean time to starting
1:00:20
up a copper mine is about 18
1:00:23
years. And so
1:00:25
if you imagine we had a
1:00:27
demand spike five years ago, 10
1:00:29
years ago, somewhere in the world,
1:00:31
there are copper mines ramping up
1:00:34
to increase production to, you know,
1:00:36
offset future demands of electric motors
1:00:38
and whatever. But they're not
1:00:40
online yet and aren't going to be online for
1:00:42
another 10, 15 years? Question
1:00:45
mark. All right. Well,
1:00:47
my pick displays, we love hacking DSI
1:00:49
from Aria. This
1:00:52
is, well, it's an
1:00:55
article about hacking DSI displays. But
1:00:57
if you haven't poked around inside
1:01:00
of, you know, a modern smartphone
1:01:03
or tablet, you might not realize that they
1:01:06
are almost all using this
1:01:08
technology. There is a little bit
1:01:10
of like an embedded display
1:01:12
port, but DSI is
1:01:14
kind of like the standard for
1:01:16
little high res screens
1:01:19
and consumer stuff and
1:01:21
learning to master that, you know,
1:01:23
in the maker and the hacker
1:01:25
community opens up an incredible world
1:01:27
of displays on the
1:01:30
second hand or, you know, replacement
1:01:32
market that are dirt cheap thanks to
1:01:34
the economies of scale. I mean, even
1:01:36
look at like the official Raspberry Pi
1:01:38
display, they went like 70 bucks for
1:01:40
this thing and it's I don't even
1:01:42
think it's 720. It's like 800 by
1:01:44
400, something ridiculous. And
1:01:49
that's just kind of the price
1:01:51
gouging that we suffer through.
1:01:53
But you know, Aria makes the point that
1:01:55
display used on the Nexus 7 tablet, which
1:01:57
is fairly old at this point. it's
1:02:00
still a 1920 by 1200 display. You
1:02:03
can get them for like 20 bucks. The trick
1:02:05
is knowing how to talk to them.
1:02:07
And that's where these kinds of efforts
1:02:09
come in that she describes in the
1:02:11
article, breaking down
1:02:13
how the communication works and these
1:02:16
efforts from different hackers
1:02:19
and hardware types to kind of crack
1:02:21
the code on a lot of these
1:02:23
things. You know, the protocol itself is
1:02:26
one thing, how the device
1:02:28
talks to the screen, but you also
1:02:30
need to know like the secret handshake,
1:02:33
for lack of a better
1:02:35
word, like the initialization routines
1:02:37
that are display dependent and
1:02:39
sniffing that does require a little bit
1:02:42
more than your $10 logic analyzer, right? Like
1:02:46
there's a little bit more that
1:02:48
goes into it, but it is
1:02:50
still doable on the hobbyist level
1:02:52
and we're saying it happened. So,
1:02:54
you know, being able to unlock
1:02:56
these displays and use them in
1:02:58
our own devices, whether it's through
1:03:00
like a little adapter board or
1:03:02
you're actually implement, you know, DSI
1:03:04
right on your own device is
1:03:06
very, very compelling. And it's only
1:03:08
gonna get better as more of
1:03:11
these screens are kind of reverse
1:03:13
engineered. And listen,
1:03:15
there's no shortage of tiny
1:03:17
high resolution screens coming out for
1:03:20
the foreseeable future, right? Everything from
1:03:23
your watch to your thermostat, the
1:03:26
billions of phones that get cranked out will
1:03:28
have this kind of technology in it. So,
1:03:30
I mean, it's really a limitless
1:03:32
playground for us to
1:03:35
pull these displays out and
1:03:37
use them for our own purposes. So half
1:03:40
the article is that and what's gone into
1:03:42
that and a little bit about, it's not
1:03:44
super deep dive into the protocol, but it
1:03:46
gives you the Reader's Digest
1:03:48
version, for lack of a
1:03:50
better term. And then Aria goes into,
1:03:52
you know, some interesting examples that
1:03:55
we've seen whether it
1:03:57
was written up through Hackaday or just, you know, in the community
1:03:59
of people. taking these DSi displays
1:04:01
and doing cool stuff with them
1:04:04
and reverse engineering them. And, you know,
1:04:06
we've got stuff like Mike Harrison going
1:04:09
back to 2013, trying
1:04:12
to hack the iPod nano screens. And that
1:04:14
kind of gives you an idea too of
1:04:16
how long these things have been out there. Like
1:04:19
this isn't some recent innovation. I mean, devices have
1:04:21
been shipping with these things for these displays
1:04:23
for over a decade now. So give you
1:04:25
an idea how many are out there waiting
1:04:27
to be utilized. I'll tell you just
1:04:29
a really cool article about this
1:04:31
technology and how it's becoming
1:04:33
more accessible to hobbyists, which
1:04:36
is always, you know,
1:04:38
always a win in my book. Yeah,
1:04:41
I think you got it right. This
1:04:43
is the dominant display technology and it's
1:04:46
one that we should get working on.
1:04:48
It's unfortunately, as Aria points out and
1:04:50
you mentioned, it's MIPI,
1:04:52
it's the M-I-P-I Alliance,
1:04:55
D-S-I, whatever it is,
1:04:57
display serial interface or something. And MIPI
1:05:00
is super closed. You can get specs
1:05:02
for this probably if you have a
1:05:04
million dollars and are willing to sign
1:05:06
an NDA. But if you're me, you
1:05:09
have to dig around into the shadier
1:05:11
corners of the internet to find the
1:05:13
relevant PDFs, which shouldn't
1:05:16
stop you. The spec isn't
1:05:18
exactly open and confounding that,
1:05:20
of course, all the different
1:05:22
manufacturers have different negotiation, initialization,
1:05:25
handshake sequences. And so what you get
1:05:27
with people like Mike Harrison and Wen
1:05:29
Ting Zhang, who did some really cool
1:05:32
work on these, is basically you have
1:05:34
to take the display you're interested in,
1:05:36
look at it in the device it's
1:05:39
installed in, and then sniff the lines
1:05:41
and see what you can get. That's
1:05:43
not trivial because at least when the
1:05:46
video data starts flowing, it's gigabytes
1:05:48
per second on four different lanes
1:05:50
or something. And so probably need
1:05:52
a scope. It probably needs to
1:05:54
be a fast scope. And you
1:05:56
probably need to get clever with
1:05:58
your triggering sequence. to begin to
1:06:01
even get a feel for what's going
1:06:03
on here. And I think that's where
1:06:05
this becomes challenging. And the other
1:06:07
method Aria suggests, and I really like
1:06:09
this one, is you find
1:06:12
a screen for which there's already Linux
1:06:14
support, you've already got the device driver
1:06:16
written, and it's open source. And so
1:06:18
if you need to find out what
1:06:20
the funny initialization sequences are, it's probably
1:06:22
in the Linux driver source code. So
1:06:25
just go look it up and you
1:06:27
can find it out. And that's a
1:06:29
lot easier than having to do the
1:06:31
reverse engineering yourself. Of course,
1:06:33
this means that for every cell phone
1:06:35
out there, somebody's gonna have to do
1:06:38
it at least once. But short of
1:06:40
open documentation coming from every cell phone
1:06:42
manufacturer, I think that's the world we're living
1:06:44
in. I do like her
1:06:46
kind of blueprint here for how to
1:06:48
get started on this. Even though it's
1:06:50
kind of hard mode right now, I
1:06:53
think some people in our audience and in
1:06:55
our community are hard mode and are gonna
1:06:57
tackle this. And the
1:07:00
benefits of one person tackling a particular
1:07:02
screen if it's shared, spread out to
1:07:04
everybody else. So I don't think this
1:07:06
is an impossible problem. And like
1:07:09
you said, and like Arya said, I
1:07:11
think the benefits of kind of cracking
1:07:13
the DSI nut are so huge that
1:07:15
it's something that would be worth putting
1:07:18
the time into. You know, it's gonna
1:07:20
take a lot of leg
1:07:22
work realistically. I mean, short
1:07:24
of somebody leaking these documents out,
1:07:27
hint, hint, if
1:07:30
you're in a position to, you know, maybe
1:07:32
release some of this info for
1:07:35
the betterment of the hardware hacking community, please
1:07:37
feel free to do so. But short
1:07:39
of that, it's just gonna take a lot
1:07:42
of research and documentation and I don't think
1:07:44
she mentions it, but I'm sure there's gotta
1:07:46
be some wiki out there that's
1:07:48
building up a lot of this info too
1:07:50
so that people can contribute to it. Or
1:07:54
maybe just the best thing is to
1:07:56
get it working in mainline Linux, right?
1:07:58
Because that's about as visible. as
1:08:00
an open source compendium of drivers as
1:08:02
the world has ever seen. So if
1:08:04
you get it working in there, you
1:08:06
know, that's the example other people can
1:08:08
work from. Right. That sounds good to
1:08:10
me. I like it. Store
1:08:13
the information in the code. All
1:08:17
right. That's the end of this week's
1:08:19
Hackaday podcast. Thanks very much for listening.
1:08:21
If you'd like to follow the links
1:08:23
or leave us comments, head on over
1:08:26
to hackaday.com/podcast. And if you see anything
1:08:28
cool or do anything cool, write us
1:08:30
tips at hackaday.com. And until next week,
1:08:33
keep on hacking. Oh, yeah,
1:08:35
that's what it feels like. Oh,
1:08:40
that is the sound of Tom Nardi with
1:08:42
a head cold. I'm not sure it's where
1:08:44
you're going with it. I'm not either. I'd
1:08:47
like to help, but I signal
1:08:49
processing. I thought you were looking for something
1:08:52
a little more specific than that. Still
1:08:54
would love to find that. I can't. I
1:08:56
can't have made it up, but
1:08:58
I guess I could. Don't get me
1:09:00
started.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More