Welcome to The Bootloader. I'm Tod Kurt.
And I'm Paul Cutler. We have a special episode for you today as we welcome our first guest,
Liz Clark of Adafruit Industries. Liz, welcome to the show.
Hey, thanks so much for having me on.
I'm glad you could be here. We've each brought two things to share. We'll chat about each
one for a couple of minutes, but no more than five. Liz, why don't you start us off?
All right. Yeah. So as Paul said, I work with Adafruit Industries. And so as a result, I
have to confess my first choice is a little biased. It's the Adafruit It's a Snap iOS
app. And I'm mildly obsessed with it. I've done a couple of Learn Guides with it recently.
And this app is done by our iOS developer, Trevor. And it basically acts as a bridge
between your iOS devices and your Adafruit I/O feeds. So if there's ever been something
on your phone that you've wanted to log to Adafruit I/O or have Adafruit I/O control
something on your phone, now you can do that. And the way that you build it up is using
Apple shortcuts. And if folks don't know, in Apple shortcuts, you can kind of make these
like mini scripts using kind of these little widgets. So you can kind of you can even use
scripting things like if statements, you can do Base64 encoding, all that kind of stuff.
So there's a lot of functionality. Recently, I've done two guides using it. The first was
I used a shortcut to send my iOS photos up to Adafruit I/O using Base64 encoding. And
then I'm using a Qualia S3, which has an ESP32 S3 on it, with a 720 by 720 round display
to take the data down from the Adafruit I/O feed, decode it, and then display it on the
display. I just think that's really awesome because otherwise you'd usually have to like
create all the bitmaps or JPEGs ahead of time so it'd be able to be displayed. But this
you can kind of do it on the fly just by like running the shortcut, sending it to the I/O
feed if it's a snap. So I'm really into that. And then the other one that I did is using
Apple HomeKit using off-the-shelf smart home devices. I used a temperature and humidity
sensor and a light sensor. And I was able to get that data into Adafruit I/O and display
it on a dashboard. And then the final thing is I got a light strip that's a matter-controlled
light strip. And I was able to set up kind of a feedback loop where if I change the value
on a feed in Adafruit I/O, it would send a text to my phone. And then I was able to use
a text automation with shortcuts so that it would then control the light strip. So basically
I'm able to change the value on the dashboard and control the light scene, which I think
is really cool. And then I was able to write some CircuitPython code to use a feather TFT
with some buttons. So I'm controlling the light strip from the feather using this kind
of bridge in between.
Oh, very cool.
Yeah. So I'm really excited about it to snap. I think Trevor's done a really awesome job
with it. He also has his own guides. He's done a health status board where he's able
to log all his workouts for the week. And he's also done a weather project. You can
send the weather from your phone for your location up to Adafruit I/O as well. And with
weather projects, you often have to have an API key and all that. But this, you avoid
all that. You're just sending the data straight to it. And then with CircuitPython, you can
just read the feed directly and display it.
I liked his project using the health kit with the health statistics. It showed how many
steps he had taken for the day. And I could totally see putting something like that on
my desk to guilt me into getting the 10,000 steps a day that I need.
Absolutely. Yeah.
Totally.
Yeah.
The whole shortcuts thing, I really want to get into it because Apple has forever had
this on the desktop, this AppleScript language, which is this weird English-like language
that you can use to have programs talk to each other, which is fascinating. And so I
think shortcuts is built on that.
I think so. Yeah.
And so by Trevor making an app that's just another shortcuts participant, everything
else that speaks shortcuts, which is every other app on the phone or on the iPad, can
now participate. And now you can do what you did, like send photos up trivially using MQTT
or whatever.
Yeah.
It's just so crazy.
Yeah. And I'd never used shortcuts before or anything like that. And once you get into
it, it kind of blows your mind because it's all the stuff you can automate. And every
app does seem to have its own functions that are available too.
Yeah. I think Apple almost mandates that you have certain shortcut capability. You can
open a file, save a file or something. When you submit an app to the App Store, it's like,
"Hey, you need to implement this to become part of the ecosystem."
Okay. Gotcha.
Well, that was a good first one. Tod, what do you have for your first one this episode?
Even though I was a big machine learning person back in the day, I've been very down on most
of the modern LLM "AI" in quotes, if you can see my fingers making air quotes, AI stuff.
But there are certain cases where I'm a big fan of it. And Raspberry Pi just introduced
a new camera module for their Raspberry Pi little computers. It's called the AI Camera.
And it plugs into the normal CSI port of Raspberry Pi, just like any other camera you'd get for
the Raspberry Pi. But it's got a little bit of extra special sauce. And some of that extra
special sauce is that it's got an embedded RP2040 microcontroller with 16 megabytes of
flash. And what that's doing is it's snooping on the camera data and you can use it to have
a pre-trained AI model loaded up into the camera module. And so you just get down the
video feed and then the machine learning results. It's like synced to the camera over I squared
C over the single cable. And so this like offloads all the hard work of doing machine
learning stuff on the Pi and puts it in this little camera module, which is pretty cool.
And philosophically, I love these where the models are running on device at the edge,
rather than where they have to make a network request out to the world that is charging
you some dollars per request. And it might go down. And so it's pretty interesting that
they're able to get the functionality and stick it in the camera module. And so you
just talk to it like a normal camera. And then if you want to talk to it like a normal
camera plus some other stuff, you can load up some machine learning model into it. Like
the two demos they give are a pose estimation. So it can just look at the video and kind
of see how a human is positioned in 3-space, like where their arm is, where their head
is. And one of the other examples is sort of the hello world of vision machine learning,
which is the object detection. So you like hold up a coffee cup to it. It'll draw a rectangle
around it and say cup. And so these are just the demos they give out. They've got a whole
repo with a bunch of other examples. And the show notes, there'll be links to all this.
There's a pretty good Hackster article that shows them trying to use it and getting started
with it. But there's also a GitHub repo, a really nice getting started guide on the Raspberry
Pi documentation website. And Adafruit has these in stock or they had them in stock.
I think they're out of stock currently. I was surprised to see that they're $70 more
than the cost of the Pi itself. But when you actually look at all the functionality, it
kind of starts to make sense. Yeah. I mean, I think one of the reasons why it's so much
more expensive, because you're thinking, oh, if it's just essentially a Raspberry Pi Pico
and a webcam module smashed together, why is that like double the price? Well, it's
like they're actually using a really good camera module. They're using this Sony FX100
module. Because with this machine learning stuff, it's really garbage in, garbage out.
So having a really good image sensor to get the best quality image you can before you
feed that through your model is really important. So they have a really good sensor on the front
end, better than the most sensors. So I was thinking of getting this module mostly just
because it's a good camera. And then playing with AI stuff, because I don't really have
a use for doing this sort of stuff. It might be fun to maybe turn on lights when I walk
in the room or something. I don't know. But yeah, so that's out now. And it's cool because
I know when I've tried to do machine learning stuff on the Pi, just running models on the
Pi is kind of a pain. Yes. So I saw that all go by when it came out, and I didn't look
too into it because I assumed it was just a camera. I didn't realize it also had the
RP2040 on there. So that's really cool. And 16 megabytes of flash.
Yeah. And if you're looking for a similar idea to have it be much cheaper, is for the
last two years now, there's been this little $10 board called the Person Sensor by Useful
Sensors. You can get them at SparkFun. They have them in stock right now. Me and Paul
talked about this back in episode three of the bootloader. And this is around Halloween
time as well. So I made at that time a little CircuitPython eyeball that would track you
around the room as it recognized your face. And so it's much simpler. It's running on
a very cheap STM32 with a really cheap little webcam sensor. But it's $10, and it can recognize
a face to tell you if a face is looking at it, which is really useful for a lot of cases.
Like if you've got some sort of UI that you want to save power and have some of it powered
down when no one's paying attention to it. But if you walk up to it, it'll light up.
It'll turn on. It'll say, "Hey, I'm here. Press this button to go." So I think having
people detection without having to know who the person is, is really useful. And both
of these cameras can do that. And if you just need that, spend $10 instead of $70.
So yeah, so that's my first one. How about you, Paul? What's your first one for this
week? My first one is the Bumpin' Sticker by Guy
DuPont. I love all of Guy's projects, probably because so many of them have a musical take
to them. I had him on the CircuitPython show a couple of years ago, and he also did a maker
chat with Liz this past August on CircuitPython Day. And I'll link to those in the show notes
as well. If you haven't come across Guy's projects before, you should definitely check
them out. The Bumpin' Sticker project is a take on the bumper stickers you might have
seen on some cars that say, "Keep honking. I'm listening to insert your favorite band
here." Except with the Bumpin' Sticker, Guy is sharing what he's listening to in real
time. He picked up an HDMI screen from AliExpress for about $60 and hooked it up with a Raspberry
Pi. He used his Last.fm account, which tracks everything you listen to, which is hooked
up to his Spotify account. This lets you use almost any streaming service with the project
as long as you have Last.fm, so it doesn't matter if you're using YouTube Music, Apple
Music, Spotify, et cetera. He then wrote a couple of TypeScript programs using Valtown
that scrapes his Last.fm page. And then lastly, he added a Particle IoT board with cellular
service. That board gets the data of what he's listening to, I think via JSON from
the Valtown TypeScript program, passes it to the Raspberry Pi, which then generates
the image and displays it on the HDMI screen as a bumper sticker on the back of his car.
He also shows how he wired it for power from the car's wiring and how he mounted all of
it and even weatherized it. All in all, it's a really cool project.
I loved that project. Just how all the pieces came together. I also think the video is an
excellent explainer on how it all worked. The video is so good.
Yeah, so good. It's excellent. Yeah, he does such a great job. And I think
in the time that he's released that in the last couple of weeks, he's already released
two new projects just this week that I'm trying to catch up on.
He even has a new Mastodon bot account that I won't spoil, but I saw him post up today.
Yes, that was the one that I saw this morning. So we'll link to that in the show notes as
well. Excellent, I haven't heard of this. There's
two aspects of his project that I really enjoyed as a nerd. One is the HDMI display he used
wasn't like your standard 4x3 or 16x9 display. It's like a 4 to 1 format. It's like this
really long rectangle. I think it's meant to be like a smart rear view mirror or something,
but it's like it's the size of a bumper shape of a bumper sticker rather than like, obviously
a screen. And so to have it change when it's on the back of his car and it changes, you're
like, wait, that wasn't a bumper sticker. That's something else.
Yeah. I can't imagine driving around Boston and
actually encountering that and trying to figure out how that was actually done.
And the other is it. Oh, yeah, sorry. Go ahead, Liz.
Speaking as someone who's around Boston, that probably wouldn't even be the weirdest thing
you see on the road that day. And the other thing is he's using this thing
called Valtown, which is a like their little slogan is, if GitHub Gists could run and AWS
Lambda was fun. It's like there's all these services out there that will run code for
you and charge you and all that kind of stuff. But Valtown is free for like trying it out
and doing some like open source stuff. And it is just like you have a function and you
say, go run this function when this happens, like when this URL comes in. And that's pretty
cool. I've been wanting to have a project that uses this because it looks like actually
kind of interesting and fun to play with. Time to dust off your JavaScript and TypeScript
skills. Yeah, totally.
All right, Liz, what's your second one for us?
So my second one's a little bit of a left turn from my previous one, but it's the nothing
to see/hear wireless fuzz pedal. So I follow quite a few guitar pedal folks on Instagram
and newest post from different noises last week is a wireless fuzz, quote unquote, that
completes the circuit by transmitting it via an infrared LED. So the circuit converts the
signal from guitar level to headphone audio level, and then again into light that's sent
out through an infrared LED into a transmitter. And what I love about different noises is
he includes a schematic for all his custom pedals. So if you scroll through his feed,
it's just schematics of pedals you could breadboard together real quick. And he includes some
video explaining the circuit and also proving that when he blocks the IR light, there is
no sound. And when he lifts it back up, you have it. And he said he got it to think about
50 feet between the LED and the transmitter. So it's still like a kind of a standard fuzz
circuit. It didn't seem to affect the sound of it. Of course, fuzz is always a little
low fidelity to begin with. But I just think it's so cool. I love when people like figure
out weird ways to transmit signals, you know, whether it's like an audio jack for a soldering
iron or, you know, anything like that. So I thought that was really cool. And in general,
his work's always really interesting.
I was looking at his Instagram page and I love, like, there's this whole community of
pedal builders and DIY electronic builders that work on these strip boards. And they
have these like way of describing how the circuit is both like how you would build it
by just saying like, this is where the parts would go on the strip board. And it's just
really beautiful because it's like the strip board is like a horizontal rows of copper
are connected and that's how the holes are. So it's sort of like breadboards, but, you
know, it's a whole 20 pin row is connected. And so you can like make these pretty complicated
circuits without really much wiring. You just have to be really clever on where your components
straddle the rows and stuff. And so it's really fun to look at them just without even knowing
what they're doing. It's like little cityscapes.
And I love the DIY guitar pedal community, similar to the maker community. They're always
sharing. They're really welcoming. And it's just folks having fun experimenting.
Very cool. Tod, what's your second one for us?
All right. Well, so I'm kind of sniping one of your topics a little bit here, Paul, because
you're the one who's got the Bambu printer. So my 3D printer, it's a Prusa Mark 2.5 S
that I've had for, I don't know, six or so years. And it works great. But, you know,
in the intervening years, 3D printers have gotten much better, like surprisingly to me,
but like the Mark 4 S Prusa and the BambuLab, they both literally print twice as fast as
my printer. Like I loaded up profiles for both and like sliced it and see how long,
like what the slicer said it was going to take. I'm like, oh, really? Come on. So I've
been, I've been looking around. I've been trying to figure out, should I go Prusa again?
Because open source, you know, but I really like having the enclosed frame. And so I'd
love to, so like the BambuLab X1 is really appealing to me because of that. And so I
stumbled upon that there is an open source firmware for the Bambu, which kind of blew
my mind. And it's sort of like, it's a, it's a thing that, that modifies the application
processor board of the X1. So the X1 apparently works kind of like a, like my Prusa with Octoprint,
where there's the Octoprint Raspberry Pi that kind of controls the user interface. And then
there's the motion control system that is controlling the actual motors and the extrusion
of plastic and stuff. And the Bambu is set up very similarly. And this open source firmware
called X1 Plus only runs on the application processor and only modifies it slightly so
that it will run an alternate firmware off of the micro SD card. So if you want to run,
if you want to go back to running stock Bambu, you still can. But if you want to run this
new open source version, you can try it out. And this new open source firmware is just
basically a different user interface. It doesn't affect the motion control aspect of the printer.
So it's still going to be fast and good and all that kind of stuff. And so, so I've been
like, huh, maybe this is the path I take for a new printer is, is I kind of do both closed
source and open source at the same time. And the reason why I discovered this is because
the people behind the X1 Plus firmware also have come out with a crowd supply campaign
for an expansion board that will let you add like more cameras or lights or other sort
of sensors and actuators that you can trigger via your G code or whatever. And so it's not,
it's not required to run the firmware, but it's like a, if you have the firmware, you
can use this cool extra little plugin board and do a bunch of other, other interesting
stuff. So that's why, that's what I'm kind of thinking right now. It's like, should I,
should I get a new printer? And should it be one of these? Should it be a Bambu with
this new crazy open source firmware, which is very intriguing. I don't know if it'll
run on your printer. Cause I, Paul, cause I think yours is a P, a P1, is that right?
I have a P1P and yeah, I did look at it. That was the first thing I looked for and nope,
this is just for the X1 Carbon. So the higher end one, I was too cheap to get the open source
version I guess. But I'm, I'm interested to hear that, that you're thinking about joining
the dark side and going with a closed source printer, even though you could argue that
the new Prusas aren't very open source to begin with either. And this is almost a luster
of two evils. Yeah. That's the thing is I've, I've been doing open source printers since
like the very first MakerBot. And so yeah, it's, it's the fact that Prusa has been going
more closed sources has even made me think of going to other printers that are not, not
open because, because of that aspect. And so yeah, it's, it's still, it's still in the
thought processes for me. One of the things that's, that's forestalled my choice is there
is a service called CraftCloud, which is a website you can go to and submit, upload an
STL and it will essentially request quotes from hundreds of 3D printer companies across
the U.S. and then it'll, it'll sort them by price and you can, and other features that
you care about. And then you can order like one or, or in my case, I ordered like a, like
a small batch of like 10 and 20 of a little synthesizer enclosure. And the quality was
great. It was from some company in Georgia and, and you know, they shipped within like,
they printed in two days and shipped within, I mean, it got here to me in like two days.
So it was like about a, like less than a week of time, physical time. Yeah, it's great.
So I'm like, well, geez, maybe I just don't get a fancy new printer. Maybe I just use
this service for some of the things I care about.
Well, that is a consideration. How many times have you heard someone buy a 3D printer and
a year later, it's just sitting there collecting dust too.
Exactly. And, and, and, and the costs are, the costs are right at the, at the, right
at the perfect place of where I'm like, ah, you know, if it was a little cheaper, I just
would use this all the time. But if it was a little more expensive, I just would buy
the 3D printer.
I would recommend that if you're looking at printing ABS, but especially ASA, which isn't
nearly as bad as ABS to get the X1 carbon. If you're only going to stick with PLA and
PETG, any other printers are good enough to print PLA these days and get the speed increases
that you're talking about. But that enclosure really makes a difference, especially for
ASA being the common one and ABS probably a little more uncommon.
Yeah. Yeah. I'm not really into those fancy plastics, but.
I had the same printer that you had, Tod, and like three or four years ago, I updated
to a Creality Ender and it was kind of amazing to see just the, the jump in performance.
Yeah.
I think you would definitely appreciate it.
Yeah, totally.
And I will say, Noe is considering the Bambu X1 right now.
Yeah. Yeah. It's so good. I mean, the thing is, is that, is that the, the Mark II, my
Mark II 5S, it has been an appliance where like I've got Octoprint running and I will
just send a print job to it knowing that like, oh yeah, I cleaned the bed three days ago
and that's clean enough for the thing I'm doing. And I just print and I can watch it
on the camera. I'm like, yep, it's printing. There's nothing wrong. For the longest time,
3D printing was never just an appliance. It was always this like little, little fiddly
thing you were messing around with the entire time, doing, doing more to it than getting
out of it. And so, yeah, it's it's kind of amazing.
It is. It really is.
All right, Paul, what's our final thing for this week, this month?
This one I'm very excited about. Steph Piper, aka Maker Queen, is a library makerspace manager
in Queensland, Australia. She's come up with a great way to visualize your skill level
and skill gaps in 45 different disciplines with her Maker Skill Trees. Similar to how
in a video game, when you get enough experience points to level up, you can put skill points
into your character to get better. This is the same concept, but for you. It can help
gauge where you are in your learning journey for each of the different skills.
Each skill is made up of 73 hex boxes. You start at the bottom with some of the basic
tasks and work your way up at both skill and complexity. You don't need to do them in order,
though they get a little harder the higher up you climb the skill tree. What's also neat
about it is that it might show some gaps that you might have in a given skill. And now you're
incentivized to try something new and fill in those gaps. Or for the person that says,
what do I do or what do I learn next? This has plenty of inspiration to draw from.
Just a few of the almost 60 disciplines are 3D modeling, 3D printing, amateur radio, coding,
cooking, crochet, Dungeons and Dragons, music, PCB design, robotics, sewing, woodworking,
and so many more. That's just a small selection. So you can see that there's things from the
maker world, the crafter world, the coding world, and all of those have Venn diagrams
that probably cross over each other in so many different ways.
Steph Piper gave a great talk on the maker skill trees at the Open Source Hardware Association
earlier this year. It's under 10 minutes long and it's worth watching to learn more about
it and I'll link to it in the show notes. She's also working on a book that uses the
skill trees, including 13 makerspace skill trees, self-reflection activities, and a life
progress dashboard for how you are doing across many of the skill trees. I think it's definitely
worth checking out and everyone, every makerspace should have a copy of this printed, I think.
That's really cool. And I am a really visual person, so I appreciate having it all laid
out. Some things that you may forget or not think about or not know about, like you said.
So it's really cool. I really like Steph Piper's work in general too.
Yeah. She made a very conscious decision that it should be printed on paper and should be
a tactile experience filling in those boxes, circling them or crossing them out or however
you want to do it and get you off the screens to do it.
I like how it's laid out as a sort of a honeycomb hexagon thing rather than a tree where there's
like a linear progression because when you're learning stuff, it's never a linear progression.
You might bust ahead really far on certain topics, but still be like down farther on
other topics. And so having it be sort of like, oh, this amorphous, you just kind of
grow from the bottom towards the top is really nice.
I agree.
Yeah.
Well, that's our show. For detailed show notes or to join our newsletter, please visit thebootloader.net
and a special thank you to Liz Clark for joining the show this episode. Until next time, stay
positive.