Microcontroller programming (Arduino, Microchip PIC, etc)

Daegerty wrote: »

Anyone here into this kind of stuff. Been reading around the forum and only been finding posts about very high level (mostly web related) programming. Nothing wrong with that but I'm just looking to see if anyone else on here is into Arduinos and what have you.

Yeah, messed around with arduinos a little bit.

Really easy to work with, nice docs, easy to get up and running, thoroughly recommend them if you want to play with microcontrollers.


Here is a video of simple project I did, with an old RC car I bought for a fiver on adverts:
http://www.youtube.com/watch?v=WoCA7rZMdpQ

I ripped out all the electronic guts of it, and replaced it with an arduino, and a motor controller and all that stuff.

It even has a laser

Webmonkey wrote: »

Ahhhh! That's exactly the stuff I want to go doing. Something that has results! I might give a start at this stuff too! So exciting

It was good fun to play with.
I did it to refresh my electronics knowledge.
Made several smaller arduino projects first, had a quick go at building a simple lego robot, and then went on to build this simple project.

Its amazing how fast you can pick up the practical electronics, when you've got a bit of a grounding in the 'theoretical waffle' :pac:
And the internet is great these days - there's so much information easily accessible out there.
I think we're really entering a create age of practical 'do it yourself' projects - its so easy to find the resources, these days.

Actually, one thing I'd recommend, is to head along to one of the hackerspaces, like 'tog' - from what I can see, they've got a good crew there doing all sorts of more interesting projects, with arduinos and the like.

fergalr wrote: »

It was good fun to play with.
I did it to refresh my electronics knowledge.
Made several smaller arduino projects first, had a quick go at building a simple lego robot, and then went on to build this simple project.

Its amazing how fast you can pick up the practical electronics, when you've got a bit of a grounding in the 'theoretical waffle' :pac:
And the internet is great these days - there's so much information easily accessible out there.
I think we're really entering a create age of practical 'do it yourself' projects - its so easy to find the resources, these days.

Actually, one thing I'd recommend, is to head along to one of the hackerspaces, like 'tog' - from what I can see, they've got a good crew there doing all sorts of more interesting projects, with arduinos and the like.

I only done the theoretical parts of electronics in my IT course, had a lot of telecoms so that's where it crept into. I've always loved to get into the practical side of it but never get the time! Maybe if I just buy one of these I'll manage to get started.

Ha yeah that "theoretical waffle", apparently it's not needed by some (/looks at another thread)

I've briefly looked at those hacker spaces alright. I'm after recently moving to Dublin so would be well up for meeting people with similar interests. Might check it out properly now some day!

Cheers!

Daegerty wrote: »

ah now not really cheating. can't say that using MPLAB C & ICD3 is any slower than an arduino to develop on with all the libraries it has. mplab is a bit of a pig of a piece of software to work with and its an expensive setup but I think you can also program arduinos in assembly easily enough, even break into assembly from the arduino IDE although I havn't tried

As I said before I haven't got around to using an arduino yet but am somewhat familiar with them. I'm just basing it on the fact that projects I've seen using arduino's have a much faster/easier coding stage than projects using PIC's. This is due to the much longer ramp up time in getting familiar with PIC's as well as PIC users having to code at a much lower level. Even using c is a fair bit closer to the bit level than what I've seen of arduino code.

Of course arduino's aren't actually cheating. I suppose it's just my dislike of moving up into the more abstract levels of coding. I'm an engineer as opposed to a coder at heart. I do think the arduino is a great tool but maybe it moves the user away from having to understand the hardware they are using. Maybe an inevitable move though since PC based coding has already done it.

The arduino has some great extensibility though like USB, LCD and ethernet shields.

Sparks wrote: »

See, if by "electronics" you mean LEDs, resistors, capacitors, and so on, cool.
If by "electronics" you really mean "microcontrollers and assembly", then dude. Go play with the electronics now. It'll make you a better coder.

Why we gave up on the idea of handing out a bunch of wire and a handful of chips and saying "go build a PC" on the first day of first year in a CS course, I'll never know. I mean, after you do assembly, pointers, arrays, linked lists, callbacks, pretty much every code-level concept that takes a while to first introduce to someone, they're all far far simpler in your mental model.

Well, until you get to C++ and pointers to references anyway

I've a lot of sympathy for this point of view, but recently I've been wondering.


By about the end of second year, in CS in TCD, we'd built most of the parts of a computer, working from the level of the transistor upwards.
Not all at once, but we'd covered each core conceptual piece - built gates using wires, adders from transistor ICs, ALUs in circuit maker, pipelines etc.

Its really nice to have a good conceptual model of the main parts in an electronic computer, when it comes to programming an electronic computer. More importantly, while I am not familiar with what exactly is happening in a modern chip, I have the confidence that I could learn it, if I need to; and I have some sort of an informed intuition to guide me, whenever I have to work at a low level.


But at the same time, I feel we did too much hardware in our course. I stopped paying detailed attention to it around 3rd year, as I felt it got needlessly detailed.
I thought spending time reading about things like software engineering and architecture, which I felt were neglected in our course, made more sense. In retrospect, I think that outlook was correct.



Recently, I've been thinking if students should be thought even less hardware than we were.
(As an aside - I absolutely don't mean dumbing down. On the contrary, the standard must rise here. It really pisses me off in the 'dumbing down' debates that people don't see this. The standard in technical disciplines is rising globally - if the goal is to maintain standards, then we've already failed.)



A digital electronic computer is only one physical manifestation of the abstract idea of a computer. Why do we put so much focus on it?

I mean, in these days, when a dev is writing Java, to be executed, massively in parallel, on Google's App Engine, what's really going on?

They are writing code for a virtual machine, that has little relation to the underlying hardware. This code is then massively distributed across a host of machines, on each of which the JVM executes. These machines, may, in turn, possibly be virtual, running on geographically distributed physical server farms.
What the underlying physical architecture is, at that point, is really starting to become irrelevant.
A good understanding of idiosyncrasies of the JVM is probably more useful.


So, that's one thing that makes me think.


We've also an increasing use of (inherently quite abstract) functional languages. Languages like Clojure and Scala are gaining traction. Haskell, even, is growing.
The intersection between computer programming, and mathematics, seems to be increasing, as we use functional languages, and execute more in virtual environments.


A lot of the cool kids these days are writing code, using very abstract functional paradigms, executing on heterogeneous and physically distributed platforms.


I wonder if perhaps the way to approach modern computer science education should involve much less of the electronic engineer, and hardware; and instead come from the direction of the mathematician, or statistician - from formal logic, algebra; and mathematical tools such as the lambda calculus etc.

Also, there's an ever increasing role played by the tools of statistics; fundamentally, computers deal with information and data, and statistics must play a role, especially as developers want to achieve goals (be they business, or scientific) that deal with the more fuzzy, more noisy real-world data.


I dunno - there's a lot of stuff to learn, to be good at CS these days.

The world is changing - we need to solve fuzzier problems, and acomplish more, using higher level tools. We need to solve harder problems, and need more powerful paradigms.

Is it still possible, in a 4 year course, to equip someone with sufficient theory to use the more powerful tools available to us, and also teach them a thorough understanding of the underlying hardware?

I wonder if perhaps a much more mathematical and statistical approach is needed.

The hardware has been becoming less and less important, for a long time.

All that said, today, when I have to work with a graph thats 10s of GB in size, every byte of representation counts.

fergalr wrote: »

We've also an increasing use of (inherently quite abstract) functional languages. Languages like Clojure and Scala are gaining traction. Haskell, even, is growing.
The intersection between computer programming, and mathematics, seems to be increasing, as we use functional languages, and execute more in virtual environments.
A lot of the cool kids these days are writing code, using very abstract functional paradigms, executing on heterogeneous and physically distributed platforms.

And meanwhile, all the code that actually runs our society (banks, hospitals, public and emergency services, city management software, embedded software, just-in-time delivery systems for food and goods, etc, etc) is all in COBOL, C, C++ and FORTRAN. Maybe with a little Java in the newer bits.

I'm all for new languages - I'm particularly fond of Python personally, but I can grok the value in LispScala and those with more unusual conceptual models - but we're not turning out enough people today that can fix the bits our society requires to keep functioning. And the people who do fix those bits today are dying out and their skillset is being lost. And the depth of those skillsets are quite dependant on knowing the hardware.

And that's only looking at the "computer" stuff and not the "appliance" stuff. Embedded controllers are the lion's share of all deployed computer hardware and computer code in the world today, and that's not going to change anytime soon. It's almost all exclusively written in assembly and C, it's only just recently started to happen in bus widths above 16 bits, it's all seriously hard-limited in cpu horsepower, memory, and every other resource you can think of, and a lot of the time, someone dies if a bug isn't caught before production. And none of the graduates I've seen in the last few years from our undergrad courses know enough to work in the area at all, and apparently, that's endemic.

We're going to see a serious manpower crisis in our industry in the next decade, one that will make the current crisis look blissful by comparison. Haskell's fun, but it won't fix that problem.

Didn't say a mini couldn't pull a plough Daegerty, just that there are better things for the job out there...