Engineering Students being taught DSP on 20 year old processors RMIT

[janoc]

I have been also criticized for teaching using an 'old' processor by some students and faculty.  In this case an 8051 derivative, albeit a recent fancy one.  So, not long ago I had a slight change of tactics: I teach using the 8051, students complete the course assignments using something else.  The options I provide: STM32, LPC824, ATmega328, PIC32, and MSP430 (new this year) or anything else they find/buy!  I require them to bread board their own system.  Most of the programming is done in C but some assembly may be also present.  The purpose of this is to teach them principles as well as to force them to RTFM (in an educational environment the F stands for 'fantastic').

That's not a bad approach how to get them to actually understand the topic - by forcing them to transfer the skills. OTOH, I am not sure I would do this on a completely beginner's course, it could be a tad overwhelming - the point being learning the subject, not battling unknown tooling/platform. However, if your students have the level already where they can handle this, great! 

The uni I have taught at used semester projects where the students had to apply the knowledge gained from the courses. Most would pick the tools they knew from the courses for the projects too but they are free to use whatever they want in most cases. So I had students do all sorts of weird and wonderful things, including some rather spectacular disasters - but hey, that's also part of the learning

[jesuscf]

That's not a bad approach how to get them to actually understand the topic - by forcing them to transfer the skills. OTOH, I am not sure I would do this on a completely beginner's course, it could be a tad overwhelming - the point being learning the subject, not battling unknown tooling/platform. However, if your students have the level already where they can handle this, great! 

The students have some experience programming in assembly.  Usually ARM, 68k, or something else depending on who taught the prerequisite course.  They also had a course in C programming.  I wouldn't do that in a completely beginner's course either!  They seem to handle it fine.  I should mention that I provide them with a getting starting guide for each of the processors.  That includes links to the documentation, IDE, compiler, flash loader, and basic examples (blinky, hello world, ADC, timer, interrupts, etc).

[MT]

I have been also criticized for teaching using an 'old' processor by some students and faculty.  In this case an 8051 derivative, albeit a recent fancy one.  So, not long ago I had a slight change of tactics: I teach using the 8051, students complete the course assignments using something else.  The options I provide: STM32, LPC824, ATmega328, PIC32, and MSP430 (new this year) or anything else they find/buy!  I require them to bread board their own system.  Most of the programming is done in C but some assembly may be also present.  The purpose of this is to teach them principles as well as to force them to RTFM (in an educational environment the F stands for 'fantastic').

That's forward looking teaching, and you could also do the opposite let students do the thing
on a blingy HAL crippled 32bitter in C then do the same solution on a bare bone 8051 in assembler.

That will induce some healthy perspective into their thinking!

[CatalinaWOW]

You are probably right - but we have a different system here. If you are studying EE&C, the courses you can choose from are ALL strictly related
to this field - you can't, for example, take courses in liberal arts, music or Greek philosophy. THE ONLY course that has nothing to do with EE
is that one. And I do not consider this to be normal in a secular (at least on paper) state.

I am sorry for you who are trapped in this educational system, and for your employers who are stuck with the result.  Over the years I ran into many engineers like this.  It was often physically painful to read their reports and documentation.  And often their decisions missed whole areas of real world impact.  The most wonderful widget in the world is useless if its function and limitations can't be communicated to the user, if the wonder and potential profit of the device cannot be communicated to the management and funding sources.  In some cases a communicator will be hired to fill out the skills of such an employee, but that really means that the work of the engineer has to be so good that it justifies two salaries.  Not many are that good.

At many universities you can take (and indeed are required to take) courses which result in a more rounded education.  Something for those not already committed to a particular university to consider.

[westfw]

already purchased an ARM processor to see if we can actually replicate the same task on modern hardware.

That's an excellent idea.  Good Luck!

I notice that there have been very few suggestions for more modern "true DSP" chips.  That's "interesting", too.

My alma mater required EE candidates to take 7 semesters of "Social Science and Humanities" classes (to ensure a proper "rounded education", you know), in addition to a certain number of elective science/engineering classes, plus some "free electives", plus the required track (a total of 5 classes/semester.)  It was tough (at the time) to pick up all the CS classes I wanted to take, because they tended not to qualify under the "requirements."  (IIRC, "Computer architecture" was OK, but "Compilers" had to come from the "free electives.")  I took Russian History, a general psychology class, "History and Sociology of Science 110: Science in Literature" (heh: Science Fiction!), a sociology class in visual perception (fascinating!), and a second psychology class focused on vision (likewise fascinating.  I got to mention the then-new Xerox Optical Mouse in my final paper!), plus the credits from HS AP English.

[Mr. Scram]

I am sorry for you who are trapped in this educational system, and for your employers who are stuck with the result.  Over the years I ran into many engineers like this.  It was often physically painful to read their reports and documentation.  And often their decisions missed whole areas of real world impact.  The most wonderful widget in the world is useless if its function and limitations can't be communicated to the user, if the wonder and potential profit of the device cannot be communicated to the management and funding sources.  In some cases a communicator will be hired to fill out the skills of such an employee, but that really means that the work of the engineer has to be so good that it justifies two salaries.  Not many are that good.

At many universities you can take (and indeed are required to take) courses which result in a more rounded education.  Something for those not already committed to a particular university to consider.

I agree with your sentiments here. In the conversation Dave had with Jack Ganssle they touch upon the trouble with the modern day curriculum. The field has become so wide that, to include all the essential parts, you cannot go too much in depth without neglecting something else. The curriculum has also been stretched from 4 to 5 years to accommodate this.

Even though it certainly does pay to make engineers more well rounded in non-engineering tasks, you have to be careful not to hurt core knowledge in the process. It also underlined the formal education only being an introduction, with the real world experience after that being integral part of your education and knowledge.

[Alex]

I need to demonstrate to the program manager...

The taught class should be about principles and problem-solving approach. The processor should have a reasonably simple architecture not to detract from the class objectives. Obviously you want to reach higher and further, so you could pitch for an environment where you can practice these concepts on 'current' technology. This could be peer-led tutorials, some 'DSP club' with a budget etc. Have a think how you could give your proposal some 'political momentum' to maximise your chances of adoption and don't forget to address supervision/liabilities.

More generally speaking, note that many moons back ARM got interested in me because I used a low-end 8-bit PIC where one would expect a much higher-end MCU.

[nctnico]

I have been also criticized for teaching using an 'old' processor by some students and faculty.  In this case an 8051 derivative, albeit a recent fancy one.  So, not long ago I had a slight change of tactics: I teach using the 8051, students complete the course assignments using something else.  The options I provide: STM32, LPC824, ATmega328, PIC32, and MSP430 (new this year) or anything else they find/buy!  I require them to bread board their own system.  Most of the programming is done in C but some assembly may be also present.  The purpose of this is to teach them principles as well as to force them to RTFM (in an educational environment the F stands for 'fantastic').

I like this approach. It is very likely students will discuss amongst themselves the pros and cons of the platforms they choose to use for their assignment so instead of getting in touch with one architecture they get a feel for several.

[NorthGuy]

Even though it certainly does pay to make engineers more well rounded in non-engineering tasks, you have to be careful not to hurt core knowledge in the process.

The education is (should be anyway) designed to set you up for life, which might be 50 years give or take. During this time, many things will change, and the more you go on, the more important the general education (or a lack of it) becomes. Philosophy is a good thing. It gives you a vision which guides you through. Details ... you can figure by yourself.

[Mr. Scram]

The education is (should be anyway) designed to set you up for life, which might be 50 years give or take. During this time, many things will change, and the more you go on, the more important the general education (or a lack of it) becomes. Philosophy is a good thing. It gives you a vision which guides you through. Details ... you can figure by yourself.

I don't think there exists, of ever has existed, an education that sets you up for life. If done properly, it can act as a catalyst and also give you a basic survival kit of skills. Education is just the beginning.

I would also argue the contrary of it becoming more and more important. After a few years, people will not care much about your formal training. What you did and learnt recently is much more important and a curious and inquisitive mind is by far the most important element.

[NorthGuy]

Education is just the beginning.

Exactly. It is important to begin well.

I would also argue the contrary of it becoming more and more important. After a few years, people will not care much about your formal training. What you did and learnt recently is much more important and a curious and inquisitive mind is by far the most important element.

Education is not formal training. It plays very important role in determining whether you get the "curious and inquisitive mind" which you value so high. The development of a human brain is such that in the age from about 18 to 25 the function is predominantly driven by emotions while rational thinking is suppressed and gets developed. What you do at this age determines the nature of your mind.

It is the same as speach. If you didn't learn to speak by the age of 10, chances of you speaking are slim to none. Rational thinking follows the same pattern. If you don't develop it by the age of 25, you don't have any chances. But if you do, you can use it to learn whatever you like rather quickly. Here's where the education comes in - it develops your brain and teaches you to think.

[Mr. Scram]

Education is not formal training. It plays very important role in determining whether you get the "curious and inquisitive mind" which you value so high. The development of a human brain is such that in the age from about 18 to 25 the function is predominantly driven by emotions while rational thinking is suppressed and gets developed. What you do at this age determines the nature of your mind.

It is the same as speach. If you didn't learn to speak by the age of 10, chances of you speaking are slim to none. Rational thinking follows the same pattern. If you don't develop it by the age of 25, you don't have any chances. But if you do, you can use it to learn whatever you like rather quickly. Here's where the education comes in - it develops your brain and teaches you to think.

I don't see how this all detracts from my previous statement that it shouldn't hurt the core knowledge in the process.

[cdev]

This doesn't really apply to microprocessors but often by the time many technologies get to start being taught in academia they are already old news and the field has by then moved on to newer things.

Technology skills are the most rapidly moving of targets imaginable.  People also have to keep using them or they will very rapidly forget much of what they know.

[NorthGuy]

I don't see how this all detracts from my previous statement that it shouldn't hurt the core knowledge in the process.

The nature of the knowledge is irrelevant. Life is long and changing. Doesn't matter if you have a degree in mathematics or in EE. Your knowledge may be drastically different, but if you had good education then you can learn new things quickly, and consequently you can take on any job. People switch careers quite often and with great success.

[Mr. Scram]

The nature of the knowledge is irrelevant. Life is long and changing. Doesn't matter if you have a degree in mathematics or in EE. Your knowledge may be drastically different, but if you had good education then you can learn new things quickly, and consequently you can take on any job. People switch careers quite often and with great success.

Again, I don't see how this detracts from my previous statements. The core knowledge should be safeguarded, for it enables exactly what you describe here.

[NorthGuy]

Again, I don't see how this detracts from my previous statements. The core knowledge should be safeguarded, for it enables exactly what you describe here.

Very well then. Looks like we agree on that. Perhaps I misunderstood what you encompass by "core knowledge".

[cdev]

As jobs get scarcer, it starts looking like a game of musical chairs and geopolitical concerns start getting in the way of common sense..

This kind of thinking puts services, 80% of all jobs, on the bargaining table.

In order to put all those jobs into play, they anticipate lots of churning..

This has been encouraged by various political entities.. and hidden by others.

Jobs in services are now internationally "tradable commodities".

[nctnico]

I don't see how this all detracts from my previous statement that it shouldn't hurt the core knowledge in the process.

The nature of the knowledge is irrelevant. Life is long and changing. Doesn't matter if you have a degree in mathematics or in EE. Your knowledge may be drastically different, but if you had good education then you can learn new things quickly, and consequently you can take on any job.

To me this translates to: if you are smart you can learn new things quickly.
Generic things like physics, math and language are skills you can use for your entire career. However more specific skills like programming languages, hardware platforms will change drastically so while necessary to land a first job these will change.

[Fire Doger]

I agree that some things are old. There are communities for newer processors where you can find a ton of information to understand your hardware better. There are options like in circuit debugging, JTag etc which can save you a lot of time. But assembly is not old, every line of code in any language in any processor ends up in assembly. Knowing it doesn't mean that you will use it but you can learn and understand any other language better.
I believe that it's the best way to feel your hardware. You have all the knowledge to build on it. You can be better if you work on it.

Then hardware comes in, if you have to set 1000 registers to do something then you won't learn. Old mcu means simplicity and most of the case same principles so you know how they did it in the past and you can understand perfectly the newer implementations, the benefits etc.

Then its up to you to learn the latest MCU. If you could learn ARM in a semester without knowing shit about it before then you can learn the same things in half that time with all that you know from old mcu. And I am talking about 100% knowledge, sure 50% ARM is better than 50% an old MCU because you lost the ability to build on it. You can't expect a uni to make you a master on anything, they just give you the tools.

[CatalinaWOW]

As a final, more focused recommendation for those who feel training on a specific processor is necessary to land a job.

Study the companies where you intend to apply for employment.  Find out what processor those companies use.  If the all use the same one, lucky you.  If not, more work in your near future.  Because the next step is to google, buy development boards, buy books and anything else you can to learn about those systems.  Practice on them.  Start small with "hello world" and "blinking light" applications.  Then do something useful.  Repeat for each processor on your list.

After this you should be as prepared as possible for those companies that want new graduates with specific processor experience.

This technique of studying what a company you are interested in working for does works wonders whether the company does microprocessor development or builds waste treatment macerators.