Guide to doing an Electronics degree in UK

[ocset]

Hi,
What do you make of this guide to doing an electronics degree in UK?..

https://massey276.wixsite.com/electronicsdegree

[Gyro]

I don't think it mentions SMPSs enough! 

[bd139]

Integrated circuit engineering....
This is  a  bit of a ‘closed shop’. The software to do it is very expensive. It’s called Cadence or something. You only learn it if you are brilliant or you know the right people. You cant work in integrated circuit engineering unless you can handle the relevant software simulators etc.

This is basically wrong. It's wide open (look for "Microelectronic Systems Engineering" which is what I did). You don't need to know the right people, just pick the right course. The software is basically free to universities so ignore that. You don't need to be brilliant - it's no harder than any other EE field. In fact I'd say it's easy once you've got above the gate-level abstraction. You can't work in any engineering discipline unless you can handle the relevant software simulators etc (it's 2018 people!)

I found it terribly boring though. This was 2 years of my life: http://www.ece.virginia.edu/~mrs8n/cadence/tutorial5.html

10…Digital Electronics....
EG   FPGA’s, VHDL etc.
This field is generally a  bot of a 'closed shop'.
You wont get good at it unless you are brilliant or know the right people.
The software  applications packages  you need to learn to be able to work in this area  are very expensive.
​

Sounds like the author should have done a BA instead?

[tggzzz]

What do you make of this guide to doing an electronics degree in UK?..

It is a load of rubbish.

One person has had some (lack of) experience, and generalises it to everybody everywhere. And that's the charitable interpretation.

Chips on shoulders and conspiracy theories both generate poor quality material.

[ocset]

The software is basically free to universities so ignore that.

Thanks, but i wonder how much it costs for someone wanting a copy of the software to use at home?
(ie the integrated cct design software)

Also, ditto the software needed to design VHDL programs...how much does that cost for a person wanting a personal copy for home?

Whilst doing an electronics degree, the thing I remember was that we never had any actual working electronics engineers coming in to the university to talk to us about what they actually did in their  job. We had sales engineers, managers and apps guys coming to talk to us, but never any actual working electronics engineers.
And the point is, that it is quite difficult to  imagine  what a VHDL software person actually does, and what an integrated circuit designer actually does…so very few people  wanted to take those kind of modules.
We did modules on Karnaugh maps....digital logic.....they seemed to be telling us that VHDL was the modernisation of all that, but it wasnt obvious.....i mean, logic gates and circuits full of them are obvious..but VHDL wasnt.

I mean C programming for microcontrollers was  an obvious thing….you could see what it was all about…but VHDL…….what was that all about?....i presumed   it was about software that needed to run very very quickly, but I wasn’t totally sure that that was the  salient point about VHDL.  It wasn’t obvious at all.
We had general modules in “semiconductor physics”, but it left you wondering what a job in that field would actually entail.
Whereas jobs making say power supplies or audio  amplifiers, or electric drives, well…..in those jobs, it was much more obvious what you would be doing.

[bd139]

Back in 2000-ish which was the last time I had to fill a PO for Cadence Virtuoso, it was about £100k per seat per year. There is no personal license. If you want to use it, you have to hit the university computer labs usually. On top of that if you want to spin a prototype, that's about £25k for a single run if you have an in house small scale fab (we did being defence sector). £60k+ outside fab if you have a decent contract.

This is why FPGAs are so popular. You can short circuit most of that cost and turnaround at the cost of speed, size and power consumption. Look at Intel / Altera / Xilinx for VHDL stuff. Cypress have ventured into this recently with their SoCs as well although those are more "configurable hardware" strapped to an MCU.

When I started ASICs were on the way out. FPGAs were cost effective and reliable enough for defence applications. Paired with an embedded PPC core, VXworks and an ADA compiler, that became the standard pretty quickly.

[ocset]

Thanks, this certainly seems to confirm that VHDL software and integrated cct software, isnt cheap enough for someone to buy for home use, so that they can assess and evaluate themselves  learning it up for a job or career.

[tggzzz]

The software is basically free to universities so ignore that.

Thanks, but i wonder how much it costs for someone wanting a copy of the software to use at home?
(ie the integrated cct design software)

A remarkably short-sighted question.

Can you design ICs at home? If not there's no point in learning the software at home.

Also, ditto the software needed to design VHDL programs...how much does that cost for a person wanting a personal copy for home?

Free, as in beer. There are some open source tools as well, but they are limited.

The FPGA manufacturers give their design tools away free, typically only for low end devices. But that's all you would be using at home. If you are using large FPGAs, you can afford their software.

BTW, if you think about VHDL as a programming tool for ICs, you will not produce good hardware. Unless, of course, you write MCU software in such a way that all statements execute simultaneously.

Go and download https://www.xilinx.com/support/download.html (or another), and try it. But be prepared for a significant learning curve, as might be imagined by an 18GB download.

Whilst doing an electronics degree, the thing I remember was that we never had any actual working electronics engineers coming in to the university to talk to us about what they actually did in their  job. We had sales engineers, managers and apps guys coming to talk to us, but never any actual working electronics engineers.

Shrug. Use initiative. Go and find them. Join a professional organisation such as the IET and go to their talks.

Don't expect to be spoonfed.

And the point is, that it is quite difficult to  imagine  what a VHDL software person actually does, and what an integrated circuit designer actually does…so very few people  wanted to take those kind of modules.

Well, if you can't imagine that, then you lack the imagination to be an engineer in the first place.

I mean C programming for microcontrollers was  an obvious thing….you could see what it was all about…but VHDL…….what was that all about?....i presumed   it was about software that needed to run very very quickly, but I wasn’t totally sure that that was the  salient point about VHDL.  It wasn’t obvious at all.

Who did you ask? What research did you do? If you aren't sufficiently curious to find someone and ask etc, then you aren't sufficiently curious to be an engineer.

We had general modules in “semiconductor physics”, but it left you wondering what a job in that field would actually entail.
Whereas jobs making say power supplies or audio  amplifiers, or electric drives, well…..in those jobs, it was much more obvious what you would be doing.

It seems that you had little imagination, ilttle initiative, and expected to be spoonfed with what you needed to know. Given the half-life of technology, learning how to learn is an essential skill.

I realised that at school, partly helped by reading an Isaac Asimov story, "Profession". 61 years later it is still relevant... https://www.abelard.org/asimov.php It also illustrates a key difference between an engineer and a technician.

[tggzzz]

Thanks, this certainly seems to confirm that VHDL software and integrated cct software, isnt cheap enough for someone to buy for home use, so that they can assess and evaluate themselves  learning it up for a job or career.

.

Choose your target hardware, then the manufacturer's VHDL/Verilog/C toolsets are free for low-end FPGAs.

As for "integrated circuit software", don't forget to include the cost of a maskset (several years salary) and to allow several months between submitting the netlist and having the ic in your hand.

[bd139]

Free for very low end. I think high end stuff was about £15k/seat/year back in the early 00's for Xilinx.

Also to note, a huge chunk of the industry is "design verification". It's pretty easy to forward-engineer something with the tools but to actually verify it does what it says on the tin is a lot more complicated. Particularly when the target is mixed mode i.e software and hardware cooperating together. You end up with verified compilers targeting verified FPGAs against verified test cases and what pops out at the end is a whole pile of empty wallets and not necessarily a project that works properly (F35 for example) even if the design verification steps pass. As tggzzz suggests, this is mainly because there are a lot of technicians on the projects

Edit: At university I got silicon fabbed. It was around £100k to turn around and 2 months. Process wasn't particularly advanced if I remember. Definitely in the tens to hundreds of uM territory and was CMOS. TSMC and Intel paid for this at the time out of their own pockets hoping they'd get some staff out of the course. We all went and did software in the end AFAIK 

[tggzzz]

Free for very low end. I think high end stuff was about £15k/seat/year back in the early 00's for Xilinx.

More than the very low end, and certainly sufficient for a hobbyist to develop and demonstrate their skills.

When I last looked, 4 years ago, Xylinx Vivado was free for the Zync 7020: dual A9 ARMs (inc 256k RAM), 85k logic cells, 4MB block RAM, 220 DSP slices. You can do a hell of a lot with that, including running Linux on one core and an RTOS on the other.

Also to note, a huge chunk of the industry is "design verification". It's pretty easy to forward-engineer something with the tools but to actually verify it does what it says on the tin is a lot more complicated. Particularly when the target is mixed mode i.e software and hardware cooperating together. You end up with verified compilers targeting verified FPGAs against verified test cases and what pops out at the end is a whole pile of empty wallets and not necessarily a project that works properly (F35 for example) even if the design verification steps pass. As tggzzz suggests, this is mainly because there are a lot of technicians on the projects

It starts with "specification elicitation", then "specification validation", then many more steps before you reach "design verification". Then rinse and repeat as the spec is modified

[bd139]

I don't think I've ever seen someone reach the end of that rabbit hole, just throw more cash down it. I'm currently persuading a fintech not to bother with it because it's not going to improve the fact that they don't actually know what the feck they are doing to start with so how can you validate it?

[tggzzz]

I don't think I've ever seen someone reach the end of that rabbit hole, just throw more cash down it. I'm currently persuading a fintech not to bother with it because it's not going to improve the fact that they don't actually know what the feck they are doing to start with so how can you validate it?

Fintechs embed business trading rules in FPGAs. The rules may or may not be correct (whatever that means), but they are fast.

I thought the HFT mob validated them by running (?ruining?) the economy with them. Either way, "mob" isn't incorrect.

[chris_leyson]

HDL design tools are free from Altera, Lattice and Xilinx.

it is quite difficult to  imagine  what a VHDL software person actually does

Well, they don't write software for a start, VHDL is a hardware description language that looks like software but it isn't, it's a set on logical instructions that describe how a piece of logic should behave.
The "instructions" are not executed on a CPLD or FPGA but tell the synthesis tool what the logic should do and then the synthesis tool generates the bit patterns for the physical interconnects on the chip. Thinking of VHDL like software is a mistake a lot of beginners make but once you start to realise what sort of logic the "instructions" generate then it makes a lot more sense. Same goes for Verilog.

[hexreader]

Buy these two products:

https://hobbycomponents.com/search?orderby=position&controller=search&orderway=desc&search_query=fpga

They are old, but stupidly cheap and toolchain is free

Just add jumper wires, LEDs and switches, and you have all you need to do practical VHDL excercises.

Seller forum gives excellent step-by-step guide to getting started with your first simple project, and clear documentation on the board



.... and why not buy one of these at the same time:
https://hobbycomponents.com/featured/274-lc-maxii-altera-epm240-cpld-development-board
Far less educational value, but fun and even more stupidly cheap - same programmer works for both boards

[bd139]

I don't think I've ever seen someone reach the end of that rabbit hole, just throw more cash down it. I'm currently persuading a fintech not to bother with it because it's not going to improve the fact that they don't actually know what the feck they are doing to start with so how can you validate it?

Fintechs embed business trading rules in FPGAs. The rules may or may not be correct (whatever that means), but they are fast.

I thought the HFT mob validated them by running (?ruining?) the economy with them. Either way, "mob" isn't incorrect.

Correct is merely either (a) we got away with it and (b) FCA didn't tear us a new arsehole.

HDL design tools are free from Altera, Lattice and Xilinx.

it is quite difficult to  imagine  what a VHDL software person actually does

Well, they don't write software for a start, VHDL is a hardware description language that looks like software but it isn't, it's a set on logical instructions that describe how a piece of logic should behave.
The "instructions" are not executed on a CPLD or FPGA but tell the synthesis tool what the logic should do and then the synthesis tool generates the bit patterns for the physical interconnects on the chip. Thinking of VHDL like software is a mistake a lot of beginners make but once you start to realise what sort of logic the "instructions" generate then it makes a lot more sense.

I (slightly) disagree there. VHDL is a domain specific language for logic. It can be compiled into something else be that silicon or C quite easily.  All is not quite straightforward behind the scenes. My favourite text on the subject is SICP. The interesting bit is here: https://mitpress.mit.edu/sites/default/files/sicp/full-text/book/book-Z-H-22.html#%_sec_3.3.4

Edit: paid my entire mortgage off thanks to reading that book.

[tggzzz]

I (slightly) disagree there. VHDL is a domain specific language for logic. It can be compiled into something else be that silicon or C quite easily.  All is not quite straightforward behind the scenes. My favourite text on the subject is SICP. The interesting bit is here: https://mitpress.mit.edu/sites/default/files/sicp/full-text/book/book-Z-H-22.html#%_sec_3.3.4

Verilog/VHDL/HiLo/etc can be used for more than just hardware, although that is clearly their forte. In particular the behavioural (i.e. non-synthesisable) parts can be used for "higher level" descriptions which are then refined into hardware. Of course just because something can be done doesn't mean it should be done. As ever, good taste is beneficial.

As for SICP, I used that bit to learn about OOPs in 1986. I implemented something similar on the only affordable OOP then was Apple's Smalltalk running glacially on a 1MB Fat Mac. That was sufficient to show me that OOP was the future.

Also when I re-implemented it in Objective-C (easy) and C++ (revolting), it enabled me to run away from C++ screaming, a decision I've never regretted

And finally it (plus Gosling's exemplary white paper) made the transition to Java in 1996 very obvious and easy.

So that book was indirectly responsible for a good bit of my career.

[hans]

https://www.eevblog.com/forum/chat/do-you-agree-this-is-how-to-approach-an-electronics-degree-in-uk/25/

[tggzzz]

https://www.eevblog.com/forum/chat/do-you-agree-this-is-how-to-approach-an-electronics-degree-in-uk/25/

Oh, I'd forgotten that. I wonder how many times treez will resurrect this topic; will he stop when he gets the "right" response?

Looks like Gyro's comment about SMPSs was on the mark.

[MrMobodies]

What do you make of this guide to doing an electronics degree in UK?..

It is a load of rubbish.

One person has had some (lack of) experience, and generalises it to everybody everywhere. And that's the charitable interpretation.

Chips on shoulders and conspiracy theories both generate poor quality material.

What I like about this board I am learning something new nearly everyday without the bs.  It is a learning experience for me. I watch the videos, observe the comments, search if I am interested in something or need to to find out things so it is a great learning resource.

When misinformation or stuff gets posted like that, part or half truths or even marketing bs, this place turns it around and especially those who are put off and told the same crap  that they have to be naturally born to able to do something without training or you know need to know certain people to be able do it etc, you get some genuine help here that it is basically rubbish and point you in some good directions so you know where to start.

[ocset]

OK thanks, so  I would update the article to say that  anybody wanting to do VHDL/FPGA should first see how they fayre in their digital logic modules  (AND, OR NOT etc etc) ….And see how they fayre in Karnaugh maps and making out Finite  State Machines. Then if they are aux fait with that, then they  could consider VHDL/FPGA as a pathway. But I would suggest that first of all they try one of the cheap VHDL hobby kits as kindly recommended by Hexreader above. If they get on well with that, then they could consider VHDL/FPGA as a specialism.
The thing I wonder is whether  VHDL is like  being an embedded software engineer. Its dead easy to write simple microcontroller programs in C and  then think you are an expert….however, as we all know, becoming an embedded software designer  and writing huge industrial programs is very different…much more difficult. I think many  on this forum would find it easy…..but many Degree students in UK would fall over. This is  shown by the big shortage of software engineers in UK.
In truth, I think to be an embedded C   software engineer in industry, I would say it would help if the student could demonstrate  significant skill in software  before they go to college….and while they are in college they would be very actively studying software and putting in large numbers of hours into it, well above what the course requires of them for a “pass”.….otherwise they could end up not getting a job in that speciality after the course.
I would have to say that in software particularly (including VHDL), it  really helps if you have close friends or family who already work in that area or have significant knowledge in it. The same  applies to general hardware, but I don’t think as much as in software or VHDL.
I also believe that generally speaking, learning how to design  general, low power  DCDC LED drivers and SMPS is much easier than learning how to be a VHDL designer.

[kony]

How about you go find actual job that is not way over your head treez, and learn from people there in small incremental steps? Because that is how vast majority of people learn, regardless of the academia creditals.

[ocset]

Well , we are trying to increase the uptake of electronics as a college subject throughout the Western World.

It is believed that many  youngsters  lack the introductory info needed to encourage them into a career/college course  in electronics.

I definitely believe that a young  child given advice  and familiarity in electronics from say a family member whilst growing up would have a much better chance of developing themselves into  electronics than a child who lacked this. The document seeks to be this “family member” for young western children.

The document likely does have inaccuracies, (eg the VHDL description), but generally, i think a child reading that would be a lot more prepared than a child who had nothing to read.
Remember,  a child with no links to anybody in electronics is going to see electronics as a foreign land.

[coppice]

Well , we are trying to increase the uptake of electronics as a college subject throughout the Western World.

Are we? Why are we doing that? Electronics jobs in the UK have completely collapsed, but they aren't exactly growing fast in the rest of the western world either. Most new jobs are in India and China, and they are doing reasonably well at expanding the number of people studying electronics for the local job market.

There used to be good opportunities in Asia for western engineers. I know. I was one of the western engineers working in Asia, That's not really true any more, as they can now source most of the skills they need locally. If you are an expert in an interesting niche things may be different, but only a few jobs are like that.

It is believed that many  youngsters  lack the introductory info needed to encourage them into a career/college course  in electronics.

In the UK they probably lack that information as there are so few electronics companies left to provide it. Don't you consider it immoral to encourage people to study for non-existent jobs? Studying electronics is fine, if you are doing it as a springboard to jobs which value things like control theory (e.g. finance), but the degree has limited value for getting a job in electronics, especially if you want to stay in the UK.

[tggzzz]

OK thanks, so  I would update the article to say that  anybody wanting to do VHDL/FPGA should first see how they fayre in their digital logic modules  (AND, OR NOT etc etc) ….And see how they fayre in Karnaugh maps and making out Finite  State Machines. Then if they are aux fait with that, then they  could consider VHDL/FPGA as a pathway. But I would suggest that first of all they try one of the cheap VHDL hobby kits as kindly recommended by Hexreader above. If they get on well with that, then they could consider VHDL/FPGA as a specialism.

"fayre"?

K-maps aren't used at the HDL level, FSMs are very useful in embedded software and high level software applications.

The differences between digital hardware and embedded software are both less and more profound than you appear to understand.

Start by considering that the first task in a hardware/software system is usually to define which functions should be in hardware, which in software, and which could be in either.

The thing I wonder is whether  VHDL is like  being an embedded software engineer. Its dead easy to write simple microcontroller programs in C and  then think you are an expert….however, as we all know, becoming an embedded software designer  and writing huge industrial programs is very different…much more difficult.

That is true of absolutely every engineering topic, and many topics beyond that. Start by considering writing a novel...

I really wonder about your background and education.

I think many  on this forum would find it easy…..but many Degree students in UK would fall over. This is  shown by the big shortage of software engineers in UK.
In truth, I think to be an embedded C   software engineer in industry, I would say it would help if the student could demonstrate  significant skill in software  before they go to college….and while they are in college they would be very actively studying software and putting in large numbers of hours into it, well above what the course requires of them for a “pass”.….otherwise they could end up not getting a job in that speciality after the course.

Absolute rubbish.

The nearest that gets to having any validity is that doing hardware/software/writing/knitting in the student's own time demonstrates to an interviewer that they like doing that.

I would have to say that in software particularly (including VHDL), it  really helps if you have close friends or family who already work in that area or have significant knowledge in it. The same  applies to general hardware, but I don’t think as much as in software or VHDL.

Absolute rubbish.

For the reason why, read the Asimov story I referred to above.

One of the few engineers I really respect (i.e. I listen carefully when we have a disagreement, since I will probably learn something) did a biochemistry degree and a one year's master in electronics. And was then designing novel high performance electronics on the boundary of what was possible.

I also believe that generally speaking, learning how to design  general, low power  DCDC LED drivers and SMPS is much easier than learning how to be a VHDL designer.

Do you do that? Kony's comment is appropriate.