The most important changes in electronic design over the past 25 years?

Flicking through modern text books for undergraduate students, I find it startling that little seems to have changed in the way of electronic design. Can anyone think of obvious examples?

Regards, Larry.

Reply to
blackhead
Loading thread data ...

Perhaps you need some newer text books? :)

I guess textbooks teach the "fundamentals", which indeed do not change. But there have been enormous changes in the *practice* of electronics.

Obvious ones would be

- the transition to surface mount for most applications.

- ubiquity of cheap, powerful microcontrollers. Meaning more and more of the functionality of a design is software.

- Complete subsystems now available as single-chip "building blocks", e.g. A/D convertors that connect direct to transducers, digital isolators, power supply sequencers.

- low cost or free development tools for microcontrollers. No more $5k in-circuit emulators or logic analysers.

- easy, online searching and availability of information, samples and production parts.

Etc etc

--

John Devereux
Reply to
John Devereux

FPGAs and logic simulation

LT Spice; more simulation, less closed-form math

Internet instead of data books

Zillions of complex-function analog chips

Digital photography

Cheap color digital scopes

High-efficiency LEDs

Math programs (Matlab, Octave)

Filter design software

DDS

Delta-Sigma converters

Serious desktop and embedded compute power

Computer based drawings and document libraries, less paper

More purchased blocks, like power converters and signal/RF bricks. That's a trend, moving us up the abstraction stack.

Some of them existed 25 years ago, but they weren't as common or affordable as today.

Undergraduate education should stick to fundamentals, so it wouldn't, or shouldn't, change much.

--
John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    
 Click to see the full signature
Reply to
John Larkin

Which textbooks did you look at?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

AoE probably :)

--

John Devereux
Reply to
John Devereux

There have been major changes. In contrast to 25 years ago I nowadays do almost complete designs on the simulator and then in real life they come out almost verbatim the same. This was theoretically possible back then as well, and I did use ECA224 in the late 80's and PSpice in the early

90's, but the available computing horsepower was not even remotely sufficient. Now it is, and that has changed things fundamentally. At least for me.
--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Fundamentals don't change, though they seem to not be emphasized anymore. Observe how many people absolutely tried to deny my loop and nodal analysis of the "Cute amplifier"... which tells me they were never actually taught how to do it... and they can't cope with math. If there's not some software for them to plug into, they can't "design" or think or withdraw their heads from up their own asses ;-)

25 years? That takes us back to 1988, CMOS was well established for digital uses, but was just barely beginning for analog applications. Most analog stuff was still made with bipolar processes.

In the mid-90's processes advanced, giving finer feature size and adding capability to mix both bipolar and CMOS... BiCMOS.

Now, which is why I think I get so much flack with analog analysis, we have a world that thinks everything can be done with a microprocessor... just write some code... you don't really need to know much of any engineering.

Getting back to math... MIT dropped the undergraduate thesis requirement for math back around 1960.... because nothing is changing in most math fundamentals. Yet our high schools keep farting around, acting like there's something new... confusing students and parents alike.

But, like I say, nothing _fundamental_ has changed, the same math and analytic methods still work just fine. There just seems to be a lot more ignorance going around :-( ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

By 1977 (36 years ago) I was doing Berkeley Spice on a VAX780. Source code was in Fortran, for which I rewrote the model for B-E capacitance... because it was initially incorrect, rising forever in increasing forward bias. It was corrected in a paper by Gummel. I probably have a copy around here somewhere.

There was, however, no schematic entry. I drew a paper schematic, numbered nodes, then created a netlist with a text editor... IIRC "VI" ??

By 1988 I had my own 80386 plus math co-processor, and MicroSim PSpice... though it took me years to abandon hand-drawn schematic for MicroSim Schematics ;-)

True enough as far as horsepower goes... today's typical laptop can outperform a VAX.

But engineering fundamentals haven't changed at all. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Reply to
Spehro Pefhany

Quoting you...

They can't "design" or think or withdraw their heads from up their own asses ;-)

There just seems to be a lot more ignorance going around :-(

I think pretty much everyone here hates me :-(

Well, the situation should be obvious. :-! or whatever.

--
John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    
 Click to see the full signature
Reply to
John Larkin

All it takes is one election gone wrong and then they do :-)

CD4000 series ICs make nice analog CMOS amps and were available. Same for the 74C-series but I've never used those on account of their high prices in Europe back then.

It's the same in code. I could do reasonable circuit simulation in 640k of RAM in the 80's. Now you need a minimum of 1G of RAM and three dozen concurrently running processes just to be able to write "Hello World".

[...]
--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

There has been a lot of quantitative advance, but far less qualitative. Processors and other digital electronics has gotten faster, although especially in processors, they have about hit the wall. I think cell phones entire RF sections are now all done in CMOS, where SiGe or GaAs technology used to be needed. memory constantly goes up in capacity.

But, real game-changing developments like core memory, the transistor, ICs, diode lasers and the like may be petering out. The transistor spurred a HUGE wave of development, and maybe it has run it's course. There are a few new devices on the horizon, the memristor or FRAM technology look like they might cure the many downsides of flash memory technology. Someday, there may be some equally revolutionary development that will start a new wave of advances, but it is not on the horizon (as far as I know) right now.

Jon

Reply to
Jon Elson

No matter what... pick any new device... it can still be analyzed using engineering fundamentals. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

WHAT? I used a VAX 11/780 and then had a MicroVax II at home, which was roughly 0.9 VAX 780 units of performance. I loved the OS (VMS) but it didn't take too long for the performance to be outstripped by the x86 architecture. Even an early Pentium would outperform the early VAXen. My current desktop is a decade-old Pentium 4,

1700 MHz, and it is roughly 1000 times faster than the VAX 780.

The VAX 780 and MicroVax II had no cache, so they ran at memory speed, and the 780's memory was on the same system bus as the peripherals, getting slowed down by DMA load. The uVax-II didn't have much memory streamlining, either, although it was on a separate bus.

I occasionally still run into this difference in performance. I have some programs that post-processed environmental data logged on the home VAX, and now logged on a Linux PC. The VAX took about

2 minutes to process the data at the end of the day. (a sample every 15 seconds all day = 5760 records) The PC takes quite a bit less than a second!

Jon

Reply to
Jon Elson

FPGAs ARE a big deal, but they've been around at least 20 years, haven't they?

Ahh, that is indeed a revolution.

Jon

Reply to
Jon Elson

Xilinx's first, primitive FPGAs came out in 1985, not much better than using MSI chips. But modern ones have MACs, multiport srams, DRAM controllers, LVDS, PLLs, SERDES, PCIe, multicore ARMs, all sorts of goodies, really system on a chip stuff. If you want to do a hardware divide in VHDL, just do it.

--
John Larkin         Highland Technology, Inc 

jlarkin at highlandtechnology dot com 
 Click to see the full signature
Reply to
John Larkin
[...]

Cool, I found a price of $141,000 for the 780 in 1977.

The VAX 11/780 appears to be the standard machine on which all the "MIPS" benchmarks are based. So a VAX 11/780 is a "1 MIPS" machine.

I have a $4 STM32F4 microcontroller on a $15 eval board on my desk right now. It is 210 MIPS :)

--

John Devereux
Reply to
John Devereux

I am not this old but I did write my schematics as text, netlist style, for PSpice entry. Kept doing that until too many people razzed me about the fact that nobody else could "see" the circuit that way. "Hey, we've got Edison electric light in here, you know?".

Very same setup here, except I opted for the Cyrix math co-processor. Paid an arm and a leg but that felt like I had installed a turbo-charger. I still have the two cloth-covered Microsim binders, with pouches. I also still have the co-processor. Maybe it rises in value when it gets to be an antique?

[...]
--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

:)

No insider knowledge, but I suspect Wins problem is that he is trying to cover the whole field, and it is advancing faster than he is writing...

--

John Devereux
Reply to
John Devereux

I think John Devereaux fingered the biggest item, which is that microprocessors are getting faster, cheaper, and smaller.

This, in turn, is driving more and more of the "brains" of systems into software rather than dedicated analog or digital hardware. More and more often, the most sensible way to view a microprocessor in some circuits is as a very versatile and flexible analog component, that happens to need some digital massaging to do its job.

This, in turn, means that the analog designer needs to be able (at least) to understand what software can do, and the software designer needs to be able to understand what the analog designer wants.

Best, of course, is to be trained and/or facile in both disciplines (smirk).

Case in point: in 1988 I was working on my Master's thesis, building a data-link radio receiver (which I found out years later that my thesis advisor didn't think was going to work). The original system design called for the data demodulation to be done on a 4x5 circuit card with switches and integrators and oscillators forming a couple of PLLs, with analog signals being sampled at 400Hz by ADCs in the microprocessor that ran the front panel.

The demodulation ended up being in the microprocessor, which sampled the audio out of the receiver section at something like 3500Hz and did all of the PLL-ing and demodulating in digital-land.

Since then arguably by biggest single body of work has been to take really high performance control loops, and move more and more of the brains of the loops from analog hardware into software, with overall gains in compactness, power consumption, thermal stability, and unit-to- unit variation, not to mention being able to do useful tweaks in the control laws involving nonlinearities that are easy to implement in code, but would require fields of piecewise-linear approximations if you even attempted the task in analog-land.

--
My liberal friends think I'm a conservative kook. 
My conservative friends think I'm a liberal kook. 
 Click to see the full signature
Reply to
Tim Wescott

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.