New EE graduate looking for infomation

[JohnRoberts]

I can also understand that electronics as a subject is no longer what it was in say 70s. It is vastly much wider of a subject now, and impossible to cover all of it. So the universities have to focus on the  areas where jobs are. At the end of the day they have to produce employable graduates.

My complaint is/was that we wasted a lot of time on doing modules like integrated engineering. I got an exemption from it in year one but ended up doing it in year 2 and 3 which was an utter waste of time. A total micky-mouse subject. Whereas they could have filled these with, say, even basic power electronics. Equally again in year one we did one module programming Lego using Robot C and another one  literally copying Arduino programs, supposedly to introduce the student to programming. Why not introduce C directly? As a result from year 2 and 3 you can feel that they are cramming a lot of stuff in a very limited time.  By the time you are in year 4 you are sprinting (as one lecturer said).

@ squareave,

The science side even at its basic indeed requires a lot more than understanding diy electronics. Although, nobody designs analogue audio using laplace you have to chew a lot of serious maths.

That has pretty much always been the case as new technology obsoletes the textbooks before the ink is wet.

Not to be Debbie downer, but designing discrete mic preamps is not a good career  path for this century.

The cream of the crop design engineers are the guys designing the innards of ICs. I took one night course in semiconductor physics and it is more physics than soldering.

Analog design is almost obsolete, but all those fancy digital chips require analog "glue" circuitry around them to work, so will never disappear.

Understand the basics, and inside those digital ICs are little analog function blocks.

Sorry this may not be helpful..

JR

 

[ruffrecords]

I was wondering why an EE would have so little knowledge of basic analog electronics but this actually makes perfect sense. Analog circuit design is really a highly specialized field at this point.

I am not convinced. Analogue is specialised it you want to design RF systems for phones or spaceships or audio but also if you want to design at transistor level in silicon - like memory, processors. Lots of people do this.

I would argue that now (as back in the 70s), every electronics job is a specialised version of some branch of electronics. That is why a degree course should include an adequate grounding in ALL the basics of electronics. I think it is the job of universities to structure their courses to achieve this. I thought they failed to do this back in the 70s and it seems as though nothing has changed.

Cheers

Ian

 

[john12ax7]

I was curious as to how things have changed so looked up an undergrad curriculum. It's not that dissimilar. Basic circuits and lab course seems a requirement for all,  then you can specialize more in various fields,  which could be circuits but also lots of other non-circuit options,  dsp,  materials, control systems, etc. There is a LOT of courses you take in an EE degree which are not analog circuits as others have mentioned.

So I would say all EE grads would have at least basic knowledge of resistors,  transistors,  opamps,  etc. But not necessarily enough to be skilled at audio design.  Though even as a circuit major most graduates probably still don't have enough hands on experience to really understand how things work.

 

[Bo Deadly]

I am not convinced. Analogue is specialised it you want to design RF systems for phones or spaceships or audio but also if you want to design at transistor level in silicon - like memory, processors. Lots of people do this.

That's not really what I meant by "basic analog electronics". The original question was about how to make a discrete amplifier preamp circuit from transistors and such. The circuitry used to marshal signals into or out of the digital domain is just basic filtering and buffering glue. The bulk of people designing analog circuits for small signal audio like mic pres, compressors, equalizers, mixers and so on work almost entirely for a handful of large companies like Harmon or Loud Technologies. It's not what I would call a growth industry. Quite the opposite. Clearly these companies are being not so slowly consolidated.

Having said that, there will always be end-points that are analog devices to get those signals to and from converters. At the very least that means microphones and power amplifiers. And despite the fact that most musicians compose on a laptop, I suspect the music industry will never be entirely digital. There will always be synths and guitars and keyboards and so on. But that's a pretty small and very competitive part of the broader audio / video industry.

If I were a newly minted EE graduate, I would definitely not look for a job designing low buck gizmos for the AV industry. I would be looking at what the latest fancy parts, chips, techniques are do a deep dive into something specific.

 

[PRyHyM]

Career and passion are different things... I desire to understand amp design... I do know and was taught circuit basics. I understand on paper op amps, bits, jfets but real life application is way different. 4 years of education yet I still look at a neve schematic from the 70s with complete confusion other than knowing the transistors are what’s supplying the gain... I get decoupling caps to but I see caps in places I don’t get. Also I’ll notice a random diode and be completely lost...

 

[JohnRoberts]

Career and passion are different things... I desire to understand amp design... I do know and was taught circuit basics. I understand on paper op amps, bits, jfets but real life application is way different. 4 years of education yet I still look at a neve schematic from the 70s with complete confusion other than knowing the transistors are what’s supplying the gain... I get decoupling caps to but I see caps in places I don’t get. Also I’ll notice a random diode and be completely lost...

Don't be confused by the complexity... look at small sections by themselves.

JR

 

[Matador]

EDIT: removed.

 

[sahib]

... I get decoupling caps to but I see caps in places I don’t get. Also I’ll notice a random diode and be completely lost...

I do not mean to be harsh on you, but this means you do not know the basics of R, C and L circuits, even though you were supposed to have ben taught. Same goes for the diode. I bet when you look at a diode what goes through your mind automatically is that it conducts in one direction. Also think about what voltage drops on the diode, hence what voltage develops across it and what you can do with it.

You are not alone in this, and this is the problem.

I would say in my experience (but your case supports my position) and also limit my views to EEE course. The whole course is based on making the student pass the exam not to really teach.  When I say this, I am not exaggerating, we had a guy in our class in year 3 who turned round and asked my class mate who was sitting next to me " what is the difference between AC and DC?". And this guy passed the exams perfectly well. May be not with very good marks, but nevertheless passed. They are all studying for the exam, they know what is coming, they memorise in parrot fashion and they pass. After the exam, all forgotten.

Maybe start with control system theory?  Many amplifier topologies stem directly from general-purpose feedback control systems once you start to recognize the pieces.

Control  theory does not teach anybody designing an amplifier. Yes, knowing it is good for you, but  if you are designing a two legged robot. Otherwise in audio design it is no more than academic. I have never designed a discrete power amp from scratch but I do not think Douglas Self designs his amplifiers using control theory. However, John and Abbey are veteran designers. They can perhaps comment on this.

 

[abbey road d enfer]

Control  theory does not teach anybody designing an amplifier. Yes, knowing it is good for you, but  if you are designing a two legged robot. Otherwise in audio design it is no more than academic. I have never designed a discrete power amp from scratch but I do not think Douglas Self designs his amplifiers using control theory. However, John and Abbey are veteran designers. They can perhaps comment on this.

Control theory is something that lurks in the back of my mind all the time, as many other things that seem not related. I agree that control theory is more adapted to compressors and gates, though. But where is the limit between an opamp-based inverter and a limiter?

 

[JohnRoberts]

However, John and Abbey are veteran designers. They can perhaps comment on this.

I don't want to make this too autobiographical but I designed my first power amp back in the early 1970s. Being mostly self taught, it was far from a scratch design but I liberally borrowed from other successful commercial amp designs. My goal was to make 4x200W not remarkable by modern standards but a big amp for  almost 50 years ago.

I started by copying the basic topology of BGW using an op amp front end and level shifting signal swing up to hit the power point. That said I couldn't use the BGW protection scheme, IIRC he (Brian Wachner RIP) used a crowbar across the PS rail to protect the amp output stage in case of fault. My four channel amp had too much power supply use a simple shut down clamp, so I looked at Phase Linear schematics for inspiration. Bob Carver probably more famous for his later work used a conventional V-I device protection scheme in the Phase Linear, based on the power transistors rated max voltage/current specifications.

I used different (more rugged) power transistors than the phase linear but doing the math on Phase Linear's  protection scheme did not look like they offered adequate protection for those devices. Their nickname "flame linear" reveals a tendency to release their smoke.

In decades of use my hybrid mashup of those two design worked reliably. The last time I checked at least two channels were still working, but I haven't used it as my main system for almost two decades.

My first scratch amp design was at Peavey in the mid 80s. The world didn't need yet another incremental conventional power amp design from me, and Jack Sondermeyer (RIP) head of analog engineering had forgotten more about amp design than I would ever know.

But that didn't stop me from doing a trick amp design. I incorporated a capacitor doubler feeding the output stage rails providing 2x voltage to handle brief transients. 2x the voltage translates to 4x the peak power. The amplifier (AMR PMA70+) was basically a 35W continuous amp, that I rated at 100W for 20 mSec. The medium term power output was limited by how fast I recharged the boost caps. I power limited the boost recharge circuitry using a PTC fuse limiting medium term power output to roughly 60W for 15 seconds. In side by side listening tests this small 35W amp sounded as loud as a several times larger conventional amp. 

I had grand schemes of developing a series of these amps extending up to silly high short term power points but sadly the technology did not scale up. For higher power points the output stage had to be sized to handle the full peak current. There were savings available from smaller heatsink and power transformer, but this saving was not enough to balance out the extra cost from increased complexity.   

Sorry about the me me me episode. You can learn from doing. Just do it.     

JR

PS; You can ask here for advice about specific circuit blocks in an existing design you don't understand. There are many here more than willing to help you.

 

[Newmarket]

I'd suggest that everyone should have "The Art of Electronics" Authors Horowitz and Hill and maybe the accompanying "Student Workbook" if that is still published. Read it then read it several times more. Don't worry about lack of audio focus  - it's the basic knowledge you need. And Douglas Self's "Small Signal Audio Design" book for the audio specifics.
Plus there's good literature from the semiconductor companies eg Texas Instruments and Analog Devices.
Become familiar with basic building blocks and concepts - inverting / non-inverting amplifiers / instrumentation and differential amplifier configurations / impedance and noise considerations.
Build some simple stuff - op amp circuits / jfet buffers etc - and see where it goes right or wrong .

 

[Matador]

Control  theory does not teach anybody designing an amplifier. Yes, knowing it is good for you, but  if you are designing a two legged robot. Otherwise in audio design it is no more than academic. I have never designed a discrete power amp from scratch but I do not think Douglas Self designs his amplifiers using control theory. However, John and Abbey are veteran designers. They can perhaps comment on this.

Feedback, stability criterion, poles and zeros in transfer functions, frequency compensation, error amplification, open-loop and closed loop gains, transient versus steady-state response?  Doesn't apply in electronic amplifiers, and only two-legged robots?  ;D

 

[sahib]

Feedback, stability criterion, poles and zeros in transfer functions, frequency compensation, error amplification, open-loop and closed loop gains, transient versus steady-state response?  Doesn't apply in electronic amplifiers, and only two-legged robots?  ;D

I know very well that they apply.

But they do not make you design an audio amplifier.

You tell me how they help you design a say push-pull stage, then we talk.

If you do not know what a push-pull stage is and how to design it, your stabilty criterion is not going to make you design it.

 

[fazer]

You can model the neve preamp and or build it on a punch board and study.  Study the effect of by pass caps and the voltage drop of a diode on the following circuit.  I graduated in 75 and had a love of tube circuits but would apply  on SS circuits.  Another author I enjoyed was Norman Crowhurst.  Very good explanation and circuits to play with.  A dvm, scope , signal gen and .....  your good to go.  Order some circuit cards PCB,s.  Stuff them or build in a breadboard and start applying.  Listening to results are very helpful. 

 

[Newmarket]

If you do not know what a push-pull stage is and how to design it, your stabilty criterion is not going to make you design it.

But having some knowledge of those concepts will help greatly in understanding whatever definition of  push-pull amplifier (or other topology) you are working with.

 

[sahib]

But having some knowledge of those concepts will help greatly in understanding whatever definition of  push-pull amplifier (or other topology) you are working with.

Yes, of course it will help you understand deeper. But first, you have to  know how to design an amplifier in the first place.

It is not control theory, it is the topological knowledge makes you design an amplifier.

 

[jacomart]

Yes, of course it will help you understand deeper. But first, you have to  know how to design an amplifier in the first place.

It is not control theory, it is the topological knowledge makes you design an amplifier.

this is correct, as we say in Italy: "Before running you have to learn to walk"

Cheers
JM

 

[Newmarket]

Yes, of course it will help you understand deeper. But first, you have to  know how to design an amplifier in the first place.

It is not control theory, it is the topological knowledge makes you design an amplifier.

Good Luck in designing any sort of amplifier - rf / audio or whatever - if you don't have an appreciation of stability criteria / feedback etc.

 

[sahib]

Good Luck in designing any sort of amplifier - rf / audio or whatever - if you don't have an appreciation of stability criteria / feedback etc.

I do not need  luck. I have already been designing audio for a long time without control theory.

You obviously have no idea what control theory is or designing audio.

here is a very simple transfer function for you.

V2(s)/V1(s) = -1/RCs

Now  draw me the circuit it describes.

 

[user 37518]

Control  theory does not teach anybody designing an amplifier. Yes, knowing it is good for you, but  if you are designing a two legged robot. Otherwise in audio design it is no more than academic. I have never designed a discrete power amp from scratch but I do not think Douglas Self designs his amplifiers using control theory. However, John and Abbey are veteran designers. They can perhaps comment on this.

Power amps require stability analysis but it is not as critical as a DOA because a power amp usually has around 20dB of gain, on the other hand, a DOA might be used with much lower gains so it is harder to make it stable the less the gain, then the trick is to make it very stable at say unity gain without making it sluggish or very bandwidth limited when used at higher gains.

What I usually use (and believe most of us do) is the Gain/Phase margin analysis of the loop gain, which seems to me to be much more intuitive than a root locus, Routh-Hurwitz stability criterion or Nyquist plot

V2(s)/V1(s) = -1/RCs

Now  draw me the circuit it describes.

Its an inverting integrator 

I agree with Newmarket that to design an RF or MW amp you MUST use a stability analysis, but usually its not the same as the one used in control theory, in my MEE thesis I designed a class F power amp at 2GHZ, I used the Rollet K factor, the determinant of the dispersion matrix and another one called the geometric mu stability factor, all of which are based on S parameters used in small signal two port analysis, if your amp does not meet the criterion for a wide set of frequencies it will most likely become an oscillator at a certain frequency. The S parameters are usually taken from real measurements using a VNA or provided by the manufacturer, and they are frequency dependent.

For a RF/MW power amp you also need the large signal parameters which are usually measured with an LSNA or a PNA-X which uses X-parameters, both equipments are in the hundreds of dollars price range, we are getting into voodoo land now... A coleague of mine used to design computer models using large signal parameters for power transistors at microwave frequencies, it is no joke, very different from your typical SPICE model, it was in fact his PhD thesis. If you don't have any fancy equipment, the old fashioned brute-force method is to use something called Load Pull and Source Pull to determine the optimum Load/Source that produces the higher efficiency or power transfer, you can even do it in specialized simulators (like ADS, Genesys or MWO) if you have a good reliable transistor model, using special analysis like "Harmonic balance".

My point is that for high frequency RF or MW you do not use the typical control theory that you would use for low freq. audio amps and so on using transfer functions in terms of the complex frequency "s" which is not the same as the S parameters I am refering to. High freq. RF/MW is much different than audio, because of the high frequencies involved conventional circuit theory such as Kirchhoff laws are no longer valid, Maxwell reigns supreme, so you must use distributed parameters and Electro Magnetic wave simulators using FEM, FDM or the Method of Moments.