How can I learn analog circuit design if I already know circuit analysis?

Thread Starter

babaliaris

Joined Nov 19, 2019
160
Hello!

Some of you might already know me from previous posts, in which I was trying to create circuits like amplifiers without knowing anything about analog circuit design.

I'm an Electrical & Computer Engineer student and I've completed all the basic subjects about circuit analysis (DC, AC, and nonlinear electronics like diodes, MOSFETs, and BJTs). Currently, I'm following the path of energy and power circuits like transmission lines, but I want to learn electronics (small power analog circuits and later digital as well) for my hobby.

Give me any circuit with known components and sources, and I can calculate all the voltages and currents across the circuit (as long as it does not result in a very big system where I can be lost in calculations).
I also have a deep understanding of how these components work individually and the mathematical models describing them.

And yet I can't design anything! I have no idea where to start, how to think when designing something, how to connect components in the order they must be connected in order to create the desired circuit, or how to choose the values of these components (resistance, capacitance, inductance).

Is there a good book, video tutorials, or a step-by-step website guide that teaches you all the stuff that you need to know when trying to build a circuit? I don't want textbooks that explain Kirchhoff's laws, Thenevin's and Norton's theorems, apparent power, Fourier transform/analysis, steady state analysis, Laplace, etc. I already know that.

People have told me that circuit design is a subject of its own and needs to be learned separately from circuit analysis.

But I can't find anything on the internet. All of them are showing you a circuit, give you the components and let you just build it. I want to know how to make it work from scratch on my own, not copy-paste the work of someone else.

I'm currently reading a book called "Practical Electronics for Inventors" but I got bored. It just explains how circuits like filters work and not how to design them yourself. Maybe I'm judging it too quickly. But it also shows a lot of math that I already know.

Do you have any suggestions?
 

WBahn

Joined Mar 31, 2012
29,976
Circuit analysis is the bedrock of circuit design, but it has to be approached differently. Instead of just analyzing a given circuit to determine how it behaves, you want to analyze it with an eye to understanding why it behaves the way it does and how it could be modified to make it behave differently in a controlled, predictable manner.

You also need to focus on understanding how circuits interact and affect each other. This is important because complexity in circuit design is controlled by being able to decompose a design into functional blocks and then designing the much-smaller block so that it has specific behaviors at its interface to other circuits.
 

Thread Starter

babaliaris

Joined Nov 19, 2019
160
This is important because complexity in circuit design is controlled by being able to decompose a design into functional blocks and then designing the much-smaller block so that it has specific behaviors at its interface to other circuits.
So basically you create smaller (basic) circuits like voltage dividers, attenuators, amps, filters etc, and then combine them together to create something different?

You also need to focus on understanding how circuits interact and affect each other.
Does this have to do with how a circuit's behavior changes once you add a load to it? For example, a voltage divider with Vs and R1=R2 will half the voltage (Vo = Vs/2, where Vo = VR2), but if you connect a load in parallel with R2 then the behavior of the whole circuit will change, meaning that Vo will no longer be Vs/2 (You need to recalculate everything again).
And the idea is to use things like voltage regulation or make sure that Rth << Rload?
I understand this, but I have no idea how to design these blocks so that once you put them together they won't affect each other.
I mean, if you are making a voltage divider or a low pass filter, how do you make sure that the internal resistance of this circuit is going to be much less than the load you are going to attach to it? And If I have multiple blocks, then what internal resistance should each of them have? Which blocks are considered loads? Things get messed up...

For example, if my circuit is composed of 3 blocks (a,b,c), and b acts as load for a and c as load for b, does this mean that Rath << Rbth << Rcth or something like that?

I need a good guide or book that explains all that plus all the basic circuits that I need to know.
 
Last edited:

WBahn

Joined Mar 31, 2012
29,976
So basically you create smaller (basic) circuits like voltage dividers, attenuators, amps, filters etc, and then combine them together to create something different?
To a large degree. Think of these (and many more) as tools in your toolbox (or building blocks, if you prefer). Most large systems can be broken down into function blocks and many of those functional blocks can be constructed using the tools you have, if you have enough of them and if you know them well enough. The rest have to be designed with more effort, but even then the knowledge you have about the rest guides you.

I understand this, but I have no idea how to design these blocks so that once you put them together they won't affect each other.
Oh, but they WILL affect each other. The key is to localize these effects so that they are well-defined at the interfaces and so that the effects are small enough to ignore, or so that they can be readily accounted for.

For example, if my circuit is composed of 3 blocks (a,b,c), and b acts as load for a and c as load for b, does this mean that Rath << Rbth << Rcth or something like that?
You might do something like that, but the better approach (usually) is to make Block B so that it has in input impedance that is much higher than the output impedance of Block A. Separate from this, make the output impedance of Block B much less than the input impedance of Block C. The key here is that the you have control over the input and output impedances of Block B separate from each other. In other situations, you want to make all of these impedance the same so that you get maximum power transfer from one block to the next. That is particularly relevant at RF frequencies where it is difficult to get amplification in all the various blocks that a signal passes through.

I need a good guide or book that explains all that plus all the basic circuits that I need to know.
Too many variations for any book to do a thorough job. The basic circuits that you need to know really well is very different for someone that does instrumentation versus someone that designs computers, versus someone that designs RF amplifiers.
 

crutschow

Joined Mar 14, 2008
34,280
plus all the basic circuits that I need to know.
That really depends upon what type of analog design your are doing.
Usually you will get a task that requires a specific type of analog circuits (frequency, gain, function, etc.), and then you look up references on those type of circuits.

I remember being rather flummoxed on my first design job when I realized that all I had really learned in school was how to analyze circuits already built, and knew essentially nothing about how to synthesize a circuit given a design requirement..
The task was to generate some basic logic circuits (using transistors since this was just at the start of IC's being developed) so I researched all I could about using transistors as switches and was able to successfully design and build the circuits.
 

MrChips

Joined Oct 2, 2009
30,705
Analysis and synthesis are two different things that require different skillsets.

Analysis requires knowledge. Synthesis requires imagination.

Imagination is more important than knowledge. For knowledge is limited to all we now know and understand while imagination embraces the entire world and all there ever will be to know and understand.
 

Jerry-Hat-Trick

Joined Aug 31, 2022
544
Search for the book "The Art of Electronics" by Paul Horowitz and Winfield Hill, - I could have included a link but there are many used copies available for sale and you don't necessarily need the latest edition.

Whilst the detailed study of how components work and the mathematical models are important as you progress towards more detailed design you can get a long way with making fairly rudimentary assumptions about what various components do. Op Amps for example are not perfect, but you can largely assume that the output will try to move to make the two inputs equal to each other if it can, and a bipolar transistor will typically have a voltage drop of 0.6V across the base/emitter whilst trying to pull current from collector to emitter equal to it's gain times the current into the base. Mostly, you'll be using Ohms Law and resistive potential dividers! Oh, and capacitors block DC but let AC through, and inductors don't like changing current.

Processors are so inexpensive and compact these days that discrete logic has become virtually redundant. The Arduino UNO is a good place to start to build a system which includes a processor although the UNO has become surprisingly expensive and there are less expensive options with greater functionality available now - follow the trend to 3.3V voltage rail.

Finally, the way to learn is to have something you want to make and try to design, build and test it. I'd recommend a hand drawn schematic first, with no component values, them calculate component values from knowing the important variables for the active components. Use SPICE only to check your design, not to randomly try different component values! It's a valuable tool but don't lean on it - understand the circuit first. As you progress you will discover that there are actually a fairly limited number of circuit concepts which are used like pieces of a jigsaw in lots of different designs.

Don't build the whole thing in one go, but build in chunks that you can test in isolation before incrementally adding the next stage. Figuring out why something doesn't work is the best way to learn and, failing that, there are always old fools like me who may be able and willing to help as long as you post something credible for us to analyse!
 

MrChips

Joined Oct 2, 2009
30,705
With knowledge of circuit analysis one can analyze a circuit once it has been designed. But how does one design a circuit in the first place?

This is where imagination is needed.

As an example, imagine as part of a larger circuit, a 2.5V stable and accurate voltage reference is required. How does one conceive such a circuit?

What are all the different ways you can imagine such a circuit?
What are the pros and cons of each design?

Remember, this is only a small portion of a larger design. One has to apply the same methodology for each portion of the circuit. Each portion serves a specific function within the overall design. Each part is a building block towards a final design.
 

BobTPH

Joined Jun 5, 2013
8,804
Opamps are used extensively in analog design because they are near perfect analog building blocks. When used with negative feedback, they are very linear and the gain is accurately settable to whatever is needed. They have high input impedance, and low output impedance and therefore can be strung together without much worry. They are a great place to start with most analog designs, those that do not push their limits. So learning opamp usage is a good place to start. Learn the standard circuits, then start thinking about putting them together.

Designing the opamps themselves would be another level if analog design.
 

dl324

Joined Mar 30, 2015
16,839
Give me any circuit with known components and sources, and I can calculate all the voltages and currents across the circuit (as long as it does not result in a very big system where I can be lost in calculations).
I also have a deep understanding of how these components work individually and the mathematical models describing them.
Can you understand why every component and value was used in those circuits?

What year are you in your studies?
 

Thread Starter

babaliaris

Joined Nov 19, 2019
160
Can you understand why every component and value was used in those circuits?
No. There was no subject that teaches that. For every single one of them (Circuit Analysis 1&2, Electronics 1&2) for a given circuit, you had to learn how to calculate all the currents and voltages across the circuit (or some of them). Also, you had to learn the mathematical analysis that derives all the formulas that you might use, the proofs that basically invent those formulas (using calculus, differential equations, Laplace and Fourier transforms, etc). Also, engineering tricks in order to make math easier (for example on inductor coupling or transformer, how to use the T equivalent circuit or ideal transformer equivalent or the per unit system).

What year are you in your studies?
Senior.

Well, the same thing is also true about programming. They taught us programming languages like C/C++, the concepts of object-oriented programming, data structures, and algorithms, but still, YOU had to learn how to think like a programmer and how to use these tools to actually start creating something useful. As a programmer, it took me like 6-8 years to truly understand programming and how to think as a programmer in order to solve different problems or create useful applications.

I still remember when I first learned python and C, literally nailed all the syntax of the language, and still when I tried to program something I was staring at my computer screen asking myself (How do I start making this now?). I literally didn't have a clue. Eventually after seeing a lot of others (youtube mostly) how they do it, and how they program a particular application, eventually I reached a state where I could sit down, think of making something (an application for example), and just do it without any help, understanding exactly every single line of code I was writing.
 
Last edited:

dl324

Joined Mar 30, 2015
16,839
There was no subject that teaches that.
That should be taught as part of circuit analysis.

Try understanding why certain components and values were used for the circuits you're able to analyze. Bill Hewlett did that to understand circuits while he was at Stanford. He didn't feel like he understood the circuits until he knew why every component and value was used.
 

MrChips

Joined Oct 2, 2009
30,705
I can give a budding mechanical engineer a boxful of mechanical components and construction tools:

1669328288338.png

They can acquire and analyze all the material composition, mechanical and structural properties of every single component. They can learn how to connect components together and how to apply the different tools.
But what can you build from it with all that knowledge?

For that, you need imagination!
 

neTC

Joined Jan 12, 2022
18
As a student you learn how to solve a problem. As an engineer you learn to define the problem. So start there, define what your design needs to do. Then break it down into smaller pieces.

If you know what all the pieces need to do then you can figure out how each can be designed. The designs for these smaller pieces might come from your existing knowledge or by exploring how others have solved the same or similar problems.You should be able to analyze different solutions to the same problem and figure out the advantages and disadvantages of each. If you can't decide.

It takes time to build the wisdom that comes from experience. You build that experience by doing. A book isn't going to be your best teacher for this. Curiosity is a better guide.
 

LowQCab

Joined Nov 6, 2012
4,022
It can be very useful to search the Internet for "other-peoples-Circuits",
because there may be many different ways of achieving a particular goal,
and probably half of them are old, proven-designs,
which "may be" really good designs,
for certain particular reasons, or for,
particular types of applications.

You will eventually start to get a feel for
how things are usually done in a particular manner, and --> WHY. <--

Also, pay attention to the promotional-blurb from Electronics-Manufacturers.
They normally make claims that are all about solving "common-problems",
usually, these are subjects that You need to
pay particular attention to when designing Circuits, or
selecting particular types of Devises, or
selecting a particular "style" or "topology" of Circuit.

Every basic Circuit design, or "building-block", may have "pluses" and "minuses",
which may vary, when used under a particular set of circumstances.
.
.
.
 

tindel

Joined Sep 16, 2012
936
It's not unusual to be where you are with this amount of experience. Undergrad teaches you how to find deterministic answers using equations.

I'd suggest one of two things if you want to be a designer:
  1. Look for a job as a test engineer that will give you exposure to a wide range of designs that you have to troubleshoot (not the designer). You'll learn how to design through watching what has worked and what hasn't worked for others.
  2. Go to grad school. Grad school usually teaches you how to use equations to make something happen, and that there is not usually a single correct answer. Does it work to the specifications? Is it good? Enough!
I've worked with many seniors and first and second year grad students (100's). I don't recall a single senior that I would be comfortable giving a design to them in a professional setting. By the time they graduate grad school, they are starting to figure it out, but usually need another couple of years before they really start designing things.

Good designers these days have experience in embedded systems, pcb layout, part selection, test, and analog design. It's rare, and I mean extremely rare, to have a senior come out of school with a good understanding of all of these to be comfortable handing them a design to go execute in a professional setting.

I do see some master's students with these skills, but we're still talking 1-in-20 students or so, and they usually worked in industry for a few years before going to grad school.

ETA: All that to say - don't stress out about this too much. You're probably exactly where most will expect you to be at your skill level. Most seniors don't get design jobs fresh out of undergrad. You'll probably be a test, product, or systems engineer depending on the company you go to.
 

dl324

Joined Mar 30, 2015
16,839
I started working at HP Labs a year after getting my associates degree and I could design circuits. Certainly I had limitations, but I had moved beyond simply analyzing circuits.

One of my co-workers was a newly minted Stanford MSEE and he was a very competent designer. He finished his MS in a year.
 
Last edited:

WBahn

Joined Mar 31, 2012
29,976
Good design skills come from lots of practice making bad design decisions.

One of the more effective ways to start gaining those skills is by designing and building things. It can be almost anything.

As an undergrad (freshman year), I designed an auxiliary electrical system for my old Ford Bronco to run extra lights, two-way comms radios, and several other items. I put in two batteries and wanted to be able to arbitrarily connect either/both the main and the auxiliary system to either battery. My first approach used regular starter solenoids as my isolators. I had no idea that such solenoids are not rated for continuous use -- but I figured it out pretty quickly when I burned a bunch of them up. That's when I discovered that they actually make battery isolators that are designed just for this kind of thing. In making this system, I also realized that I wanted some things, like the clock in the radio, to be powered at all times from either battery regardless of how the system was configured, but that I didn't want a dead battery to drain the other battery. That's when I discovered that I could use diodes to keep that from happening (using them in a way that I came up with on my own, but turned out to be patently obvious and commonplace later on). I also had a bunch of fuses that I wanted to be able to check with indicator lamps by switching a single ground connection to enable/disable them, but discovered that without the ground connection there were sneak paths between them. So again came my recent discovery of diodes to the rescue, but I needed a way of putting several dozen of them in my control box and so I learned to make a printed circuit board. At another point I discovered that running all of the power to my three window defrosters, at 20 A each, through a long piece of 10 AWG wire was not a particularly wise decision -- at which point I discovered that they make this thing called an ampacity table to guide these decisions. Through a number of stages, most involving missteps, I ended up with something that I was pretty proud of and that worked pretty well for a couple decades.

In my sophomore year the school's drama club needed a way to ring a telephone on stage during a play. I had made a code practice oscillator in high school that used a 555 timer (though my understanding of the circuit was pretty non-existent at the time, it was just a schematic that I implemented). But that seemed to do something close to what was needed, so I started with that as a basis and built a circuit that used a transistor and a relay to buzz the ringer in the phone. It worked, but the transistor kept failing quickly. Had no idea what was causing that -- but a semester later when we covered inductance in my Physics II class and how a classic automotive ignition system worked, the notion of inductive kickback, how it could destroy components, and how to deal with it made a lot of sense. They also wanted me to design a circuit so that they could dim a bunch of lamps on stage all together from a small control box off stage and so I learned about using triacs to do that. It wasn't great, but it worked. About a year later, when I was working at Taco Bell, the manager wanted me to design a "hospitality timer" that would remind the crew to go check/clean the dining room every fifteen minutes. The only way, at that time, that I knew how to do that was with a 555 timer (since we had now covered that in a physics department lab course). That led to me discovering the limitations of the 555 timer for intervals that long and to the CMOS version, which was better, and how to choose better capacitors. That was the first design I ever did that I got paid for (all of $50 in 1989) and, with what I learned just a couple semesters later, was a pretty pathetic design.

In my junior year I designed and built the buzzer systems for the annual Physics Bowl that required capturing the order of responses from three teams and locking out teams that had already responded with a wrong answer and only letting other teams respond. What would have been a simple task using a PIC or other MCU (I had NO idea that such things existed yet) was done using nine (or was it twelve?) quad NAND gates and a 555 timer. I didn't know about designing sequential circuits yet, so I designed logic for one channel that could communicate with the adjacent channels. It worked beautifully and actually became the basis of some fully-asynchronous logic circuits that went on clockless IC designs two decades later. However, doing this design is when I discovered that TTL logic doesn't actually go to 5 V like I expected and that it couldn't drive my LEDs by sourcing current (so I modified the logic to sink the LED current instead). To optimize my logic I also discovered that I could change the polarity of various logic signals (i.e., make some active-HI and others active-LO) -- once again I thought I had invented something revolutionary, only to discover a semester or so later that this was bread-and-butter stuff. Another thing I discovered was that 9 V batteries couldn't feed this circuit and all those LEDs for more than an hour or so, and as a result I was running around during the middle of the competition buying 9 V batteries from every store I could get to. The Radio Shack Enercell Battery Handbook became a good friend as a result of that.

These experiences, and a few others, made it so that I was able to design two measurement systems for NIST while an undergraduate co-op student, was in a position to successfully design and build a critical-current measurement system for high-temperature superconductors for my senior design project and be hired by one of my professors, while waiting for my graduation ceremony to start, to design the electronics for a radon measurement system they hoped to commercialize, and be hired to spec, design, and build an expansion of the servohydraulic measurement facilities at NIST as a first-year grad student as a freelance contractor .

It's difficult to say how much "design" was taught in my curriculum (undergraduate or graduate) -- there was certainly a fair amount, but I would have to say that it would likely have been pretty ineffective if it weren't for all of the extracurricular design projects I undertook, most of them unpaid and just for fun.

Another thing that I noticed (and have seen several times since) that took me by surprise was that all of the senior design projects in the Physics department were solo efforts and nearly all of them were very successful, despite being pretty darn complicated. In the Engineering department, on the other hand, the norm was teams of four to six people doing pretty mundane projects and frequently ending up with little to show for it. While I think that there is a qualitative difference between your run-of-the-mill physics major and the corresponding engineering major, I think a bigger factor was that a physics education emphasizes learning fundamental concepts and how to apply them to solve larger problems with very little of the "memorize all of the equations on this sheet and how to turn the crank according to a set of steps" approach that was pretty common in many engineering courses that I took. When you come right down to it -- design is all about problem solving, so an education that emphasizes generic problem solving skills is going to set you up well for learning to design solutions to problems.
 

tindel

Joined Sep 16, 2012
936
@WBahn, I know you personally, so I know this applies to you. @dl324, I don't know you personally, so maybe these points don't apply to you, but the fact that you got an engineering job with an AE degree tells me you are either the rare person that accomplishes this or that you're likely about WBahn's age.

Frankly, I don't disagree that there are exceptions to my previous comments, but also realize you come from an era where you had some distinct advantages.
  1. You could take your TV apart and see how it worked.
  2. When your TV broke, you had a chance to fix it.
  3. You could go down to RadioShack and buy all the components you need to build a transistor radio or equivalent while talking shop with experience folks about your projects and ideas.
  4. Design when you entered the workforce was analog OR digital, and not so-much mixed signal.
  5. When you entered the workforce, clock rates were still entered in the 1's of MHz and still pretty forgiving.
I submit that both of you had already had a lot of EE experience before you went to undergrad. Today's engineers do not have these luxuries coming out of school. To be an effective design engineer you need depth of knowledge in digital, analog, hardware, software, pcb, high-speed, etc., etc. to design even simple things. The bar has been raised a lot.

My experience, having worked with many graduating EE's, over the last few years is that most of them have no business designing circuits until they are coming out of grad school. Maybe there are some opportunities to design a circuit for a low-risk application with oversight from some other engineers, or an opportunity to be a 'design engineer' that is just repurposing someone else's heritage design, but a true design engineer fresh out of school is a very, very rare case. Of the 100's of student's I've advised over the last 3-4 years, I know of 1 person that got a design job out of school with no prior related work experience. He deserved it - very bright guy and a good designer.

One of my co-workers was a newly minted Stanford MSEE and he was a very competent designer. He finished his MS in a year.
You're making my point for me. It takes going through grad school to be a competent design engineer, because that's where most EE's learn to think for themselves. Also recognize that Stanford is NOT most EE schools - by an order of magnitude most of the time. Stanford and MIT are the two EE schools that other schools can't touch. They attract the people that don't want to do what has done before, and are, at their very core, design engineers and disruptors.

My point, again, for @babaliaris is that: don't be hard on yourself if you don't know how to design and aren't comfortable in that role yet. Find a job doing something else, and when you're ready to make the switch, make the switch.

As an aside, my experience was that I started out as a test engineer 'designing' a battery simulator unit (which was just a repurposed design of a much more experienced design engineer, with 40 years of experience, that I just made into a smaller form factor, studying under him). I did that for a year or two and then went on to work as a 'design engineer' where I was using heritage designs to accomplish new things but had very little opportunity for any real 'design'. Eventually the company I was working for had a need to really design something new, and I was in the right spot at the right time, and had the right experience, to design the new circuit board and not look back. It just so happens that that new design is sitting on the surface of Mars today and I've designed a lot more since that time.
 
Top