How to Design Circuits

[Xenon_K]

Hi everyone,
I'm currently trying to understand how to design my own circuits. I’ve chosen the LT1243 PWM IC and want to use it to generate a PWM signal. I’m planning to simulate everything in LTspice.

I know that there’s usually an example circuit available in the datasheet. Is that what you're supposed to use? Or how do you actually know which components to connect to which pins, and with what values?

Is it possible to figure all of that out just from the datasheet, or is there a specific method to follow?

Also, can anyone recommend a good book or video that explains this kind of stuff – especially how to read datasheets?

Thanks!

[temperance]

The IC you mention is current mode PWM controller. Those can't be used to generate a PWM signal.

[Zero999]

The data sheet is a good place to start.

You need a model to simulate it in LTSpice. If you can't find one, then you'll need to make your own, which can be a huge task and isn't easy to get right. The model will never be 100% perfect, but it just needs to model the characteristics relevant to your circuit.

Here's an example with the SG3525A I used in another thread.


^{PWM SG3525A.asc}
https://www.eevblog.com/forum/beginners/pwm-generator/msg5860931/#msg5860931

[tggzzz]

Hi everyone,
I'm currently trying to understand how to design my own circuits.
...
I know that there’s usually an example circuit available in the datasheet. Is that what you're supposed to use? Or how do you actually know which components to connect to which pins, and with what values?

Is it possible to figure all of that out just from the datasheet, or is there a specific method to follow?

Depends on what you mean by "design".

If you just want to create a module which is very similar to a device's prime market, then copying the circuit in the datasheet and application notes will probably be sufficient. That should, with a good understanding of basic circuit and control theory, enable you to tweak the design.

If you want to use it in a way the manufacturers didn't intend, or you need to workaround some limitation, then you will need a deeper and broader understanding of circuit theory.

Also, can anyone recommend a good book or video that explains this kind of stuff – especially how to read datasheets?

The key point about reading datasheets of anything (electronics, mechanical, software, project requirements...) is to work out what they don't say. That takes experience, and experience takes time and understanding.

No, it isn't easy; nothing interesting and worthwhile is easy. But it can be done.

Video: no, since most on yootoob are there to stroke the creators' ego, and most creators are crap teachers. Except for the cases where moving pictures are essential, they are a bad way to convey subtle concepts.

Books: very much depends on what you already do and don't know. If you have suitable theoretical understanding and some experience of designing and testing circuits (e.g. an undergrad electronics course), then TAoE and x-chapters are excellent. But I suspect you are not at that stage.

[Xenon_K]

Thank you very much for the helpful answers. I do have a "good" understanding of electrical engineering—at least in certain areas. I'm nearly finished with my electrical engineering degree, but here in Germany, the education is almost entirely theoretical and lacks practical experience. That's why I haven't gained much hands-on knowledge yet.

What matters most to me is not just copying circuits, but being able to build them myself. Otherwise, it somehow feels like cheating. I've set a goal for myself this year to design a rectifier, a buck converter, and maybe even a simple inverter from scratch. I want to understand and simulate everything in LTspice first, then move on to designing the layout in KiCad.

It was really hard to find a proper starting point, but by now I feel like I'm slowly getting into it. For the buck converter, for example, I don’t want to just use an all-in-one IC that does everything. I want to solve it piece by piece—PWM signal, control loop, gate driver, etc.—using discrete components so I can really understand how it all works.

However, I’ve had trouble finding the right components. Are there any good databases for this, or is it always just a long and tedious search to find suitable parts? I’ll take a look at the book and experiment a bit. Thanks again for your help!

[Andy Chee]

For the buck converter, for example, I don’t want to just use an all-in-one IC that does everything. I want to solve it piece by piece—PWM signal, control loop, gate driver, etc.—using discrete components so I can really understand how it all works.

How about following the functional block diagram of, say LM2576, and then constructing the discrete versions of each block?



It's not as if you are attempting to invent an entirely new converter topology.... or are you???

[Xenon_K]

Thank you very much for the idea, that actually sounds very reasonable. I’ll give it a try and see how it goes. I’m not trying to invent anything new, I just want to understand how it works and maybe tackle more complex projects in the future.

[tggzzz]

Thank you very much for the helpful answers. I do have a "good" understanding of electrical engineering—at least in certain areas. I'm nearly finished with my electrical engineering degree, but here in Germany, the education is almost entirely theoretical and lacks practical experience. That's why I haven't gained much hands-on knowledge yet.

You are welcome. It is a pleasure to give hints to someone who thinks and listens.

It is good you have the theoretical knowledge; too many people think practical experience is better than theoretical knowledge. It isn't; the two are complementary. When I was interviewing young candidates, I wanted to see someone who did things that weren't in their course. That indicated they were interested in the subject, and it enabled us to discuss why they chose a stretch goal and what they would do differently next time. If they couldn't give some answers to the latter, then either it wasn't a stretch goal or they weren't good at looking for opportunities to improve.

It sounds like TAoE and TAoE X-Chapters would be useful to you. You can download a sample chapter on this page https://artofelectronics.net/ and helpfully chapter 9 is 100 pages about "Voltage Regulation and Power Conversion". Having read that, I suspect you will want to buy the two books

While that is an excellent blend of theory and practice, I would recommend you paying careful attention to the PCB layouts in your chosen device's data sheets, application notes, and possibly evaluation boards.

Having good documentation is one good reason for selecting one device in preference to another.

Have fun, safely.

[Zero999]

For the buck converter, for example, I don’t want to just use an all-in-one IC that does everything. I want to solve it piece by piece—PWM signal, control loop, gate driver, etc.—using discrete components so I can really understand how it all works.

How about following the functional block diagram of, say LM2576, and then constructing the discrete versions of each block?

(Attachment Link)

It's not as if you are attempting to invent an entirely new converter topology.... or are you???

That's a good start, but there are plenty of traps, for example the fixed gain error amplifier's aren't specified in the data sheet, so how would you go about creating a model for it?

[Andy Chee]

For the buck converter, for example, I don’t want to just use an all-in-one IC that does everything. I want to solve it piece by piece—PWM signal, control loop, gate driver, etc.—using discrete components so I can really understand how it all works.

How about following the functional block diagram of, say LM2576, and then constructing the discrete versions of each block?

(Attachment Link)

It's not as if you are attempting to invent an entirely new converter topology.... or are you???

That's a good start, but there are plenty of traps, for example the fixed gain error amplifier's aren't specified in the data sheet, so how would you go about creating a model for it?

In the scope of this thread, that might be asking the wrong question.

A better question might be, is there another device datasheet with a better described/detailed block diagram?

[tooki]

Hi everyone,
I'm currently trying to understand how to design my own circuits. I’ve chosen the LT1243 PWM IC and want to use it to generate a PWM signal. I’m planning to simulate everything in LTspice.

I know that there’s usually an example circuit available in the datasheet. Is that what you're supposed to use? Or how do you actually know which components to connect to which pins, and with what values?

Is it possible to figure all of that out just from the datasheet, or is there a specific method to follow?

Also, can anyone recommend a good book or video that explains this kind of stuff – especially how to read datasheets?

Fundamentally, I see two ways of learning about the design process: watching others, and soing it yourself.

For the first, I highly suggest watching videos that show the design process in detail. EEVblog itself did some years ago, for the μcurrent and for a power supply. The power supply series is a great example IMHO. Another great example is Leo’s ultimate continuity tester: https://youtu.be/N2M-p-OGvPg?si=mBVyzCeuvkTHvez2

Next, look at old service manuals that include principles of operation. Old HP and Keithley test gear, for example. Similarly, old electronics magazines (including ELV’s now-free archive, and the hundreds of publications available on worldradiohistory.com) are treasure troves.

The old HP Journal is another gem with lots of deep dives into the design process.

Another great resource are chip manufacturers’ reference designs. I don’t mean the example circuits shown in the datasheets, but the actual sample product designs they publish. Those are great at showing how to integrate components to create a whole system.

Continuing on the approach of “see how others do it”, study other existing designs. Look at real products and see how they do things. Reverse engineer the circuits.

And of course then there’s doing it yourself and experiencing the process. You learn about pitfalls you didn’t know existed, about the process of selecting components, and how to tie them together. (Electronics education tends to focus on building blocks, but not so much on system integration, I think.)



I’ll also be a bit of a dissenting voice in that I find AoE to be a very informative book, but not necessarily the most helpful. It goes into some topics in great detail and entirely skips others, and it won’t really help with what I think you’re struggling with. I disagree with the recommendation to avoid YouTube; sure, some creators are just self-serving, but there are some truly great channels out there producing top-notch electronics design training. I learned a HUGE amount from EEVblog videos, probably as much as I learned from my formal electronics education.

[Zero999]

For the buck converter, for example, I don’t want to just use an all-in-one IC that does everything. I want to solve it piece by piece—PWM signal, control loop, gate driver, etc.—using discrete components so I can really understand how it all works.

How about following the functional block diagram of, say LM2576, and then constructing the discrete versions of each block?

(Attachment Link)

It's not as if you are attempting to invent an entirely new converter topology.... or are you???

That's a good start, but there are plenty of traps, for example the fixed gain error amplifier's aren't specified in the data sheet, so how would you go about creating a model for it?

In the scope of this thread, that might be asking the wrong question.

A better question might be, is there another device datasheet with a better described/detailed block diagram?

I couldn't find one.

The characteristics of the error amplifier are fairly important. Get it wrong and it won't model the device behaviour correctly. A circuit which is stable in the simulator, might oscillate or have poor transient response in real life.

EDIT:
Is this thread about designing a circuit or a SPICE model? They are two completely different things. A SPICE model just has to predict the behaviour of a component. It doesn't have to look anything like the internal circuitry, although that can often be the case.  SPICE is a useful tool to design circuits, but it has its limitations and shouldn't replace building it in real life.

[tggzzz]

I’ll also be a bit of a dissenting voice in that I find AoE to be a very informative book, but not necessarily the most helpful. It goes into some topics in great detail and entirely skips others, and it won’t really help with what I think you’re struggling with. I disagree with the recommendation to avoid YouTube; sure, some creators are just self-serving, but there are some truly great channels out there producing top-notch electronics design training. I learned a HUGE amount from EEVblog videos, probably as much as I learned from my formal electronics education.

No book, including TAoE is suitable for all people and all purposes. Hence my initial caution in mentioning it. Nonetheless, it is one that seems applicable to the OP at his stage. Other publications will also be helpful in complementary ways. The issue is to quickly determine a small set of useful sources.

The problem with yoootoob videos is threefold fold: the huge number of them, the vast variation in quality (most are dreadful), and you can't quickly and accurately scan most of them to determine whether or not a particular video might/will be useful. That leads to a lot of time wasted before finding the very useful nuggets of information. They are principally useful where moving pictures are necessary, and/or for disassembly instructions (and occasionally operating instructions) of specific devices.

Contrariwise, books are expensive to produce so rubbish is weeded out, the ToC and index allow speedy assessment of scope, and chapters can be speedread to verify that assessment.

[tggzzz]

SPICE is a useful tool to design circuits, but it has its limitations and shouldn't replace building it in real life.

In some cases it should replace building the circuit.

Examples include:

• monte-carlo analysis of a component's tolerance
• determining ideal behaviour applicable in all circumstances, without being distracted by a specific component's foibles. The latter can be a quagmire, the former can indicate the latter would/wouldn't be a waste of effort

A key question with all simulations is "what is included and excluded from the model". All models are wrong, but some are useful.

[tooki]

The problem with yoootoob videos is threefold fold: the huge number of them, the vast variation in quality (most are dreadful), and you can't quickly and accurately scan most of them to determine whether or not a particular video might/will be useful. That leads to a lot of time wasted before finding the very useful nuggets of information.

One can ask for recommendations…

They are principally useful where moving pictures are necessary, and/or for disassembly instructions (and occasionally operating instructions) of specific devices.

I totally disagree with this characterization. There are many channels that go into much more general theory, whether using an example device/circuit or not.

I mean… have you actually watched the EEVblog theory and design videos?!? You know, that YouTube channel that’s somehow connected with this forum? They’re basically university-level electrical engineering lectures, but with a bit more real-world applicability.

Not to mention that many aspects of theory are better explained when moving pictures can be used.

It sounds to me like you prefer to learn from reading, but it’s important to remember that there are many different learning styles, and many people do benefit from a person explaining it with spoken words and visuals.

Contrariwise, books are expensive to produce so rubbish is weeded out, the ToC and index allow speedy assessment of scope, and chapters can be speedread to verify that assessment.

Wellllll… there are still a lot of junk books out there. (And that was before AI-generated slop came on the scene…) It’s quite telling how the number of books (or authors) that gets recommended for a given topic often can be counted on one hand.

[tggzzz]

The problem with yoootoob videos is threefold fold: the huge number of them, the vast variation in quality (most are dreadful), and you can't quickly and accurately scan most of them to determine whether or not a particular video might/will be useful. That leads to a lot of time wasted before finding the very useful nuggets of information.

One can ask for recommendations…

They are principally useful where moving pictures are necessary, and/or for disassembly instructions (and occasionally operating instructions) of specific devices.

I totally disagree with this characterization. There are many channels that go into much more general theory, whether using an example device/circuit or not.

I mean… have you actually watched the EEVblog theory and design videos?!? You know, that YouTube channel that’s somehow connected with this forum? They’re basically university-level electrical engineering lectures, but with a bit more real-world applicability.

There's some truth there, but any utility is largely negated by the too small probability that I can find a useful video within reasonable time.

No I haven't looked at the EEVBlog theory and design videos, mainly because I've looked at the thoughtful/useful/rare table of contents and decided they aren't of interest. That's also what I do with text and illustrations.

Not to mention that many aspects of theory are better explained when moving pictures can be used.

Some can, but it is rare. Most yootoob videos fail in that respect.

It sounds to me like you prefer to learn from reading, but it’s important to remember that there are many different learning styles, and many people do benefit from a person explaining it with spoken words and visuals.

I like videos where moving pictures offer something than cannot be shown in text plus illustrations.

I don't like watching/reading a whole video/book before I know that. Speedy triage is essential

Contrariwise, books are expensive to produce so rubbish is weeded out, the ToC and index allow speedy assessment of scope, and chapters can be speedread to verify that assessment.

Wellllll… there are still a lot of junk books out there. (And that was before AI-generated slop came on the scene…) It’s quite telling how the number of books (or authors) that gets recommended for a given topic often can be counted on one hand.

Agreed, especially w.r.t. AI slop

Both books and videos can be egoboos or CV enhancers. But there are fewer books and smaller proportion of books to wade through, and they can be quickly discounted.

When I was young the key skill was to be able to find and slowly/carefully devour snippets of knowledge. Nowadays the key skill is the exact opposite: to quickly determine what to ignore. Videos make that difficult.

[SteveThackery]

I wonder if we've emphasised enough the importance of learning the theory aspects. Electronic engineering is a degree-level (or higher) subject with some pretty serious maths involved and it requires a commitment to study hard.

My concern with introducing SPICE so early is that it encourages the "poke and hope" design methodology, whereby you lift a circuit from a data sheet and then frig about with the component values until SPICE says it works. Is this really "design"?

Yes, yes, I appreciate it depends on the end objectives. Maybe the latter approach is OK for hobbyists.

[tggzzz]

I wonder if we've emphasised enough the importance of learning the theory aspects. Electronic engineering is a degree-level (or higher) subject with some pretty serious maths involved and it requires a commitment to study hard.

My concern with introducing SPICE so early is that it encourages the "poke and hope" design methodology, whereby you lift a circuit from a data sheet and then frig about with the component values until SPICE says it works. Is this really "design"?

Yes, yes, I appreciate it depends on the end objectives. Maybe the latter approach is OK for hobbyists.

Just so.

The problem is very acute in the AI and commercial software environments, where a common attitude is "it is working because it passes the unit tests".

That completely ignores the fundamental concepts "you can't test quality into a product" and the quality of the tests themselves.