Recently I came across this Design Flow Chart: http://www.williamson-labs.com/design.htm Even though it looks rather complicated (I would not recommend following it step by step) it does contain some very interesting thoughts. For example, if you click the “No Cook Books” box (not far from the top of the chart):

“…The first thing you don't do is read how others did it. The first thing you should do is several preliminary designs, using only yourself as a resource… You will get insights into the problem that using "cookbook" solutions prevent…”

After several other steps Williamson finally recommends to research other’s designs. This is quite contrary to what I used to do. Trying to understand the “state of the art” was my very first step when a new project started. However, I never thought of this as a “cookbook” approach, and I was never interested in copying anything without understanding how it worked. I just wanted to make sure that I did not miss the best ideas from the experts in the field before I tried to come up with my own ideas.

So what is your approach to a new design ? And are there any other steps in the Flow Chart that you find especially valuable (or steps that you use and that are missing from the chart) ?

 

The one thing I *don't* do is look at other similar designs. I've found that doing this tends to psychologically box me into a solution that may not be the best.

I will look at others' results (not the design), look for the weaknesses, and develop my own original solution that addresses the weaknesses.

Only if I am really stuck with a particular issue will I "cheat" and look for others' solutions.

Granted, if someone came to me and asked me to invent a wheel, it wouldn't make sense to not 'discover' how other wheels are made. But, generally, in that case, I'd tell them to just go buy a wheel that's made already...

EDIT: In retrospect, the flowchart at your link agrees with me!

 

I 100% agree with not looking at other designs first.. Why.. because you get that design stuck in your head and keep referring back to that..Many times you even carry over their bad decisions into yours. You need to always take a fresh look at a problem..

If you know what you are doing there is no need to look at other designs. But when you get a problem you don't know how to tackle looking at others can help alot.. but then you aren't designing..you are copying.

 

I think creating the solution by yourself first and then looking for other solutions is better than the other way, as it makes you think about the design a lot more than the other way.

 

Being creative is a good aim, but it does not always make good sense to "re-invent the wheel". While it is undesirable to repeat other peoples mistakes by copying them, it is also a pity not to take advantage of methods which were developed to avoid even earlier mistakes.

This applies with particular force when dealing with things which are unfamiliar. It is always tempting to think that we don't need to rely on the crutch of other people's efforts, but pride can come before a fall.

I'm not sure if the biggest offenders here are real beginners, or would-be "experts" who fail to recognise they are not so experienced in the areas required by the task in hand, and would do well to look at what how these things have been done before.

 

I tend to go with Adjusters post. If you are a beginner it is dumb not to look at prior art. If you want to invent something from scratch it is always your choice, but you may be missing out on elegant solutions, or ever worse, off the shelf technology. In general you are not going to be able to match the price of something off the shelf (though there are exceptions).

There is a big difference between designing something professionally or just for kicks, and it is important to recognize and understand the differences. I do the latter all the time.

 

I always look at how people solved problems. Or most of the time I try to find out why there isn't a solution yet.
This is true for the majority of things I want to do and I work on.

First, we all have to admit that there are always people who have more experience than us. Also people often become experts in a certain field, they are specialized, so why not having a look at what they did. Especially research and development papers from universities or companies are interesting. There is so much to learn.

Second, it's a waste of time to re-invent everything from scratch. We don't do it in other areas so why would we do it in electronics? Unless you have unlimited financial ressources and you are unemployed you may loose time AND money not looking at existing technology.

If you have a certain project in mind you take every help/information available to get there.
If existing technology is too expensive try to make it cheaper, if it consumes to much energy, make it less power-hungry, if it doesn't exist - this is where the fun starts.

 

Thanks a lot for all your comments so far. When I started this thread, I was pretty sure that Williamson's approach was a "minority opinion", so it was really interesting to see that the first three posters agree with him.

Yes, it is probably true that if you first see another guys schematic (or code or whatever), you can't get it out of your head anymore, and then you can't start a "fresh" approach with your own ideas anymore.. On the other hand, I see the danger that you could "fall in love" with your own design, and if you later learn that another solution is superior you might have trouble to admit it ?

Also, perhaps one little clarification is in order: I do in no way propose to blindly copy anything. I'm just talking about the case, where a solution together with a detailed explanation is available, so that you can really understand why somebody has choosen it (and how you can modify it if necessary). I would not call this cheating, you have just learnt something new. Or, even better, if you don't find a ready solution but a method that will tell you how to arrive at a solution.

EDIT: I had not seen the answers by Bill Marsden and praondevou yet when I started to type this. I'm getting too slow - bedtime for me and tomorrow I'll reply to them. Thanks again for your answers !

 

Most of the designs can't be copied anyway, if you want to incorporate them in to your own design.To many variations.

I don't think a general answer can be given to your question. It depends on the project.

 

Who is the guy in your avatar? Is he pointing a gun at us?

 

I guess you're not a James Bond fan.

 

I guess you're not a James Bond fan.

Ok got it. I looked up "Blofeld".

 

If you can get a nice design with all the comments that explain why and how, then sure read it and learn.
But if you are making something that is not as generic and has several interacting parts to it, then I think you should try to decompose it into sections and think hard about all the what-ifs of the design and then start searching if the whole or its parts have already been made, read the abstract papers that you will probably find, and consider what options are best for you.

But that is all assuming you have "average+" knowledge in the field of what you´re making, if you are doing something you don´t really know much about then searching first is the better option.

 

I tend to go with Adjusters post. If you are a beginner it is dumb not to look at prior art. If you want to invent something from scratch it is always your choice, but you may be missing out on elegant solutions, or ever worse, off the shelf technology. In general you are not going to be able to match the price of something off the shelf (though there are exceptions).

There is a big difference between designing something professionally or just for kicks, and it is important to recognize and understand the differences. I do the latter all the time.

Researching whether an off the shelf solution is available is very important, but I would argue that this is outside the scope of the design process, it has to be done before. OK ,maybe this is just nitpicking, I guess it should be included in the text of the “Educate the Customer” box at the top of the chart. The sad thing is, it completely takes away all the fun and you learn absolutely nothing. As you said, the question is whether it is your profession. If it is a hobby, and you have spent a month to build a cool gadget and then someone comes along and tells you that you could have bought it for less than $10, I would simply ignore him. Of course, as a professional it would be simply unethical to waste the money of your boss or your customer just because designing is more fun.

Apart from this, I pretty much agree with everything that you, Adjuster and praondevou have said, also with kubeek's post no 13. Maybe as a compromise it might be a good idea to use one day or so at the start of the project to think about a truly original solution. But I would not try several approaches before researching the “state of the art” first.

 

Who is the guy in your avatar? Is he pointing a gun at us?

Well, he’s a supervillain, he has to point a gun at somebody. It’s a little confusing because there are several Blofelds with completely different looks and personalities in the James Bond movies. This is the one from “Diamonds Are Forever”, and he’s a pretty cool guy – apart from the fact that he attacks the world with a space based laser

I’ll stop now before I completely derail my own thread. One day I’ll start a thread about it in Off Topic.

 

I usually look at what is needed, then make a few different ways of doing it. If it is something that's somewhat simple and been done a lot, I'll look at a cookbook afterward to see if the design I came up with is better. Though I will look at the datasheet to see how an IC was tested and some applications of it before designing, but it's common sense to always look at the datasheet on a component you'll be using.

We get lots of questions here where people ask about a certain part of a circuit and why it won't work (without showing the rest of the circuit). What they are actually doing, is trying to "copy and paste" 4 different cookbook designs together, but they always forget to tell us that until they are questioned over and over.

Once they fess up, it's usually easier to put together something that does what they want to, without making all the extra stuff from the cookbook circuits that isn't needed.

 

The given flowchart is better then some internal procedures I have been asked to work to as it understands design is an iterative process, but the day I stop reading app notes and reference designs is the last day I could consider myself an engineer.

Hey, if I can solve *MY* problem (which is developing a circuit for a product at a cost to make my company a profit) by doing a copy and paste of an existing design then I am happy and DONE. When I am doing research reading data sheets I always look at the "suggested circuits," and those are notes from god (the IC designer) on how to use his product best.

Now there is good cook booking and bad cook booking. A good one (of course *I* did it) was when we got a quote to build some "hybrids" (raw silicon chips inside a header) for a power device , two power transistors, two resistors and a cap all in a TO-3 package. We quoted a fair price for a small order without any NRE. When the bid was accepted I looked at the requirements once more, the brain cell fired, and I found an obsolete Motorola darlington that exactly matched the spec, or more probably the spec was written to cover that part. We found some on the market, cleaned, marked, and tested them and made a good sale.

Here is a bad cook book: a previous boss had his "Golden Book of Electronic Circuits." He would literally and blindly copy pages from it into his circuits. I saw one of his circuits one day, something like some gain and an offset subtraction, something else making a total of 3 op amps. I combined them all into a single stage and presented it to him the next day. He looked incredulously at it and literally looked scared when he asked "but is it linear?"

Hrmmm, an op amp and a collection of resistors. I didn't see any non-linear components there. He agreed it seemed to work in his circuit simulator but just couldn't dare to use it. I've also seen this guy use a resistive divider to make a voltage, then use an op amp unity gain stage to buffer it to directly drive an op amp input.

Bad cook booking is when you blindly copy something you don't really understand. Good cook booking is when if you don't find exactly what you need you adapt it to be so, or toss it out and insert your own.

App notes and sample designs are just dancing lessons. Once the music starts if you're still counting 1-2-3 inside your head you still ain't dancing.

 

Here is a bad cook book: a previous boss had his "Golden Book of Electronic Circuits." He would literally and blindly copy pages from it into his circuits. I saw one of his circuits one day, something like some gain and an offset subtraction, something else making a total of 3 op amps. I combined them all into a single stage and presented it to him the next day. He looked incredulously at it and literally looked scared when he asked "but is it linear?"

Hrmmm, an op amp and a collection of resistors. I didn't see any non-linear components there. He agreed it seemed to work in his circuit simulator but just couldn't dare to use it. I've also seen this guy use a resistive divider to make a voltage, then use an op amp unity gain stage to buffer it to directly drive an op amp input.

Bad cook booking is when you blindly copy something you don't really understand. Good cook booking is when if you don't find exactly what you need you adapt it to be so, or toss it out and insert your own.

I don't think I could cope with that guy. If he simulates it and sees that it is linear, but still uses a design that costs probably 4x more after board space and extra components, he's eating profit.

Cookbooks are good if you understand what the circuit is doing, if you don't, and you copy/paste verbatim like each cookbook entry is a black box, you'll more than likely run into issues.

 

Blofeld already has an excellent design procedure;
1. Kidnap scientists
2. Force scientists to do the design
3. Kill scientists so they can't talk

 

I do not care for the first statement in the flow chart on the web link “Educate the Customer”.

It makes it sound that the customer is stupid. It should read “What are the customers’ requirements and needs”. Then you can explain (not educate) to the customer what you can do to bring the design to life. Hold design reviews with the customers to make sure that you are in agreement with each other at different phases of his or her design.